Index: head/contrib/jemalloc/ChangeLog =================================================================== --- head/contrib/jemalloc/ChangeLog (revision 289899) +++ head/contrib/jemalloc/ChangeLog (revision 289900) @@ -1,774 +1,800 @@ Following are change highlights associated with official releases. Important bug fixes are all mentioned, but some internal enhancements are omitted here for brevity. Much more detail can be found in the git revision history: https://github.com/jemalloc/jemalloc +* 4.0.4 (October 24, 2015) + + This bugfix release fixes another xallocx() regression. No other regressions + have come to light in over a month, so this is likely a good starting point + for people who prefer to wait for "dot one" releases with all the major issues + shaken out. + + Bug fixes: + - Fix xallocx(..., MALLOCX_ZERO to zero the last full trailing page of large + allocations that have been randomly assigned an offset of 0 when + --enable-cache-oblivious configure option is enabled. + +* 4.0.3 (September 24, 2015) + + This bugfix release continues the trend of xallocx() and heap profiling fixes. + + Bug fixes: + - Fix xallocx(..., MALLOCX_ZERO) to zero all trailing bytes of large + allocations when --enable-cache-oblivious configure option is enabled. + - Fix xallocx(..., MALLOCX_ZERO) to zero trailing bytes of huge allocations + when resizing from/to a size class that is not a multiple of the chunk size. + - Fix prof_tctx_dump_iter() to filter out nodes that were created after heap + profile dumping started. + - Work around a potentially bad thread-specific data initialization + interaction with NPTL (glibc's pthreads implementation). + * 4.0.2 (September 21, 2015) This bugfix release addresses a few bugs specific to heap profiling. Bug fixes: - Fix ixallocx_prof_sample() to never modify nor create sampled small allocations. xallocx() is in general incapable of moving small allocations, so this fix removes buggy code without loss of generality. - Fix irallocx_prof_sample() to always allocate large regions, even when alignment is non-zero. - Fix prof_alloc_rollback() to read tdata from thread-specific data rather than dereferencing a potentially invalid tctx. * 4.0.1 (September 15, 2015) This is a bugfix release that is somewhat high risk due to the amount of refactoring required to address deep xallocx() problems. As a side effect of these fixes, xallocx() now tries harder to partially fulfill requests for optional extra space. Note that a couple of minor heap profiling optimizations are included, but these are better thought of as performance fixes that were integral to disovering most of the other bugs. Optimizations: - Avoid a chunk metadata read in arena_prof_tctx_set(), since it is in the fast path when heap profiling is enabled. Additionally, split a special case out into arena_prof_tctx_reset(), which also avoids chunk metadata reads. - Optimize irallocx_prof() to optimistically update the sampler state. The prior implementation appears to have been a holdover from when rallocx()/xallocx() functionality was combined as rallocm(). Bug fixes: - Fix TLS configuration such that it is enabled by default for platforms on which it works correctly. - Fix arenas_cache_cleanup() and arena_get_hard() to handle allocation/deallocation within the application's thread-specific data cleanup functions even after arenas_cache is torn down. - Fix xallocx() bugs related to size+extra exceeding HUGE_MAXCLASS. - Fix chunk purge hook calls for in-place huge shrinking reallocation to specify the old chunk size rather than the new chunk size. This bug caused no correctness issues for the default chunk purge function, but was visible to custom functions set via the "arena..chunk_hooks" mallctl. - Fix heap profiling bugs: + Fix heap profiling to distinguish among otherwise identical sample sites with interposed resets (triggered via the "prof.reset" mallctl). This bug could cause data structure corruption that would most likely result in a segfault. + Fix irealloc_prof() to prof_alloc_rollback() on OOM. + Make one call to prof_active_get_unlocked() per allocation event, and use the result throughout the relevant functions that handle an allocation event. Also add a missing check in prof_realloc(). These fixes protect allocation events against concurrent prof_active changes. + Fix ixallocx_prof() to pass usize_max and zero to ixallocx_prof_sample() in the correct order. + Fix prof_realloc() to call prof_free_sampled_object() after calling prof_malloc_sample_object(). Prior to this fix, if tctx and old_tctx were the same, the tctx could have been prematurely destroyed. - Fix portability bugs: + Don't bitshift by negative amounts when encoding/decoding run sizes in chunk header maps. This affected systems with page sizes greater than 8 KiB. + Rename index_t to szind_t to avoid an existing type on Solaris. + Add JEMALLOC_CXX_THROW to the memalign() function prototype, in order to match glibc and avoid compilation errors when including both jemalloc/jemalloc.h and malloc.h in C++ code. + Don't assume that /bin/sh is appropriate when running size_classes.sh during configuration. + Consider __sparcv9 a synonym for __sparc64__ when defining LG_QUANTUM. + Link tests to librt if it contains clock_gettime(2). * 4.0.0 (August 17, 2015) This version contains many speed and space optimizations, both minor and major. The major themes are generalization, unification, and simplification. Although many of these optimizations cause no visible behavior change, their cumulative effect is substantial. New features: - Normalize size class spacing to be consistent across the complete size range. By default there are four size classes per size doubling, but this is now configurable via the --with-lg-size-class-group option. Also add the --with-lg-page, --with-lg-page-sizes, --with-lg-quantum, and --with-lg-tiny-min options, which can be used to tweak page and size class settings. Impacts: + Worst case performance for incrementally growing/shrinking reallocation is improved because there are far fewer size classes, and therefore copying happens less often. + Internal fragmentation is limited to 20% for all but the smallest size classes (those less than four times the quantum). (1B + 4 KiB) and (1B + 4 MiB) previously suffered nearly 50% internal fragmentation. + Chunk fragmentation tends to be lower because there are fewer distinct run sizes to pack. - Add support for explicit tcaches. The "tcache.create", "tcache.flush", and "tcache.destroy" mallctls control tcache lifetime and flushing, and the MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to the *allocx() API control which tcache is used for each operation. - Implement per thread heap profiling, as well as the ability to enable/disable heap profiling on a per thread basis. Add the "prof.reset", "prof.lg_sample", "thread.prof.name", "thread.prof.active", "opt.prof_thread_active_init", "prof.thread_active_init", and "thread.prof.active" mallctls. - Add support for per arena application-specified chunk allocators, configured via the "arena..chunk_hooks" mallctl. - Refactor huge allocation to be managed by arenas, so that arenas now function as general purpose independent allocators. This is important in the context of user-specified chunk allocators, aside from the scalability benefits. Related new statistics: + The "stats.arenas..huge.allocated", "stats.arenas..huge.nmalloc", "stats.arenas..huge.ndalloc", and "stats.arenas..huge.nrequests" mallctls provide high level per arena huge allocation statistics. + The "arenas.nhchunks", "arenas.hchunk..size", "stats.arenas..hchunks..nmalloc", "stats.arenas..hchunks..ndalloc", "stats.arenas..hchunks..nrequests", and "stats.arenas..hchunks..curhchunks" mallctls provide per size class statistics. - Add the 'util' column to malloc_stats_print() output, which reports the proportion of available regions that are currently in use for each small size class. - Add "alloc" and "free" modes for for junk filling (see the "opt.junk" mallctl), so that it is possible to separately enable junk filling for allocation versus deallocation. - Add the jemalloc-config script, which provides information about how jemalloc was configured, and how to integrate it into application builds. - Add metadata statistics, which are accessible via the "stats.metadata", "stats.arenas..metadata.mapped", and "stats.arenas..metadata.allocated" mallctls. - Add the "stats.resident" mallctl, which reports the upper limit of physically resident memory mapped by the allocator. - Add per arena control over unused dirty page purging, via the "arenas.lg_dirty_mult", "arena..lg_dirty_mult", and "stats.arenas..lg_dirty_mult" mallctls. - Add the "prof.gdump" mallctl, which makes it possible to toggle the gdump feature on/off during program execution. - Add sdallocx(), which implements sized deallocation. The primary optimization over dallocx() is the removal of a metadata read, which often suffers an L1 cache miss. - Add missing header includes in jemalloc/jemalloc.h, so that applications only have to #include . - Add support for additional platforms: + Bitrig + Cygwin + DragonFlyBSD + iOS + OpenBSD + OpenRISC/or1k Optimizations: - Maintain dirty runs in per arena LRUs rather than in per arena trees of dirty-run-containing chunks. In practice this change significantly reduces dirty page purging volume. - Integrate whole chunks into the unused dirty page purging machinery. This reduces the cost of repeated huge allocation/deallocation, because it effectively introduces a cache of chunks. - Split the arena chunk map into two separate arrays, in order to increase cache locality for the frequently accessed bits. - Move small run metadata out of runs, into arena chunk headers. This reduces run fragmentation, smaller runs reduce external fragmentation for small size classes, and packed (less uniformly aligned) metadata layout improves CPU cache set distribution. - Randomly distribute large allocation base pointer alignment relative to page boundaries in order to more uniformly utilize CPU cache sets. This can be disabled via the --disable-cache-oblivious configure option, and queried via the "config.cache_oblivious" mallctl. - Micro-optimize the fast paths for the public API functions. - Refactor thread-specific data to reside in a single structure. This assures that only a single TLS read is necessary per call into the public API. - Implement in-place huge allocation growing and shrinking. - Refactor rtree (radix tree for chunk lookups) to be lock-free, and make additional optimizations that reduce maximum lookup depth to one or two levels. This resolves what was a concurrency bottleneck for per arena huge allocation, because a global data structure is critical for determining which arenas own which huge allocations. Incompatible changes: - Replace --enable-cc-silence with --disable-cc-silence to suppress spurious warnings by default. - Assure that the constness of malloc_usable_size()'s return type matches that of the system implementation. - Change the heap profile dump format to support per thread heap profiling, rename pprof to jeprof, and enhance it with the --thread= option. As a result, the bundled jeprof must now be used rather than the upstream (gperftools) pprof. - Disable "opt.prof_final" by default, in order to avoid atexit(3), which can internally deadlock on some platforms. - Change the "arenas.nlruns" mallctl type from size_t to unsigned. - Replace the "stats.arenas..bins..allocated" mallctl with "stats.arenas..bins..curregs". - Ignore MALLOC_CONF in set{uid,gid,cap} binaries. - Ignore MALLOCX_ARENA(a) in dallocx(), in favor of using the MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to control tcache usage. Removed features: - Remove the *allocm() API, which is superseded by the *allocx() API. - Remove the --enable-dss options, and make dss non-optional on all platforms which support sbrk(2). - Remove the "arenas.purge" mallctl, which was obsoleted by the "arena..purge" mallctl in 3.1.0. - Remove the unnecessary "opt.valgrind" mallctl; jemalloc automatically detects whether it is running inside Valgrind. - Remove the "stats.huge.allocated", "stats.huge.nmalloc", and "stats.huge.ndalloc" mallctls. - Remove the --enable-mremap option. - Remove the "stats.chunks.current", "stats.chunks.total", and "stats.chunks.high" mallctls. Bug fixes: - Fix the cactive statistic to decrease (rather than increase) when active memory decreases. This regression was first released in 3.5.0. - Fix OOM handling in memalign() and valloc(). A variant of this bug existed in all releases since 2.0.0, which introduced these functions. - Fix an OOM-related regression in arena_tcache_fill_small(), which could cause cache corruption on OOM. This regression was present in all releases from 2.2.0 through 3.6.0. - Fix size class overflow handling for malloc(), posix_memalign(), memalign(), calloc(), and realloc() when profiling is enabled. - Fix the "arena..dss" mallctl to return an error if "primary" or "secondary" precedence is specified, but sbrk(2) is not supported. - Fix fallback lg_floor() implementations to handle extremely large inputs. - Ensure the default purgeable zone is after the default zone on OS X. - Fix latent bugs in atomic_*(). - Fix the "arena..dss" mallctl to handle read-only calls. - Fix tls_model configuration to enable the initial-exec model when possible. - Mark malloc_conf as a weak symbol so that the application can override it. - Correctly detect glibc's adaptive pthread mutexes. - Fix the --without-export configure option. * 3.6.0 (March 31, 2014) This version contains a critical bug fix for a regression present in 3.5.0 and 3.5.1. Bug fixes: - Fix a regression in arena_chunk_alloc() that caused crashes during small/large allocation if chunk allocation failed. In the absence of this bug, chunk allocation failure would result in allocation failure, e.g. NULL return from malloc(). This regression was introduced in 3.5.0. - Fix backtracing for gcc intrinsics-based backtracing by specifying -fno-omit-frame-pointer to gcc. Note that the application (and all the libraries it links to) must also be compiled with this option for backtracing to be reliable. - Use dss allocation precedence for huge allocations as well as small/large allocations. - Fix test assertion failure message formatting. This bug did not manifest on x86_64 systems because of implementation subtleties in va_list. - Fix inconsequential test failures for hash and SFMT code. New features: - Support heap profiling on FreeBSD. This feature depends on the proc filesystem being mounted during heap profile dumping. * 3.5.1 (February 25, 2014) This version primarily addresses minor bugs in test code. Bug fixes: - Configure Solaris/Illumos to use MADV_FREE. - Fix junk filling for mremap(2)-based huge reallocation. This is only relevant if configuring with the --enable-mremap option specified. - Avoid compilation failure if 'restrict' C99 keyword is not supported by the compiler. - Add a configure test for SSE2 rather than assuming it is usable on i686 systems. This fixes test compilation errors, especially on 32-bit Linux systems. - Fix mallctl argument size mismatches (size_t vs. uint64_t) in the stats unit test. - Fix/remove flawed alignment-related overflow tests. - Prevent compiler optimizations that could change backtraces in the prof_accum unit test. * 3.5.0 (January 22, 2014) This version focuses on refactoring and automated testing, though it also includes some non-trivial heap profiling optimizations not mentioned below. New features: - Add the *allocx() API, which is a successor to the experimental *allocm() API. The *allocx() functions are slightly simpler to use because they have fewer parameters, they directly return the results of primary interest, and mallocx()/rallocx() avoid the strict aliasing pitfall that allocm()/rallocm() share with posix_memalign(). Note that *allocm() is slated for removal in the next non-bugfix release. - Add support for LinuxThreads. Bug fixes: - Unless heap profiling is enabled, disable floating point code and don't link with libm. This, in combination with e.g. EXTRA_CFLAGS=-mno-sse on x64 systems, makes it possible to completely disable floating point register use. Some versions of glibc neglect to save/restore caller-saved floating point registers during dynamic lazy symbol loading, and the symbol loading code uses whatever malloc the application happens to have linked/loaded with, the result being potential floating point register corruption. - Report ENOMEM rather than EINVAL if an OOM occurs during heap profiling backtrace creation in imemalign(). This bug impacted posix_memalign() and aligned_alloc(). - Fix a file descriptor leak in a prof_dump_maps() error path. - Fix prof_dump() to close the dump file descriptor for all relevant error paths. - Fix rallocm() to use the arena specified by the ALLOCM_ARENA(s) flag for allocation, not just deallocation. - Fix a data race for large allocation stats counters. - Fix a potential infinite loop during thread exit. This bug occurred on Solaris, and could affect other platforms with similar pthreads TSD implementations. - Don't junk-fill reallocations unless usable size changes. This fixes a violation of the *allocx()/*allocm() semantics. - Fix growing large reallocation to junk fill new space. - Fix huge deallocation to junk fill when munmap is disabled. - Change the default private namespace prefix from empty to je_, and change --with-private-namespace-prefix so that it prepends an additional prefix rather than replacing je_. This reduces the likelihood of applications which statically link jemalloc experiencing symbol name collisions. - Add missing private namespace mangling (relevant when --with-private-namespace is specified). - Add and use JEMALLOC_INLINE_C so that static inline functions are marked as static even for debug builds. - Add a missing mutex unlock in a malloc_init_hard() error path. In practice this error path is never executed. - Fix numerous bugs in malloc_strotumax() error handling/reporting. These bugs had no impact except for malformed inputs. - Fix numerous bugs in malloc_snprintf(). These bugs were not exercised by existing calls, so they had no impact. * 3.4.1 (October 20, 2013) Bug fixes: - Fix a race in the "arenas.extend" mallctl that could cause memory corruption of internal data structures and subsequent crashes. - Fix Valgrind integration flaws that caused Valgrind warnings about reads of uninitialized memory in: + arena chunk headers + internal zero-initialized data structures (relevant to tcache and prof code) - Preserve errno during the first allocation. A readlink(2) call during initialization fails unless /etc/malloc.conf exists, so errno was typically set during the first allocation prior to this fix. - Fix compilation warnings reported by gcc 4.8.1. * 3.4.0 (June 2, 2013) This version is essentially a small bugfix release, but the addition of aarch64 support requires that the minor version be incremented. Bug fixes: - Fix race-triggered deadlocks in chunk_record(). These deadlocks were typically triggered by multiple threads concurrently deallocating huge objects. New features: - Add support for the aarch64 architecture. * 3.3.1 (March 6, 2013) This version fixes bugs that are typically encountered only when utilizing custom run-time options. Bug fixes: - Fix a locking order bug that could cause deadlock during fork if heap profiling were enabled. - Fix a chunk recycling bug that could cause the allocator to lose track of whether a chunk was zeroed. On FreeBSD, NetBSD, and OS X, it could cause corruption if allocating via sbrk(2) (unlikely unless running with the "dss:primary" option specified). This was completely harmless on Linux unless using mlockall(2) (and unlikely even then, unless the --disable-munmap configure option or the "dss:primary" option was specified). This regression was introduced in 3.1.0 by the mlockall(2)/madvise(2) interaction fix. - Fix TLS-related memory corruption that could occur during thread exit if the thread never allocated memory. Only the quarantine and prof facilities were susceptible. - Fix two quarantine bugs: + Internal reallocation of the quarantined object array leaked the old array. + Reallocation failure for internal reallocation of the quarantined object array (very unlikely) resulted in memory corruption. - Fix Valgrind integration to annotate all internally allocated memory in a way that keeps Valgrind happy about internal data structure access. - Fix building for s390 systems. * 3.3.0 (January 23, 2013) This version includes a few minor performance improvements in addition to the listed new features and bug fixes. New features: - Add clipping support to lg_chunk option processing. - Add the --enable-ivsalloc option. - Add the --without-export option. - Add the --disable-zone-allocator option. Bug fixes: - Fix "arenas.extend" mallctl to output the number of arenas. - Fix chunk_recycle() to unconditionally inform Valgrind that returned memory is undefined. - Fix build break on FreeBSD related to alloca.h. * 3.2.0 (November 9, 2012) In addition to a couple of bug fixes, this version modifies page run allocation and dirty page purging algorithms in order to better control page-level virtual memory fragmentation. Incompatible changes: - Change the "opt.lg_dirty_mult" default from 5 to 3 (32:1 to 8:1). Bug fixes: - Fix dss/mmap allocation precedence code to use recyclable mmap memory only after primary dss allocation fails. - Fix deadlock in the "arenas.purge" mallctl. This regression was introduced in 3.1.0 by the addition of the "arena..purge" mallctl. * 3.1.0 (October 16, 2012) New features: - Auto-detect whether running inside Valgrind, thus removing the need to manually specify MALLOC_CONF=valgrind:true. - Add the "arenas.extend" mallctl, which allows applications to create manually managed arenas. - Add the ALLOCM_ARENA() flag for {,r,d}allocm(). - Add the "opt.dss", "arena..dss", and "stats.arenas..dss" mallctls, which provide control over dss/mmap precedence. - Add the "arena..purge" mallctl, which obsoletes "arenas.purge". - Define LG_QUANTUM for hppa. Incompatible changes: - Disable tcache by default if running inside Valgrind, in order to avoid making unallocated objects appear reachable to Valgrind. - Drop const from malloc_usable_size() argument on Linux. Bug fixes: - Fix heap profiling crash if sampled object is freed via realloc(p, 0). - Remove const from __*_hook variable declarations, so that glibc can modify them during process forking. - Fix mlockall(2)/madvise(2) interaction. - Fix fork(2)-related deadlocks. - Fix error return value for "thread.tcache.enabled" mallctl. * 3.0.0 (May 11, 2012) Although this version adds some major new features, the primary focus is on internal code cleanup that facilitates maintainability and portability, most of which is not reflected in the ChangeLog. This is the first release to incorporate substantial contributions from numerous other developers, and the result is a more broadly useful allocator (see the git revision history for contribution details). Note that the license has been unified, thanks to Facebook granting a license under the same terms as the other copyright holders (see COPYING). New features: - Implement Valgrind support, redzones, and quarantine. - Add support for additional platforms: + FreeBSD + Mac OS X Lion + MinGW + Windows (no support yet for replacing the system malloc) - Add support for additional architectures: + MIPS + SH4 + Tilera - Add support for cross compiling. - Add nallocm(), which rounds a request size up to the nearest size class without actually allocating. - Implement aligned_alloc() (blame C11). - Add the "thread.tcache.enabled" mallctl. - Add the "opt.prof_final" mallctl. - Update pprof (from gperftools 2.0). - Add the --with-mangling option. - Add the --disable-experimental option. - Add the --disable-munmap option, and make it the default on Linux. - Add the --enable-mremap option, which disables use of mremap(2) by default. Incompatible changes: - Enable stats by default. - Enable fill by default. - Disable lazy locking by default. - Rename the "tcache.flush" mallctl to "thread.tcache.flush". - Rename the "arenas.pagesize" mallctl to "arenas.page". - Change the "opt.lg_prof_sample" default from 0 to 19 (1 B to 512 KiB). - Change the "opt.prof_accum" default from true to false. Removed features: - Remove the swap feature, including the "config.swap", "swap.avail", "swap.prezeroed", "swap.nfds", and "swap.fds" mallctls. - Remove highruns statistics, including the "stats.arenas..bins..highruns" and "stats.arenas..lruns..highruns" mallctls. - As part of small size class refactoring, remove the "opt.lg_[qc]space_max", "arenas.cacheline", "arenas.subpage", "arenas.[tqcs]space_{min,max}", and "arenas.[tqcs]bins" mallctls. - Remove the "arenas.chunksize" mallctl. - Remove the "opt.lg_prof_tcmax" option. - Remove the "opt.lg_prof_bt_max" option. - Remove the "opt.lg_tcache_gc_sweep" option. - Remove the --disable-tiny option, including the "config.tiny" mallctl. - Remove the --enable-dynamic-page-shift configure option. - Remove the --enable-sysv configure option. Bug fixes: - Fix a statistics-related bug in the "thread.arena" mallctl that could cause invalid statistics and crashes. - Work around TLS deallocation via free() on Linux. This bug could cause write-after-free memory corruption. - Fix a potential deadlock that could occur during interval- and growth-triggered heap profile dumps. - Fix large calloc() zeroing bugs due to dropping chunk map unzeroed flags. - Fix chunk_alloc_dss() to stop claiming memory is zeroed. This bug could cause memory corruption and crashes with --enable-dss specified. - Fix fork-related bugs that could cause deadlock in children between fork and exec. - Fix malloc_stats_print() to honor 'b' and 'l' in the opts parameter. - Fix realloc(p, 0) to act like free(p). - Do not enforce minimum alignment in memalign(). - Check for NULL pointer in malloc_usable_size(). - Fix an off-by-one heap profile statistics bug that could be observed in interval- and growth-triggered heap profiles. - Fix the "epoch" mallctl to update cached stats even if the passed in epoch is 0. - Fix bin->runcur management to fix a layout policy bug. This bug did not affect correctness. - Fix a bug in choose_arena_hard() that potentially caused more arenas to be initialized than necessary. - Add missing "opt.lg_tcache_max" mallctl implementation. - Use glibc allocator hooks to make mixed allocator usage less likely. - Fix build issues for --disable-tcache. - Don't mangle pthread_create() when --with-private-namespace is specified. * 2.2.5 (November 14, 2011) Bug fixes: - Fix huge_ralloc() race when using mremap(2). This is a serious bug that could cause memory corruption and/or crashes. - Fix huge_ralloc() to maintain chunk statistics. - Fix malloc_stats_print(..., "a") output. * 2.2.4 (November 5, 2011) Bug fixes: - Initialize arenas_tsd before using it. This bug existed for 2.2.[0-3], as well as for --disable-tls builds in earlier releases. - Do not assume a 4 KiB page size in test/rallocm.c. * 2.2.3 (August 31, 2011) This version fixes numerous bugs related to heap profiling. Bug fixes: - Fix a prof-related race condition. This bug could cause memory corruption, but only occurred in non-default configurations (prof_accum:false). - Fix off-by-one backtracing issues (make sure that prof_alloc_prep() is excluded from backtraces). - Fix a prof-related bug in realloc() (only triggered by OOM errors). - Fix prof-related bugs in allocm() and rallocm(). - Fix prof_tdata_cleanup() for --disable-tls builds. - Fix a relative include path, to fix objdir builds. * 2.2.2 (July 30, 2011) Bug fixes: - Fix a build error for --disable-tcache. - Fix assertions in arena_purge() (for real this time). - Add the --with-private-namespace option. This is a workaround for symbol conflicts that can inadvertently arise when using static libraries. * 2.2.1 (March 30, 2011) Bug fixes: - Implement atomic operations for x86/x64. This fixes compilation failures for versions of gcc that are still in wide use. - Fix an assertion in arena_purge(). * 2.2.0 (March 22, 2011) This version incorporates several improvements to algorithms and data structures that tend to reduce fragmentation and increase speed. New features: - Add the "stats.cactive" mallctl. - Update pprof (from google-perftools 1.7). - Improve backtracing-related configuration logic, and add the --disable-prof-libgcc option. Bug fixes: - Change default symbol visibility from "internal", to "hidden", which decreases the overhead of library-internal function calls. - Fix symbol visibility so that it is also set on OS X. - Fix a build dependency regression caused by the introduction of the .pic.o suffix for PIC object files. - Add missing checks for mutex initialization failures. - Don't use libgcc-based backtracing except on x64, where it is known to work. - Fix deadlocks on OS X that were due to memory allocation in pthread_mutex_lock(). - Heap profiling-specific fixes: + Fix memory corruption due to integer overflow in small region index computation, when using a small enough sample interval that profiling context pointers are stored in small run headers. + Fix a bootstrap ordering bug that only occurred with TLS disabled. + Fix a rallocm() rsize bug. + Fix error detection bugs for aligned memory allocation. * 2.1.3 (March 14, 2011) Bug fixes: - Fix a cpp logic regression (due to the "thread.{de,}allocatedp" mallctl fix for OS X in 2.1.2). - Fix a "thread.arena" mallctl bug. - Fix a thread cache stats merging bug. * 2.1.2 (March 2, 2011) Bug fixes: - Fix "thread.{de,}allocatedp" mallctl for OS X. - Add missing jemalloc.a to build system. * 2.1.1 (January 31, 2011) Bug fixes: - Fix aligned huge reallocation (affected allocm()). - Fix the ALLOCM_LG_ALIGN macro definition. - Fix a heap dumping deadlock. - Fix a "thread.arena" mallctl bug. * 2.1.0 (December 3, 2010) This version incorporates some optimizations that can't quite be considered bug fixes. New features: - Use Linux's mremap(2) for huge object reallocation when possible. - Avoid locking in mallctl*() when possible. - Add the "thread.[de]allocatedp" mallctl's. - Convert the manual page source from roff to DocBook, and generate both roff and HTML manuals. Bug fixes: - Fix a crash due to incorrect bootstrap ordering. This only impacted --enable-debug --enable-dss configurations. - Fix a minor statistics bug for mallctl("swap.avail", ...). * 2.0.1 (October 29, 2010) Bug fixes: - Fix a race condition in heap profiling that could cause undefined behavior if "opt.prof_accum" were disabled. - Add missing mutex unlocks for some OOM error paths in the heap profiling code. - Fix a compilation error for non-C99 builds. * 2.0.0 (October 24, 2010) This version focuses on the experimental *allocm() API, and on improved run-time configuration/introspection. Nonetheless, numerous performance improvements are also included. New features: - Implement the experimental {,r,s,d}allocm() API, which provides a superset of the functionality available via malloc(), calloc(), posix_memalign(), realloc(), malloc_usable_size(), and free(). These functions can be used to allocate/reallocate aligned zeroed memory, ask for optional extra memory during reallocation, prevent object movement during reallocation, etc. - Replace JEMALLOC_OPTIONS/JEMALLOC_PROF_PREFIX with MALLOC_CONF, which is more human-readable, and more flexible. For example: JEMALLOC_OPTIONS=AJP is now: MALLOC_CONF=abort:true,fill:true,stats_print:true - Port to Apple OS X. Sponsored by Mozilla. - Make it possible for the application to control thread-->arena mappings via the "thread.arena" mallctl. - Add compile-time support for all TLS-related functionality via pthreads TSD. This is mainly of interest for OS X, which does not support TLS, but has a TSD implementation with similar performance. - Override memalign() and valloc() if they are provided by the system. - Add the "arenas.purge" mallctl, which can be used to synchronously purge all dirty unused pages. - Make cumulative heap profiling data optional, so that it is possible to limit the amount of memory consumed by heap profiling data structures. - Add per thread allocation counters that can be accessed via the "thread.allocated" and "thread.deallocated" mallctls. Incompatible changes: - Remove JEMALLOC_OPTIONS and malloc_options (see MALLOC_CONF above). - Increase default backtrace depth from 4 to 128 for heap profiling. - Disable interval-based profile dumps by default. Bug fixes: - Remove bad assertions in fork handler functions. These assertions could cause aborts for some combinations of configure settings. - Fix strerror_r() usage to deal with non-standard semantics in GNU libc. - Fix leak context reporting. This bug tended to cause the number of contexts to be underreported (though the reported number of objects and bytes were correct). - Fix a realloc() bug for large in-place growing reallocation. This bug could cause memory corruption, but it was hard to trigger. - Fix an allocation bug for small allocations that could be triggered if multiple threads raced to create a new run of backing pages. - Enhance the heap profiler to trigger samples based on usable size, rather than request size. - Fix a heap profiling bug due to sometimes losing track of requested object size for sampled objects. * 1.0.3 (August 12, 2010) Bug fixes: - Fix the libunwind-based implementation of stack backtracing (used for heap profiling). This bug could cause zero-length backtraces to be reported. - Add a missing mutex unlock in library initialization code. If multiple threads raced to initialize malloc, some of them could end up permanently blocked. * 1.0.2 (May 11, 2010) Bug fixes: - Fix junk filling of large objects, which could cause memory corruption. - Add MAP_NORESERVE support for chunk mapping, because otherwise virtual memory limits could cause swap file configuration to fail. Contributed by Jordan DeLong. * 1.0.1 (April 14, 2010) Bug fixes: - Fix compilation when --enable-fill is specified. - Fix threads-related profiling bugs that affected accuracy and caused memory to be leaked during thread exit. - Fix dirty page purging race conditions that could cause crashes. - Fix crash in tcache flushing code during thread destruction. * 1.0.0 (April 11, 2010) This release focuses on speed and run-time introspection. Numerous algorithmic improvements make this release substantially faster than its predecessors. New features: - Implement autoconf-based configuration system. - Add mallctl*(), for the purposes of introspection and run-time configuration. - Make it possible for the application to manually flush a thread's cache, via the "tcache.flush" mallctl. - Base maximum dirty page count on proportion of active memory. - Compute various additional run-time statistics, including per size class statistics for large objects. - Expose malloc_stats_print(), which can be called repeatedly by the application. - Simplify the malloc_message() signature to only take one string argument, and incorporate an opaque data pointer argument for use by the application in combination with malloc_stats_print(). - Add support for allocation backed by one or more swap files, and allow the application to disable over-commit if swap files are in use. - Implement allocation profiling and leak checking. Removed features: - Remove the dynamic arena rebalancing code, since thread-specific caching reduces its utility. Bug fixes: - Modify chunk allocation to work when address space layout randomization (ASLR) is in use. - Fix thread cleanup bugs related to TLS destruction. - Handle 0-size allocation requests in posix_memalign(). - Fix a chunk leak. The leaked chunks were never touched, so this impacted virtual memory usage, but not physical memory usage. * linux_2008082[78]a (August 27/28, 2008) These snapshot releases are the simple result of incorporating Linux-specific support into the FreeBSD malloc sources. -------------------------------------------------------------------------------- vim:filetype=text:textwidth=80 Index: head/contrib/jemalloc/FREEBSD-diffs =================================================================== --- head/contrib/jemalloc/FREEBSD-diffs (revision 289899) +++ head/contrib/jemalloc/FREEBSD-diffs (revision 289900) @@ -1,491 +1,491 @@ diff --git a/doc/jemalloc.xml.in b/doc/jemalloc.xml.in -index 8fc774b..fdbef95 100644 +index 26a5e14..2a801b7 100644 --- a/doc/jemalloc.xml.in +++ b/doc/jemalloc.xml.in @@ -53,11 +53,23 @@ This manual describes jemalloc @jemalloc_version@. More information can be found at the jemalloc website. + + The following configuration options are enabled in libc's built-in + jemalloc: , + , , + , , + , , and + . Additionally, + is enabled in development versions of + FreeBSD (controlled by the MALLOC_PRODUCTION make + variable). + SYNOPSIS - #include <jemalloc/jemalloc.h> + #include <stdlib.h> +#include <malloc_np.h> Standard API @@ -2759,4 +2771,18 @@ malloc_conf = "lg_chunk:24";]]> The posix_memalign function conforms to IEEE Std 1003.1-2001 (“POSIX.1”). + + HISTORY + The malloc_usable_size and + posix_memalign functions first appeared in + FreeBSD 7.0. + + The aligned_alloc, + malloc_stats_print, and + mallctl* functions first appeared in + FreeBSD 10.0. + + The *allocx functions first appeared + in FreeBSD 11.0. + diff --git a/include/jemalloc/internal/jemalloc_internal.h.in b/include/jemalloc/internal/jemalloc_internal.h.in -index 8536a3e..0c2a81f 100644 +index 654cd08..ad5382d 100644 --- a/include/jemalloc/internal/jemalloc_internal.h.in +++ b/include/jemalloc/internal/jemalloc_internal.h.in @@ -8,6 +8,9 @@ #include #endif +#include "un-namespace.h" +#include "libc_private.h" + #define JEMALLOC_NO_DEMANGLE #ifdef JEMALLOC_JET # define JEMALLOC_N(n) jet_##n @@ -42,13 +45,7 @@ static const bool config_fill = false #endif ; -static const bool config_lazy_lock = -#ifdef JEMALLOC_LAZY_LOCK - true -#else - false -#endif - ; +static const bool config_lazy_lock = true; static const bool config_prof = #ifdef JEMALLOC_PROF true diff --git a/include/jemalloc/internal/jemalloc_internal_decls.h b/include/jemalloc/internal/jemalloc_internal_decls.h index a601d6e..e7094b2 100644 --- a/include/jemalloc/internal/jemalloc_internal_decls.h +++ b/include/jemalloc/internal/jemalloc_internal_decls.h @@ -1,6 +1,9 @@ #ifndef JEMALLOC_INTERNAL_DECLS_H #define JEMALLOC_INTERNAL_DECLS_H +#include "libc_private.h" +#include "namespace.h" + #include #ifdef _WIN32 # include diff --git a/include/jemalloc/internal/mutex.h b/include/jemalloc/internal/mutex.h index f051f29..561378f 100644 --- a/include/jemalloc/internal/mutex.h +++ b/include/jemalloc/internal/mutex.h @@ -47,15 +47,13 @@ struct malloc_mutex_s { #ifdef JEMALLOC_LAZY_LOCK extern bool isthreaded; -#else -# undef isthreaded /* Undo private_namespace.h definition. */ -# define isthreaded true #endif bool malloc_mutex_init(malloc_mutex_t *mutex); void malloc_mutex_prefork(malloc_mutex_t *mutex); void malloc_mutex_postfork_parent(malloc_mutex_t *mutex); void malloc_mutex_postfork_child(malloc_mutex_t *mutex); +bool malloc_mutex_first_thread(void); bool mutex_boot(void); #endif /* JEMALLOC_H_EXTERNS */ diff --git a/include/jemalloc/internal/private_symbols.txt b/include/jemalloc/internal/private_symbols.txt index a90021a..34904bf 100644 --- a/include/jemalloc/internal/private_symbols.txt +++ b/include/jemalloc/internal/private_symbols.txt @@ -280,7 +280,6 @@ iralloct_realign isalloc isdalloct isqalloc -isthreaded ivsalloc ixalloc jemalloc_postfork_child diff --git a/include/jemalloc/jemalloc_FreeBSD.h b/include/jemalloc/jemalloc_FreeBSD.h new file mode 100644 index 0000000..737542e --- /dev/null +++ b/include/jemalloc/jemalloc_FreeBSD.h @@ -0,0 +1,142 @@ +/* + * Override settings that were generated in jemalloc_defs.h as necessary. + */ + +#undef JEMALLOC_OVERRIDE_VALLOC + +#ifndef MALLOC_PRODUCTION +#define JEMALLOC_DEBUG +#endif + +/* + * The following are architecture-dependent, so conditionally define them for + * each supported architecture. + */ +#undef JEMALLOC_TLS_MODEL +#undef STATIC_PAGE_SHIFT +#undef LG_SIZEOF_PTR +#undef LG_SIZEOF_INT +#undef LG_SIZEOF_LONG +#undef LG_SIZEOF_INTMAX_T + +#ifdef __i386__ +# define LG_SIZEOF_PTR 2 +# define JEMALLOC_TLS_MODEL __attribute__((tls_model("initial-exec"))) +#endif +#ifdef __ia64__ +# define LG_SIZEOF_PTR 3 +#endif +#ifdef __sparc64__ +# define LG_SIZEOF_PTR 3 +# define JEMALLOC_TLS_MODEL __attribute__((tls_model("initial-exec"))) +#endif +#ifdef __amd64__ +# define LG_SIZEOF_PTR 3 +# define JEMALLOC_TLS_MODEL __attribute__((tls_model("initial-exec"))) +#endif +#ifdef __arm__ +# define LG_SIZEOF_PTR 2 +#endif +#ifdef __aarch64__ +# define LG_SIZEOF_PTR 3 +#endif +#ifdef __mips__ +#ifdef __mips_n64 +# define LG_SIZEOF_PTR 3 +#else +# define LG_SIZEOF_PTR 2 +#endif +#endif +#ifdef __powerpc64__ +# define LG_SIZEOF_PTR 3 +#elif defined(__powerpc__) +# define LG_SIZEOF_PTR 2 +#endif + +#ifndef JEMALLOC_TLS_MODEL +# define JEMALLOC_TLS_MODEL /* Default. */ +#endif + +#define STATIC_PAGE_SHIFT PAGE_SHIFT +#define LG_SIZEOF_INT 2 +#define LG_SIZEOF_LONG LG_SIZEOF_PTR +#define LG_SIZEOF_INTMAX_T 3 + +#undef CPU_SPINWAIT +#include +#include +#define CPU_SPINWAIT cpu_spinwait() + +/* Disable lazy-lock machinery, mangle isthreaded, and adjust its type. */ +#undef JEMALLOC_LAZY_LOCK +extern int __isthreaded; +#define isthreaded ((bool)__isthreaded) + +/* Mangle. */ +#undef je_malloc +#undef je_calloc +#undef je_realloc +#undef je_free +#undef je_posix_memalign +#undef je_aligned_alloc +#undef je_malloc_usable_size +#undef je_mallocx +#undef je_rallocx +#undef je_xallocx +#undef je_sallocx +#undef je_dallocx +#undef je_nallocx +#undef je_allocm +#undef je_rallocm +#undef je_sallocm +#undef je_dallocm +#undef je_nallocm +#define je_malloc __malloc +#define je_calloc __calloc +#define je_realloc __realloc +#define je_free __free +#define je_posix_memalign __posix_memalign +#define je_aligned_alloc __aligned_alloc +#define je_malloc_usable_size __malloc_usable_size +#define je_mallocx __mallocx +#define je_rallocx __rallocx +#define je_xallocx __xallocx +#define je_sallocx __sallocx +#define je_dallocx __dallocx +#define je_nallocx __nallocx +#define je_allocm __allocm +#define je_rallocm __rallocm +#define je_sallocm __sallocm +#define je_dallocm __dallocm +#define je_nallocm __nallocm +#define open _open +#define read _read +#define write _write +#define close _close +#define pthread_mutex_lock _pthread_mutex_lock +#define pthread_mutex_unlock _pthread_mutex_unlock + +#ifdef JEMALLOC_C_ +/* + * Define 'weak' symbols so that an application can have its own versions + * of malloc, calloc, realloc, free, et al. + */ +__weak_reference(__malloc, malloc); +__weak_reference(__calloc, calloc); +__weak_reference(__realloc, realloc); +__weak_reference(__free, free); +__weak_reference(__posix_memalign, posix_memalign); +__weak_reference(__aligned_alloc, aligned_alloc); +__weak_reference(__malloc_usable_size, malloc_usable_size); +__weak_reference(__mallocx, mallocx); +__weak_reference(__rallocx, rallocx); +__weak_reference(__xallocx, xallocx); +__weak_reference(__sallocx, sallocx); +__weak_reference(__dallocx, dallocx); +__weak_reference(__nallocx, nallocx); +__weak_reference(__allocm, allocm); +__weak_reference(__rallocm, rallocm); +__weak_reference(__sallocm, sallocm); +__weak_reference(__dallocm, dallocm); +__weak_reference(__nallocm, nallocm); +#endif diff --git a/include/jemalloc/jemalloc_rename.sh b/include/jemalloc/jemalloc_rename.sh index f943891..47d032c 100755 --- a/include/jemalloc/jemalloc_rename.sh +++ b/include/jemalloc/jemalloc_rename.sh @@ -19,4 +19,6 @@ done cat <: */ +const char *__malloc_options_1_0 = NULL; +__sym_compat(_malloc_options, __malloc_options_1_0, FBSD_1.0); + /* Runtime configuration options. */ const char *je_malloc_conf JEMALLOC_ATTR(weak); bool opt_abort = @@ -2490,6 +2494,107 @@ je_malloc_usable_size(JEMALLOC_USABLE_SIZE_CONST void *ptr) */ /******************************************************************************/ /* + * Begin compatibility functions. + */ + +#define ALLOCM_LG_ALIGN(la) (la) +#define ALLOCM_ALIGN(a) (ffsl(a)-1) +#define ALLOCM_ZERO ((int)0x40) +#define ALLOCM_NO_MOVE ((int)0x80) + +#define ALLOCM_SUCCESS 0 +#define ALLOCM_ERR_OOM 1 +#define ALLOCM_ERR_NOT_MOVED 2 + +int +je_allocm(void **ptr, size_t *rsize, size_t size, int flags) +{ + void *p; + + assert(ptr != NULL); + + p = je_mallocx(size, flags); + if (p == NULL) + return (ALLOCM_ERR_OOM); + if (rsize != NULL) + *rsize = isalloc(p, config_prof); + *ptr = p; + return (ALLOCM_SUCCESS); +} + +int +je_rallocm(void **ptr, size_t *rsize, size_t size, size_t extra, int flags) +{ + int ret; + bool no_move = flags & ALLOCM_NO_MOVE; + + assert(ptr != NULL); + assert(*ptr != NULL); + assert(size != 0); + assert(SIZE_T_MAX - size >= extra); + + if (no_move) { + size_t usize = je_xallocx(*ptr, size, extra, flags); + ret = (usize >= size) ? ALLOCM_SUCCESS : ALLOCM_ERR_NOT_MOVED; + if (rsize != NULL) + *rsize = usize; + } else { + void *p = je_rallocx(*ptr, size+extra, flags); + if (p != NULL) { + *ptr = p; + ret = ALLOCM_SUCCESS; + } else + ret = ALLOCM_ERR_OOM; + if (rsize != NULL) + *rsize = isalloc(*ptr, config_prof); + } + return (ret); +} + +int +je_sallocm(const void *ptr, size_t *rsize, int flags) +{ + + assert(rsize != NULL); + *rsize = je_sallocx(ptr, flags); + return (ALLOCM_SUCCESS); +} + +int +je_dallocm(void *ptr, int flags) +{ + + je_dallocx(ptr, flags); + return (ALLOCM_SUCCESS); +} + +int +je_nallocm(size_t *rsize, size_t size, int flags) +{ + size_t usize; + + usize = je_nallocx(size, flags); + if (usize == 0) + return (ALLOCM_ERR_OOM); + if (rsize != NULL) + *rsize = usize; + return (ALLOCM_SUCCESS); +} + +#undef ALLOCM_LG_ALIGN +#undef ALLOCM_ALIGN +#undef ALLOCM_ZERO +#undef ALLOCM_NO_MOVE + +#undef ALLOCM_SUCCESS +#undef ALLOCM_ERR_OOM +#undef ALLOCM_ERR_NOT_MOVED + +/* + * End compatibility functions. + */ +/******************************************************************************/ +/* * The following functions are used by threading libraries for protection of * malloc during fork(). */ @@ -2590,4 +2695,11 @@ jemalloc_postfork_child(void) ctl_postfork_child(); } +void +_malloc_first_thread(void) +{ + + (void)malloc_mutex_first_thread(); +} + /******************************************************************************/ diff --git a/src/mutex.c b/src/mutex.c index 2d47af9..934d5aa 100644 --- a/src/mutex.c +++ b/src/mutex.c @@ -66,6 +66,17 @@ pthread_create(pthread_t *__restrict thread, #ifdef JEMALLOC_MUTEX_INIT_CB JEMALLOC_EXPORT int _pthread_mutex_init_calloc_cb(pthread_mutex_t *mutex, void *(calloc_cb)(size_t, size_t)); + +#pragma weak _pthread_mutex_init_calloc_cb +int +_pthread_mutex_init_calloc_cb(pthread_mutex_t *mutex, + void *(calloc_cb)(size_t, size_t)) +{ + + return (((int (*)(pthread_mutex_t *, void *(*)(size_t, size_t))) + __libc_interposing[INTERPOS__pthread_mutex_init_calloc_cb])(mutex, + calloc_cb)); +} #endif bool @@ -137,7 +148,7 @@ malloc_mutex_postfork_child(malloc_mutex_t *mutex) } bool -mutex_boot(void) +malloc_mutex_first_thread(void) { #ifdef JEMALLOC_MUTEX_INIT_CB @@ -151,3 +162,14 @@ mutex_boot(void) #endif return (false); } + +bool +mutex_boot(void) +{ + +#ifndef JEMALLOC_MUTEX_INIT_CB + return (malloc_mutex_first_thread()); +#else + return (false); +#endif +} diff --git a/src/util.c b/src/util.c index 4cb0d6c..25b61c2 100644 --- a/src/util.c +++ b/src/util.c @@ -58,6 +58,22 @@ wrtmessage(void *cbopaque, const char *s) JEMALLOC_EXPORT void (*je_malloc_message)(void *, const char *s); +JEMALLOC_ATTR(visibility("hidden")) +void +wrtmessage_1_0(const char *s1, const char *s2, const char *s3, + const char *s4) +{ + + wrtmessage(NULL, s1); + wrtmessage(NULL, s2); + wrtmessage(NULL, s3); + wrtmessage(NULL, s4); +} + +void (*__malloc_message_1_0)(const char *s1, const char *s2, const char *s3, + const char *s4) = wrtmessage_1_0; +__sym_compat(_malloc_message, __malloc_message_1_0, FBSD_1.0); + /* * Wrapper around malloc_message() that avoids the need for * je_malloc_message(...) throughout the code. Index: head/contrib/jemalloc/VERSION =================================================================== --- head/contrib/jemalloc/VERSION (revision 289899) +++ head/contrib/jemalloc/VERSION (revision 289900) @@ -1 +1 @@ -4.0.2-0-g486d249fb4715fd3de679b6c2a04f7e657883111 +4.0.4-0-g91010a9e2ebfc84b1ac1ed7fdde3bfed4f65f180 Index: head/contrib/jemalloc/doc/jemalloc.3 =================================================================== --- head/contrib/jemalloc/doc/jemalloc.3 (revision 289899) +++ head/contrib/jemalloc/doc/jemalloc.3 (revision 289900) @@ -1,1993 +1,1993 @@ '\" t .\" Title: JEMALLOC .\" Author: Jason Evans .\" Generator: DocBook XSL Stylesheets v1.76.1 -.\" Date: 09/21/2015 +.\" Date: 10/24/2015 .\" Manual: User Manual -.\" Source: jemalloc 4.0.2-0-g486d249fb4715fd3de679b6c2a04f7e657883111 +.\" Source: jemalloc 4.0.4-0-g91010a9e2ebfc84b1ac1ed7fdde3bfed4f65f180 .\" Language: English .\" -.TH "JEMALLOC" "3" "09/21/2015" "jemalloc 4.0.2-0-g486d249fb471" "User Manual" +.TH "JEMALLOC" "3" "10/24/2015" "jemalloc 4.0.4-0-g91010a9e2ebf" "User Manual" .\" ----------------------------------------------------------------- .\" * Define some portability stuff .\" ----------------------------------------------------------------- .\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ .\" http://bugs.debian.org/507673 .\" http://lists.gnu.org/archive/html/groff/2009-02/msg00013.html .\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ .ie \n(.g .ds Aq \(aq .el .ds Aq ' .\" ----------------------------------------------------------------- .\" * set default formatting .\" ----------------------------------------------------------------- .\" disable hyphenation .nh .\" disable justification (adjust text to left margin only) .ad l .\" ----------------------------------------------------------------- .\" * MAIN CONTENT STARTS HERE * .\" ----------------------------------------------------------------- .SH "NAME" jemalloc \- general purpose memory allocation functions .SH "LIBRARY" .PP -This manual describes jemalloc 4\&.0\&.2\-0\-g486d249fb4715fd3de679b6c2a04f7e657883111\&. More information can be found at the +This manual describes jemalloc 4\&.0\&.4\-0\-g91010a9e2ebfc84b1ac1ed7fdde3bfed4f65f180\&. More information can be found at the \m[blue]\fBjemalloc website\fR\m[]\&\s-2\u[1]\d\s+2\&. .PP The following configuration options are enabled in libc\*(Aqs built\-in jemalloc: \fB\-\-enable\-fill\fR, \fB\-\-enable\-lazy\-lock\fR, \fB\-\-enable\-munmap\fR, \fB\-\-enable\-stats\fR, \fB\-\-enable\-tcache\fR, \fB\-\-enable\-tls\fR, \fB\-\-enable\-utrace\fR, and \fB\-\-enable\-xmalloc\fR\&. Additionally, \fB\-\-enable\-debug\fR is enabled in development versions of FreeBSD (controlled by the \fBMALLOC_PRODUCTION\fR make variable)\&. .SH "SYNOPSIS" .sp .ft B .nf #include #include .fi .ft .SS "Standard API" .HP \w'void\ *malloc('u .BI "void *malloc(size_t\ " "size" ");" .HP \w'void\ *calloc('u .BI "void *calloc(size_t\ " "number" ", size_t\ " "size" ");" .HP \w'int\ posix_memalign('u .BI "int posix_memalign(void\ **" "ptr" ", size_t\ " "alignment" ", size_t\ " "size" ");" .HP \w'void\ *aligned_alloc('u .BI "void *aligned_alloc(size_t\ " "alignment" ", size_t\ " "size" ");" .HP \w'void\ *realloc('u .BI "void *realloc(void\ *" "ptr" ", size_t\ " "size" ");" .HP \w'void\ free('u .BI "void free(void\ *" "ptr" ");" .SS "Non\-standard API" .HP \w'void\ *mallocx('u .BI "void *mallocx(size_t\ " "size" ", int\ " "flags" ");" .HP \w'void\ *rallocx('u .BI "void *rallocx(void\ *" "ptr" ", size_t\ " "size" ", int\ " "flags" ");" .HP \w'size_t\ xallocx('u .BI "size_t xallocx(void\ *" "ptr" ", size_t\ " "size" ", size_t\ " "extra" ", int\ " "flags" ");" .HP \w'size_t\ sallocx('u .BI "size_t sallocx(void\ *" "ptr" ", int\ " "flags" ");" .HP \w'void\ dallocx('u .BI "void dallocx(void\ *" "ptr" ", int\ " "flags" ");" .HP \w'void\ sdallocx('u .BI "void sdallocx(void\ *" "ptr" ", size_t\ " "size" ", int\ " "flags" ");" .HP \w'size_t\ nallocx('u .BI "size_t nallocx(size_t\ " "size" ", int\ " "flags" ");" .HP \w'int\ mallctl('u .BI "int mallctl(const\ char\ *" "name" ", void\ *" "oldp" ", size_t\ *" "oldlenp" ", void\ *" "newp" ", size_t\ " "newlen" ");" .HP \w'int\ mallctlnametomib('u .BI "int mallctlnametomib(const\ char\ *" "name" ", size_t\ *" "mibp" ", size_t\ *" "miblenp" ");" .HP \w'int\ mallctlbymib('u .BI "int mallctlbymib(const\ size_t\ *" "mib" ", size_t\ " "miblen" ", void\ *" "oldp" ", size_t\ *" "oldlenp" ", void\ *" "newp" ", size_t\ " "newlen" ");" .HP \w'void\ malloc_stats_print('u .BI "void malloc_stats_print(void\ " "(*write_cb)" "\ (void\ *,\ const\ char\ *), void\ *" "cbopaque" ", const\ char\ *" "opts" ");" .HP \w'size_t\ malloc_usable_size('u .BI "size_t malloc_usable_size(const\ void\ *" "ptr" ");" .HP \w'void\ (*malloc_message)('u .BI "void (*malloc_message)(void\ *" "cbopaque" ", const\ char\ *" "s" ");" .PP const char *\fImalloc_conf\fR; .SH "DESCRIPTION" .SS "Standard API" .PP The \fBmalloc\fR\fB\fR function allocates \fIsize\fR bytes of uninitialized memory\&. The allocated space is suitably aligned (after possible pointer coercion) for storage of any type of object\&. .PP The \fBcalloc\fR\fB\fR function allocates space for \fInumber\fR objects, each \fIsize\fR bytes in length\&. The result is identical to calling \fBmalloc\fR\fB\fR with an argument of \fInumber\fR * \fIsize\fR, with the exception that the allocated memory is explicitly initialized to zero bytes\&. .PP The \fBposix_memalign\fR\fB\fR function allocates \fIsize\fR bytes of memory such that the allocation\*(Aqs base address is a multiple of \fIalignment\fR, and returns the allocation in the value pointed to by \fIptr\fR\&. The requested \fIalignment\fR must be a power of 2 at least as large as sizeof(\fBvoid *\fR)\&. .PP The \fBaligned_alloc\fR\fB\fR function allocates \fIsize\fR bytes of memory such that the allocation\*(Aqs base address is a multiple of \fIalignment\fR\&. The requested \fIalignment\fR must be a power of 2\&. Behavior is undefined if \fIsize\fR is not an integral multiple of \fIalignment\fR\&. .PP The \fBrealloc\fR\fB\fR function changes the size of the previously allocated memory referenced by \fIptr\fR to \fIsize\fR bytes\&. The contents of the memory are unchanged up to the lesser of the new and old sizes\&. If the new size is larger, the contents of the newly allocated portion of the memory are undefined\&. Upon success, the memory referenced by \fIptr\fR is freed and a pointer to the newly allocated memory is returned\&. Note that \fBrealloc\fR\fB\fR may move the memory allocation, resulting in a different return value than \fIptr\fR\&. If \fIptr\fR is \fBNULL\fR, the \fBrealloc\fR\fB\fR function behaves identically to \fBmalloc\fR\fB\fR for the specified size\&. .PP The \fBfree\fR\fB\fR function causes the allocated memory referenced by \fIptr\fR to be made available for future allocations\&. If \fIptr\fR is \fBNULL\fR, no action occurs\&. .SS "Non\-standard API" .PP The \fBmallocx\fR\fB\fR, \fBrallocx\fR\fB\fR, \fBxallocx\fR\fB\fR, \fBsallocx\fR\fB\fR, \fBdallocx\fR\fB\fR, \fBsdallocx\fR\fB\fR, and \fBnallocx\fR\fB\fR functions all have a \fIflags\fR argument that can be used to specify options\&. The functions only check the options that are contextually relevant\&. Use bitwise or (|) operations to specify one or more of the following: .PP \fBMALLOCX_LG_ALIGN(\fR\fB\fIla\fR\fR\fB) \fR .RS 4 Align the memory allocation to start at an address that is a multiple of (1 << \fIla\fR)\&. This macro does not validate that \fIla\fR is within the valid range\&. .RE .PP \fBMALLOCX_ALIGN(\fR\fB\fIa\fR\fR\fB) \fR .RS 4 Align the memory allocation to start at an address that is a multiple of \fIa\fR, where \fIa\fR is a power of two\&. This macro does not validate that \fIa\fR is a power of 2\&. .RE .PP \fBMALLOCX_ZERO\fR .RS 4 Initialize newly allocated memory to contain zero bytes\&. In the growing reallocation case, the real size prior to reallocation defines the boundary between untouched bytes and those that are initialized to contain zero bytes\&. If this macro is absent, newly allocated memory is uninitialized\&. .RE .PP \fBMALLOCX_TCACHE(\fR\fB\fItc\fR\fR\fB) \fR .RS 4 Use the thread\-specific cache (tcache) specified by the identifier \fItc\fR, which must have been acquired via the "tcache\&.create" mallctl\&. This macro does not validate that \fItc\fR specifies a valid identifier\&. .RE .PP \fBMALLOCX_TCACHE_NONE\fR .RS 4 Do not use a thread\-specific cache (tcache)\&. Unless \fBMALLOCX_TCACHE(\fR\fB\fItc\fR\fR\fB)\fR or \fBMALLOCX_TCACHE_NONE\fR is specified, an automatically managed tcache will be used under many circumstances\&. This macro cannot be used in the same \fIflags\fR argument as \fBMALLOCX_TCACHE(\fR\fB\fItc\fR\fR\fB)\fR\&. .RE .PP \fBMALLOCX_ARENA(\fR\fB\fIa\fR\fR\fB) \fR .RS 4 Use the arena specified by the index \fIa\fR\&. This macro has no effect for regions that were allocated via an arena other than the one specified\&. This macro does not validate that \fIa\fR specifies an arena index in the valid range\&. .RE .PP The \fBmallocx\fR\fB\fR function allocates at least \fIsize\fR bytes of memory, and returns a pointer to the base address of the allocation\&. Behavior is undefined if \fIsize\fR is \fB0\fR, or if request size overflows due to size class and/or alignment constraints\&. .PP The \fBrallocx\fR\fB\fR function resizes the allocation at \fIptr\fR to be at least \fIsize\fR bytes, and returns a pointer to the base address of the resulting allocation, which may or may not have moved from its original location\&. Behavior is undefined if \fIsize\fR is \fB0\fR, or if request size overflows due to size class and/or alignment constraints\&. .PP The \fBxallocx\fR\fB\fR function resizes the allocation at \fIptr\fR in place to be at least \fIsize\fR bytes, and returns the real size of the allocation\&. If \fIextra\fR is non\-zero, an attempt is made to resize the allocation to be at least (\fIsize\fR + \fIextra\fR) bytes, though inability to allocate the extra byte(s) will not by itself result in failure to resize\&. Behavior is undefined if \fIsize\fR is \fB0\fR, or if (\fIsize\fR + \fIextra\fR > \fBSIZE_T_MAX\fR)\&. .PP The \fBsallocx\fR\fB\fR function returns the real size of the allocation at \fIptr\fR\&. .PP The \fBdallocx\fR\fB\fR function causes the memory referenced by \fIptr\fR to be made available for future allocations\&. .PP The \fBsdallocx\fR\fB\fR function is an extension of \fBdallocx\fR\fB\fR with a \fIsize\fR parameter to allow the caller to pass in the allocation size as an optimization\&. The minimum valid input size is the original requested size of the allocation, and the maximum valid input size is the corresponding value returned by \fBnallocx\fR\fB\fR or \fBsallocx\fR\fB\fR\&. .PP The \fBnallocx\fR\fB\fR function allocates no memory, but it performs the same size computation as the \fBmallocx\fR\fB\fR function, and returns the real size of the allocation that would result from the equivalent \fBmallocx\fR\fB\fR function call\&. Behavior is undefined if \fIsize\fR is \fB0\fR, or if request size overflows due to size class and/or alignment constraints\&. .PP The \fBmallctl\fR\fB\fR function provides a general interface for introspecting the memory allocator, as well as setting modifiable parameters and triggering actions\&. The period\-separated \fIname\fR argument specifies a location in a tree\-structured namespace; see the MALLCTL NAMESPACE section for documentation on the tree contents\&. To read a value, pass a pointer via \fIoldp\fR to adequate space to contain the value, and a pointer to its length via \fIoldlenp\fR; otherwise pass \fBNULL\fR and \fBNULL\fR\&. Similarly, to write a value, pass a pointer to the value via \fInewp\fR, and its length via \fInewlen\fR; otherwise pass \fBNULL\fR and \fB0\fR\&. .PP The \fBmallctlnametomib\fR\fB\fR function provides a way to avoid repeated name lookups for applications that repeatedly query the same portion of the namespace, by translating a name to a \(lqManagement Information Base\(rq (MIB) that can be passed repeatedly to \fBmallctlbymib\fR\fB\fR\&. Upon successful return from \fBmallctlnametomib\fR\fB\fR, \fImibp\fR contains an array of \fI*miblenp\fR integers, where \fI*miblenp\fR is the lesser of the number of components in \fIname\fR and the input value of \fI*miblenp\fR\&. Thus it is possible to pass a \fI*miblenp\fR that is smaller than the number of period\-separated name components, which results in a partial MIB that can be used as the basis for constructing a complete MIB\&. For name components that are integers (e\&.g\&. the 2 in "arenas\&.bin\&.2\&.size"), the corresponding MIB component will always be that integer\&. Therefore, it is legitimate to construct code like the following: .sp .if n \{\ .RS 4 .\} .nf unsigned nbins, i; size_t mib[4]; size_t len, miblen; len = sizeof(nbins); mallctl("arenas\&.nbins", &nbins, &len, NULL, 0); miblen = 4; mallctlnametomib("arenas\&.bin\&.0\&.size", mib, &miblen); for (i = 0; i < nbins; i++) { size_t bin_size; mib[2] = i; len = sizeof(bin_size); mallctlbymib(mib, miblen, &bin_size, &len, NULL, 0); /* Do something with bin_size\&.\&.\&. */ } .fi .if n \{\ .RE .\} .PP The \fBmalloc_stats_print\fR\fB\fR function writes human\-readable summary statistics via the \fIwrite_cb\fR callback function pointer and \fIcbopaque\fR data passed to \fIwrite_cb\fR, or \fBmalloc_message\fR\fB\fR if \fIwrite_cb\fR is \fBNULL\fR\&. This function can be called repeatedly\&. General information that never changes during execution can be omitted by specifying "g" as a character within the \fIopts\fR string\&. Note that \fBmalloc_message\fR\fB\fR uses the \fBmallctl*\fR\fB\fR functions internally, so inconsistent statistics can be reported if multiple threads use these functions simultaneously\&. If \fB\-\-enable\-stats\fR is specified during configuration, \(lqm\(rq and \(lqa\(rq can be specified to omit merged arena and per arena statistics, respectively; \(lqb\(rq, \(lql\(rq, and \(lqh\(rq can be specified to omit per size class statistics for bins, large objects, and huge objects, respectively\&. Unrecognized characters are silently ignored\&. Note that thread caching may prevent some statistics from being completely up to date, since extra locking would be required to merge counters that track thread cache operations\&. .PP The \fBmalloc_usable_size\fR\fB\fR function returns the usable size of the allocation pointed to by \fIptr\fR\&. The return value may be larger than the size that was requested during allocation\&. The \fBmalloc_usable_size\fR\fB\fR function is not a mechanism for in\-place \fBrealloc\fR\fB\fR; rather it is provided solely as a tool for introspection purposes\&. Any discrepancy between the requested allocation size and the size reported by \fBmalloc_usable_size\fR\fB\fR should not be depended on, since such behavior is entirely implementation\-dependent\&. .SH "TUNING" .PP Once, when the first call is made to one of the memory allocation routines, the allocator initializes its internals based in part on various options that can be specified at compile\- or run\-time\&. .PP The string pointed to by the global variable \fImalloc_conf\fR, the \(lqname\(rq of the file referenced by the symbolic link named /etc/malloc\&.conf, and the value of the environment variable \fBMALLOC_CONF\fR, will be interpreted, in that order, from left to right as options\&. Note that \fImalloc_conf\fR may be read before \fBmain\fR\fB\fR is entered, so the declaration of \fImalloc_conf\fR should specify an initializer that contains the final value to be read by jemalloc\&. \fImalloc_conf\fR is a compile\-time setting, whereas /etc/malloc\&.conf and \fBMALLOC_CONF\fR can be safely set any time prior to program invocation\&. .PP An options string is a comma\-separated list of option:value pairs\&. There is one key corresponding to each "opt\&.*" mallctl (see the MALLCTL NAMESPACE section for options documentation)\&. For example, abort:true,narenas:1 sets the "opt\&.abort" and "opt\&.narenas" options\&. Some options have boolean values (true/false), others have integer values (base 8, 10, or 16, depending on prefix), and yet others have raw string values\&. .SH "IMPLEMENTATION NOTES" .PP Traditionally, allocators have used \fBsbrk\fR(2) to obtain memory, which is suboptimal for several reasons, including race conditions, increased fragmentation, and artificial limitations on maximum usable memory\&. If \fBsbrk\fR(2) is supported by the operating system, this allocator uses both \fBmmap\fR(2) and \fBsbrk\fR(2), in that order of preference; otherwise only \fBmmap\fR(2) is used\&. .PP This allocator uses multiple arenas in order to reduce lock contention for threaded programs on multi\-processor systems\&. This works well with regard to threading scalability, but incurs some costs\&. There is a small fixed per\-arena overhead, and additionally, arenas manage memory completely independently of each other, which means a small fixed increase in overall memory fragmentation\&. These overheads are not generally an issue, given the number of arenas normally used\&. Note that using substantially more arenas than the default is not likely to improve performance, mainly due to reduced cache performance\&. However, it may make sense to reduce the number of arenas if an application does not make much use of the allocation functions\&. .PP In addition to multiple arenas, unless \fB\-\-disable\-tcache\fR is specified during configuration, this allocator supports thread\-specific caching for small and large objects, in order to make it possible to completely avoid synchronization for most allocation requests\&. Such caching allows very fast allocation in the common case, but it increases memory usage and fragmentation, since a bounded number of objects can remain allocated in each thread cache\&. .PP Memory is conceptually broken into equal\-sized chunks, where the chunk size is a power of two that is greater than the page size\&. Chunks are always aligned to multiples of the chunk size\&. This alignment makes it possible to find metadata for user objects very quickly\&. .PP User objects are broken into three categories according to size: small, large, and huge\&. Small and large objects are managed entirely by arenas; huge objects are additionally aggregated in a single data structure that is shared by all threads\&. Huge objects are typically used by applications infrequently enough that this single data structure is not a scalability issue\&. .PP Each chunk that is managed by an arena tracks its contents as runs of contiguous pages (unused, backing a set of small objects, or backing one large object)\&. The combination of chunk alignment and chunk page maps makes it possible to determine all metadata regarding small and large allocations in constant time\&. .PP Small objects are managed in groups by page runs\&. Each run maintains a bitmap to track which regions are in use\&. Allocation requests that are no more than half the quantum (8 or 16, depending on architecture) are rounded up to the nearest power of two that is at least sizeof(\fBdouble\fR)\&. All other object size classes are multiples of the quantum, spaced such that there are four size classes for each doubling in size, which limits internal fragmentation to approximately 20% for all but the smallest size classes\&. Small size classes are smaller than four times the page size, large size classes are smaller than the chunk size (see the "opt\&.lg_chunk" option), and huge size classes extend from the chunk size up to one size class less than the full address space size\&. .PP Allocations are packed tightly together, which can be an issue for multi\-threaded applications\&. If you need to assure that allocations do not suffer from cacheline sharing, round your allocation requests up to the nearest multiple of the cacheline size, or specify cacheline alignment when allocating\&. .PP The \fBrealloc\fR\fB\fR, \fBrallocx\fR\fB\fR, and \fBxallocx\fR\fB\fR functions may resize allocations without moving them under limited circumstances\&. Unlike the \fB*allocx\fR\fB\fR API, the standard API does not officially round up the usable size of an allocation to the nearest size class, so technically it is necessary to call \fBrealloc\fR\fB\fR to grow e\&.g\&. a 9\-byte allocation to 16 bytes, or shrink a 16\-byte allocation to 9 bytes\&. Growth and shrinkage trivially succeeds in place as long as the pre\-size and post\-size both round up to the same size class\&. No other API guarantees are made regarding in\-place resizing, but the current implementation also tries to resize large and huge allocations in place, as long as the pre\-size and post\-size are both large or both huge\&. In such cases shrinkage always succeeds for large size classes, but for huge size classes the chunk allocator must support splitting (see "arena\&.\&.chunk_hooks")\&. Growth only succeeds if the trailing memory is currently available, and additionally for huge size classes the chunk allocator must support merging\&. .PP Assuming 2 MiB chunks, 4 KiB pages, and a 16\-byte quantum on a 64\-bit system, the size classes in each category are as shown in Table 1\&. .sp .it 1 an-trap .nr an-no-space-flag 1 .nr an-break-flag 1 .br .B Table\ \&1.\ \&Size classes .TS allbox tab(:); lB rB lB. T{ Category T}:T{ Spacing T}:T{ Size T} .T& l r l ^ r l ^ r l ^ r l ^ r l ^ r l ^ r l ^ r l ^ r l l r l ^ r l ^ r l ^ r l ^ r l ^ r l ^ r l ^ r l l r l ^ r l ^ r l ^ r l ^ r l ^ r l ^ r l. T{ Small T}:T{ lg T}:T{ [8] T} :T{ 16 T}:T{ [16, 32, 48, 64, 80, 96, 112, 128] T} :T{ 32 T}:T{ [160, 192, 224, 256] T} :T{ 64 T}:T{ [320, 384, 448, 512] T} :T{ 128 T}:T{ [640, 768, 896, 1024] T} :T{ 256 T}:T{ [1280, 1536, 1792, 2048] T} :T{ 512 T}:T{ [2560, 3072, 3584, 4096] T} :T{ 1 KiB T}:T{ [5 KiB, 6 KiB, 7 KiB, 8 KiB] T} :T{ 2 KiB T}:T{ [10 KiB, 12 KiB, 14 KiB] T} T{ Large T}:T{ 2 KiB T}:T{ [16 KiB] T} :T{ 4 KiB T}:T{ [20 KiB, 24 KiB, 28 KiB, 32 KiB] T} :T{ 8 KiB T}:T{ [40 KiB, 48 KiB, 54 KiB, 64 KiB] T} :T{ 16 KiB T}:T{ [80 KiB, 96 KiB, 112 KiB, 128 KiB] T} :T{ 32 KiB T}:T{ [160 KiB, 192 KiB, 224 KiB, 256 KiB] T} :T{ 64 KiB T}:T{ [320 KiB, 384 KiB, 448 KiB, 512 KiB] T} :T{ 128 KiB T}:T{ [640 KiB, 768 KiB, 896 KiB, 1 MiB] T} :T{ 256 KiB T}:T{ [1280 KiB, 1536 KiB, 1792 KiB] T} T{ Huge T}:T{ 256 KiB T}:T{ [2 MiB] T} :T{ 512 KiB T}:T{ [2560 KiB, 3 MiB, 3584 KiB, 4 MiB] T} :T{ 1 MiB T}:T{ [5 MiB, 6 MiB, 7 MiB, 8 MiB] T} :T{ 2 MiB T}:T{ [10 MiB, 12 MiB, 14 MiB, 16 MiB] T} :T{ 4 MiB T}:T{ [20 MiB, 24 MiB, 28 MiB, 32 MiB] T} :T{ 8 MiB T}:T{ [40 MiB, 48 MiB, 56 MiB, 64 MiB] T} :T{ \&.\&.\&. T}:T{ \&.\&.\&. T} .TE .sp 1 .SH "MALLCTL NAMESPACE" .PP The following names are defined in the namespace accessible via the \fBmallctl*\fR\fB\fR functions\&. Value types are specified in parentheses, their readable/writable statuses are encoded as rw, r\-, \-w, or \-\-, and required build configuration flags follow, if any\&. A name element encoded as or indicates an integer component, where the integer varies from 0 to some upper value that must be determined via introspection\&. In the case of "stats\&.arenas\&.\&.*", equal to "arenas\&.narenas" can be used to access the summation of statistics from all arenas\&. Take special note of the "epoch" mallctl, which controls refreshing of cached dynamic statistics\&. .PP "version" (\fBconst char *\fR) r\- .RS 4 Return the jemalloc version string\&. .RE .PP "epoch" (\fBuint64_t\fR) rw .RS 4 If a value is passed in, refresh the data from which the \fBmallctl*\fR\fB\fR functions report values, and increment the epoch\&. Return the current epoch\&. This is useful for detecting whether another thread caused a refresh\&. .RE .PP "config\&.cache_oblivious" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-cache\-oblivious\fR was specified during build configuration\&. .RE .PP "config\&.debug" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-debug\fR was specified during build configuration\&. .RE .PP "config\&.fill" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-fill\fR was specified during build configuration\&. .RE .PP "config\&.lazy_lock" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-lazy\-lock\fR was specified during build configuration\&. .RE .PP "config\&.munmap" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-munmap\fR was specified during build configuration\&. .RE .PP "config\&.prof" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-prof\fR was specified during build configuration\&. .RE .PP "config\&.prof_libgcc" (\fBbool\fR) r\- .RS 4 \fB\-\-disable\-prof\-libgcc\fR was not specified during build configuration\&. .RE .PP "config\&.prof_libunwind" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-prof\-libunwind\fR was specified during build configuration\&. .RE .PP "config\&.stats" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-stats\fR was specified during build configuration\&. .RE .PP "config\&.tcache" (\fBbool\fR) r\- .RS 4 \fB\-\-disable\-tcache\fR was not specified during build configuration\&. .RE .PP "config\&.tls" (\fBbool\fR) r\- .RS 4 \fB\-\-disable\-tls\fR was not specified during build configuration\&. .RE .PP "config\&.utrace" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-utrace\fR was specified during build configuration\&. .RE .PP "config\&.valgrind" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-valgrind\fR was specified during build configuration\&. .RE .PP "config\&.xmalloc" (\fBbool\fR) r\- .RS 4 \fB\-\-enable\-xmalloc\fR was specified during build configuration\&. .RE .PP "opt\&.abort" (\fBbool\fR) r\- .RS 4 Abort\-on\-warning enabled/disabled\&. If true, most warnings are fatal\&. The process will call \fBabort\fR(3) in these cases\&. This option is disabled by default unless \fB\-\-enable\-debug\fR is specified during configuration, in which case it is enabled by default\&. .RE .PP "opt\&.dss" (\fBconst char *\fR) r\- .RS 4 dss (\fBsbrk\fR(2)) allocation precedence as related to \fBmmap\fR(2) allocation\&. The following settings are supported if \fBsbrk\fR(2) is supported by the operating system: \(lqdisabled\(rq, \(lqprimary\(rq, and \(lqsecondary\(rq; otherwise only \(lqdisabled\(rq is supported\&. The default is \(lqsecondary\(rq if \fBsbrk\fR(2) is supported by the operating system; \(lqdisabled\(rq otherwise\&. .RE .PP "opt\&.lg_chunk" (\fBsize_t\fR) r\- .RS 4 Virtual memory chunk size (log base 2)\&. If a chunk size outside the supported size range is specified, the size is silently clipped to the minimum/maximum supported size\&. The default chunk size is 2 MiB (2^21)\&. .RE .PP "opt\&.narenas" (\fBsize_t\fR) r\- .RS 4 Maximum number of arenas to use for automatic multiplexing of threads and arenas\&. The default is four times the number of CPUs, or one if there is a single CPU\&. .RE .PP "opt\&.lg_dirty_mult" (\fBssize_t\fR) r\- .RS 4 Per\-arena minimum ratio (log base 2) of active to dirty pages\&. Some dirty unused pages may be allowed to accumulate, within the limit set by the ratio (or one chunk worth of dirty pages, whichever is greater), before informing the kernel about some of those pages via \fBmadvise\fR(2) or a similar system call\&. This provides the kernel with sufficient information to recycle dirty pages if physical memory becomes scarce and the pages remain unused\&. The default minimum ratio is 8:1 (2^3:1); an option value of \-1 will disable dirty page purging\&. See "arenas\&.lg_dirty_mult" and "arena\&.\&.lg_dirty_mult" for related dynamic control options\&. .RE .PP "opt\&.stats_print" (\fBbool\fR) r\- .RS 4 Enable/disable statistics printing at exit\&. If enabled, the \fBmalloc_stats_print\fR\fB\fR function is called at program exit via an \fBatexit\fR(3) function\&. If \fB\-\-enable\-stats\fR is specified during configuration, this has the potential to cause deadlock for a multi\-threaded process that exits while one or more threads are executing in the memory allocation functions\&. Furthermore, \fBatexit\fR\fB\fR may allocate memory during application initialization and then deadlock internally when jemalloc in turn calls \fBatexit\fR\fB\fR, so this option is not univerally usable (though the application can register its own \fBatexit\fR\fB\fR function with equivalent functionality)\&. Therefore, this option should only be used with care; it is primarily intended as a performance tuning aid during application development\&. This option is disabled by default\&. .RE .PP "opt\&.junk" (\fBconst char *\fR) r\- [\fB\-\-enable\-fill\fR] .RS 4 Junk filling\&. If set to "alloc", each byte of uninitialized allocated memory will be initialized to 0xa5\&. If set to "free", all deallocated memory will be initialized to 0x5a\&. If set to "true", both allocated and deallocated memory will be initialized, and if set to "false", junk filling be disabled entirely\&. This is intended for debugging and will impact performance negatively\&. This option is "false" by default unless \fB\-\-enable\-debug\fR is specified during configuration, in which case it is "true" by default unless running inside \m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2\&. .RE .PP "opt\&.quarantine" (\fBsize_t\fR) r\- [\fB\-\-enable\-fill\fR] .RS 4 Per thread quarantine size in bytes\&. If non\-zero, each thread maintains a FIFO object quarantine that stores up to the specified number of bytes of memory\&. The quarantined memory is not freed until it is released from quarantine, though it is immediately junk\-filled if the "opt\&.junk" option is enabled\&. This feature is of particular use in combination with \m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2, which can detect attempts to access quarantined objects\&. This is intended for debugging and will impact performance negatively\&. The default quarantine size is 0 unless running inside Valgrind, in which case the default is 16 MiB\&. .RE .PP "opt\&.redzone" (\fBbool\fR) r\- [\fB\-\-enable\-fill\fR] .RS 4 Redzones enabled/disabled\&. If enabled, small allocations have redzones before and after them\&. Furthermore, if the "opt\&.junk" option is enabled, the redzones are checked for corruption during deallocation\&. However, the primary intended purpose of this feature is to be used in combination with \m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2, which needs redzones in order to do effective buffer overflow/underflow detection\&. This option is intended for debugging and will impact performance negatively\&. This option is disabled by default unless running inside Valgrind\&. .RE .PP "opt\&.zero" (\fBbool\fR) r\- [\fB\-\-enable\-fill\fR] .RS 4 Zero filling enabled/disabled\&. If enabled, each byte of uninitialized allocated memory will be initialized to 0\&. Note that this initialization only happens once for each byte, so \fBrealloc\fR\fB\fR and \fBrallocx\fR\fB\fR calls do not zero memory that was previously allocated\&. This is intended for debugging and will impact performance negatively\&. This option is disabled by default\&. .RE .PP "opt\&.utrace" (\fBbool\fR) r\- [\fB\-\-enable\-utrace\fR] .RS 4 Allocation tracing based on \fButrace\fR(2) enabled/disabled\&. This option is disabled by default\&. .RE .PP "opt\&.xmalloc" (\fBbool\fR) r\- [\fB\-\-enable\-xmalloc\fR] .RS 4 Abort\-on\-out\-of\-memory enabled/disabled\&. If enabled, rather than returning failure for any allocation function, display a diagnostic message on \fBSTDERR_FILENO\fR and cause the program to drop core (using \fBabort\fR(3))\&. If an application is designed to depend on this behavior, set the option at compile time by including the following in the source code: .sp .if n \{\ .RS 4 .\} .nf malloc_conf = "xmalloc:true"; .fi .if n \{\ .RE .\} .sp This option is disabled by default\&. .RE .PP "opt\&.tcache" (\fBbool\fR) r\- [\fB\-\-enable\-tcache\fR] .RS 4 Thread\-specific caching (tcache) enabled/disabled\&. When there are multiple threads, each thread uses a tcache for objects up to a certain size\&. Thread\-specific caching allows many allocations to be satisfied without performing any thread synchronization, at the cost of increased memory use\&. See the "opt\&.lg_tcache_max" option for related tuning information\&. This option is enabled by default unless running inside \m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2, in which case it is forcefully disabled\&. .RE .PP "opt\&.lg_tcache_max" (\fBsize_t\fR) r\- [\fB\-\-enable\-tcache\fR] .RS 4 Maximum size class (log base 2) to cache in the thread\-specific cache (tcache)\&. At a minimum, all small size classes are cached, and at a maximum all large size classes are cached\&. The default maximum is 32 KiB (2^15)\&. .RE .PP "opt\&.prof" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Memory profiling enabled/disabled\&. If enabled, profile memory allocation activity\&. See the "opt\&.prof_active" option for on\-the\-fly activation/deactivation\&. See the "opt\&.lg_prof_sample" option for probabilistic sampling control\&. See the "opt\&.prof_accum" option for control of cumulative sample reporting\&. See the "opt\&.lg_prof_interval" option for information on interval\-triggered profile dumping, the "opt\&.prof_gdump" option for information on high\-water\-triggered profile dumping, and the "opt\&.prof_final" option for final profile dumping\&. Profile output is compatible with the \fBjeprof\fR command, which is based on the \fBpprof\fR that is developed as part of the \m[blue]\fBgperftools package\fR\m[]\&\s-2\u[3]\d\s+2\&. .RE .PP "opt\&.prof_prefix" (\fBconst char *\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Filename prefix for profile dumps\&. If the prefix is set to the empty string, no automatic dumps will occur; this is primarily useful for disabling the automatic final heap dump (which also disables leak reporting, if enabled)\&. The default prefix is jeprof\&. .RE .PP "opt\&.prof_active" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Profiling activated/deactivated\&. This is a secondary control mechanism that makes it possible to start the application with profiling enabled (see the "opt\&.prof" option) but inactive, then toggle profiling at any time during program execution with the "prof\&.active" mallctl\&. This option is enabled by default\&. .RE .PP "opt\&.prof_thread_active_init" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Initial setting for "thread\&.prof\&.active" in newly created threads\&. The initial setting for newly created threads can also be changed during execution via the "prof\&.thread_active_init" mallctl\&. This option is enabled by default\&. .RE .PP "opt\&.lg_prof_sample" (\fBsize_t\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Average interval (log base 2) between allocation samples, as measured in bytes of allocation activity\&. Increasing the sampling interval decreases profile fidelity, but also decreases the computational overhead\&. The default sample interval is 512 KiB (2^19 B)\&. .RE .PP "opt\&.prof_accum" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Reporting of cumulative object/byte counts in profile dumps enabled/disabled\&. If this option is enabled, every unique backtrace must be stored for the duration of execution\&. Depending on the application, this can impose a large memory overhead, and the cumulative counts are not always of interest\&. This option is disabled by default\&. .RE .PP "opt\&.lg_prof_interval" (\fBssize_t\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Average interval (log base 2) between memory profile dumps, as measured in bytes of allocation activity\&. The actual interval between dumps may be sporadic because decentralized allocation counters are used to avoid synchronization bottlenecks\&. Profiles are dumped to files named according to the pattern \&.\&.\&.i\&.heap, where is controlled by the "opt\&.prof_prefix" option\&. By default, interval\-triggered profile dumping is disabled (encoded as \-1)\&. .RE .PP "opt\&.prof_gdump" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Set the initial state of "prof\&.gdump", which when enabled triggers a memory profile dump every time the total virtual memory exceeds the previous maximum\&. This option is disabled by default\&. .RE .PP "opt\&.prof_final" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Use an \fBatexit\fR(3) function to dump final memory usage to a file named according to the pattern \&.\&.\&.f\&.heap, where is controlled by the "opt\&.prof_prefix" option\&. Note that \fBatexit\fR\fB\fR may allocate memory during application initialization and then deadlock internally when jemalloc in turn calls \fBatexit\fR\fB\fR, so this option is not univerally usable (though the application can register its own \fBatexit\fR\fB\fR function with equivalent functionality)\&. This option is disabled by default\&. .RE .PP "opt\&.prof_leak" (\fBbool\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Leak reporting enabled/disabled\&. If enabled, use an \fBatexit\fR(3) function to report memory leaks detected by allocation sampling\&. See the "opt\&.prof" option for information on analyzing heap profile output\&. This option is disabled by default\&. .RE .PP "thread\&.arena" (\fBunsigned\fR) rw .RS 4 Get or set the arena associated with the calling thread\&. If the specified arena was not initialized beforehand (see the "arenas\&.initialized" mallctl), it will be automatically initialized as a side effect of calling this interface\&. .RE .PP "thread\&.allocated" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Get the total number of bytes ever allocated by the calling thread\&. This counter has the potential to wrap around; it is up to the application to appropriately interpret the counter in such cases\&. .RE .PP "thread\&.allocatedp" (\fBuint64_t *\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Get a pointer to the the value that is returned by the "thread\&.allocated" mallctl\&. This is useful for avoiding the overhead of repeated \fBmallctl*\fR\fB\fR calls\&. .RE .PP "thread\&.deallocated" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Get the total number of bytes ever deallocated by the calling thread\&. This counter has the potential to wrap around; it is up to the application to appropriately interpret the counter in such cases\&. .RE .PP "thread\&.deallocatedp" (\fBuint64_t *\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Get a pointer to the the value that is returned by the "thread\&.deallocated" mallctl\&. This is useful for avoiding the overhead of repeated \fBmallctl*\fR\fB\fR calls\&. .RE .PP "thread\&.tcache\&.enabled" (\fBbool\fR) rw [\fB\-\-enable\-tcache\fR] .RS 4 Enable/disable calling thread\*(Aqs tcache\&. The tcache is implicitly flushed as a side effect of becoming disabled (see "thread\&.tcache\&.flush")\&. .RE .PP "thread\&.tcache\&.flush" (\fBvoid\fR) \-\- [\fB\-\-enable\-tcache\fR] .RS 4 Flush calling thread\*(Aqs thread\-specific cache (tcache)\&. This interface releases all cached objects and internal data structures associated with the calling thread\*(Aqs tcache\&. Ordinarily, this interface need not be called, since automatic periodic incremental garbage collection occurs, and the thread cache is automatically discarded when a thread exits\&. However, garbage collection is triggered by allocation activity, so it is possible for a thread that stops allocating/deallocating to retain its cache indefinitely, in which case the developer may find manual flushing useful\&. .RE .PP "thread\&.prof\&.name" (\fBconst char *\fR) r\- or \-w [\fB\-\-enable\-prof\fR] .RS 4 -Get/set the descriptive name associated with the calling thread in memory profile dumps\&. An internal copy of the name string is created, so the input string need not be maintained after this interface completes execution\&. The output string of this interface should be copied for non\-ephemeral uses, because multiple implementation details can cause asynchronous string deallocation\&. Furthermore, each invocation of this interface can only read or write; simultaneous read/write is not supported due to string lifetime limitations\&. The name string must nil\-terminated and comprised only of characters in the sets recognized by +Get/set the descriptive name associated with the calling thread in memory profile dumps\&. An internal copy of the name string is created, so the input string need not be maintained after this interface completes execution\&. The output string of this interface should be copied for non\-ephemeral uses, because multiple implementation details can cause asynchronous string deallocation\&. Furthermore, each invocation of this interface can only read or write; simultaneous read/write is not supported due to string lifetime limitations\&. The name string must be nil\-terminated and comprised only of characters in the sets recognized by \fBisgraph\fR(3) and \fBisblank\fR(3)\&. .RE .PP "thread\&.prof\&.active" (\fBbool\fR) rw [\fB\-\-enable\-prof\fR] .RS 4 Control whether sampling is currently active for the calling thread\&. This is an activation mechanism in addition to "prof\&.active"; both must be active for the calling thread to sample\&. This flag is enabled by default\&. .RE .PP "tcache\&.create" (\fBunsigned\fR) r\- [\fB\-\-enable\-tcache\fR] .RS 4 Create an explicit thread\-specific cache (tcache) and return an identifier that can be passed to the \fBMALLOCX_TCACHE(\fR\fB\fItc\fR\fR\fB)\fR macro to explicitly use the specified cache rather than the automatically managed one that is used by default\&. Each explicit cache can be used by only one thread at a time; the application must assure that this constraint holds\&. .RE .PP "tcache\&.flush" (\fBunsigned\fR) \-w [\fB\-\-enable\-tcache\fR] .RS 4 Flush the specified thread\-specific cache (tcache)\&. The same considerations apply to this interface as to "thread\&.tcache\&.flush", except that the tcache will never be automatically be discarded\&. .RE .PP "tcache\&.destroy" (\fBunsigned\fR) \-w [\fB\-\-enable\-tcache\fR] .RS 4 Flush the specified thread\-specific cache (tcache) and make the identifier available for use during a future tcache creation\&. .RE .PP "arena\&.\&.purge" (\fBvoid\fR) \-\- .RS 4 Purge unused dirty pages for arena , or for all arenas if equals "arenas\&.narenas"\&. .RE .PP "arena\&.\&.dss" (\fBconst char *\fR) rw .RS 4 Set the precedence of dss allocation as related to mmap allocation for arena , or for all arenas if equals "arenas\&.narenas"\&. See "opt\&.dss" for supported settings\&. .RE .PP "arena\&.\&.lg_dirty_mult" (\fBssize_t\fR) rw .RS 4 Current per\-arena minimum ratio (log base 2) of active to dirty pages for arena \&. Each time this interface is set and the ratio is increased, pages are synchronously purged as necessary to impose the new ratio\&. See "opt\&.lg_dirty_mult" for additional information\&. .RE .PP "arena\&.\&.chunk_hooks" (\fBchunk_hooks_t\fR) rw .RS 4 Get or set the chunk management hook functions for arena \&. The functions must be capable of operating on all extant chunks associated with arena , usually by passing unknown chunks to the replaced functions\&. In practice, it is feasible to control allocation for arenas created via "arenas\&.extend" such that all chunks originate from an application\-supplied chunk allocator (by setting custom chunk hook functions just after arena creation), but the automatically created arenas may have already created chunks prior to the application having an opportunity to take over chunk allocation\&. .sp .if n \{\ .RS 4 .\} .nf typedef struct { chunk_alloc_t *alloc; chunk_dalloc_t *dalloc; chunk_commit_t *commit; chunk_decommit_t *decommit; chunk_purge_t *purge; chunk_split_t *split; chunk_merge_t *merge; } chunk_hooks_t; .fi .if n \{\ .RE .\} .sp The \fBchunk_hooks_t\fR structure comprises function pointers which are described individually below\&. jemalloc uses these functions to manage chunk lifetime, which starts off with allocation of mapped committed memory, in the simplest case followed by deallocation\&. However, there are performance and platform reasons to retain chunks for later reuse\&. Cleanup attempts cascade from deallocation to decommit to purging, which gives the chunk management functions opportunities to reject the most permanent cleanup operations in favor of less permanent (and often less costly) operations\&. The chunk splitting and merging operations can also be opted out of, but this is mainly intended to support platforms on which virtual memory mappings provided by the operating system kernel do not automatically coalesce and split, e\&.g\&. Windows\&. .HP \w'typedef\ void\ *(chunk_alloc_t)('u .BI "typedef void *(chunk_alloc_t)(void\ *" "chunk" ", size_t\ " "size" ", size_t\ " "alignment" ", bool\ *" "zero" ", bool\ *" "commit" ", unsigned\ " "arena_ind" ");" .sp .if n \{\ .RS 4 .\} .nf .fi .if n \{\ .RE .\} .sp A chunk allocation function conforms to the \fBchunk_alloc_t\fR type and upon success returns a pointer to \fIsize\fR bytes of mapped memory on behalf of arena \fIarena_ind\fR such that the chunk\*(Aqs base address is a multiple of \fIalignment\fR, as well as setting \fI*zero\fR to indicate whether the chunk is zeroed and \fI*commit\fR to indicate whether the chunk is committed\&. Upon error the function returns \fBNULL\fR and leaves \fI*zero\fR and \fI*commit\fR unmodified\&. The \fIsize\fR parameter is always a multiple of the chunk size\&. The \fIalignment\fR parameter is always a power of two at least as large as the chunk size\&. Zeroing is mandatory if \fI*zero\fR is true upon function entry\&. Committing is mandatory if \fI*commit\fR is true upon function entry\&. If \fIchunk\fR is not \fBNULL\fR, the returned pointer must be \fIchunk\fR on success or \fBNULL\fR on error\&. Committed memory may be committed in absolute terms as on a system that does not overcommit, or in implicit terms as on a system that overcommits and satisfies physical memory needs on demand via soft page faults\&. Note that replacing the default chunk allocation function makes the arena\*(Aqs "arena\&.\&.dss" setting irrelevant\&. .HP \w'typedef\ bool\ (chunk_dalloc_t)('u .BI "typedef bool (chunk_dalloc_t)(void\ *" "chunk" ", size_t\ " "size" ", bool\ " "committed" ", unsigned\ " "arena_ind" ");" .sp .if n \{\ .RS 4 .\} .nf .fi .if n \{\ .RE .\} .sp A chunk deallocation function conforms to the \fBchunk_dalloc_t\fR type and deallocates a \fIchunk\fR of given \fIsize\fR with \fIcommitted\fR/decommited memory as indicated, on behalf of arena \fIarena_ind\fR, returning false upon success\&. If the function returns true, this indicates opt\-out from deallocation; the virtual memory mapping associated with the chunk remains mapped, in the same commit state, and available for future use, in which case it will be automatically retained for later reuse\&. .HP \w'typedef\ bool\ (chunk_commit_t)('u .BI "typedef bool (chunk_commit_t)(void\ *" "chunk" ", size_t\ " "size" ", size_t\ " "offset" ", size_t\ " "length" ", unsigned\ " "arena_ind" ");" .sp .if n \{\ .RS 4 .\} .nf .fi .if n \{\ .RE .\} .sp A chunk commit function conforms to the \fBchunk_commit_t\fR type and commits zeroed physical memory to back pages within a \fIchunk\fR of given \fIsize\fR at \fIoffset\fR bytes, extending for \fIlength\fR on behalf of arena \fIarena_ind\fR, returning false upon success\&. Committed memory may be committed in absolute terms as on a system that does not overcommit, or in implicit terms as on a system that overcommits and satisfies physical memory needs on demand via soft page faults\&. If the function returns true, this indicates insufficient physical memory to satisfy the request\&. .HP \w'typedef\ bool\ (chunk_decommit_t)('u .BI "typedef bool (chunk_decommit_t)(void\ *" "chunk" ", size_t\ " "size" ", size_t\ " "offset" ", size_t\ " "length" ", unsigned\ " "arena_ind" ");" .sp .if n \{\ .RS 4 .\} .nf .fi .if n \{\ .RE .\} .sp A chunk decommit function conforms to the \fBchunk_decommit_t\fR type and decommits any physical memory that is backing pages within a \fIchunk\fR of given \fIsize\fR at \fIoffset\fR bytes, extending for \fIlength\fR on behalf of arena \fIarena_ind\fR, returning false upon success, in which case the pages will be committed via the chunk commit function before being reused\&. If the function returns true, this indicates opt\-out from decommit; the memory remains committed and available for future use, in which case it will be automatically retained for later reuse\&. .HP \w'typedef\ bool\ (chunk_purge_t)('u .BI "typedef bool (chunk_purge_t)(void\ *" "chunk" ", size_t" "size" ", size_t\ " "offset" ", size_t\ " "length" ", unsigned\ " "arena_ind" ");" .sp .if n \{\ .RS 4 .\} .nf .fi .if n \{\ .RE .\} .sp A chunk purge function conforms to the \fBchunk_purge_t\fR type and optionally discards physical pages within the virtual memory mapping associated with \fIchunk\fR of given \fIsize\fR at \fIoffset\fR bytes, extending for \fIlength\fR on behalf of arena \fIarena_ind\fR, returning false if pages within the purged virtual memory range will be zero\-filled the next time they are accessed\&. .HP \w'typedef\ bool\ (chunk_split_t)('u .BI "typedef bool (chunk_split_t)(void\ *" "chunk" ", size_t\ " "size" ", size_t\ " "size_a" ", size_t\ " "size_b" ", bool\ " "committed" ", unsigned\ " "arena_ind" ");" .sp .if n \{\ .RS 4 .\} .nf .fi .if n \{\ .RE .\} .sp A chunk split function conforms to the \fBchunk_split_t\fR type and optionally splits \fIchunk\fR of given \fIsize\fR into two adjacent chunks, the first of \fIsize_a\fR bytes, and the second of \fIsize_b\fR bytes, operating on \fIcommitted\fR/decommitted memory as indicated, on behalf of arena \fIarena_ind\fR, returning false upon success\&. If the function returns true, this indicates that the chunk remains unsplit and therefore should continue to be operated on as a whole\&. .HP \w'typedef\ bool\ (chunk_merge_t)('u .BI "typedef bool (chunk_merge_t)(void\ *" "chunk_a" ", size_t\ " "size_a" ", void\ *" "chunk_b" ", size_t\ " "size_b" ", bool\ " "committed" ", unsigned\ " "arena_ind" ");" .sp .if n \{\ .RS 4 .\} .nf .fi .if n \{\ .RE .\} .sp A chunk merge function conforms to the \fBchunk_merge_t\fR type and optionally merges adjacent chunks, \fIchunk_a\fR of given \fIsize_a\fR and \fIchunk_b\fR of given \fIsize_b\fR into one contiguous chunk, operating on \fIcommitted\fR/decommitted memory as indicated, on behalf of arena \fIarena_ind\fR, returning false upon success\&. If the function returns true, this indicates that the chunks remain distinct mappings and therefore should continue to be operated on independently\&. .RE .PP "arenas\&.narenas" (\fBunsigned\fR) r\- .RS 4 Current limit on number of arenas\&. .RE .PP "arenas\&.initialized" (\fBbool *\fR) r\- .RS 4 An array of "arenas\&.narenas" booleans\&. Each boolean indicates whether the corresponding arena is initialized\&. .RE .PP "arenas\&.lg_dirty_mult" (\fBssize_t\fR) rw .RS 4 Current default per\-arena minimum ratio (log base 2) of active to dirty pages, used to initialize "arena\&.\&.lg_dirty_mult" during arena creation\&. See "opt\&.lg_dirty_mult" for additional information\&. .RE .PP "arenas\&.quantum" (\fBsize_t\fR) r\- .RS 4 Quantum size\&. .RE .PP "arenas\&.page" (\fBsize_t\fR) r\- .RS 4 Page size\&. .RE .PP "arenas\&.tcache_max" (\fBsize_t\fR) r\- [\fB\-\-enable\-tcache\fR] .RS 4 Maximum thread\-cached size class\&. .RE .PP "arenas\&.nbins" (\fBunsigned\fR) r\- .RS 4 Number of bin size classes\&. .RE .PP "arenas\&.nhbins" (\fBunsigned\fR) r\- [\fB\-\-enable\-tcache\fR] .RS 4 Total number of thread cache bin size classes\&. .RE .PP "arenas\&.bin\&.\&.size" (\fBsize_t\fR) r\- .RS 4 Maximum size supported by size class\&. .RE .PP "arenas\&.bin\&.\&.nregs" (\fBuint32_t\fR) r\- .RS 4 Number of regions per page run\&. .RE .PP "arenas\&.bin\&.\&.run_size" (\fBsize_t\fR) r\- .RS 4 Number of bytes per page run\&. .RE .PP "arenas\&.nlruns" (\fBunsigned\fR) r\- .RS 4 Total number of large size classes\&. .RE .PP "arenas\&.lrun\&.\&.size" (\fBsize_t\fR) r\- .RS 4 Maximum size supported by this large size class\&. .RE .PP "arenas\&.nhchunks" (\fBunsigned\fR) r\- .RS 4 Total number of huge size classes\&. .RE .PP "arenas\&.hchunk\&.\&.size" (\fBsize_t\fR) r\- .RS 4 Maximum size supported by this huge size class\&. .RE .PP "arenas\&.extend" (\fBunsigned\fR) r\- .RS 4 Extend the array of arenas by appending a new arena, and returning the new arena index\&. .RE .PP "prof\&.thread_active_init" (\fBbool\fR) rw [\fB\-\-enable\-prof\fR] .RS 4 Control the initial setting for "thread\&.prof\&.active" in newly created threads\&. See the "opt\&.prof_thread_active_init" option for additional information\&. .RE .PP "prof\&.active" (\fBbool\fR) rw [\fB\-\-enable\-prof\fR] .RS 4 Control whether sampling is currently active\&. See the "opt\&.prof_active" option for additional information, as well as the interrelated "thread\&.prof\&.active" mallctl\&. .RE .PP "prof\&.dump" (\fBconst char *\fR) \-w [\fB\-\-enable\-prof\fR] .RS 4 Dump a memory profile to the specified file, or if NULL is specified, to a file according to the pattern \&.\&.\&.m\&.heap, where is controlled by the "opt\&.prof_prefix" option\&. .RE .PP "prof\&.gdump" (\fBbool\fR) rw [\fB\-\-enable\-prof\fR] .RS 4 When enabled, trigger a memory profile dump every time the total virtual memory exceeds the previous maximum\&. Profiles are dumped to files named according to the pattern \&.\&.\&.u\&.heap, where is controlled by the "opt\&.prof_prefix" option\&. .RE .PP "prof\&.reset" (\fBsize_t\fR) \-w [\fB\-\-enable\-prof\fR] .RS 4 Reset all memory profile statistics, and optionally update the sample rate (see "opt\&.lg_prof_sample" and "prof\&.lg_sample")\&. .RE .PP "prof\&.lg_sample" (\fBsize_t\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Get the current sample rate (see "opt\&.lg_prof_sample")\&. .RE .PP "prof\&.interval" (\fBuint64_t\fR) r\- [\fB\-\-enable\-prof\fR] .RS 4 Average number of bytes allocated between inverval\-based profile dumps\&. See the "opt\&.lg_prof_interval" option for additional information\&. .RE .PP "stats\&.cactive" (\fBsize_t *\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Pointer to a counter that contains an approximate count of the current number of bytes in active pages\&. The estimate may be high, but never low, because each arena rounds up when computing its contribution to the counter\&. Note that the "epoch" mallctl has no bearing on this counter\&. Furthermore, counter consistency is maintained via atomic operations, so it is necessary to use an atomic operation in order to guarantee a consistent read when dereferencing the pointer\&. .RE .PP "stats\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Total number of bytes allocated by the application\&. .RE .PP "stats\&.active" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Total number of bytes in active pages allocated by the application\&. This is a multiple of the page size, and greater than or equal to "stats\&.allocated"\&. This does not include "stats\&.arenas\&.\&.pdirty", nor pages entirely devoted to allocator metadata\&. .RE .PP "stats\&.metadata" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Total number of bytes dedicated to metadata, which comprise base allocations used for bootstrap\-sensitive internal allocator data structures, arena chunk headers (see "stats\&.arenas\&.\&.metadata\&.mapped"), and internal allocations (see "stats\&.arenas\&.\&.metadata\&.allocated")\&. .RE .PP "stats\&.resident" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Maximum number of bytes in physically resident data pages mapped by the allocator, comprising all pages dedicated to allocator metadata, pages backing active allocations, and unused dirty pages\&. This is a maximum rather than precise because pages may not actually be physically resident if they correspond to demand\-zeroed virtual memory that has not yet been touched\&. This is a multiple of the page size, and is larger than "stats\&.active"\&. .RE .PP "stats\&.mapped" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Total number of bytes in active chunks mapped by the allocator\&. This is a multiple of the chunk size, and is larger than "stats\&.active"\&. This does not include inactive chunks, even those that contain unused dirty pages, which means that there is no strict ordering between this and "stats\&.resident"\&. .RE .PP "stats\&.arenas\&.\&.dss" (\fBconst char *\fR) r\- .RS 4 dss (\fBsbrk\fR(2)) allocation precedence as related to \fBmmap\fR(2) allocation\&. See "opt\&.dss" for details\&. .RE .PP "stats\&.arenas\&.\&.lg_dirty_mult" (\fBssize_t\fR) r\- .RS 4 Minimum ratio (log base 2) of active to dirty pages\&. See "opt\&.lg_dirty_mult" for details\&. .RE .PP "stats\&.arenas\&.\&.nthreads" (\fBunsigned\fR) r\- .RS 4 Number of threads currently assigned to arena\&. .RE .PP "stats\&.arenas\&.\&.pactive" (\fBsize_t\fR) r\- .RS 4 Number of pages in active runs\&. .RE .PP "stats\&.arenas\&.\&.pdirty" (\fBsize_t\fR) r\- .RS 4 Number of pages within unused runs that are potentially dirty, and for which \fBmadvise\fR\fB\fI\&.\&.\&.\fR\fR\fB \fR\fB\fI\fBMADV_DONTNEED\fR\fR\fR or similar has not been called\&. .RE .PP "stats\&.arenas\&.\&.mapped" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of mapped bytes\&. .RE .PP "stats\&.arenas\&.\&.metadata\&.mapped" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of mapped bytes in arena chunk headers, which track the states of the non\-metadata pages\&. .RE .PP "stats\&.arenas\&.\&.metadata\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of bytes dedicated to internal allocations\&. Internal allocations differ from application\-originated allocations in that they are for internal use, and that they are omitted from heap profiles\&. This statistic is reported separately from "stats\&.metadata" and "stats\&.arenas\&.\&.metadata\&.mapped" because it overlaps with e\&.g\&. the "stats\&.allocated" and "stats\&.active" statistics, whereas the other metadata statistics do not\&. .RE .PP "stats\&.arenas\&.\&.npurge" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of dirty page purge sweeps performed\&. .RE .PP "stats\&.arenas\&.\&.nmadvise" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of \fBmadvise\fR\fB\fI\&.\&.\&.\fR\fR\fB \fR\fB\fI\fBMADV_DONTNEED\fR\fR\fR or similar calls made to purge dirty pages\&. .RE .PP "stats\&.arenas\&.\&.purged" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of pages purged\&. .RE .PP "stats\&.arenas\&.\&.small\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of bytes currently allocated by small objects\&. .RE .PP "stats\&.arenas\&.\&.small\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocation requests served by small bins\&. .RE .PP "stats\&.arenas\&.\&.small\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of small objects returned to bins\&. .RE .PP "stats\&.arenas\&.\&.small\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of small allocation requests\&. .RE .PP "stats\&.arenas\&.\&.large\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of bytes currently allocated by large objects\&. .RE .PP "stats\&.arenas\&.\&.large\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of large allocation requests served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.large\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of large deallocation requests served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.large\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of large allocation requests\&. .RE .PP "stats\&.arenas\&.\&.huge\&.allocated" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Number of bytes currently allocated by huge objects\&. .RE .PP "stats\&.arenas\&.\&.huge\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of huge allocation requests served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.huge\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of huge deallocation requests served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.huge\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of huge allocation requests\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocations served by bin\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocations returned to bin\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocation requests\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.curregs" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Current number of regions for this size class\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.nfills" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR \fB\-\-enable\-tcache\fR] .RS 4 Cumulative number of tcache fills\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.nflushes" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR \fB\-\-enable\-tcache\fR] .RS 4 Cumulative number of tcache flushes\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.nruns" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of runs created\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.nreruns" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of times the current run from which to allocate changed\&. .RE .PP "stats\&.arenas\&.\&.bins\&.\&.curruns" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Current number of runs\&. .RE .PP "stats\&.arenas\&.\&.lruns\&.\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocation requests for this size class served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.lruns\&.\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of deallocation requests for this size class served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.lruns\&.\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocation requests for this size class\&. .RE .PP "stats\&.arenas\&.\&.lruns\&.\&.curruns" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Current number of runs for this size class\&. .RE .PP "stats\&.arenas\&.\&.hchunks\&.\&.nmalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocation requests for this size class served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.hchunks\&.\&.ndalloc" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of deallocation requests for this size class served directly by the arena\&. .RE .PP "stats\&.arenas\&.\&.hchunks\&.\&.nrequests" (\fBuint64_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Cumulative number of allocation requests for this size class\&. .RE .PP "stats\&.arenas\&.\&.hchunks\&.\&.curhchunks" (\fBsize_t\fR) r\- [\fB\-\-enable\-stats\fR] .RS 4 Current number of huge allocations for this size class\&. .RE .SH "DEBUGGING MALLOC PROBLEMS" .PP When debugging, it is a good idea to configure/build jemalloc with the \fB\-\-enable\-debug\fR and \fB\-\-enable\-fill\fR options, and recompile the program with suitable options and symbols for debugger support\&. When so configured, jemalloc incorporates a wide variety of run\-time assertions that catch application errors such as double\-free, write\-after\-free, etc\&. .PP Programs often accidentally depend on \(lquninitialized\(rq memory actually being filled with zero bytes\&. Junk filling (see the "opt\&.junk" option) tends to expose such bugs in the form of obviously incorrect results and/or coredumps\&. Conversely, zero filling (see the "opt\&.zero" option) eliminates the symptoms of such bugs\&. Between these two options, it is usually possible to quickly detect, diagnose, and eliminate such bugs\&. .PP This implementation does not provide much detail about the problems it detects, because the performance impact for storing such information would be prohibitive\&. However, jemalloc does integrate with the most excellent \m[blue]\fBValgrind\fR\m[]\&\s-2\u[2]\d\s+2 tool if the \fB\-\-enable\-valgrind\fR configuration option is enabled\&. .SH "DIAGNOSTIC MESSAGES" .PP If any of the memory allocation/deallocation functions detect an error or warning condition, a message will be printed to file descriptor \fBSTDERR_FILENO\fR\&. Errors will result in the process dumping core\&. If the "opt\&.abort" option is set, most warnings are treated as errors\&. .PP The \fImalloc_message\fR variable allows the programmer to override the function which emits the text strings forming the errors and warnings if for some reason the \fBSTDERR_FILENO\fR file descriptor is not suitable for this\&. \fBmalloc_message\fR\fB\fR takes the \fIcbopaque\fR pointer argument that is \fBNULL\fR unless overridden by the arguments in a call to \fBmalloc_stats_print\fR\fB\fR, followed by a string pointer\&. Please note that doing anything which tries to allocate memory in this function is likely to result in a crash or deadlock\&. .PP All messages are prefixed by \(lq:\(rq\&. .SH "RETURN VALUES" .SS "Standard API" .PP The \fBmalloc\fR\fB\fR and \fBcalloc\fR\fB\fR functions return a pointer to the allocated memory if successful; otherwise a \fBNULL\fR pointer is returned and \fIerrno\fR is set to ENOMEM\&. .PP The \fBposix_memalign\fR\fB\fR function returns the value 0 if successful; otherwise it returns an error value\&. The \fBposix_memalign\fR\fB\fR function will fail if: .PP EINVAL .RS 4 The \fIalignment\fR parameter is not a power of 2 at least as large as sizeof(\fBvoid *\fR)\&. .RE .PP ENOMEM .RS 4 Memory allocation error\&. .RE .PP The \fBaligned_alloc\fR\fB\fR function returns a pointer to the allocated memory if successful; otherwise a \fBNULL\fR pointer is returned and \fIerrno\fR is set\&. The \fBaligned_alloc\fR\fB\fR function will fail if: .PP EINVAL .RS 4 The \fIalignment\fR parameter is not a power of 2\&. .RE .PP ENOMEM .RS 4 Memory allocation error\&. .RE .PP The \fBrealloc\fR\fB\fR function returns a pointer, possibly identical to \fIptr\fR, to the allocated memory if successful; otherwise a \fBNULL\fR pointer is returned, and \fIerrno\fR is set to ENOMEM if the error was the result of an allocation failure\&. The \fBrealloc\fR\fB\fR function always leaves the original buffer intact when an error occurs\&. .PP The \fBfree\fR\fB\fR function returns no value\&. .SS "Non\-standard API" .PP The \fBmallocx\fR\fB\fR and \fBrallocx\fR\fB\fR functions return a pointer to the allocated memory if successful; otherwise a \fBNULL\fR pointer is returned to indicate insufficient contiguous memory was available to service the allocation request\&. .PP The \fBxallocx\fR\fB\fR function returns the real size of the resulting resized allocation pointed to by \fIptr\fR, which is a value less than \fIsize\fR if the allocation could not be adequately grown in place\&. .PP The \fBsallocx\fR\fB\fR function returns the real size of the allocation pointed to by \fIptr\fR\&. .PP The \fBnallocx\fR\fB\fR returns the real size that would result from a successful equivalent \fBmallocx\fR\fB\fR function call, or zero if insufficient memory is available to perform the size computation\&. .PP The \fBmallctl\fR\fB\fR, \fBmallctlnametomib\fR\fB\fR, and \fBmallctlbymib\fR\fB\fR functions return 0 on success; otherwise they return an error value\&. The functions will fail if: .PP EINVAL .RS 4 \fInewp\fR is not \fBNULL\fR, and \fInewlen\fR is too large or too small\&. Alternatively, \fI*oldlenp\fR is too large or too small; in this case as much data as possible are read despite the error\&. .RE .PP ENOENT .RS 4 \fIname\fR or \fImib\fR specifies an unknown/invalid value\&. .RE .PP EPERM .RS 4 Attempt to read or write void value, or attempt to write read\-only value\&. .RE .PP EAGAIN .RS 4 A memory allocation failure occurred\&. .RE .PP EFAULT .RS 4 An interface with side effects failed in some way not directly related to \fBmallctl*\fR\fB\fR read/write processing\&. .RE .PP The \fBmalloc_usable_size\fR\fB\fR function returns the usable size of the allocation pointed to by \fIptr\fR\&. .SH "ENVIRONMENT" .PP The following environment variable affects the execution of the allocation functions: .PP \fBMALLOC_CONF\fR .RS 4 If the environment variable \fBMALLOC_CONF\fR is set, the characters it contains will be interpreted as options\&. .RE .SH "EXAMPLES" .PP To dump core whenever a problem occurs: .sp .if n \{\ .RS 4 .\} .nf ln \-s \*(Aqabort:true\*(Aq /etc/malloc\&.conf .fi .if n \{\ .RE .\} .PP To specify in the source a chunk size that is 16 MiB: .sp .if n \{\ .RS 4 .\} .nf malloc_conf = "lg_chunk:24"; .fi .if n \{\ .RE .\} .SH "SEE ALSO" .PP \fBmadvise\fR(2), \fBmmap\fR(2), \fBsbrk\fR(2), \fButrace\fR(2), \fBalloca\fR(3), \fBatexit\fR(3), \fBgetpagesize\fR(3) .SH "STANDARDS" .PP The \fBmalloc\fR\fB\fR, \fBcalloc\fR\fB\fR, \fBrealloc\fR\fB\fR, and \fBfree\fR\fB\fR functions conform to ISO/IEC 9899:1990 (\(lqISO C90\(rq)\&. .PP The \fBposix_memalign\fR\fB\fR function conforms to IEEE Std 1003\&.1\-2001 (\(lqPOSIX\&.1\(rq)\&. .SH "HISTORY" .PP The \fBmalloc_usable_size\fR\fB\fR and \fBposix_memalign\fR\fB\fR functions first appeared in FreeBSD 7\&.0\&. .PP The \fBaligned_alloc\fR\fB\fR, \fBmalloc_stats_print\fR\fB\fR, and \fBmallctl*\fR\fB\fR functions first appeared in FreeBSD 10\&.0\&. .PP The \fB*allocx\fR\fB\fR functions first appeared in FreeBSD 11\&.0\&. .SH "AUTHOR" .PP \fBJason Evans\fR .RS 4 .RE .SH "NOTES" .IP " 1." 4 jemalloc website .RS 4 \%http://www.canonware.com/jemalloc/ .RE .IP " 2." 4 Valgrind .RS 4 \%http://valgrind.org/ .RE .IP " 3." 4 gperftools package .RS 4 \%http://code.google.com/p/gperftools/ .RE Index: head/contrib/jemalloc/include/jemalloc/internal/jemalloc_internal.h =================================================================== --- head/contrib/jemalloc/include/jemalloc/internal/jemalloc_internal.h (revision 289899) +++ head/contrib/jemalloc/include/jemalloc/internal/jemalloc_internal.h (revision 289900) @@ -1,1131 +1,1135 @@ #ifndef JEMALLOC_INTERNAL_H #define JEMALLOC_INTERNAL_H #include "jemalloc_internal_defs.h" #include "jemalloc/internal/jemalloc_internal_decls.h" #ifdef JEMALLOC_UTRACE #include #endif #include "un-namespace.h" #include "libc_private.h" #define JEMALLOC_NO_DEMANGLE #ifdef JEMALLOC_JET # define JEMALLOC_N(n) jet_##n # include "jemalloc/internal/public_namespace.h" # define JEMALLOC_NO_RENAME # include "../jemalloc.h" # undef JEMALLOC_NO_RENAME #else # define JEMALLOC_N(n) __je_##n # include "../jemalloc.h" #endif #include "jemalloc/internal/private_namespace.h" static const bool config_debug = #ifdef JEMALLOC_DEBUG true #else false #endif ; static const bool have_dss = #ifdef JEMALLOC_DSS true #else false #endif ; static const bool config_fill = #ifdef JEMALLOC_FILL true #else false #endif ; static const bool config_lazy_lock = true; static const bool config_prof = #ifdef JEMALLOC_PROF true #else false #endif ; static const bool config_prof_libgcc = #ifdef JEMALLOC_PROF_LIBGCC true #else false #endif ; static const bool config_prof_libunwind = #ifdef JEMALLOC_PROF_LIBUNWIND true #else false #endif ; static const bool maps_coalesce = #ifdef JEMALLOC_MAPS_COALESCE true #else false #endif ; static const bool config_munmap = #ifdef JEMALLOC_MUNMAP true #else false #endif ; static const bool config_stats = #ifdef JEMALLOC_STATS true #else false #endif ; static const bool config_tcache = #ifdef JEMALLOC_TCACHE true #else false #endif ; static const bool config_tls = #ifdef JEMALLOC_TLS true #else false #endif ; static const bool config_utrace = #ifdef JEMALLOC_UTRACE true #else false #endif ; static const bool config_valgrind = #ifdef JEMALLOC_VALGRIND true #else false #endif ; static const bool config_xmalloc = #ifdef JEMALLOC_XMALLOC true #else false #endif ; static const bool config_ivsalloc = #ifdef JEMALLOC_IVSALLOC true #else false #endif ; static const bool config_cache_oblivious = #ifdef JEMALLOC_CACHE_OBLIVIOUS true #else false #endif ; #ifdef JEMALLOC_C11ATOMICS #include #endif #ifdef JEMALLOC_ATOMIC9 #include #endif #if (defined(JEMALLOC_OSATOMIC) || defined(JEMALLOC_OSSPIN)) #include #endif #ifdef JEMALLOC_ZONE #include #include #include #include #endif #define RB_COMPACT #include "jemalloc/internal/rb.h" #include "jemalloc/internal/qr.h" #include "jemalloc/internal/ql.h" /* * jemalloc can conceptually be broken into components (arena, tcache, etc.), * but there are circular dependencies that cannot be broken without * substantial performance degradation. In order to reduce the effect on * visual code flow, read the header files in multiple passes, with one of the * following cpp variables defined during each pass: * * JEMALLOC_H_TYPES : Preprocessor-defined constants and psuedo-opaque data * types. * JEMALLOC_H_STRUCTS : Data structures. * JEMALLOC_H_EXTERNS : Extern data declarations and function prototypes. * JEMALLOC_H_INLINES : Inline functions. */ /******************************************************************************/ #define JEMALLOC_H_TYPES #include "jemalloc/internal/jemalloc_internal_macros.h" /* Size class index type. */ typedef unsigned szind_t; /* * Flags bits: * * a: arena * t: tcache * 0: unused * z: zero * n: alignment * * aaaaaaaa aaaatttt tttttttt 0znnnnnn */ #define MALLOCX_ARENA_MASK ((int)~0xfffff) #define MALLOCX_ARENA_MAX 0xffe #define MALLOCX_TCACHE_MASK ((int)~0xfff000ffU) #define MALLOCX_TCACHE_MAX 0xffd #define MALLOCX_LG_ALIGN_MASK ((int)0x3f) /* Use MALLOCX_ALIGN_GET() if alignment may not be specified in flags. */ #define MALLOCX_ALIGN_GET_SPECIFIED(flags) \ (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK)) #define MALLOCX_ALIGN_GET(flags) \ (MALLOCX_ALIGN_GET_SPECIFIED(flags) & (SIZE_T_MAX-1)) #define MALLOCX_ZERO_GET(flags) \ ((bool)(flags & MALLOCX_ZERO)) #define MALLOCX_TCACHE_GET(flags) \ (((unsigned)((flags & MALLOCX_TCACHE_MASK) >> 8)) - 2) #define MALLOCX_ARENA_GET(flags) \ (((unsigned)(((unsigned)flags) >> 20)) - 1) /* Smallest size class to support. */ #define TINY_MIN (1U << LG_TINY_MIN) /* * Minimum allocation alignment is 2^LG_QUANTUM bytes (ignoring tiny size * classes). */ #ifndef LG_QUANTUM # if (defined(__i386__) || defined(_M_IX86)) # define LG_QUANTUM 4 # endif # ifdef __ia64__ # define LG_QUANTUM 4 # endif # ifdef __alpha__ # define LG_QUANTUM 4 # endif # if (defined(__sparc64__) || defined(__sparcv9)) # define LG_QUANTUM 4 # endif # if (defined(__amd64__) || defined(__x86_64__) || defined(_M_X64)) # define LG_QUANTUM 4 # endif # ifdef __arm__ # define LG_QUANTUM 3 # endif # ifdef __aarch64__ # define LG_QUANTUM 4 # endif # ifdef __hppa__ # define LG_QUANTUM 4 # endif # ifdef __mips__ # define LG_QUANTUM 3 # endif # ifdef __or1k__ # define LG_QUANTUM 3 # endif # ifdef __powerpc__ # define LG_QUANTUM 4 # endif # ifdef __s390__ # define LG_QUANTUM 4 # endif # ifdef __SH4__ # define LG_QUANTUM 4 # endif # ifdef __tile__ # define LG_QUANTUM 4 # endif # ifdef __le32__ # define LG_QUANTUM 4 # endif # ifndef LG_QUANTUM # error "Unknown minimum alignment for architecture; specify via " "--with-lg-quantum" # endif #endif #define QUANTUM ((size_t)(1U << LG_QUANTUM)) #define QUANTUM_MASK (QUANTUM - 1) /* Return the smallest quantum multiple that is >= a. */ #define QUANTUM_CEILING(a) \ (((a) + QUANTUM_MASK) & ~QUANTUM_MASK) #define LONG ((size_t)(1U << LG_SIZEOF_LONG)) #define LONG_MASK (LONG - 1) /* Return the smallest long multiple that is >= a. */ #define LONG_CEILING(a) \ (((a) + LONG_MASK) & ~LONG_MASK) #define SIZEOF_PTR (1U << LG_SIZEOF_PTR) #define PTR_MASK (SIZEOF_PTR - 1) /* Return the smallest (void *) multiple that is >= a. */ #define PTR_CEILING(a) \ (((a) + PTR_MASK) & ~PTR_MASK) /* * Maximum size of L1 cache line. This is used to avoid cache line aliasing. * In addition, this controls the spacing of cacheline-spaced size classes. * * CACHELINE cannot be based on LG_CACHELINE because __declspec(align()) can * only handle raw constants. */ #define LG_CACHELINE 6 #define CACHELINE 64 #define CACHELINE_MASK (CACHELINE - 1) /* Return the smallest cacheline multiple that is >= s. */ #define CACHELINE_CEILING(s) \ (((s) + CACHELINE_MASK) & ~CACHELINE_MASK) /* Page size. LG_PAGE is determined by the configure script. */ #ifdef PAGE_MASK # undef PAGE_MASK #endif #define PAGE ((size_t)(1U << LG_PAGE)) #define PAGE_MASK ((size_t)(PAGE - 1)) +/* Return the page base address for the page containing address a. */ +#define PAGE_ADDR2BASE(a) \ + ((void *)((uintptr_t)(a) & ~PAGE_MASK)) + /* Return the smallest pagesize multiple that is >= s. */ #define PAGE_CEILING(s) \ (((s) + PAGE_MASK) & ~PAGE_MASK) /* Return the nearest aligned address at or below a. */ #define ALIGNMENT_ADDR2BASE(a, alignment) \ ((void *)((uintptr_t)(a) & (-(alignment)))) /* Return the offset between a and the nearest aligned address at or below a. */ #define ALIGNMENT_ADDR2OFFSET(a, alignment) \ ((size_t)((uintptr_t)(a) & (alignment - 1))) /* Return the smallest alignment multiple that is >= s. */ #define ALIGNMENT_CEILING(s, alignment) \ (((s) + (alignment - 1)) & (-(alignment))) /* Declare a variable-length array. */ #if __STDC_VERSION__ < 199901L # ifdef _MSC_VER # include # define alloca _alloca # else # ifdef JEMALLOC_HAS_ALLOCA_H # include # else # include # endif # endif # define VARIABLE_ARRAY(type, name, count) \ type *name = alloca(sizeof(type) * (count)) #else # define VARIABLE_ARRAY(type, name, count) type name[(count)] #endif #include "jemalloc/internal/valgrind.h" #include "jemalloc/internal/util.h" #include "jemalloc/internal/atomic.h" #include "jemalloc/internal/prng.h" #include "jemalloc/internal/ckh.h" #include "jemalloc/internal/size_classes.h" #include "jemalloc/internal/stats.h" #include "jemalloc/internal/ctl.h" #include "jemalloc/internal/mutex.h" #include "jemalloc/internal/tsd.h" #include "jemalloc/internal/mb.h" #include "jemalloc/internal/extent.h" #include "jemalloc/internal/arena.h" #include "jemalloc/internal/bitmap.h" #include "jemalloc/internal/base.h" #include "jemalloc/internal/rtree.h" #include "jemalloc/internal/pages.h" #include "jemalloc/internal/chunk.h" #include "jemalloc/internal/huge.h" #include "jemalloc/internal/tcache.h" #include "jemalloc/internal/hash.h" #include "jemalloc/internal/quarantine.h" #include "jemalloc/internal/prof.h" #undef JEMALLOC_H_TYPES /******************************************************************************/ #define JEMALLOC_H_STRUCTS #include "jemalloc/internal/valgrind.h" #include "jemalloc/internal/util.h" #include "jemalloc/internal/atomic.h" #include "jemalloc/internal/prng.h" #include "jemalloc/internal/ckh.h" #include "jemalloc/internal/size_classes.h" #include "jemalloc/internal/stats.h" #include "jemalloc/internal/ctl.h" #include "jemalloc/internal/mutex.h" #include "jemalloc/internal/mb.h" #include "jemalloc/internal/bitmap.h" #define JEMALLOC_ARENA_STRUCTS_A #include "jemalloc/internal/arena.h" #undef JEMALLOC_ARENA_STRUCTS_A #include "jemalloc/internal/extent.h" #define JEMALLOC_ARENA_STRUCTS_B #include "jemalloc/internal/arena.h" #undef JEMALLOC_ARENA_STRUCTS_B #include "jemalloc/internal/base.h" #include "jemalloc/internal/rtree.h" #include "jemalloc/internal/pages.h" #include "jemalloc/internal/chunk.h" #include "jemalloc/internal/huge.h" #include "jemalloc/internal/tcache.h" #include "jemalloc/internal/hash.h" #include "jemalloc/internal/quarantine.h" #include "jemalloc/internal/prof.h" #include "jemalloc/internal/tsd.h" #undef JEMALLOC_H_STRUCTS /******************************************************************************/ #define JEMALLOC_H_EXTERNS extern bool opt_abort; extern const char *opt_junk; extern bool opt_junk_alloc; extern bool opt_junk_free; extern size_t opt_quarantine; extern bool opt_redzone; extern bool opt_utrace; extern bool opt_xmalloc; extern bool opt_zero; extern size_t opt_narenas; extern bool in_valgrind; /* Number of CPUs. */ extern unsigned ncpus; /* * index2size_tab encodes the same information as could be computed (at * unacceptable cost in some code paths) by index2size_compute(). */ extern size_t const index2size_tab[NSIZES]; /* * size2index_tab is a compact lookup table that rounds request sizes up to * size classes. In order to reduce cache footprint, the table is compressed, * and all accesses are via size2index(). */ extern uint8_t const size2index_tab[]; arena_t *a0get(void); void *a0malloc(size_t size); void a0dalloc(void *ptr); void *bootstrap_malloc(size_t size); void *bootstrap_calloc(size_t num, size_t size); void bootstrap_free(void *ptr); arena_t *arenas_extend(unsigned ind); arena_t *arena_init(unsigned ind); unsigned narenas_total_get(void); arena_t *arena_get_hard(tsd_t *tsd, unsigned ind, bool init_if_missing); arena_t *arena_choose_hard(tsd_t *tsd); void arena_migrate(tsd_t *tsd, unsigned oldind, unsigned newind); unsigned arena_nbound(unsigned ind); void thread_allocated_cleanup(tsd_t *tsd); void thread_deallocated_cleanup(tsd_t *tsd); void arena_cleanup(tsd_t *tsd); void arenas_cache_cleanup(tsd_t *tsd); void narenas_cache_cleanup(tsd_t *tsd); void arenas_cache_bypass_cleanup(tsd_t *tsd); void jemalloc_prefork(void); void jemalloc_postfork_parent(void); void jemalloc_postfork_child(void); #include "jemalloc/internal/valgrind.h" #include "jemalloc/internal/util.h" #include "jemalloc/internal/atomic.h" #include "jemalloc/internal/prng.h" #include "jemalloc/internal/ckh.h" #include "jemalloc/internal/size_classes.h" #include "jemalloc/internal/stats.h" #include "jemalloc/internal/ctl.h" #include "jemalloc/internal/mutex.h" #include "jemalloc/internal/mb.h" #include "jemalloc/internal/bitmap.h" #include "jemalloc/internal/extent.h" #include "jemalloc/internal/arena.h" #include "jemalloc/internal/base.h" #include "jemalloc/internal/rtree.h" #include "jemalloc/internal/pages.h" #include "jemalloc/internal/chunk.h" #include "jemalloc/internal/huge.h" #include "jemalloc/internal/tcache.h" #include "jemalloc/internal/hash.h" #include "jemalloc/internal/quarantine.h" #include "jemalloc/internal/prof.h" #include "jemalloc/internal/tsd.h" #undef JEMALLOC_H_EXTERNS /******************************************************************************/ #define JEMALLOC_H_INLINES #include "jemalloc/internal/valgrind.h" #include "jemalloc/internal/util.h" #include "jemalloc/internal/atomic.h" #include "jemalloc/internal/prng.h" #include "jemalloc/internal/ckh.h" #include "jemalloc/internal/size_classes.h" #include "jemalloc/internal/stats.h" #include "jemalloc/internal/ctl.h" #include "jemalloc/internal/mutex.h" #include "jemalloc/internal/tsd.h" #include "jemalloc/internal/mb.h" #include "jemalloc/internal/extent.h" #include "jemalloc/internal/base.h" #include "jemalloc/internal/rtree.h" #include "jemalloc/internal/pages.h" #include "jemalloc/internal/chunk.h" #include "jemalloc/internal/huge.h" #ifndef JEMALLOC_ENABLE_INLINE szind_t size2index_compute(size_t size); szind_t size2index_lookup(size_t size); szind_t size2index(size_t size); size_t index2size_compute(szind_t index); size_t index2size_lookup(szind_t index); size_t index2size(szind_t index); size_t s2u_compute(size_t size); size_t s2u_lookup(size_t size); size_t s2u(size_t size); size_t sa2u(size_t size, size_t alignment); arena_t *arena_choose(tsd_t *tsd, arena_t *arena); arena_t *arena_get(tsd_t *tsd, unsigned ind, bool init_if_missing, bool refresh_if_missing); #endif #if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_C_)) JEMALLOC_INLINE szind_t size2index_compute(size_t size) { #if (NTBINS != 0) if (size <= (ZU(1) << LG_TINY_MAXCLASS)) { size_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1; size_t lg_ceil = lg_floor(pow2_ceil(size)); return (lg_ceil < lg_tmin ? 0 : lg_ceil - lg_tmin); } #endif { size_t x = unlikely(ZI(size) < 0) ? ((size<<1) ? (ZU(1)<<(LG_SIZEOF_PTR+3)) : ((ZU(1)<<(LG_SIZEOF_PTR+3))-1)) : lg_floor((size<<1)-1); size_t shift = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM) ? 0 : x - (LG_SIZE_CLASS_GROUP + LG_QUANTUM); size_t grp = shift << LG_SIZE_CLASS_GROUP; size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1) ? LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1; size_t delta_inverse_mask = ZI(-1) << lg_delta; size_t mod = ((((size-1) & delta_inverse_mask) >> lg_delta)) & ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1); size_t index = NTBINS + grp + mod; return (index); } } JEMALLOC_ALWAYS_INLINE szind_t size2index_lookup(size_t size) { assert(size <= LOOKUP_MAXCLASS); { size_t ret = ((size_t)(size2index_tab[(size-1) >> LG_TINY_MIN])); assert(ret == size2index_compute(size)); return (ret); } } JEMALLOC_ALWAYS_INLINE szind_t size2index(size_t size) { assert(size > 0); if (likely(size <= LOOKUP_MAXCLASS)) return (size2index_lookup(size)); return (size2index_compute(size)); } JEMALLOC_INLINE size_t index2size_compute(szind_t index) { #if (NTBINS > 0) if (index < NTBINS) return (ZU(1) << (LG_TINY_MAXCLASS - NTBINS + 1 + index)); #endif { size_t reduced_index = index - NTBINS; size_t grp = reduced_index >> LG_SIZE_CLASS_GROUP; size_t mod = reduced_index & ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1); size_t grp_size_mask = ~((!!grp)-1); size_t grp_size = ((ZU(1) << (LG_QUANTUM + (LG_SIZE_CLASS_GROUP-1))) << grp) & grp_size_mask; size_t shift = (grp == 0) ? 1 : grp; size_t lg_delta = shift + (LG_QUANTUM-1); size_t mod_size = (mod+1) << lg_delta; size_t usize = grp_size + mod_size; return (usize); } } JEMALLOC_ALWAYS_INLINE size_t index2size_lookup(szind_t index) { size_t ret = (size_t)index2size_tab[index]; assert(ret == index2size_compute(index)); return (ret); } JEMALLOC_ALWAYS_INLINE size_t index2size(szind_t index) { assert(index < NSIZES); return (index2size_lookup(index)); } JEMALLOC_ALWAYS_INLINE size_t s2u_compute(size_t size) { #if (NTBINS > 0) if (size <= (ZU(1) << LG_TINY_MAXCLASS)) { size_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1; size_t lg_ceil = lg_floor(pow2_ceil(size)); return (lg_ceil < lg_tmin ? (ZU(1) << lg_tmin) : (ZU(1) << lg_ceil)); } #endif { size_t x = unlikely(ZI(size) < 0) ? ((size<<1) ? (ZU(1)<<(LG_SIZEOF_PTR+3)) : ((ZU(1)<<(LG_SIZEOF_PTR+3))-1)) : lg_floor((size<<1)-1); size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1) ? LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1; size_t delta = ZU(1) << lg_delta; size_t delta_mask = delta - 1; size_t usize = (size + delta_mask) & ~delta_mask; return (usize); } } JEMALLOC_ALWAYS_INLINE size_t s2u_lookup(size_t size) { size_t ret = index2size_lookup(size2index_lookup(size)); assert(ret == s2u_compute(size)); return (ret); } /* * Compute usable size that would result from allocating an object with the * specified size. */ JEMALLOC_ALWAYS_INLINE size_t s2u(size_t size) { assert(size > 0); if (likely(size <= LOOKUP_MAXCLASS)) return (s2u_lookup(size)); return (s2u_compute(size)); } /* * Compute usable size that would result from allocating an object with the * specified size and alignment. */ JEMALLOC_ALWAYS_INLINE size_t sa2u(size_t size, size_t alignment) { size_t usize; assert(alignment != 0 && ((alignment - 1) & alignment) == 0); /* Try for a small size class. */ if (size <= SMALL_MAXCLASS && alignment < PAGE) { /* * Round size up to the nearest multiple of alignment. * * This done, we can take advantage of the fact that for each * small size class, every object is aligned at the smallest * power of two that is non-zero in the base two representation * of the size. For example: * * Size | Base 2 | Minimum alignment * -----+----------+------------------ * 96 | 1100000 | 32 * 144 | 10100000 | 32 * 192 | 11000000 | 64 */ usize = s2u(ALIGNMENT_CEILING(size, alignment)); if (usize < LARGE_MINCLASS) return (usize); } /* Try for a large size class. */ if (likely(size <= large_maxclass) && likely(alignment < chunksize)) { /* * We can't achieve subpage alignment, so round up alignment * to the minimum that can actually be supported. */ alignment = PAGE_CEILING(alignment); /* Make sure result is a large size class. */ usize = (size <= LARGE_MINCLASS) ? LARGE_MINCLASS : s2u(size); /* * Calculate the size of the over-size run that arena_palloc() * would need to allocate in order to guarantee the alignment. */ if (usize + large_pad + alignment - PAGE <= arena_maxrun) return (usize); } /* Huge size class. Beware of size_t overflow. */ /* * We can't achieve subchunk alignment, so round up alignment to the * minimum that can actually be supported. */ alignment = CHUNK_CEILING(alignment); if (alignment == 0) { /* size_t overflow. */ return (0); } /* Make sure result is a huge size class. */ if (size <= chunksize) usize = chunksize; else { usize = s2u(size); if (usize < size) { /* size_t overflow. */ return (0); } } /* * Calculate the multi-chunk mapping that huge_palloc() would need in * order to guarantee the alignment. */ if (usize + alignment - PAGE < usize) { /* size_t overflow. */ return (0); } return (usize); } /* Choose an arena based on a per-thread value. */ JEMALLOC_INLINE arena_t * arena_choose(tsd_t *tsd, arena_t *arena) { arena_t *ret; if (arena != NULL) return (arena); if (unlikely((ret = tsd_arena_get(tsd)) == NULL)) ret = arena_choose_hard(tsd); return (ret); } JEMALLOC_INLINE arena_t * arena_get(tsd_t *tsd, unsigned ind, bool init_if_missing, bool refresh_if_missing) { arena_t *arena; arena_t **arenas_cache = tsd_arenas_cache_get(tsd); /* init_if_missing requires refresh_if_missing. */ assert(!init_if_missing || refresh_if_missing); if (unlikely(arenas_cache == NULL)) { /* arenas_cache hasn't been initialized yet. */ return (arena_get_hard(tsd, ind, init_if_missing)); } if (unlikely(ind >= tsd_narenas_cache_get(tsd))) { /* * ind is invalid, cache is old (too small), or arena to be * initialized. */ return (refresh_if_missing ? arena_get_hard(tsd, ind, init_if_missing) : NULL); } arena = arenas_cache[ind]; if (likely(arena != NULL) || !refresh_if_missing) return (arena); return (arena_get_hard(tsd, ind, init_if_missing)); } #endif #include "jemalloc/internal/bitmap.h" /* * Include portions of arena.h interleaved with tcache.h in order to resolve * circular dependencies. */ #define JEMALLOC_ARENA_INLINE_A #include "jemalloc/internal/arena.h" #undef JEMALLOC_ARENA_INLINE_A #include "jemalloc/internal/tcache.h" #define JEMALLOC_ARENA_INLINE_B #include "jemalloc/internal/arena.h" #undef JEMALLOC_ARENA_INLINE_B #include "jemalloc/internal/hash.h" #include "jemalloc/internal/quarantine.h" #ifndef JEMALLOC_ENABLE_INLINE arena_t *iaalloc(const void *ptr); size_t isalloc(const void *ptr, bool demote); void *iallocztm(tsd_t *tsd, size_t size, bool zero, tcache_t *tcache, bool is_metadata, arena_t *arena); void *imalloct(tsd_t *tsd, size_t size, tcache_t *tcache, arena_t *arena); void *imalloc(tsd_t *tsd, size_t size); void *icalloct(tsd_t *tsd, size_t size, tcache_t *tcache, arena_t *arena); void *icalloc(tsd_t *tsd, size_t size); void *ipallocztm(tsd_t *tsd, size_t usize, size_t alignment, bool zero, tcache_t *tcache, bool is_metadata, arena_t *arena); void *ipalloct(tsd_t *tsd, size_t usize, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena); void *ipalloc(tsd_t *tsd, size_t usize, size_t alignment, bool zero); size_t ivsalloc(const void *ptr, bool demote); size_t u2rz(size_t usize); size_t p2rz(const void *ptr); void idalloctm(tsd_t *tsd, void *ptr, tcache_t *tcache, bool is_metadata); void idalloct(tsd_t *tsd, void *ptr, tcache_t *tcache); void idalloc(tsd_t *tsd, void *ptr); void iqalloc(tsd_t *tsd, void *ptr, tcache_t *tcache); void isdalloct(tsd_t *tsd, void *ptr, size_t size, tcache_t *tcache); void isqalloc(tsd_t *tsd, void *ptr, size_t size, tcache_t *tcache); void *iralloct_realign(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t extra, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena); void *iralloct(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena); void *iralloc(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t alignment, bool zero); bool ixalloc(void *ptr, size_t oldsize, size_t size, size_t extra, size_t alignment, bool zero); #endif #if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_C_)) JEMALLOC_ALWAYS_INLINE arena_t * iaalloc(const void *ptr) { assert(ptr != NULL); return (arena_aalloc(ptr)); } /* * Typical usage: * void *ptr = [...] * size_t sz = isalloc(ptr, config_prof); */ JEMALLOC_ALWAYS_INLINE size_t isalloc(const void *ptr, bool demote) { assert(ptr != NULL); /* Demotion only makes sense if config_prof is true. */ assert(config_prof || !demote); return (arena_salloc(ptr, demote)); } JEMALLOC_ALWAYS_INLINE void * iallocztm(tsd_t *tsd, size_t size, bool zero, tcache_t *tcache, bool is_metadata, arena_t *arena) { void *ret; assert(size != 0); ret = arena_malloc(tsd, arena, size, zero, tcache); if (config_stats && is_metadata && likely(ret != NULL)) { arena_metadata_allocated_add(iaalloc(ret), isalloc(ret, config_prof)); } return (ret); } JEMALLOC_ALWAYS_INLINE void * imalloct(tsd_t *tsd, size_t size, tcache_t *tcache, arena_t *arena) { return (iallocztm(tsd, size, false, tcache, false, arena)); } JEMALLOC_ALWAYS_INLINE void * imalloc(tsd_t *tsd, size_t size) { return (iallocztm(tsd, size, false, tcache_get(tsd, true), false, NULL)); } JEMALLOC_ALWAYS_INLINE void * icalloct(tsd_t *tsd, size_t size, tcache_t *tcache, arena_t *arena) { return (iallocztm(tsd, size, true, tcache, false, arena)); } JEMALLOC_ALWAYS_INLINE void * icalloc(tsd_t *tsd, size_t size) { return (iallocztm(tsd, size, true, tcache_get(tsd, true), false, NULL)); } JEMALLOC_ALWAYS_INLINE void * ipallocztm(tsd_t *tsd, size_t usize, size_t alignment, bool zero, tcache_t *tcache, bool is_metadata, arena_t *arena) { void *ret; assert(usize != 0); assert(usize == sa2u(usize, alignment)); ret = arena_palloc(tsd, arena, usize, alignment, zero, tcache); assert(ALIGNMENT_ADDR2BASE(ret, alignment) == ret); if (config_stats && is_metadata && likely(ret != NULL)) { arena_metadata_allocated_add(iaalloc(ret), isalloc(ret, config_prof)); } return (ret); } JEMALLOC_ALWAYS_INLINE void * ipalloct(tsd_t *tsd, size_t usize, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena) { return (ipallocztm(tsd, usize, alignment, zero, tcache, false, arena)); } JEMALLOC_ALWAYS_INLINE void * ipalloc(tsd_t *tsd, size_t usize, size_t alignment, bool zero) { return (ipallocztm(tsd, usize, alignment, zero, tcache_get(tsd, NULL), false, NULL)); } JEMALLOC_ALWAYS_INLINE size_t ivsalloc(const void *ptr, bool demote) { extent_node_t *node; /* Return 0 if ptr is not within a chunk managed by jemalloc. */ node = chunk_lookup(ptr, false); if (node == NULL) return (0); /* Only arena chunks should be looked up via interior pointers. */ assert(extent_node_addr_get(node) == ptr || extent_node_achunk_get(node)); return (isalloc(ptr, demote)); } JEMALLOC_INLINE size_t u2rz(size_t usize) { size_t ret; if (usize <= SMALL_MAXCLASS) { szind_t binind = size2index(usize); ret = arena_bin_info[binind].redzone_size; } else ret = 0; return (ret); } JEMALLOC_INLINE size_t p2rz(const void *ptr) { size_t usize = isalloc(ptr, false); return (u2rz(usize)); } JEMALLOC_ALWAYS_INLINE void idalloctm(tsd_t *tsd, void *ptr, tcache_t *tcache, bool is_metadata) { assert(ptr != NULL); if (config_stats && is_metadata) { arena_metadata_allocated_sub(iaalloc(ptr), isalloc(ptr, config_prof)); } arena_dalloc(tsd, ptr, tcache); } JEMALLOC_ALWAYS_INLINE void idalloct(tsd_t *tsd, void *ptr, tcache_t *tcache) { idalloctm(tsd, ptr, tcache, false); } JEMALLOC_ALWAYS_INLINE void idalloc(tsd_t *tsd, void *ptr) { idalloctm(tsd, ptr, tcache_get(tsd, false), false); } JEMALLOC_ALWAYS_INLINE void iqalloc(tsd_t *tsd, void *ptr, tcache_t *tcache) { if (config_fill && unlikely(opt_quarantine)) quarantine(tsd, ptr); else idalloctm(tsd, ptr, tcache, false); } JEMALLOC_ALWAYS_INLINE void isdalloct(tsd_t *tsd, void *ptr, size_t size, tcache_t *tcache) { arena_sdalloc(tsd, ptr, size, tcache); } JEMALLOC_ALWAYS_INLINE void isqalloc(tsd_t *tsd, void *ptr, size_t size, tcache_t *tcache) { if (config_fill && unlikely(opt_quarantine)) quarantine(tsd, ptr); else isdalloct(tsd, ptr, size, tcache); } JEMALLOC_ALWAYS_INLINE void * iralloct_realign(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t extra, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena) { void *p; size_t usize, copysize; usize = sa2u(size + extra, alignment); if (usize == 0) return (NULL); p = ipalloct(tsd, usize, alignment, zero, tcache, arena); if (p == NULL) { if (extra == 0) return (NULL); /* Try again, without extra this time. */ usize = sa2u(size, alignment); if (usize == 0) return (NULL); p = ipalloct(tsd, usize, alignment, zero, tcache, arena); if (p == NULL) return (NULL); } /* * Copy at most size bytes (not size+extra), since the caller has no * expectation that the extra bytes will be reliably preserved. */ copysize = (size < oldsize) ? size : oldsize; memcpy(p, ptr, copysize); isqalloc(tsd, ptr, oldsize, tcache); return (p); } JEMALLOC_ALWAYS_INLINE void * iralloct(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena) { assert(ptr != NULL); assert(size != 0); if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1)) != 0) { /* * Existing object alignment is inadequate; allocate new space * and copy. */ return (iralloct_realign(tsd, ptr, oldsize, size, 0, alignment, zero, tcache, arena)); } return (arena_ralloc(tsd, arena, ptr, oldsize, size, alignment, zero, tcache)); } JEMALLOC_ALWAYS_INLINE void * iralloc(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t alignment, bool zero) { return (iralloct(tsd, ptr, oldsize, size, alignment, zero, tcache_get(tsd, true), NULL)); } JEMALLOC_ALWAYS_INLINE bool ixalloc(void *ptr, size_t oldsize, size_t size, size_t extra, size_t alignment, bool zero) { assert(ptr != NULL); assert(size != 0); if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1)) != 0) { /* Existing object alignment is inadequate. */ return (true); } return (arena_ralloc_no_move(ptr, oldsize, size, extra, zero)); } #endif #include "jemalloc/internal/prof.h" #undef JEMALLOC_H_INLINES /******************************************************************************/ #endif /* JEMALLOC_INTERNAL_H */ Index: head/contrib/jemalloc/include/jemalloc/jemalloc.h =================================================================== --- head/contrib/jemalloc/include/jemalloc/jemalloc.h (revision 289899) +++ head/contrib/jemalloc/include/jemalloc/jemalloc.h (revision 289900) @@ -1,372 +1,372 @@ #ifndef JEMALLOC_H_ #define JEMALLOC_H_ #ifdef __cplusplus extern "C" { #endif /* Defined if __attribute__((...)) syntax is supported. */ #define JEMALLOC_HAVE_ATTR /* Defined if alloc_size attribute is supported. */ /* #undef JEMALLOC_HAVE_ATTR_ALLOC_SIZE */ /* Defined if format(gnu_printf, ...) attribute is supported. */ /* #undef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF */ /* Defined if format(printf, ...) attribute is supported. */ #define JEMALLOC_HAVE_ATTR_FORMAT_PRINTF /* * Define overrides for non-standard allocator-related functions if they are * present on the system. */ /* #undef JEMALLOC_OVERRIDE_MEMALIGN */ #define JEMALLOC_OVERRIDE_VALLOC /* * At least Linux omits the "const" in: * * size_t malloc_usable_size(const void *ptr); * * Match the operating system's prototype. */ #define JEMALLOC_USABLE_SIZE_CONST const /* * If defined, specify throw() for the public function prototypes when compiling * with C++. The only justification for this is to match the prototypes that * glibc defines. */ /* #undef JEMALLOC_USE_CXX_THROW */ /* sizeof(void *) == 2^LG_SIZEOF_PTR. */ #define LG_SIZEOF_PTR 3 /* * Name mangling for public symbols is controlled by --with-mangling and * --with-jemalloc-prefix. With default settings the je_ prefix is stripped by * these macro definitions. */ #ifndef JEMALLOC_NO_RENAME # define je_malloc_conf malloc_conf # define je_malloc_message malloc_message # define je_malloc malloc # define je_calloc calloc # define je_posix_memalign posix_memalign # define je_aligned_alloc aligned_alloc # define je_realloc realloc # define je_free free # define je_mallocx mallocx # define je_rallocx rallocx # define je_xallocx xallocx # define je_sallocx sallocx # define je_dallocx dallocx # define je_sdallocx sdallocx # define je_nallocx nallocx # define je_mallctl mallctl # define je_mallctlnametomib mallctlnametomib # define je_mallctlbymib mallctlbymib # define je_malloc_stats_print malloc_stats_print # define je_malloc_usable_size malloc_usable_size # define je_valloc valloc #endif #include "jemalloc_FreeBSD.h" #include #include #include #include #include -#define JEMALLOC_VERSION "4.0.2-0-g486d249fb4715fd3de679b6c2a04f7e657883111" +#define JEMALLOC_VERSION "4.0.4-0-g91010a9e2ebfc84b1ac1ed7fdde3bfed4f65f180" #define JEMALLOC_VERSION_MAJOR 4 #define JEMALLOC_VERSION_MINOR 0 -#define JEMALLOC_VERSION_BUGFIX 2 +#define JEMALLOC_VERSION_BUGFIX 4 #define JEMALLOC_VERSION_NREV 0 -#define JEMALLOC_VERSION_GID "486d249fb4715fd3de679b6c2a04f7e657883111" +#define JEMALLOC_VERSION_GID "91010a9e2ebfc84b1ac1ed7fdde3bfed4f65f180" # define MALLOCX_LG_ALIGN(la) (la) # if LG_SIZEOF_PTR == 2 # define MALLOCX_ALIGN(a) (ffs(a)-1) # else # define MALLOCX_ALIGN(a) \ ((a < (size_t)INT_MAX) ? ffs(a)-1 : ffs(a>>32)+31) # endif # define MALLOCX_ZERO ((int)0x40) /* * Bias tcache index bits so that 0 encodes "automatic tcache management", and 1 * encodes MALLOCX_TCACHE_NONE. */ # define MALLOCX_TCACHE(tc) ((int)(((tc)+2) << 8)) # define MALLOCX_TCACHE_NONE MALLOCX_TCACHE(-1) /* * Bias arena index bits so that 0 encodes "use an automatically chosen arena". */ # define MALLOCX_ARENA(a) ((int)(((a)+1) << 20)) #if defined(__cplusplus) && defined(JEMALLOC_USE_CXX_THROW) # define JEMALLOC_CXX_THROW throw() #else # define JEMALLOC_CXX_THROW #endif #ifdef JEMALLOC_HAVE_ATTR # define JEMALLOC_ATTR(s) __attribute__((s)) # define JEMALLOC_ALIGNED(s) JEMALLOC_ATTR(aligned(s)) # ifdef JEMALLOC_HAVE_ATTR_ALLOC_SIZE # define JEMALLOC_ALLOC_SIZE(s) JEMALLOC_ATTR(alloc_size(s)) # define JEMALLOC_ALLOC_SIZE2(s1, s2) JEMALLOC_ATTR(alloc_size(s1, s2)) # else # define JEMALLOC_ALLOC_SIZE(s) # define JEMALLOC_ALLOC_SIZE2(s1, s2) # endif # ifndef JEMALLOC_EXPORT # define JEMALLOC_EXPORT JEMALLOC_ATTR(visibility("default")) # endif # ifdef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF # define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(gnu_printf, s, i)) # elif defined(JEMALLOC_HAVE_ATTR_FORMAT_PRINTF) # define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(printf, s, i)) # else # define JEMALLOC_FORMAT_PRINTF(s, i) # endif # define JEMALLOC_NOINLINE JEMALLOC_ATTR(noinline) # define JEMALLOC_NOTHROW JEMALLOC_ATTR(nothrow) # define JEMALLOC_SECTION(s) JEMALLOC_ATTR(section(s)) # define JEMALLOC_RESTRICT_RETURN # define JEMALLOC_ALLOCATOR #elif _MSC_VER # define JEMALLOC_ATTR(s) # define JEMALLOC_ALIGNED(s) __declspec(align(s)) # define JEMALLOC_ALLOC_SIZE(s) # define JEMALLOC_ALLOC_SIZE2(s1, s2) # ifndef JEMALLOC_EXPORT # ifdef DLLEXPORT # define JEMALLOC_EXPORT __declspec(dllexport) # else # define JEMALLOC_EXPORT __declspec(dllimport) # endif # endif # define JEMALLOC_FORMAT_PRINTF(s, i) # define JEMALLOC_NOINLINE __declspec(noinline) # ifdef __cplusplus # define JEMALLOC_NOTHROW __declspec(nothrow) # else # define JEMALLOC_NOTHROW # endif # define JEMALLOC_SECTION(s) __declspec(allocate(s)) # define JEMALLOC_RESTRICT_RETURN __declspec(restrict) # if _MSC_VER >= 1900 && !defined(__EDG__) # define JEMALLOC_ALLOCATOR __declspec(allocator) # else # define JEMALLOC_ALLOCATOR # endif #else # define JEMALLOC_ATTR(s) # define JEMALLOC_ALIGNED(s) # define JEMALLOC_ALLOC_SIZE(s) # define JEMALLOC_ALLOC_SIZE2(s1, s2) # define JEMALLOC_EXPORT # define JEMALLOC_FORMAT_PRINTF(s, i) # define JEMALLOC_NOINLINE # define JEMALLOC_NOTHROW # define JEMALLOC_SECTION(s) # define JEMALLOC_RESTRICT_RETURN # define JEMALLOC_ALLOCATOR #endif /* * The je_ prefix on the following public symbol declarations is an artifact * of namespace management, and should be omitted in application code unless * JEMALLOC_NO_DEMANGLE is defined (see jemalloc_mangle.h). */ extern JEMALLOC_EXPORT const char *je_malloc_conf; extern JEMALLOC_EXPORT void (*je_malloc_message)(void *cbopaque, const char *s); JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_malloc(size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1); JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_calloc(size_t num, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE2(1, 2); JEMALLOC_EXPORT int JEMALLOC_NOTHROW je_posix_memalign(void **memptr, size_t alignment, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(nonnull(1)); JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_aligned_alloc(size_t alignment, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(2); JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_realloc(void *ptr, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ALLOC_SIZE(2); JEMALLOC_EXPORT void JEMALLOC_NOTHROW je_free(void *ptr) JEMALLOC_CXX_THROW; JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_mallocx(size_t size, int flags) JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1); JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_rallocx(void *ptr, size_t size, int flags) JEMALLOC_ALLOC_SIZE(2); JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW je_xallocx(void *ptr, size_t size, size_t extra, int flags); JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW je_sallocx(const void *ptr, int flags) JEMALLOC_ATTR(pure); JEMALLOC_EXPORT void JEMALLOC_NOTHROW je_dallocx(void *ptr, int flags); JEMALLOC_EXPORT void JEMALLOC_NOTHROW je_sdallocx(void *ptr, size_t size, int flags); JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW je_nallocx(size_t size, int flags) JEMALLOC_ATTR(pure); JEMALLOC_EXPORT int JEMALLOC_NOTHROW je_mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp, size_t newlen); JEMALLOC_EXPORT int JEMALLOC_NOTHROW je_mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp); JEMALLOC_EXPORT int JEMALLOC_NOTHROW je_mallctlbymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen); JEMALLOC_EXPORT void JEMALLOC_NOTHROW je_malloc_stats_print( void (*write_cb)(void *, const char *), void *je_cbopaque, const char *opts); JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW je_malloc_usable_size( JEMALLOC_USABLE_SIZE_CONST void *ptr) JEMALLOC_CXX_THROW; #ifdef JEMALLOC_OVERRIDE_MEMALIGN JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_memalign(size_t alignment, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc); #endif #ifdef JEMALLOC_OVERRIDE_VALLOC JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN void JEMALLOC_NOTHROW *je_valloc(size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc); #endif /* * void * * chunk_alloc(void *new_addr, size_t size, size_t alignment, bool *zero, * bool *commit, unsigned arena_ind); */ typedef void *(chunk_alloc_t)(void *, size_t, size_t, bool *, bool *, unsigned); /* * bool * chunk_dalloc(void *chunk, size_t size, bool committed, unsigned arena_ind); */ typedef bool (chunk_dalloc_t)(void *, size_t, bool, unsigned); /* * bool * chunk_commit(void *chunk, size_t size, size_t offset, size_t length, * unsigned arena_ind); */ typedef bool (chunk_commit_t)(void *, size_t, size_t, size_t, unsigned); /* * bool * chunk_decommit(void *chunk, size_t size, size_t offset, size_t length, * unsigned arena_ind); */ typedef bool (chunk_decommit_t)(void *, size_t, size_t, size_t, unsigned); /* * bool * chunk_purge(void *chunk, size_t size, size_t offset, size_t length, * unsigned arena_ind); */ typedef bool (chunk_purge_t)(void *, size_t, size_t, size_t, unsigned); /* * bool * chunk_split(void *chunk, size_t size, size_t size_a, size_t size_b, * bool committed, unsigned arena_ind); */ typedef bool (chunk_split_t)(void *, size_t, size_t, size_t, bool, unsigned); /* * bool * chunk_merge(void *chunk_a, size_t size_a, void *chunk_b, size_t size_b, * bool committed, unsigned arena_ind); */ typedef bool (chunk_merge_t)(void *, size_t, void *, size_t, bool, unsigned); typedef struct { chunk_alloc_t *alloc; chunk_dalloc_t *dalloc; chunk_commit_t *commit; chunk_decommit_t *decommit; chunk_purge_t *purge; chunk_split_t *split; chunk_merge_t *merge; } chunk_hooks_t; /* * By default application code must explicitly refer to mangled symbol names, * so that it is possible to use jemalloc in conjunction with another allocator * in the same application. Define JEMALLOC_MANGLE in order to cause automatic * name mangling that matches the API prefixing that happened as a result of * --with-mangling and/or --with-jemalloc-prefix configuration settings. */ #ifdef JEMALLOC_MANGLE # ifndef JEMALLOC_NO_DEMANGLE # define JEMALLOC_NO_DEMANGLE # endif # define malloc_conf je_malloc_conf # define malloc_message je_malloc_message # define malloc je_malloc # define calloc je_calloc # define posix_memalign je_posix_memalign # define aligned_alloc je_aligned_alloc # define realloc je_realloc # define free je_free # define mallocx je_mallocx # define rallocx je_rallocx # define xallocx je_xallocx # define sallocx je_sallocx # define dallocx je_dallocx # define sdallocx je_sdallocx # define nallocx je_nallocx # define mallctl je_mallctl # define mallctlnametomib je_mallctlnametomib # define mallctlbymib je_mallctlbymib # define malloc_stats_print je_malloc_stats_print # define malloc_usable_size je_malloc_usable_size # define valloc je_valloc #endif /* * The je_* macros can be used as stable alternative names for the * public jemalloc API if JEMALLOC_NO_DEMANGLE is defined. This is primarily * meant for use in jemalloc itself, but it can be used by application code to * provide isolation from the name mangling specified via --with-mangling * and/or --with-jemalloc-prefix. */ #ifndef JEMALLOC_NO_DEMANGLE # undef je_malloc_conf # undef je_malloc_message # undef je_malloc # undef je_calloc # undef je_posix_memalign # undef je_aligned_alloc # undef je_realloc # undef je_free # undef je_mallocx # undef je_rallocx # undef je_xallocx # undef je_sallocx # undef je_dallocx # undef je_sdallocx # undef je_nallocx # undef je_mallctl # undef je_mallctlnametomib # undef je_mallctlbymib # undef je_malloc_stats_print # undef je_malloc_usable_size # undef je_valloc #endif #ifdef __cplusplus } #endif #endif /* JEMALLOC_H_ */ Index: head/contrib/jemalloc/src/arena.c =================================================================== --- head/contrib/jemalloc/src/arena.c (revision 289899) +++ head/contrib/jemalloc/src/arena.c (revision 289900) @@ -1,3308 +1,3324 @@ #define JEMALLOC_ARENA_C_ #include "jemalloc/internal/jemalloc_internal.h" /******************************************************************************/ /* Data. */ ssize_t opt_lg_dirty_mult = LG_DIRTY_MULT_DEFAULT; static ssize_t lg_dirty_mult_default; arena_bin_info_t arena_bin_info[NBINS]; size_t map_bias; size_t map_misc_offset; size_t arena_maxrun; /* Max run size for arenas. */ size_t large_maxclass; /* Max large size class. */ static size_t small_maxrun; /* Max run size used for small size classes. */ static bool *small_run_tab; /* Valid small run page multiples. */ unsigned nlclasses; /* Number of large size classes. */ unsigned nhclasses; /* Number of huge size classes. */ /******************************************************************************/ /* * Function prototypes for static functions that are referenced prior to * definition. */ static void arena_purge(arena_t *arena, bool all); static void arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned, bool decommitted); static void arena_dalloc_bin_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, arena_bin_t *bin); static void arena_bin_lower_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, arena_bin_t *bin); /******************************************************************************/ #define CHUNK_MAP_KEY ((uintptr_t)0x1U) JEMALLOC_INLINE_C arena_chunk_map_misc_t * arena_miscelm_key_create(size_t size) { return ((arena_chunk_map_misc_t *)(arena_mapbits_size_encode(size) | CHUNK_MAP_KEY)); } JEMALLOC_INLINE_C bool arena_miscelm_is_key(const arena_chunk_map_misc_t *miscelm) { return (((uintptr_t)miscelm & CHUNK_MAP_KEY) != 0); } #undef CHUNK_MAP_KEY JEMALLOC_INLINE_C size_t arena_miscelm_key_size_get(const arena_chunk_map_misc_t *miscelm) { assert(arena_miscelm_is_key(miscelm)); return (arena_mapbits_size_decode((uintptr_t)miscelm)); } JEMALLOC_INLINE_C size_t arena_miscelm_size_get(arena_chunk_map_misc_t *miscelm) { arena_chunk_t *chunk; size_t pageind, mapbits; assert(!arena_miscelm_is_key(miscelm)); chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(miscelm); pageind = arena_miscelm_to_pageind(miscelm); mapbits = arena_mapbits_get(chunk, pageind); return (arena_mapbits_size_decode(mapbits)); } JEMALLOC_INLINE_C int arena_run_comp(arena_chunk_map_misc_t *a, arena_chunk_map_misc_t *b) { uintptr_t a_miscelm = (uintptr_t)a; uintptr_t b_miscelm = (uintptr_t)b; assert(a != NULL); assert(b != NULL); return ((a_miscelm > b_miscelm) - (a_miscelm < b_miscelm)); } /* Generate red-black tree functions. */ rb_gen(static UNUSED, arena_run_tree_, arena_run_tree_t, arena_chunk_map_misc_t, rb_link, arena_run_comp) static size_t run_quantize(size_t size) { size_t qsize; assert(size != 0); assert(size == PAGE_CEILING(size)); /* Don't change sizes that are valid small run sizes. */ if (size <= small_maxrun && small_run_tab[size >> LG_PAGE]) return (size); /* * Round down to the nearest run size that can actually be requested * during normal large allocation. Add large_pad so that cache index * randomization can offset the allocation from the page boundary. */ qsize = index2size(size2index(size - large_pad + 1) - 1) + large_pad; if (qsize <= SMALL_MAXCLASS + large_pad) return (run_quantize(size - large_pad)); assert(qsize <= size); return (qsize); } static size_t run_quantize_next(size_t size) { size_t large_run_size_next; assert(size != 0); assert(size == PAGE_CEILING(size)); /* * Return the next quantized size greater than the input size. * Quantized sizes comprise the union of run sizes that back small * region runs, and run sizes that back large regions with no explicit * alignment constraints. */ if (size > SMALL_MAXCLASS) { large_run_size_next = PAGE_CEILING(index2size(size2index(size - large_pad) + 1) + large_pad); } else large_run_size_next = SIZE_T_MAX; if (size >= small_maxrun) return (large_run_size_next); while (true) { size += PAGE; assert(size <= small_maxrun); if (small_run_tab[size >> LG_PAGE]) { if (large_run_size_next < size) return (large_run_size_next); return (size); } } } static size_t run_quantize_first(size_t size) { size_t qsize = run_quantize(size); if (qsize < size) { /* * Skip a quantization that may have an adequately large run, * because under-sized runs may be mixed in. This only happens * when an unusual size is requested, i.e. for aligned * allocation, and is just one of several places where linear * search would potentially find sufficiently aligned available * memory somewhere lower. */ qsize = run_quantize_next(size); } return (qsize); } JEMALLOC_INLINE_C int arena_avail_comp(arena_chunk_map_misc_t *a, arena_chunk_map_misc_t *b) { int ret; uintptr_t a_miscelm = (uintptr_t)a; size_t a_qsize = run_quantize(arena_miscelm_is_key(a) ? arena_miscelm_key_size_get(a) : arena_miscelm_size_get(a)); size_t b_qsize = run_quantize(arena_miscelm_size_get(b)); /* * Compare based on quantized size rather than size, in order to sort * equally useful runs only by address. */ ret = (a_qsize > b_qsize) - (a_qsize < b_qsize); if (ret == 0) { if (!arena_miscelm_is_key(a)) { uintptr_t b_miscelm = (uintptr_t)b; ret = (a_miscelm > b_miscelm) - (a_miscelm < b_miscelm); } else { /* * Treat keys as if they are lower than anything else. */ ret = -1; } } return (ret); } /* Generate red-black tree functions. */ rb_gen(static UNUSED, arena_avail_tree_, arena_avail_tree_t, arena_chunk_map_misc_t, rb_link, arena_avail_comp) static void arena_avail_insert(arena_t *arena, arena_chunk_t *chunk, size_t pageind, size_t npages) { assert(npages == (arena_mapbits_unallocated_size_get(chunk, pageind) >> LG_PAGE)); arena_avail_tree_insert(&arena->runs_avail, arena_miscelm_get(chunk, pageind)); } static void arena_avail_remove(arena_t *arena, arena_chunk_t *chunk, size_t pageind, size_t npages) { assert(npages == (arena_mapbits_unallocated_size_get(chunk, pageind) >> LG_PAGE)); arena_avail_tree_remove(&arena->runs_avail, arena_miscelm_get(chunk, pageind)); } static void arena_run_dirty_insert(arena_t *arena, arena_chunk_t *chunk, size_t pageind, size_t npages) { arena_chunk_map_misc_t *miscelm = arena_miscelm_get(chunk, pageind); assert(npages == (arena_mapbits_unallocated_size_get(chunk, pageind) >> LG_PAGE)); assert(arena_mapbits_dirty_get(chunk, pageind) == CHUNK_MAP_DIRTY); assert(arena_mapbits_dirty_get(chunk, pageind+npages-1) == CHUNK_MAP_DIRTY); qr_new(&miscelm->rd, rd_link); qr_meld(&arena->runs_dirty, &miscelm->rd, rd_link); arena->ndirty += npages; } static void arena_run_dirty_remove(arena_t *arena, arena_chunk_t *chunk, size_t pageind, size_t npages) { arena_chunk_map_misc_t *miscelm = arena_miscelm_get(chunk, pageind); assert(npages == (arena_mapbits_unallocated_size_get(chunk, pageind) >> LG_PAGE)); assert(arena_mapbits_dirty_get(chunk, pageind) == CHUNK_MAP_DIRTY); assert(arena_mapbits_dirty_get(chunk, pageind+npages-1) == CHUNK_MAP_DIRTY); qr_remove(&miscelm->rd, rd_link); assert(arena->ndirty >= npages); arena->ndirty -= npages; } static size_t arena_chunk_dirty_npages(const extent_node_t *node) { return (extent_node_size_get(node) >> LG_PAGE); } void arena_chunk_cache_maybe_insert(arena_t *arena, extent_node_t *node, bool cache) { if (cache) { extent_node_dirty_linkage_init(node); extent_node_dirty_insert(node, &arena->runs_dirty, &arena->chunks_cache); arena->ndirty += arena_chunk_dirty_npages(node); } } void arena_chunk_cache_maybe_remove(arena_t *arena, extent_node_t *node, bool dirty) { if (dirty) { extent_node_dirty_remove(node); assert(arena->ndirty >= arena_chunk_dirty_npages(node)); arena->ndirty -= arena_chunk_dirty_npages(node); } } JEMALLOC_INLINE_C void * arena_run_reg_alloc(arena_run_t *run, arena_bin_info_t *bin_info) { void *ret; unsigned regind; arena_chunk_map_misc_t *miscelm; void *rpages; assert(run->nfree > 0); assert(!bitmap_full(run->bitmap, &bin_info->bitmap_info)); regind = bitmap_sfu(run->bitmap, &bin_info->bitmap_info); miscelm = arena_run_to_miscelm(run); rpages = arena_miscelm_to_rpages(miscelm); ret = (void *)((uintptr_t)rpages + (uintptr_t)bin_info->reg0_offset + (uintptr_t)(bin_info->reg_interval * regind)); run->nfree--; return (ret); } JEMALLOC_INLINE_C void arena_run_reg_dalloc(arena_run_t *run, void *ptr) { arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE; size_t mapbits = arena_mapbits_get(chunk, pageind); szind_t binind = arena_ptr_small_binind_get(ptr, mapbits); arena_bin_info_t *bin_info = &arena_bin_info[binind]; unsigned regind = arena_run_regind(run, bin_info, ptr); assert(run->nfree < bin_info->nregs); /* Freeing an interior pointer can cause assertion failure. */ assert(((uintptr_t)ptr - ((uintptr_t)arena_miscelm_to_rpages(arena_run_to_miscelm(run)) + (uintptr_t)bin_info->reg0_offset)) % (uintptr_t)bin_info->reg_interval == 0); assert((uintptr_t)ptr >= (uintptr_t)arena_miscelm_to_rpages(arena_run_to_miscelm(run)) + (uintptr_t)bin_info->reg0_offset); /* Freeing an unallocated pointer can cause assertion failure. */ assert(bitmap_get(run->bitmap, &bin_info->bitmap_info, regind)); bitmap_unset(run->bitmap, &bin_info->bitmap_info, regind); run->nfree++; } JEMALLOC_INLINE_C void arena_run_zero(arena_chunk_t *chunk, size_t run_ind, size_t npages) { JEMALLOC_VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), (npages << LG_PAGE)); memset((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), 0, (npages << LG_PAGE)); } JEMALLOC_INLINE_C void arena_run_page_mark_zeroed(arena_chunk_t *chunk, size_t run_ind) { JEMALLOC_VALGRIND_MAKE_MEM_DEFINED((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), PAGE); } JEMALLOC_INLINE_C void arena_run_page_validate_zeroed(arena_chunk_t *chunk, size_t run_ind) { size_t i; UNUSED size_t *p = (size_t *)((uintptr_t)chunk + (run_ind << LG_PAGE)); arena_run_page_mark_zeroed(chunk, run_ind); for (i = 0; i < PAGE / sizeof(size_t); i++) assert(p[i] == 0); } static void arena_cactive_update(arena_t *arena, size_t add_pages, size_t sub_pages) { if (config_stats) { ssize_t cactive_diff = CHUNK_CEILING((arena->nactive + add_pages - sub_pages) << LG_PAGE) - CHUNK_CEILING(arena->nactive << LG_PAGE); if (cactive_diff != 0) stats_cactive_add(cactive_diff); } } static void arena_run_split_remove(arena_t *arena, arena_chunk_t *chunk, size_t run_ind, size_t flag_dirty, size_t flag_decommitted, size_t need_pages) { size_t total_pages, rem_pages; assert(flag_dirty == 0 || flag_decommitted == 0); total_pages = arena_mapbits_unallocated_size_get(chunk, run_ind) >> LG_PAGE; assert(arena_mapbits_dirty_get(chunk, run_ind+total_pages-1) == flag_dirty); assert(need_pages <= total_pages); rem_pages = total_pages - need_pages; arena_avail_remove(arena, chunk, run_ind, total_pages); if (flag_dirty != 0) arena_run_dirty_remove(arena, chunk, run_ind, total_pages); arena_cactive_update(arena, need_pages, 0); arena->nactive += need_pages; /* Keep track of trailing unused pages for later use. */ if (rem_pages > 0) { size_t flags = flag_dirty | flag_decommitted; size_t flag_unzeroed_mask = (flags == 0) ? CHUNK_MAP_UNZEROED : 0; arena_mapbits_unallocated_set(chunk, run_ind+need_pages, (rem_pages << LG_PAGE), flags | (arena_mapbits_unzeroed_get(chunk, run_ind+need_pages) & flag_unzeroed_mask)); arena_mapbits_unallocated_set(chunk, run_ind+total_pages-1, (rem_pages << LG_PAGE), flags | (arena_mapbits_unzeroed_get(chunk, run_ind+total_pages-1) & flag_unzeroed_mask)); if (flag_dirty != 0) { arena_run_dirty_insert(arena, chunk, run_ind+need_pages, rem_pages); } arena_avail_insert(arena, chunk, run_ind+need_pages, rem_pages); } } static bool arena_run_split_large_helper(arena_t *arena, arena_run_t *run, size_t size, bool remove, bool zero) { arena_chunk_t *chunk; arena_chunk_map_misc_t *miscelm; size_t flag_dirty, flag_decommitted, run_ind, need_pages; size_t flag_unzeroed_mask; chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); miscelm = arena_run_to_miscelm(run); run_ind = arena_miscelm_to_pageind(miscelm); flag_dirty = arena_mapbits_dirty_get(chunk, run_ind); flag_decommitted = arena_mapbits_decommitted_get(chunk, run_ind); need_pages = (size >> LG_PAGE); assert(need_pages > 0); if (flag_decommitted != 0 && arena->chunk_hooks.commit(chunk, chunksize, run_ind << LG_PAGE, size, arena->ind)) return (true); if (remove) { arena_run_split_remove(arena, chunk, run_ind, flag_dirty, flag_decommitted, need_pages); } if (zero) { if (flag_decommitted != 0) { /* The run is untouched, and therefore zeroed. */ JEMALLOC_VALGRIND_MAKE_MEM_DEFINED((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), (need_pages << LG_PAGE)); } else if (flag_dirty != 0) { /* The run is dirty, so all pages must be zeroed. */ arena_run_zero(chunk, run_ind, need_pages); } else { /* * The run is clean, so some pages may be zeroed (i.e. * never before touched). */ size_t i; for (i = 0; i < need_pages; i++) { if (arena_mapbits_unzeroed_get(chunk, run_ind+i) != 0) arena_run_zero(chunk, run_ind+i, 1); else if (config_debug) { arena_run_page_validate_zeroed(chunk, run_ind+i); } else { arena_run_page_mark_zeroed(chunk, run_ind+i); } } } } else { JEMALLOC_VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), (need_pages << LG_PAGE)); } /* * Set the last element first, in case the run only contains one page * (i.e. both statements set the same element). */ flag_unzeroed_mask = (flag_dirty | flag_decommitted) == 0 ? CHUNK_MAP_UNZEROED : 0; arena_mapbits_large_set(chunk, run_ind+need_pages-1, 0, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, run_ind+need_pages-1))); arena_mapbits_large_set(chunk, run_ind, size, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, run_ind))); return (false); } static bool arena_run_split_large(arena_t *arena, arena_run_t *run, size_t size, bool zero) { return (arena_run_split_large_helper(arena, run, size, true, zero)); } static bool arena_run_init_large(arena_t *arena, arena_run_t *run, size_t size, bool zero) { return (arena_run_split_large_helper(arena, run, size, false, zero)); } static bool arena_run_split_small(arena_t *arena, arena_run_t *run, size_t size, szind_t binind) { arena_chunk_t *chunk; arena_chunk_map_misc_t *miscelm; size_t flag_dirty, flag_decommitted, run_ind, need_pages, i; assert(binind != BININD_INVALID); chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); miscelm = arena_run_to_miscelm(run); run_ind = arena_miscelm_to_pageind(miscelm); flag_dirty = arena_mapbits_dirty_get(chunk, run_ind); flag_decommitted = arena_mapbits_decommitted_get(chunk, run_ind); need_pages = (size >> LG_PAGE); assert(need_pages > 0); if (flag_decommitted != 0 && arena->chunk_hooks.commit(chunk, chunksize, run_ind << LG_PAGE, size, arena->ind)) return (true); arena_run_split_remove(arena, chunk, run_ind, flag_dirty, flag_decommitted, need_pages); for (i = 0; i < need_pages; i++) { size_t flag_unzeroed = arena_mapbits_unzeroed_get(chunk, run_ind+i); arena_mapbits_small_set(chunk, run_ind+i, i, binind, flag_unzeroed); if (config_debug && flag_dirty == 0 && flag_unzeroed == 0) arena_run_page_validate_zeroed(chunk, run_ind+i); } JEMALLOC_VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), (need_pages << LG_PAGE)); return (false); } static arena_chunk_t * arena_chunk_init_spare(arena_t *arena) { arena_chunk_t *chunk; assert(arena->spare != NULL); chunk = arena->spare; arena->spare = NULL; assert(arena_mapbits_allocated_get(chunk, map_bias) == 0); assert(arena_mapbits_allocated_get(chunk, chunk_npages-1) == 0); assert(arena_mapbits_unallocated_size_get(chunk, map_bias) == arena_maxrun); assert(arena_mapbits_unallocated_size_get(chunk, chunk_npages-1) == arena_maxrun); assert(arena_mapbits_dirty_get(chunk, map_bias) == arena_mapbits_dirty_get(chunk, chunk_npages-1)); return (chunk); } static bool arena_chunk_register(arena_t *arena, arena_chunk_t *chunk, bool zero) { /* * The extent node notion of "committed" doesn't directly apply to * arena chunks. Arbitrarily mark them as committed. The commit state * of runs is tracked individually, and upon chunk deallocation the * entire chunk is in a consistent commit state. */ extent_node_init(&chunk->node, arena, chunk, chunksize, zero, true); extent_node_achunk_set(&chunk->node, true); return (chunk_register(chunk, &chunk->node)); } static arena_chunk_t * arena_chunk_alloc_internal_hard(arena_t *arena, chunk_hooks_t *chunk_hooks, bool *zero, bool *commit) { arena_chunk_t *chunk; malloc_mutex_unlock(&arena->lock); chunk = (arena_chunk_t *)chunk_alloc_wrapper(arena, chunk_hooks, NULL, chunksize, chunksize, zero, commit); if (chunk != NULL && !*commit) { /* Commit header. */ if (chunk_hooks->commit(chunk, chunksize, 0, map_bias << LG_PAGE, arena->ind)) { chunk_dalloc_wrapper(arena, chunk_hooks, (void *)chunk, chunksize, *commit); chunk = NULL; } } if (chunk != NULL && arena_chunk_register(arena, chunk, *zero)) { if (!*commit) { /* Undo commit of header. */ chunk_hooks->decommit(chunk, chunksize, 0, map_bias << LG_PAGE, arena->ind); } chunk_dalloc_wrapper(arena, chunk_hooks, (void *)chunk, chunksize, *commit); chunk = NULL; } malloc_mutex_lock(&arena->lock); return (chunk); } static arena_chunk_t * arena_chunk_alloc_internal(arena_t *arena, bool *zero, bool *commit) { arena_chunk_t *chunk; chunk_hooks_t chunk_hooks = CHUNK_HOOKS_INITIALIZER; chunk = chunk_alloc_cache(arena, &chunk_hooks, NULL, chunksize, chunksize, zero, true); if (chunk != NULL) { if (arena_chunk_register(arena, chunk, *zero)) { chunk_dalloc_cache(arena, &chunk_hooks, chunk, chunksize, true); return (NULL); } *commit = true; } if (chunk == NULL) { chunk = arena_chunk_alloc_internal_hard(arena, &chunk_hooks, zero, commit); } if (config_stats && chunk != NULL) { arena->stats.mapped += chunksize; arena->stats.metadata_mapped += (map_bias << LG_PAGE); } return (chunk); } static arena_chunk_t * arena_chunk_init_hard(arena_t *arena) { arena_chunk_t *chunk; bool zero, commit; size_t flag_unzeroed, flag_decommitted, i; assert(arena->spare == NULL); zero = false; commit = false; chunk = arena_chunk_alloc_internal(arena, &zero, &commit); if (chunk == NULL) return (NULL); /* * Initialize the map to contain one maximal free untouched run. Mark * the pages as zeroed if chunk_alloc() returned a zeroed or decommitted * chunk. */ flag_unzeroed = (zero || !commit) ? 0 : CHUNK_MAP_UNZEROED; flag_decommitted = commit ? 0 : CHUNK_MAP_DECOMMITTED; arena_mapbits_unallocated_set(chunk, map_bias, arena_maxrun, flag_unzeroed | flag_decommitted); /* * There is no need to initialize the internal page map entries unless * the chunk is not zeroed. */ if (!zero) { JEMALLOC_VALGRIND_MAKE_MEM_UNDEFINED( (void *)arena_bitselm_get(chunk, map_bias+1), (size_t)((uintptr_t) arena_bitselm_get(chunk, chunk_npages-1) - (uintptr_t)arena_bitselm_get(chunk, map_bias+1))); for (i = map_bias+1; i < chunk_npages-1; i++) arena_mapbits_internal_set(chunk, i, flag_unzeroed); } else { JEMALLOC_VALGRIND_MAKE_MEM_DEFINED((void *)arena_bitselm_get(chunk, map_bias+1), (size_t)((uintptr_t) arena_bitselm_get(chunk, chunk_npages-1) - (uintptr_t)arena_bitselm_get(chunk, map_bias+1))); if (config_debug) { for (i = map_bias+1; i < chunk_npages-1; i++) { assert(arena_mapbits_unzeroed_get(chunk, i) == flag_unzeroed); } } } arena_mapbits_unallocated_set(chunk, chunk_npages-1, arena_maxrun, flag_unzeroed); return (chunk); } static arena_chunk_t * arena_chunk_alloc(arena_t *arena) { arena_chunk_t *chunk; if (arena->spare != NULL) chunk = arena_chunk_init_spare(arena); else { chunk = arena_chunk_init_hard(arena); if (chunk == NULL) return (NULL); } /* Insert the run into the runs_avail tree. */ arena_avail_insert(arena, chunk, map_bias, chunk_npages-map_bias); return (chunk); } static void arena_chunk_dalloc(arena_t *arena, arena_chunk_t *chunk) { assert(arena_mapbits_allocated_get(chunk, map_bias) == 0); assert(arena_mapbits_allocated_get(chunk, chunk_npages-1) == 0); assert(arena_mapbits_unallocated_size_get(chunk, map_bias) == arena_maxrun); assert(arena_mapbits_unallocated_size_get(chunk, chunk_npages-1) == arena_maxrun); assert(arena_mapbits_dirty_get(chunk, map_bias) == arena_mapbits_dirty_get(chunk, chunk_npages-1)); assert(arena_mapbits_decommitted_get(chunk, map_bias) == arena_mapbits_decommitted_get(chunk, chunk_npages-1)); /* * Remove run from the runs_avail tree, so that the arena does not use * it. */ arena_avail_remove(arena, chunk, map_bias, chunk_npages-map_bias); if (arena->spare != NULL) { arena_chunk_t *spare = arena->spare; chunk_hooks_t chunk_hooks = CHUNK_HOOKS_INITIALIZER; bool committed; arena->spare = chunk; if (arena_mapbits_dirty_get(spare, map_bias) != 0) { arena_run_dirty_remove(arena, spare, map_bias, chunk_npages-map_bias); } chunk_deregister(spare, &spare->node); committed = (arena_mapbits_decommitted_get(spare, map_bias) == 0); if (!committed) { /* * Decommit the header. Mark the chunk as decommitted * even if header decommit fails, since treating a * partially committed chunk as committed has a high * potential for causing later access of decommitted * memory. */ chunk_hooks = chunk_hooks_get(arena); chunk_hooks.decommit(spare, chunksize, 0, map_bias << LG_PAGE, arena->ind); } chunk_dalloc_cache(arena, &chunk_hooks, (void *)spare, chunksize, committed); if (config_stats) { arena->stats.mapped -= chunksize; arena->stats.metadata_mapped -= (map_bias << LG_PAGE); } } else arena->spare = chunk; } static void arena_huge_malloc_stats_update(arena_t *arena, size_t usize) { szind_t index = size2index(usize) - nlclasses - NBINS; cassert(config_stats); arena->stats.nmalloc_huge++; arena->stats.allocated_huge += usize; arena->stats.hstats[index].nmalloc++; arena->stats.hstats[index].curhchunks++; } static void arena_huge_malloc_stats_update_undo(arena_t *arena, size_t usize) { szind_t index = size2index(usize) - nlclasses - NBINS; cassert(config_stats); arena->stats.nmalloc_huge--; arena->stats.allocated_huge -= usize; arena->stats.hstats[index].nmalloc--; arena->stats.hstats[index].curhchunks--; } static void arena_huge_dalloc_stats_update(arena_t *arena, size_t usize) { szind_t index = size2index(usize) - nlclasses - NBINS; cassert(config_stats); arena->stats.ndalloc_huge++; arena->stats.allocated_huge -= usize; arena->stats.hstats[index].ndalloc++; arena->stats.hstats[index].curhchunks--; } static void arena_huge_dalloc_stats_update_undo(arena_t *arena, size_t usize) { szind_t index = size2index(usize) - nlclasses - NBINS; cassert(config_stats); arena->stats.ndalloc_huge--; arena->stats.allocated_huge += usize; arena->stats.hstats[index].ndalloc--; arena->stats.hstats[index].curhchunks++; } static void arena_huge_ralloc_stats_update(arena_t *arena, size_t oldsize, size_t usize) { arena_huge_dalloc_stats_update(arena, oldsize); arena_huge_malloc_stats_update(arena, usize); } static void arena_huge_ralloc_stats_update_undo(arena_t *arena, size_t oldsize, size_t usize) { arena_huge_dalloc_stats_update_undo(arena, oldsize); arena_huge_malloc_stats_update_undo(arena, usize); } extent_node_t * arena_node_alloc(arena_t *arena) { extent_node_t *node; malloc_mutex_lock(&arena->node_cache_mtx); node = ql_last(&arena->node_cache, ql_link); if (node == NULL) { malloc_mutex_unlock(&arena->node_cache_mtx); return (base_alloc(sizeof(extent_node_t))); } ql_tail_remove(&arena->node_cache, extent_node_t, ql_link); malloc_mutex_unlock(&arena->node_cache_mtx); return (node); } void arena_node_dalloc(arena_t *arena, extent_node_t *node) { malloc_mutex_lock(&arena->node_cache_mtx); ql_elm_new(node, ql_link); ql_tail_insert(&arena->node_cache, node, ql_link); malloc_mutex_unlock(&arena->node_cache_mtx); } static void * arena_chunk_alloc_huge_hard(arena_t *arena, chunk_hooks_t *chunk_hooks, size_t usize, size_t alignment, bool *zero, size_t csize) { void *ret; bool commit = true; ret = chunk_alloc_wrapper(arena, chunk_hooks, NULL, csize, alignment, zero, &commit); if (ret == NULL) { /* Revert optimistic stats updates. */ malloc_mutex_lock(&arena->lock); if (config_stats) { arena_huge_malloc_stats_update_undo(arena, usize); arena->stats.mapped -= usize; } arena->nactive -= (usize >> LG_PAGE); malloc_mutex_unlock(&arena->lock); } return (ret); } void * arena_chunk_alloc_huge(arena_t *arena, size_t usize, size_t alignment, bool *zero) { void *ret; chunk_hooks_t chunk_hooks = CHUNK_HOOKS_INITIALIZER; size_t csize = CHUNK_CEILING(usize); malloc_mutex_lock(&arena->lock); /* Optimistically update stats. */ if (config_stats) { arena_huge_malloc_stats_update(arena, usize); arena->stats.mapped += usize; } arena->nactive += (usize >> LG_PAGE); ret = chunk_alloc_cache(arena, &chunk_hooks, NULL, csize, alignment, zero, true); malloc_mutex_unlock(&arena->lock); if (ret == NULL) { ret = arena_chunk_alloc_huge_hard(arena, &chunk_hooks, usize, alignment, zero, csize); } if (config_stats && ret != NULL) stats_cactive_add(usize); return (ret); } void arena_chunk_dalloc_huge(arena_t *arena, void *chunk, size_t usize) { chunk_hooks_t chunk_hooks = CHUNK_HOOKS_INITIALIZER; size_t csize; csize = CHUNK_CEILING(usize); malloc_mutex_lock(&arena->lock); if (config_stats) { arena_huge_dalloc_stats_update(arena, usize); arena->stats.mapped -= usize; stats_cactive_sub(usize); } arena->nactive -= (usize >> LG_PAGE); chunk_dalloc_cache(arena, &chunk_hooks, chunk, csize, true); malloc_mutex_unlock(&arena->lock); } void arena_chunk_ralloc_huge_similar(arena_t *arena, void *chunk, size_t oldsize, size_t usize) { assert(CHUNK_CEILING(oldsize) == CHUNK_CEILING(usize)); assert(oldsize != usize); malloc_mutex_lock(&arena->lock); if (config_stats) arena_huge_ralloc_stats_update(arena, oldsize, usize); if (oldsize < usize) { size_t udiff = usize - oldsize; arena->nactive += udiff >> LG_PAGE; if (config_stats) stats_cactive_add(udiff); } else { size_t udiff = oldsize - usize; arena->nactive -= udiff >> LG_PAGE; if (config_stats) stats_cactive_sub(udiff); } malloc_mutex_unlock(&arena->lock); } void arena_chunk_ralloc_huge_shrink(arena_t *arena, void *chunk, size_t oldsize, size_t usize) { size_t udiff = oldsize - usize; size_t cdiff = CHUNK_CEILING(oldsize) - CHUNK_CEILING(usize); malloc_mutex_lock(&arena->lock); if (config_stats) { arena_huge_ralloc_stats_update(arena, oldsize, usize); if (cdiff != 0) { arena->stats.mapped -= cdiff; stats_cactive_sub(udiff); } } arena->nactive -= udiff >> LG_PAGE; if (cdiff != 0) { chunk_hooks_t chunk_hooks = CHUNK_HOOKS_INITIALIZER; void *nchunk = (void *)((uintptr_t)chunk + CHUNK_CEILING(usize)); chunk_dalloc_cache(arena, &chunk_hooks, nchunk, cdiff, true); } malloc_mutex_unlock(&arena->lock); } static bool arena_chunk_ralloc_huge_expand_hard(arena_t *arena, chunk_hooks_t *chunk_hooks, void *chunk, size_t oldsize, size_t usize, bool *zero, void *nchunk, size_t udiff, size_t cdiff) { bool err; bool commit = true; err = (chunk_alloc_wrapper(arena, chunk_hooks, nchunk, cdiff, chunksize, zero, &commit) == NULL); if (err) { /* Revert optimistic stats updates. */ malloc_mutex_lock(&arena->lock); if (config_stats) { arena_huge_ralloc_stats_update_undo(arena, oldsize, usize); arena->stats.mapped -= cdiff; } arena->nactive -= (udiff >> LG_PAGE); malloc_mutex_unlock(&arena->lock); } else if (chunk_hooks->merge(chunk, CHUNK_CEILING(oldsize), nchunk, cdiff, true, arena->ind)) { chunk_dalloc_arena(arena, chunk_hooks, nchunk, cdiff, *zero, true); err = true; } return (err); } bool arena_chunk_ralloc_huge_expand(arena_t *arena, void *chunk, size_t oldsize, size_t usize, bool *zero) { bool err; chunk_hooks_t chunk_hooks = chunk_hooks_get(arena); void *nchunk = (void *)((uintptr_t)chunk + CHUNK_CEILING(oldsize)); size_t udiff = usize - oldsize; size_t cdiff = CHUNK_CEILING(usize) - CHUNK_CEILING(oldsize); malloc_mutex_lock(&arena->lock); /* Optimistically update stats. */ if (config_stats) { arena_huge_ralloc_stats_update(arena, oldsize, usize); arena->stats.mapped += cdiff; } arena->nactive += (udiff >> LG_PAGE); err = (chunk_alloc_cache(arena, &arena->chunk_hooks, nchunk, cdiff, chunksize, zero, true) == NULL); malloc_mutex_unlock(&arena->lock); if (err) { err = arena_chunk_ralloc_huge_expand_hard(arena, &chunk_hooks, chunk, oldsize, usize, zero, nchunk, udiff, cdiff); } else if (chunk_hooks.merge(chunk, CHUNK_CEILING(oldsize), nchunk, cdiff, true, arena->ind)) { chunk_dalloc_arena(arena, &chunk_hooks, nchunk, cdiff, *zero, true); err = true; } if (config_stats && !err) stats_cactive_add(udiff); return (err); } /* * Do first-best-fit run selection, i.e. select the lowest run that best fits. * Run sizes are quantized, so not all candidate runs are necessarily exactly * the same size. */ static arena_run_t * arena_run_first_best_fit(arena_t *arena, size_t size) { size_t search_size = run_quantize_first(size); arena_chunk_map_misc_t *key = arena_miscelm_key_create(search_size); arena_chunk_map_misc_t *miscelm = arena_avail_tree_nsearch(&arena->runs_avail, key); if (miscelm == NULL) return (NULL); return (&miscelm->run); } static arena_run_t * arena_run_alloc_large_helper(arena_t *arena, size_t size, bool zero) { arena_run_t *run = arena_run_first_best_fit(arena, s2u(size)); if (run != NULL) { if (arena_run_split_large(arena, run, size, zero)) run = NULL; } return (run); } static arena_run_t * arena_run_alloc_large(arena_t *arena, size_t size, bool zero) { arena_chunk_t *chunk; arena_run_t *run; assert(size <= arena_maxrun); assert(size == PAGE_CEILING(size)); /* Search the arena's chunks for the lowest best fit. */ run = arena_run_alloc_large_helper(arena, size, zero); if (run != NULL) return (run); /* * No usable runs. Create a new chunk from which to allocate the run. */ chunk = arena_chunk_alloc(arena); if (chunk != NULL) { run = &arena_miscelm_get(chunk, map_bias)->run; if (arena_run_split_large(arena, run, size, zero)) run = NULL; return (run); } /* * arena_chunk_alloc() failed, but another thread may have made * sufficient memory available while this one dropped arena->lock in * arena_chunk_alloc(), so search one more time. */ return (arena_run_alloc_large_helper(arena, size, zero)); } static arena_run_t * arena_run_alloc_small_helper(arena_t *arena, size_t size, szind_t binind) { arena_run_t *run = arena_run_first_best_fit(arena, size); if (run != NULL) { if (arena_run_split_small(arena, run, size, binind)) run = NULL; } return (run); } static arena_run_t * arena_run_alloc_small(arena_t *arena, size_t size, szind_t binind) { arena_chunk_t *chunk; arena_run_t *run; assert(size <= arena_maxrun); assert(size == PAGE_CEILING(size)); assert(binind != BININD_INVALID); /* Search the arena's chunks for the lowest best fit. */ run = arena_run_alloc_small_helper(arena, size, binind); if (run != NULL) return (run); /* * No usable runs. Create a new chunk from which to allocate the run. */ chunk = arena_chunk_alloc(arena); if (chunk != NULL) { run = &arena_miscelm_get(chunk, map_bias)->run; if (arena_run_split_small(arena, run, size, binind)) run = NULL; return (run); } /* * arena_chunk_alloc() failed, but another thread may have made * sufficient memory available while this one dropped arena->lock in * arena_chunk_alloc(), so search one more time. */ return (arena_run_alloc_small_helper(arena, size, binind)); } static bool arena_lg_dirty_mult_valid(ssize_t lg_dirty_mult) { return (lg_dirty_mult >= -1 && lg_dirty_mult < (ssize_t)(sizeof(size_t) << 3)); } ssize_t arena_lg_dirty_mult_get(arena_t *arena) { ssize_t lg_dirty_mult; malloc_mutex_lock(&arena->lock); lg_dirty_mult = arena->lg_dirty_mult; malloc_mutex_unlock(&arena->lock); return (lg_dirty_mult); } bool arena_lg_dirty_mult_set(arena_t *arena, ssize_t lg_dirty_mult) { if (!arena_lg_dirty_mult_valid(lg_dirty_mult)) return (true); malloc_mutex_lock(&arena->lock); arena->lg_dirty_mult = lg_dirty_mult; arena_maybe_purge(arena); malloc_mutex_unlock(&arena->lock); return (false); } void arena_maybe_purge(arena_t *arena) { /* Don't purge if the option is disabled. */ if (arena->lg_dirty_mult < 0) return; /* Don't recursively purge. */ if (arena->purging) return; /* * Iterate, since preventing recursive purging could otherwise leave too * many dirty pages. */ while (true) { size_t threshold = (arena->nactive >> arena->lg_dirty_mult); if (threshold < chunk_npages) threshold = chunk_npages; /* * Don't purge unless the number of purgeable pages exceeds the * threshold. */ if (arena->ndirty <= threshold) return; arena_purge(arena, false); } } static size_t arena_dirty_count(arena_t *arena) { size_t ndirty = 0; arena_runs_dirty_link_t *rdelm; extent_node_t *chunkselm; for (rdelm = qr_next(&arena->runs_dirty, rd_link), chunkselm = qr_next(&arena->chunks_cache, cc_link); rdelm != &arena->runs_dirty; rdelm = qr_next(rdelm, rd_link)) { size_t npages; if (rdelm == &chunkselm->rd) { npages = extent_node_size_get(chunkselm) >> LG_PAGE; chunkselm = qr_next(chunkselm, cc_link); } else { arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE( rdelm); arena_chunk_map_misc_t *miscelm = arena_rd_to_miscelm(rdelm); size_t pageind = arena_miscelm_to_pageind(miscelm); assert(arena_mapbits_allocated_get(chunk, pageind) == 0); assert(arena_mapbits_large_get(chunk, pageind) == 0); assert(arena_mapbits_dirty_get(chunk, pageind) != 0); npages = arena_mapbits_unallocated_size_get(chunk, pageind) >> LG_PAGE; } ndirty += npages; } return (ndirty); } static size_t arena_compute_npurge(arena_t *arena, bool all) { size_t npurge; /* * Compute the minimum number of pages that this thread should try to * purge. */ if (!all) { size_t threshold = (arena->nactive >> arena->lg_dirty_mult); threshold = threshold < chunk_npages ? chunk_npages : threshold; npurge = arena->ndirty - threshold; } else npurge = arena->ndirty; return (npurge); } static size_t arena_stash_dirty(arena_t *arena, chunk_hooks_t *chunk_hooks, bool all, size_t npurge, arena_runs_dirty_link_t *purge_runs_sentinel, extent_node_t *purge_chunks_sentinel) { arena_runs_dirty_link_t *rdelm, *rdelm_next; extent_node_t *chunkselm; size_t nstashed = 0; /* Stash at least npurge pages. */ for (rdelm = qr_next(&arena->runs_dirty, rd_link), chunkselm = qr_next(&arena->chunks_cache, cc_link); rdelm != &arena->runs_dirty; rdelm = rdelm_next) { size_t npages; rdelm_next = qr_next(rdelm, rd_link); if (rdelm == &chunkselm->rd) { extent_node_t *chunkselm_next; bool zero; UNUSED void *chunk; chunkselm_next = qr_next(chunkselm, cc_link); /* * Allocate. chunkselm remains valid due to the * dalloc_node=false argument to chunk_alloc_cache(). */ zero = false; chunk = chunk_alloc_cache(arena, chunk_hooks, extent_node_addr_get(chunkselm), extent_node_size_get(chunkselm), chunksize, &zero, false); assert(chunk == extent_node_addr_get(chunkselm)); assert(zero == extent_node_zeroed_get(chunkselm)); extent_node_dirty_insert(chunkselm, purge_runs_sentinel, purge_chunks_sentinel); npages = extent_node_size_get(chunkselm) >> LG_PAGE; chunkselm = chunkselm_next; } else { arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(rdelm); arena_chunk_map_misc_t *miscelm = arena_rd_to_miscelm(rdelm); size_t pageind = arena_miscelm_to_pageind(miscelm); arena_run_t *run = &miscelm->run; size_t run_size = arena_mapbits_unallocated_size_get(chunk, pageind); npages = run_size >> LG_PAGE; assert(pageind + npages <= chunk_npages); assert(arena_mapbits_dirty_get(chunk, pageind) == arena_mapbits_dirty_get(chunk, pageind+npages-1)); /* * If purging the spare chunk's run, make it available * prior to allocation. */ if (chunk == arena->spare) arena_chunk_alloc(arena); /* Temporarily allocate the free dirty run. */ arena_run_split_large(arena, run, run_size, false); /* Stash. */ if (false) qr_new(rdelm, rd_link); /* Redundant. */ else { assert(qr_next(rdelm, rd_link) == rdelm); assert(qr_prev(rdelm, rd_link) == rdelm); } qr_meld(purge_runs_sentinel, rdelm, rd_link); } nstashed += npages; if (!all && nstashed >= npurge) break; } return (nstashed); } static size_t arena_purge_stashed(arena_t *arena, chunk_hooks_t *chunk_hooks, arena_runs_dirty_link_t *purge_runs_sentinel, extent_node_t *purge_chunks_sentinel) { size_t npurged, nmadvise; arena_runs_dirty_link_t *rdelm; extent_node_t *chunkselm; if (config_stats) nmadvise = 0; npurged = 0; malloc_mutex_unlock(&arena->lock); for (rdelm = qr_next(purge_runs_sentinel, rd_link), chunkselm = qr_next(purge_chunks_sentinel, cc_link); rdelm != purge_runs_sentinel; rdelm = qr_next(rdelm, rd_link)) { size_t npages; if (rdelm == &chunkselm->rd) { /* * Don't actually purge the chunk here because 1) * chunkselm is embedded in the chunk and must remain * valid, and 2) we deallocate the chunk in * arena_unstash_purged(), where it is destroyed, * decommitted, or purged, depending on chunk * deallocation policy. */ size_t size = extent_node_size_get(chunkselm); npages = size >> LG_PAGE; chunkselm = qr_next(chunkselm, cc_link); } else { size_t pageind, run_size, flag_unzeroed, flags, i; bool decommitted; arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(rdelm); arena_chunk_map_misc_t *miscelm = arena_rd_to_miscelm(rdelm); pageind = arena_miscelm_to_pageind(miscelm); run_size = arena_mapbits_large_size_get(chunk, pageind); npages = run_size >> LG_PAGE; assert(pageind + npages <= chunk_npages); assert(!arena_mapbits_decommitted_get(chunk, pageind)); assert(!arena_mapbits_decommitted_get(chunk, pageind+npages-1)); decommitted = !chunk_hooks->decommit(chunk, chunksize, pageind << LG_PAGE, npages << LG_PAGE, arena->ind); if (decommitted) { flag_unzeroed = 0; flags = CHUNK_MAP_DECOMMITTED; } else { flag_unzeroed = chunk_purge_wrapper(arena, chunk_hooks, chunk, chunksize, pageind << LG_PAGE, run_size) ? CHUNK_MAP_UNZEROED : 0; flags = flag_unzeroed; } arena_mapbits_large_set(chunk, pageind+npages-1, 0, flags); arena_mapbits_large_set(chunk, pageind, run_size, flags); /* * Set the unzeroed flag for internal pages, now that * chunk_purge_wrapper() has returned whether the pages * were zeroed as a side effect of purging. This chunk * map modification is safe even though the arena mutex * isn't currently owned by this thread, because the run * is marked as allocated, thus protecting it from being * modified by any other thread. As long as these * writes don't perturb the first and last elements' * CHUNK_MAP_ALLOCATED bits, behavior is well defined. */ for (i = 1; i < npages-1; i++) { arena_mapbits_internal_set(chunk, pageind+i, flag_unzeroed); } } npurged += npages; if (config_stats) nmadvise++; } malloc_mutex_lock(&arena->lock); if (config_stats) { arena->stats.nmadvise += nmadvise; arena->stats.purged += npurged; } return (npurged); } static void arena_unstash_purged(arena_t *arena, chunk_hooks_t *chunk_hooks, arena_runs_dirty_link_t *purge_runs_sentinel, extent_node_t *purge_chunks_sentinel) { arena_runs_dirty_link_t *rdelm, *rdelm_next; extent_node_t *chunkselm; /* Deallocate chunks/runs. */ for (rdelm = qr_next(purge_runs_sentinel, rd_link), chunkselm = qr_next(purge_chunks_sentinel, cc_link); rdelm != purge_runs_sentinel; rdelm = rdelm_next) { rdelm_next = qr_next(rdelm, rd_link); if (rdelm == &chunkselm->rd) { extent_node_t *chunkselm_next = qr_next(chunkselm, cc_link); void *addr = extent_node_addr_get(chunkselm); size_t size = extent_node_size_get(chunkselm); bool zeroed = extent_node_zeroed_get(chunkselm); bool committed = extent_node_committed_get(chunkselm); extent_node_dirty_remove(chunkselm); arena_node_dalloc(arena, chunkselm); chunkselm = chunkselm_next; chunk_dalloc_arena(arena, chunk_hooks, addr, size, zeroed, committed); } else { arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(rdelm); arena_chunk_map_misc_t *miscelm = arena_rd_to_miscelm(rdelm); size_t pageind = arena_miscelm_to_pageind(miscelm); bool decommitted = (arena_mapbits_decommitted_get(chunk, pageind) != 0); arena_run_t *run = &miscelm->run; qr_remove(rdelm, rd_link); arena_run_dalloc(arena, run, false, true, decommitted); } } } static void arena_purge(arena_t *arena, bool all) { chunk_hooks_t chunk_hooks = chunk_hooks_get(arena); size_t npurge, npurgeable, npurged; arena_runs_dirty_link_t purge_runs_sentinel; extent_node_t purge_chunks_sentinel; arena->purging = true; /* * Calls to arena_dirty_count() are disabled even for debug builds * because overhead grows nonlinearly as memory usage increases. */ if (false && config_debug) { size_t ndirty = arena_dirty_count(arena); assert(ndirty == arena->ndirty); } assert((arena->nactive >> arena->lg_dirty_mult) < arena->ndirty || all); if (config_stats) arena->stats.npurge++; npurge = arena_compute_npurge(arena, all); qr_new(&purge_runs_sentinel, rd_link); extent_node_dirty_linkage_init(&purge_chunks_sentinel); npurgeable = arena_stash_dirty(arena, &chunk_hooks, all, npurge, &purge_runs_sentinel, &purge_chunks_sentinel); assert(npurgeable >= npurge); npurged = arena_purge_stashed(arena, &chunk_hooks, &purge_runs_sentinel, &purge_chunks_sentinel); assert(npurged == npurgeable); arena_unstash_purged(arena, &chunk_hooks, &purge_runs_sentinel, &purge_chunks_sentinel); arena->purging = false; } void arena_purge_all(arena_t *arena) { malloc_mutex_lock(&arena->lock); arena_purge(arena, true); malloc_mutex_unlock(&arena->lock); } static void arena_run_coalesce(arena_t *arena, arena_chunk_t *chunk, size_t *p_size, size_t *p_run_ind, size_t *p_run_pages, size_t flag_dirty, size_t flag_decommitted) { size_t size = *p_size; size_t run_ind = *p_run_ind; size_t run_pages = *p_run_pages; /* Try to coalesce forward. */ if (run_ind + run_pages < chunk_npages && arena_mapbits_allocated_get(chunk, run_ind+run_pages) == 0 && arena_mapbits_dirty_get(chunk, run_ind+run_pages) == flag_dirty && arena_mapbits_decommitted_get(chunk, run_ind+run_pages) == flag_decommitted) { size_t nrun_size = arena_mapbits_unallocated_size_get(chunk, run_ind+run_pages); size_t nrun_pages = nrun_size >> LG_PAGE; /* * Remove successor from runs_avail; the coalesced run is * inserted later. */ assert(arena_mapbits_unallocated_size_get(chunk, run_ind+run_pages+nrun_pages-1) == nrun_size); assert(arena_mapbits_dirty_get(chunk, run_ind+run_pages+nrun_pages-1) == flag_dirty); assert(arena_mapbits_decommitted_get(chunk, run_ind+run_pages+nrun_pages-1) == flag_decommitted); arena_avail_remove(arena, chunk, run_ind+run_pages, nrun_pages); /* * If the successor is dirty, remove it from the set of dirty * pages. */ if (flag_dirty != 0) { arena_run_dirty_remove(arena, chunk, run_ind+run_pages, nrun_pages); } size += nrun_size; run_pages += nrun_pages; arena_mapbits_unallocated_size_set(chunk, run_ind, size); arena_mapbits_unallocated_size_set(chunk, run_ind+run_pages-1, size); } /* Try to coalesce backward. */ if (run_ind > map_bias && arena_mapbits_allocated_get(chunk, run_ind-1) == 0 && arena_mapbits_dirty_get(chunk, run_ind-1) == flag_dirty && arena_mapbits_decommitted_get(chunk, run_ind-1) == flag_decommitted) { size_t prun_size = arena_mapbits_unallocated_size_get(chunk, run_ind-1); size_t prun_pages = prun_size >> LG_PAGE; run_ind -= prun_pages; /* * Remove predecessor from runs_avail; the coalesced run is * inserted later. */ assert(arena_mapbits_unallocated_size_get(chunk, run_ind) == prun_size); assert(arena_mapbits_dirty_get(chunk, run_ind) == flag_dirty); assert(arena_mapbits_decommitted_get(chunk, run_ind) == flag_decommitted); arena_avail_remove(arena, chunk, run_ind, prun_pages); /* * If the predecessor is dirty, remove it from the set of dirty * pages. */ if (flag_dirty != 0) { arena_run_dirty_remove(arena, chunk, run_ind, prun_pages); } size += prun_size; run_pages += prun_pages; arena_mapbits_unallocated_size_set(chunk, run_ind, size); arena_mapbits_unallocated_size_set(chunk, run_ind+run_pages-1, size); } *p_size = size; *p_run_ind = run_ind; *p_run_pages = run_pages; } static size_t arena_run_size_get(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, size_t run_ind) { size_t size; assert(run_ind >= map_bias); assert(run_ind < chunk_npages); if (arena_mapbits_large_get(chunk, run_ind) != 0) { size = arena_mapbits_large_size_get(chunk, run_ind); assert(size == PAGE || arena_mapbits_large_size_get(chunk, run_ind+(size>>LG_PAGE)-1) == 0); } else { arena_bin_info_t *bin_info = &arena_bin_info[run->binind]; size = bin_info->run_size; } return (size); } static bool arena_run_decommit(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run) { arena_chunk_map_misc_t *miscelm = arena_run_to_miscelm(run); size_t run_ind = arena_miscelm_to_pageind(miscelm); size_t offset = run_ind << LG_PAGE; size_t length = arena_run_size_get(arena, chunk, run, run_ind); return (arena->chunk_hooks.decommit(chunk, chunksize, offset, length, arena->ind)); } static void arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned, bool decommitted) { arena_chunk_t *chunk; arena_chunk_map_misc_t *miscelm; size_t size, run_ind, run_pages, flag_dirty, flag_decommitted; chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); miscelm = arena_run_to_miscelm(run); run_ind = arena_miscelm_to_pageind(miscelm); assert(run_ind >= map_bias); assert(run_ind < chunk_npages); size = arena_run_size_get(arena, chunk, run, run_ind); run_pages = (size >> LG_PAGE); arena_cactive_update(arena, 0, run_pages); arena->nactive -= run_pages; /* * The run is dirty if the caller claims to have dirtied it, as well as * if it was already dirty before being allocated and the caller * doesn't claim to have cleaned it. */ assert(arena_mapbits_dirty_get(chunk, run_ind) == arena_mapbits_dirty_get(chunk, run_ind+run_pages-1)); if (!cleaned && !decommitted && arena_mapbits_dirty_get(chunk, run_ind) != 0) dirty = true; flag_dirty = dirty ? CHUNK_MAP_DIRTY : 0; flag_decommitted = decommitted ? CHUNK_MAP_DECOMMITTED : 0; /* Mark pages as unallocated in the chunk map. */ if (dirty || decommitted) { size_t flags = flag_dirty | flag_decommitted; arena_mapbits_unallocated_set(chunk, run_ind, size, flags); arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size, flags); } else { arena_mapbits_unallocated_set(chunk, run_ind, size, arena_mapbits_unzeroed_get(chunk, run_ind)); arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size, arena_mapbits_unzeroed_get(chunk, run_ind+run_pages-1)); } arena_run_coalesce(arena, chunk, &size, &run_ind, &run_pages, flag_dirty, flag_decommitted); /* Insert into runs_avail, now that coalescing is complete. */ assert(arena_mapbits_unallocated_size_get(chunk, run_ind) == arena_mapbits_unallocated_size_get(chunk, run_ind+run_pages-1)); assert(arena_mapbits_dirty_get(chunk, run_ind) == arena_mapbits_dirty_get(chunk, run_ind+run_pages-1)); assert(arena_mapbits_decommitted_get(chunk, run_ind) == arena_mapbits_decommitted_get(chunk, run_ind+run_pages-1)); arena_avail_insert(arena, chunk, run_ind, run_pages); if (dirty) arena_run_dirty_insert(arena, chunk, run_ind, run_pages); /* Deallocate chunk if it is now completely unused. */ if (size == arena_maxrun) { assert(run_ind == map_bias); assert(run_pages == (arena_maxrun >> LG_PAGE)); arena_chunk_dalloc(arena, chunk); } /* * It is okay to do dirty page processing here even if the chunk was * deallocated above, since in that case it is the spare. Waiting * until after possible chunk deallocation to do dirty processing * allows for an old spare to be fully deallocated, thus decreasing the * chances of spuriously crossing the dirty page purging threshold. */ if (dirty) arena_maybe_purge(arena); } static void arena_run_dalloc_decommit(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run) { bool committed = arena_run_decommit(arena, chunk, run); arena_run_dalloc(arena, run, committed, false, !committed); } static void arena_run_trim_head(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, size_t oldsize, size_t newsize) { arena_chunk_map_misc_t *miscelm = arena_run_to_miscelm(run); size_t pageind = arena_miscelm_to_pageind(miscelm); size_t head_npages = (oldsize - newsize) >> LG_PAGE; size_t flag_dirty = arena_mapbits_dirty_get(chunk, pageind); size_t flag_decommitted = arena_mapbits_decommitted_get(chunk, pageind); size_t flag_unzeroed_mask = (flag_dirty | flag_decommitted) == 0 ? CHUNK_MAP_UNZEROED : 0; assert(oldsize > newsize); /* * Update the chunk map so that arena_run_dalloc() can treat the * leading run as separately allocated. Set the last element of each * run first, in case of single-page runs. */ assert(arena_mapbits_large_size_get(chunk, pageind) == oldsize); arena_mapbits_large_set(chunk, pageind+head_npages-1, 0, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind+head_npages-1))); arena_mapbits_large_set(chunk, pageind, oldsize-newsize, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind))); if (config_debug) { UNUSED size_t tail_npages = newsize >> LG_PAGE; assert(arena_mapbits_large_size_get(chunk, pageind+head_npages+tail_npages-1) == 0); assert(arena_mapbits_dirty_get(chunk, pageind+head_npages+tail_npages-1) == flag_dirty); } arena_mapbits_large_set(chunk, pageind+head_npages, newsize, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind+head_npages))); arena_run_dalloc(arena, run, false, false, (flag_decommitted != 0)); } static void arena_run_trim_tail(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, size_t oldsize, size_t newsize, bool dirty) { arena_chunk_map_misc_t *miscelm = arena_run_to_miscelm(run); size_t pageind = arena_miscelm_to_pageind(miscelm); size_t head_npages = newsize >> LG_PAGE; size_t flag_dirty = arena_mapbits_dirty_get(chunk, pageind); size_t flag_decommitted = arena_mapbits_decommitted_get(chunk, pageind); size_t flag_unzeroed_mask = (flag_dirty | flag_decommitted) == 0 ? CHUNK_MAP_UNZEROED : 0; arena_chunk_map_misc_t *tail_miscelm; arena_run_t *tail_run; assert(oldsize > newsize); /* * Update the chunk map so that arena_run_dalloc() can treat the * trailing run as separately allocated. Set the last element of each * run first, in case of single-page runs. */ assert(arena_mapbits_large_size_get(chunk, pageind) == oldsize); arena_mapbits_large_set(chunk, pageind+head_npages-1, 0, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind+head_npages-1))); arena_mapbits_large_set(chunk, pageind, newsize, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind))); if (config_debug) { UNUSED size_t tail_npages = (oldsize - newsize) >> LG_PAGE; assert(arena_mapbits_large_size_get(chunk, pageind+head_npages+tail_npages-1) == 0); assert(arena_mapbits_dirty_get(chunk, pageind+head_npages+tail_npages-1) == flag_dirty); } arena_mapbits_large_set(chunk, pageind+head_npages, oldsize-newsize, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind+head_npages))); tail_miscelm = arena_miscelm_get(chunk, pageind + head_npages); tail_run = &tail_miscelm->run; arena_run_dalloc(arena, tail_run, dirty, false, (flag_decommitted != 0)); } static arena_run_t * arena_bin_runs_first(arena_bin_t *bin) { arena_chunk_map_misc_t *miscelm = arena_run_tree_first(&bin->runs); if (miscelm != NULL) return (&miscelm->run); return (NULL); } static void arena_bin_runs_insert(arena_bin_t *bin, arena_run_t *run) { arena_chunk_map_misc_t *miscelm = arena_run_to_miscelm(run); assert(arena_run_tree_search(&bin->runs, miscelm) == NULL); arena_run_tree_insert(&bin->runs, miscelm); } static void arena_bin_runs_remove(arena_bin_t *bin, arena_run_t *run) { arena_chunk_map_misc_t *miscelm = arena_run_to_miscelm(run); assert(arena_run_tree_search(&bin->runs, miscelm) != NULL); arena_run_tree_remove(&bin->runs, miscelm); } static arena_run_t * arena_bin_nonfull_run_tryget(arena_bin_t *bin) { arena_run_t *run = arena_bin_runs_first(bin); if (run != NULL) { arena_bin_runs_remove(bin, run); if (config_stats) bin->stats.reruns++; } return (run); } static arena_run_t * arena_bin_nonfull_run_get(arena_t *arena, arena_bin_t *bin) { arena_run_t *run; szind_t binind; arena_bin_info_t *bin_info; /* Look for a usable run. */ run = arena_bin_nonfull_run_tryget(bin); if (run != NULL) return (run); /* No existing runs have any space available. */ binind = arena_bin_index(arena, bin); bin_info = &arena_bin_info[binind]; /* Allocate a new run. */ malloc_mutex_unlock(&bin->lock); /******************************/ malloc_mutex_lock(&arena->lock); run = arena_run_alloc_small(arena, bin_info->run_size, binind); if (run != NULL) { /* Initialize run internals. */ run->binind = binind; run->nfree = bin_info->nregs; bitmap_init(run->bitmap, &bin_info->bitmap_info); } malloc_mutex_unlock(&arena->lock); /********************************/ malloc_mutex_lock(&bin->lock); if (run != NULL) { if (config_stats) { bin->stats.nruns++; bin->stats.curruns++; } return (run); } /* * arena_run_alloc_small() failed, but another thread may have made * sufficient memory available while this one dropped bin->lock above, * so search one more time. */ run = arena_bin_nonfull_run_tryget(bin); if (run != NULL) return (run); return (NULL); } /* Re-fill bin->runcur, then call arena_run_reg_alloc(). */ static void * arena_bin_malloc_hard(arena_t *arena, arena_bin_t *bin) { szind_t binind; arena_bin_info_t *bin_info; arena_run_t *run; binind = arena_bin_index(arena, bin); bin_info = &arena_bin_info[binind]; bin->runcur = NULL; run = arena_bin_nonfull_run_get(arena, bin); if (bin->runcur != NULL && bin->runcur->nfree > 0) { /* * Another thread updated runcur while this one ran without the * bin lock in arena_bin_nonfull_run_get(). */ void *ret; assert(bin->runcur->nfree > 0); ret = arena_run_reg_alloc(bin->runcur, bin_info); if (run != NULL) { arena_chunk_t *chunk; /* * arena_run_alloc_small() may have allocated run, or * it may have pulled run from the bin's run tree. * Therefore it is unsafe to make any assumptions about * how run has previously been used, and * arena_bin_lower_run() must be called, as if a region * were just deallocated from the run. */ chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); if (run->nfree == bin_info->nregs) arena_dalloc_bin_run(arena, chunk, run, bin); else arena_bin_lower_run(arena, chunk, run, bin); } return (ret); } if (run == NULL) return (NULL); bin->runcur = run; assert(bin->runcur->nfree > 0); return (arena_run_reg_alloc(bin->runcur, bin_info)); } void arena_tcache_fill_small(arena_t *arena, tcache_bin_t *tbin, szind_t binind, uint64_t prof_accumbytes) { unsigned i, nfill; arena_bin_t *bin; assert(tbin->ncached == 0); if (config_prof && arena_prof_accum(arena, prof_accumbytes)) prof_idump(); bin = &arena->bins[binind]; malloc_mutex_lock(&bin->lock); for (i = 0, nfill = (tcache_bin_info[binind].ncached_max >> tbin->lg_fill_div); i < nfill; i++) { arena_run_t *run; void *ptr; if ((run = bin->runcur) != NULL && run->nfree > 0) ptr = arena_run_reg_alloc(run, &arena_bin_info[binind]); else ptr = arena_bin_malloc_hard(arena, bin); if (ptr == NULL) { /* * OOM. tbin->avail isn't yet filled down to its first * element, so the successful allocations (if any) must * be moved to the base of tbin->avail before bailing * out. */ if (i > 0) { memmove(tbin->avail, &tbin->avail[nfill - i], i * sizeof(void *)); } break; } if (config_fill && unlikely(opt_junk_alloc)) { arena_alloc_junk_small(ptr, &arena_bin_info[binind], true); } /* Insert such that low regions get used first. */ tbin->avail[nfill - 1 - i] = ptr; } if (config_stats) { bin->stats.nmalloc += i; bin->stats.nrequests += tbin->tstats.nrequests; bin->stats.curregs += i; bin->stats.nfills++; tbin->tstats.nrequests = 0; } malloc_mutex_unlock(&bin->lock); tbin->ncached = i; } void arena_alloc_junk_small(void *ptr, arena_bin_info_t *bin_info, bool zero) { if (zero) { size_t redzone_size = bin_info->redzone_size; memset((void *)((uintptr_t)ptr - redzone_size), 0xa5, redzone_size); memset((void *)((uintptr_t)ptr + bin_info->reg_size), 0xa5, redzone_size); } else { memset((void *)((uintptr_t)ptr - bin_info->redzone_size), 0xa5, bin_info->reg_interval); } } #ifdef JEMALLOC_JET #undef arena_redzone_corruption #define arena_redzone_corruption JEMALLOC_N(arena_redzone_corruption_impl) #endif static void arena_redzone_corruption(void *ptr, size_t usize, bool after, size_t offset, uint8_t byte) { malloc_printf(": Corrupt redzone %zu byte%s %s %p " "(size %zu), byte=%#x\n", offset, (offset == 1) ? "" : "s", after ? "after" : "before", ptr, usize, byte); } #ifdef JEMALLOC_JET #undef arena_redzone_corruption #define arena_redzone_corruption JEMALLOC_N(arena_redzone_corruption) arena_redzone_corruption_t *arena_redzone_corruption = JEMALLOC_N(arena_redzone_corruption_impl); #endif static void arena_redzones_validate(void *ptr, arena_bin_info_t *bin_info, bool reset) { bool error = false; if (opt_junk_alloc) { size_t size = bin_info->reg_size; size_t redzone_size = bin_info->redzone_size; size_t i; for (i = 1; i <= redzone_size; i++) { uint8_t *byte = (uint8_t *)((uintptr_t)ptr - i); if (*byte != 0xa5) { error = true; arena_redzone_corruption(ptr, size, false, i, *byte); if (reset) *byte = 0xa5; } } for (i = 0; i < redzone_size; i++) { uint8_t *byte = (uint8_t *)((uintptr_t)ptr + size + i); if (*byte != 0xa5) { error = true; arena_redzone_corruption(ptr, size, true, i, *byte); if (reset) *byte = 0xa5; } } } if (opt_abort && error) abort(); } #ifdef JEMALLOC_JET #undef arena_dalloc_junk_small #define arena_dalloc_junk_small JEMALLOC_N(arena_dalloc_junk_small_impl) #endif void arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info) { size_t redzone_size = bin_info->redzone_size; arena_redzones_validate(ptr, bin_info, false); memset((void *)((uintptr_t)ptr - redzone_size), 0x5a, bin_info->reg_interval); } #ifdef JEMALLOC_JET #undef arena_dalloc_junk_small #define arena_dalloc_junk_small JEMALLOC_N(arena_dalloc_junk_small) arena_dalloc_junk_small_t *arena_dalloc_junk_small = JEMALLOC_N(arena_dalloc_junk_small_impl); #endif void arena_quarantine_junk_small(void *ptr, size_t usize) { szind_t binind; arena_bin_info_t *bin_info; cassert(config_fill); assert(opt_junk_free); assert(opt_quarantine); assert(usize <= SMALL_MAXCLASS); binind = size2index(usize); bin_info = &arena_bin_info[binind]; arena_redzones_validate(ptr, bin_info, true); } void * arena_malloc_small(arena_t *arena, size_t size, bool zero) { void *ret; arena_bin_t *bin; arena_run_t *run; szind_t binind; binind = size2index(size); assert(binind < NBINS); bin = &arena->bins[binind]; size = index2size(binind); malloc_mutex_lock(&bin->lock); if ((run = bin->runcur) != NULL && run->nfree > 0) ret = arena_run_reg_alloc(run, &arena_bin_info[binind]); else ret = arena_bin_malloc_hard(arena, bin); if (ret == NULL) { malloc_mutex_unlock(&bin->lock); return (NULL); } if (config_stats) { bin->stats.nmalloc++; bin->stats.nrequests++; bin->stats.curregs++; } malloc_mutex_unlock(&bin->lock); if (config_prof && !isthreaded && arena_prof_accum(arena, size)) prof_idump(); if (!zero) { if (config_fill) { if (unlikely(opt_junk_alloc)) { arena_alloc_junk_small(ret, &arena_bin_info[binind], false); } else if (unlikely(opt_zero)) memset(ret, 0, size); } JEMALLOC_VALGRIND_MAKE_MEM_UNDEFINED(ret, size); } else { if (config_fill && unlikely(opt_junk_alloc)) { arena_alloc_junk_small(ret, &arena_bin_info[binind], true); } JEMALLOC_VALGRIND_MAKE_MEM_UNDEFINED(ret, size); memset(ret, 0, size); } return (ret); } void * arena_malloc_large(arena_t *arena, size_t size, bool zero) { void *ret; size_t usize; uintptr_t random_offset; arena_run_t *run; arena_chunk_map_misc_t *miscelm; UNUSED bool idump; /* Large allocation. */ usize = s2u(size); malloc_mutex_lock(&arena->lock); if (config_cache_oblivious) { uint64_t r; /* * Compute a uniformly distributed offset within the first page * that is a multiple of the cacheline size, e.g. [0 .. 63) * 64 * for 4 KiB pages and 64-byte cachelines. */ prng64(r, LG_PAGE - LG_CACHELINE, arena->offset_state, UINT64_C(6364136223846793009), UINT64_C(1442695040888963409)); random_offset = ((uintptr_t)r) << LG_CACHELINE; } else random_offset = 0; run = arena_run_alloc_large(arena, usize + large_pad, zero); if (run == NULL) { malloc_mutex_unlock(&arena->lock); return (NULL); } miscelm = arena_run_to_miscelm(run); ret = (void *)((uintptr_t)arena_miscelm_to_rpages(miscelm) + random_offset); if (config_stats) { szind_t index = size2index(usize) - NBINS; arena->stats.nmalloc_large++; arena->stats.nrequests_large++; arena->stats.allocated_large += usize; arena->stats.lstats[index].nmalloc++; arena->stats.lstats[index].nrequests++; arena->stats.lstats[index].curruns++; } if (config_prof) idump = arena_prof_accum_locked(arena, usize); malloc_mutex_unlock(&arena->lock); if (config_prof && idump) prof_idump(); if (!zero) { if (config_fill) { if (unlikely(opt_junk_alloc)) memset(ret, 0xa5, usize); else if (unlikely(opt_zero)) memset(ret, 0, usize); } } return (ret); } /* Only handles large allocations that require more than page alignment. */ static void * arena_palloc_large(tsd_t *tsd, arena_t *arena, size_t usize, size_t alignment, bool zero) { void *ret; size_t alloc_size, leadsize, trailsize; arena_run_t *run; arena_chunk_t *chunk; arena_chunk_map_misc_t *miscelm; void *rpages; assert(usize == PAGE_CEILING(usize)); arena = arena_choose(tsd, arena); if (unlikely(arena == NULL)) return (NULL); alignment = PAGE_CEILING(alignment); alloc_size = usize + large_pad + alignment - PAGE; malloc_mutex_lock(&arena->lock); run = arena_run_alloc_large(arena, alloc_size, false); if (run == NULL) { malloc_mutex_unlock(&arena->lock); return (NULL); } chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); miscelm = arena_run_to_miscelm(run); rpages = arena_miscelm_to_rpages(miscelm); leadsize = ALIGNMENT_CEILING((uintptr_t)rpages, alignment) - (uintptr_t)rpages; assert(alloc_size >= leadsize + usize); trailsize = alloc_size - leadsize - usize - large_pad; if (leadsize != 0) { arena_chunk_map_misc_t *head_miscelm = miscelm; arena_run_t *head_run = run; miscelm = arena_miscelm_get(chunk, arena_miscelm_to_pageind(head_miscelm) + (leadsize >> LG_PAGE)); run = &miscelm->run; arena_run_trim_head(arena, chunk, head_run, alloc_size, alloc_size - leadsize); } if (trailsize != 0) { arena_run_trim_tail(arena, chunk, run, usize + large_pad + trailsize, usize + large_pad, false); } if (arena_run_init_large(arena, run, usize + large_pad, zero)) { size_t run_ind = arena_miscelm_to_pageind(arena_run_to_miscelm(run)); bool dirty = (arena_mapbits_dirty_get(chunk, run_ind) != 0); bool decommitted = (arena_mapbits_decommitted_get(chunk, run_ind) != 0); assert(decommitted); /* Cause of OOM. */ arena_run_dalloc(arena, run, dirty, false, decommitted); malloc_mutex_unlock(&arena->lock); return (NULL); } ret = arena_miscelm_to_rpages(miscelm); if (config_stats) { szind_t index = size2index(usize) - NBINS; arena->stats.nmalloc_large++; arena->stats.nrequests_large++; arena->stats.allocated_large += usize; arena->stats.lstats[index].nmalloc++; arena->stats.lstats[index].nrequests++; arena->stats.lstats[index].curruns++; } malloc_mutex_unlock(&arena->lock); if (config_fill && !zero) { if (unlikely(opt_junk_alloc)) memset(ret, 0xa5, usize); else if (unlikely(opt_zero)) memset(ret, 0, usize); } return (ret); } void * arena_palloc(tsd_t *tsd, arena_t *arena, size_t usize, size_t alignment, bool zero, tcache_t *tcache) { void *ret; if (usize <= SMALL_MAXCLASS && (alignment < PAGE || (alignment == PAGE && (usize & PAGE_MASK) == 0))) { /* Small; alignment doesn't require special run placement. */ ret = arena_malloc(tsd, arena, usize, zero, tcache); } else if (usize <= large_maxclass && alignment <= PAGE) { /* * Large; alignment doesn't require special run placement. * However, the cached pointer may be at a random offset from * the base of the run, so do some bit manipulation to retrieve * the base. */ ret = arena_malloc(tsd, arena, usize, zero, tcache); if (config_cache_oblivious) ret = (void *)((uintptr_t)ret & ~PAGE_MASK); } else { if (likely(usize <= large_maxclass)) { ret = arena_palloc_large(tsd, arena, usize, alignment, zero); } else if (likely(alignment <= chunksize)) ret = huge_malloc(tsd, arena, usize, zero, tcache); else { ret = huge_palloc(tsd, arena, usize, alignment, zero, tcache); } } return (ret); } void arena_prof_promoted(const void *ptr, size_t size) { arena_chunk_t *chunk; size_t pageind; szind_t binind; cassert(config_prof); assert(ptr != NULL); assert(CHUNK_ADDR2BASE(ptr) != ptr); assert(isalloc(ptr, false) == LARGE_MINCLASS); assert(isalloc(ptr, true) == LARGE_MINCLASS); assert(size <= SMALL_MAXCLASS); chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr); pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE; binind = size2index(size); assert(binind < NBINS); arena_mapbits_large_binind_set(chunk, pageind, binind); assert(isalloc(ptr, false) == LARGE_MINCLASS); assert(isalloc(ptr, true) == size); } static void arena_dissociate_bin_run(arena_chunk_t *chunk, arena_run_t *run, arena_bin_t *bin) { /* Dissociate run from bin. */ if (run == bin->runcur) bin->runcur = NULL; else { szind_t binind = arena_bin_index(extent_node_arena_get( &chunk->node), bin); arena_bin_info_t *bin_info = &arena_bin_info[binind]; if (bin_info->nregs != 1) { /* * This block's conditional is necessary because if the * run only contains one region, then it never gets * inserted into the non-full runs tree. */ arena_bin_runs_remove(bin, run); } } } static void arena_dalloc_bin_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, arena_bin_t *bin) { assert(run != bin->runcur); assert(arena_run_tree_search(&bin->runs, arena_run_to_miscelm(run)) == NULL); malloc_mutex_unlock(&bin->lock); /******************************/ malloc_mutex_lock(&arena->lock); arena_run_dalloc_decommit(arena, chunk, run); malloc_mutex_unlock(&arena->lock); /****************************/ malloc_mutex_lock(&bin->lock); if (config_stats) bin->stats.curruns--; } static void arena_bin_lower_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, arena_bin_t *bin) { /* * Make sure that if bin->runcur is non-NULL, it refers to the lowest * non-full run. It is okay to NULL runcur out rather than proactively * keeping it pointing at the lowest non-full run. */ if ((uintptr_t)run < (uintptr_t)bin->runcur) { /* Switch runcur. */ if (bin->runcur->nfree > 0) arena_bin_runs_insert(bin, bin->runcur); bin->runcur = run; if (config_stats) bin->stats.reruns++; } else arena_bin_runs_insert(bin, run); } static void arena_dalloc_bin_locked_impl(arena_t *arena, arena_chunk_t *chunk, void *ptr, arena_chunk_map_bits_t *bitselm, bool junked) { size_t pageind, rpages_ind; arena_run_t *run; arena_bin_t *bin; arena_bin_info_t *bin_info; szind_t binind; pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE; rpages_ind = pageind - arena_mapbits_small_runind_get(chunk, pageind); run = &arena_miscelm_get(chunk, rpages_ind)->run; binind = run->binind; bin = &arena->bins[binind]; bin_info = &arena_bin_info[binind]; if (!junked && config_fill && unlikely(opt_junk_free)) arena_dalloc_junk_small(ptr, bin_info); arena_run_reg_dalloc(run, ptr); if (run->nfree == bin_info->nregs) { arena_dissociate_bin_run(chunk, run, bin); arena_dalloc_bin_run(arena, chunk, run, bin); } else if (run->nfree == 1 && run != bin->runcur) arena_bin_lower_run(arena, chunk, run, bin); if (config_stats) { bin->stats.ndalloc++; bin->stats.curregs--; } } void arena_dalloc_bin_junked_locked(arena_t *arena, arena_chunk_t *chunk, void *ptr, arena_chunk_map_bits_t *bitselm) { arena_dalloc_bin_locked_impl(arena, chunk, ptr, bitselm, true); } void arena_dalloc_bin(arena_t *arena, arena_chunk_t *chunk, void *ptr, size_t pageind, arena_chunk_map_bits_t *bitselm) { arena_run_t *run; arena_bin_t *bin; size_t rpages_ind; rpages_ind = pageind - arena_mapbits_small_runind_get(chunk, pageind); run = &arena_miscelm_get(chunk, rpages_ind)->run; bin = &arena->bins[run->binind]; malloc_mutex_lock(&bin->lock); arena_dalloc_bin_locked_impl(arena, chunk, ptr, bitselm, false); malloc_mutex_unlock(&bin->lock); } void arena_dalloc_small(arena_t *arena, arena_chunk_t *chunk, void *ptr, size_t pageind) { arena_chunk_map_bits_t *bitselm; if (config_debug) { /* arena_ptr_small_binind_get() does extra sanity checking. */ assert(arena_ptr_small_binind_get(ptr, arena_mapbits_get(chunk, pageind)) != BININD_INVALID); } bitselm = arena_bitselm_get(chunk, pageind); arena_dalloc_bin(arena, chunk, ptr, pageind, bitselm); } #ifdef JEMALLOC_JET #undef arena_dalloc_junk_large #define arena_dalloc_junk_large JEMALLOC_N(arena_dalloc_junk_large_impl) #endif void arena_dalloc_junk_large(void *ptr, size_t usize) { if (config_fill && unlikely(opt_junk_free)) memset(ptr, 0x5a, usize); } #ifdef JEMALLOC_JET #undef arena_dalloc_junk_large #define arena_dalloc_junk_large JEMALLOC_N(arena_dalloc_junk_large) arena_dalloc_junk_large_t *arena_dalloc_junk_large = JEMALLOC_N(arena_dalloc_junk_large_impl); #endif static void arena_dalloc_large_locked_impl(arena_t *arena, arena_chunk_t *chunk, void *ptr, bool junked) { size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE; arena_chunk_map_misc_t *miscelm = arena_miscelm_get(chunk, pageind); arena_run_t *run = &miscelm->run; if (config_fill || config_stats) { size_t usize = arena_mapbits_large_size_get(chunk, pageind) - large_pad; if (!junked) arena_dalloc_junk_large(ptr, usize); if (config_stats) { szind_t index = size2index(usize) - NBINS; arena->stats.ndalloc_large++; arena->stats.allocated_large -= usize; arena->stats.lstats[index].ndalloc++; arena->stats.lstats[index].curruns--; } } arena_run_dalloc_decommit(arena, chunk, run); } void arena_dalloc_large_junked_locked(arena_t *arena, arena_chunk_t *chunk, void *ptr) { arena_dalloc_large_locked_impl(arena, chunk, ptr, true); } void arena_dalloc_large(arena_t *arena, arena_chunk_t *chunk, void *ptr) { malloc_mutex_lock(&arena->lock); arena_dalloc_large_locked_impl(arena, chunk, ptr, false); malloc_mutex_unlock(&arena->lock); } static void arena_ralloc_large_shrink(arena_t *arena, arena_chunk_t *chunk, void *ptr, size_t oldsize, size_t size) { size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE; arena_chunk_map_misc_t *miscelm = arena_miscelm_get(chunk, pageind); arena_run_t *run = &miscelm->run; assert(size < oldsize); /* * Shrink the run, and make trailing pages available for other * allocations. */ malloc_mutex_lock(&arena->lock); arena_run_trim_tail(arena, chunk, run, oldsize + large_pad, size + large_pad, true); if (config_stats) { szind_t oldindex = size2index(oldsize) - NBINS; szind_t index = size2index(size) - NBINS; arena->stats.ndalloc_large++; arena->stats.allocated_large -= oldsize; arena->stats.lstats[oldindex].ndalloc++; arena->stats.lstats[oldindex].curruns--; arena->stats.nmalloc_large++; arena->stats.nrequests_large++; arena->stats.allocated_large += size; arena->stats.lstats[index].nmalloc++; arena->stats.lstats[index].nrequests++; arena->stats.lstats[index].curruns++; } malloc_mutex_unlock(&arena->lock); } static bool arena_ralloc_large_grow(arena_t *arena, arena_chunk_t *chunk, void *ptr, size_t oldsize, size_t usize_min, size_t usize_max, bool zero) { size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE; size_t npages = (oldsize + large_pad) >> LG_PAGE; size_t followsize; assert(oldsize == arena_mapbits_large_size_get(chunk, pageind) - large_pad); /* Try to extend the run. */ malloc_mutex_lock(&arena->lock); if (pageind+npages >= chunk_npages || arena_mapbits_allocated_get(chunk, pageind+npages) != 0) goto label_fail; followsize = arena_mapbits_unallocated_size_get(chunk, pageind+npages); if (oldsize + followsize >= usize_min) { /* * The next run is available and sufficiently large. Split the * following run, then merge the first part with the existing * allocation. */ arena_run_t *run; size_t usize, splitsize, size, flag_dirty, flag_unzeroed_mask; usize = usize_max; while (oldsize + followsize < usize) usize = index2size(size2index(usize)-1); assert(usize >= usize_min); assert(usize >= oldsize); splitsize = usize - oldsize; if (splitsize == 0) goto label_fail; run = &arena_miscelm_get(chunk, pageind+npages)->run; if (arena_run_split_large(arena, run, splitsize, zero)) goto label_fail; + if (config_cache_oblivious && zero) { + /* + * Zero the trailing bytes of the original allocation's + * last page, since they are in an indeterminate state. + * There will always be trailing bytes, because ptr's + * offset from the beginning of the run is a multiple of + * CACHELINE in [0 .. PAGE). + */ + void *zbase = (void *)((uintptr_t)ptr + oldsize); + void *zpast = PAGE_ADDR2BASE((void *)((uintptr_t)zbase + + PAGE)); + size_t nzero = (uintptr_t)zpast - (uintptr_t)zbase; + assert(nzero > 0); + memset(zbase, 0, nzero); + } + size = oldsize + splitsize; npages = (size + large_pad) >> LG_PAGE; /* * Mark the extended run as dirty if either portion of the run * was dirty before allocation. This is rather pedantic, * because there's not actually any sequence of events that * could cause the resulting run to be passed to * arena_run_dalloc() with the dirty argument set to false * (which is when dirty flag consistency would really matter). */ flag_dirty = arena_mapbits_dirty_get(chunk, pageind) | arena_mapbits_dirty_get(chunk, pageind+npages-1); flag_unzeroed_mask = flag_dirty == 0 ? CHUNK_MAP_UNZEROED : 0; arena_mapbits_large_set(chunk, pageind, size + large_pad, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind))); arena_mapbits_large_set(chunk, pageind+npages-1, 0, flag_dirty | (flag_unzeroed_mask & arena_mapbits_unzeroed_get(chunk, pageind+npages-1))); if (config_stats) { szind_t oldindex = size2index(oldsize) - NBINS; szind_t index = size2index(size) - NBINS; arena->stats.ndalloc_large++; arena->stats.allocated_large -= oldsize; arena->stats.lstats[oldindex].ndalloc++; arena->stats.lstats[oldindex].curruns--; arena->stats.nmalloc_large++; arena->stats.nrequests_large++; arena->stats.allocated_large += size; arena->stats.lstats[index].nmalloc++; arena->stats.lstats[index].nrequests++; arena->stats.lstats[index].curruns++; } malloc_mutex_unlock(&arena->lock); return (false); } label_fail: malloc_mutex_unlock(&arena->lock); return (true); } #ifdef JEMALLOC_JET #undef arena_ralloc_junk_large #define arena_ralloc_junk_large JEMALLOC_N(arena_ralloc_junk_large_impl) #endif static void arena_ralloc_junk_large(void *ptr, size_t old_usize, size_t usize) { if (config_fill && unlikely(opt_junk_free)) { memset((void *)((uintptr_t)ptr + usize), 0x5a, old_usize - usize); } } #ifdef JEMALLOC_JET #undef arena_ralloc_junk_large #define arena_ralloc_junk_large JEMALLOC_N(arena_ralloc_junk_large) arena_ralloc_junk_large_t *arena_ralloc_junk_large = JEMALLOC_N(arena_ralloc_junk_large_impl); #endif /* * Try to resize a large allocation, in order to avoid copying. This will * always fail if growing an object, and the following run is already in use. */ static bool arena_ralloc_large(void *ptr, size_t oldsize, size_t usize_min, size_t usize_max, bool zero) { arena_chunk_t *chunk; arena_t *arena; if (oldsize == usize_max) { /* Current size class is compatible and maximal. */ return (false); } chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr); arena = extent_node_arena_get(&chunk->node); if (oldsize < usize_max) { bool ret = arena_ralloc_large_grow(arena, chunk, ptr, oldsize, usize_min, usize_max, zero); if (config_fill && !ret && !zero) { if (unlikely(opt_junk_alloc)) { memset((void *)((uintptr_t)ptr + oldsize), 0xa5, isalloc(ptr, config_prof) - oldsize); } else if (unlikely(opt_zero)) { memset((void *)((uintptr_t)ptr + oldsize), 0, isalloc(ptr, config_prof) - oldsize); } } return (ret); } assert(oldsize > usize_max); /* Fill before shrinking in order avoid a race. */ arena_ralloc_junk_large(ptr, oldsize, usize_max); arena_ralloc_large_shrink(arena, chunk, ptr, oldsize, usize_max); return (false); } bool arena_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra, bool zero) { size_t usize_min, usize_max; usize_min = s2u(size); usize_max = s2u(size + extra); if (likely(oldsize <= large_maxclass && usize_min <= large_maxclass)) { /* * Avoid moving the allocation if the size class can be left the * same. */ if (oldsize <= SMALL_MAXCLASS) { assert(arena_bin_info[size2index(oldsize)].reg_size == oldsize); if ((usize_max <= SMALL_MAXCLASS && size2index(usize_max) == size2index(oldsize)) || (size <= oldsize && usize_max >= oldsize)) return (false); } else { if (usize_max > SMALL_MAXCLASS) { if (!arena_ralloc_large(ptr, oldsize, usize_min, usize_max, zero)) return (false); } } /* Reallocation would require a move. */ return (true); } else { return (huge_ralloc_no_move(ptr, oldsize, usize_min, usize_max, zero)); } } static void * arena_ralloc_move_helper(tsd_t *tsd, arena_t *arena, size_t usize, size_t alignment, bool zero, tcache_t *tcache) { if (alignment == 0) return (arena_malloc(tsd, arena, usize, zero, tcache)); usize = sa2u(usize, alignment); if (usize == 0) return (NULL); return (ipalloct(tsd, usize, alignment, zero, tcache, arena)); } void * arena_ralloc(tsd_t *tsd, arena_t *arena, void *ptr, size_t oldsize, size_t size, size_t alignment, bool zero, tcache_t *tcache) { void *ret; size_t usize; usize = s2u(size); if (usize == 0) return (NULL); if (likely(usize <= large_maxclass)) { size_t copysize; /* Try to avoid moving the allocation. */ if (!arena_ralloc_no_move(ptr, oldsize, usize, 0, zero)) return (ptr); /* * size and oldsize are different enough that we need to move * the object. In that case, fall back to allocating new space * and copying. */ ret = arena_ralloc_move_helper(tsd, arena, usize, alignment, zero, tcache); if (ret == NULL) return (NULL); /* * Junk/zero-filling were already done by * ipalloc()/arena_malloc(). */ copysize = (usize < oldsize) ? usize : oldsize; JEMALLOC_VALGRIND_MAKE_MEM_UNDEFINED(ret, copysize); memcpy(ret, ptr, copysize); isqalloc(tsd, ptr, oldsize, tcache); } else { ret = huge_ralloc(tsd, arena, ptr, oldsize, usize, alignment, zero, tcache); } return (ret); } dss_prec_t arena_dss_prec_get(arena_t *arena) { dss_prec_t ret; malloc_mutex_lock(&arena->lock); ret = arena->dss_prec; malloc_mutex_unlock(&arena->lock); return (ret); } bool arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec) { if (!have_dss) return (dss_prec != dss_prec_disabled); malloc_mutex_lock(&arena->lock); arena->dss_prec = dss_prec; malloc_mutex_unlock(&arena->lock); return (false); } ssize_t arena_lg_dirty_mult_default_get(void) { return ((ssize_t)atomic_read_z((size_t *)&lg_dirty_mult_default)); } bool arena_lg_dirty_mult_default_set(ssize_t lg_dirty_mult) { if (!arena_lg_dirty_mult_valid(lg_dirty_mult)) return (true); atomic_write_z((size_t *)&lg_dirty_mult_default, (size_t)lg_dirty_mult); return (false); } void arena_stats_merge(arena_t *arena, const char **dss, ssize_t *lg_dirty_mult, size_t *nactive, size_t *ndirty, arena_stats_t *astats, malloc_bin_stats_t *bstats, malloc_large_stats_t *lstats, malloc_huge_stats_t *hstats) { unsigned i; malloc_mutex_lock(&arena->lock); *dss = dss_prec_names[arena->dss_prec]; *lg_dirty_mult = arena->lg_dirty_mult; *nactive += arena->nactive; *ndirty += arena->ndirty; astats->mapped += arena->stats.mapped; astats->npurge += arena->stats.npurge; astats->nmadvise += arena->stats.nmadvise; astats->purged += arena->stats.purged; astats->metadata_mapped += arena->stats.metadata_mapped; astats->metadata_allocated += arena_metadata_allocated_get(arena); astats->allocated_large += arena->stats.allocated_large; astats->nmalloc_large += arena->stats.nmalloc_large; astats->ndalloc_large += arena->stats.ndalloc_large; astats->nrequests_large += arena->stats.nrequests_large; astats->allocated_huge += arena->stats.allocated_huge; astats->nmalloc_huge += arena->stats.nmalloc_huge; astats->ndalloc_huge += arena->stats.ndalloc_huge; for (i = 0; i < nlclasses; i++) { lstats[i].nmalloc += arena->stats.lstats[i].nmalloc; lstats[i].ndalloc += arena->stats.lstats[i].ndalloc; lstats[i].nrequests += arena->stats.lstats[i].nrequests; lstats[i].curruns += arena->stats.lstats[i].curruns; } for (i = 0; i < nhclasses; i++) { hstats[i].nmalloc += arena->stats.hstats[i].nmalloc; hstats[i].ndalloc += arena->stats.hstats[i].ndalloc; hstats[i].curhchunks += arena->stats.hstats[i].curhchunks; } malloc_mutex_unlock(&arena->lock); for (i = 0; i < NBINS; i++) { arena_bin_t *bin = &arena->bins[i]; malloc_mutex_lock(&bin->lock); bstats[i].nmalloc += bin->stats.nmalloc; bstats[i].ndalloc += bin->stats.ndalloc; bstats[i].nrequests += bin->stats.nrequests; bstats[i].curregs += bin->stats.curregs; if (config_tcache) { bstats[i].nfills += bin->stats.nfills; bstats[i].nflushes += bin->stats.nflushes; } bstats[i].nruns += bin->stats.nruns; bstats[i].reruns += bin->stats.reruns; bstats[i].curruns += bin->stats.curruns; malloc_mutex_unlock(&bin->lock); } } arena_t * arena_new(unsigned ind) { arena_t *arena; unsigned i; arena_bin_t *bin; /* * Allocate arena, arena->lstats, and arena->hstats contiguously, mainly * because there is no way to clean up if base_alloc() OOMs. */ if (config_stats) { arena = (arena_t *)base_alloc(CACHELINE_CEILING(sizeof(arena_t)) + QUANTUM_CEILING(nlclasses * sizeof(malloc_large_stats_t) + nhclasses) * sizeof(malloc_huge_stats_t)); } else arena = (arena_t *)base_alloc(sizeof(arena_t)); if (arena == NULL) return (NULL); arena->ind = ind; arena->nthreads = 0; if (malloc_mutex_init(&arena->lock)) return (NULL); if (config_stats) { memset(&arena->stats, 0, sizeof(arena_stats_t)); arena->stats.lstats = (malloc_large_stats_t *)((uintptr_t)arena + CACHELINE_CEILING(sizeof(arena_t))); memset(arena->stats.lstats, 0, nlclasses * sizeof(malloc_large_stats_t)); arena->stats.hstats = (malloc_huge_stats_t *)((uintptr_t)arena + CACHELINE_CEILING(sizeof(arena_t)) + QUANTUM_CEILING(nlclasses * sizeof(malloc_large_stats_t))); memset(arena->stats.hstats, 0, nhclasses * sizeof(malloc_huge_stats_t)); if (config_tcache) ql_new(&arena->tcache_ql); } if (config_prof) arena->prof_accumbytes = 0; if (config_cache_oblivious) { /* * A nondeterministic seed based on the address of arena reduces * the likelihood of lockstep non-uniform cache index * utilization among identical concurrent processes, but at the * cost of test repeatability. For debug builds, instead use a * deterministic seed. */ arena->offset_state = config_debug ? ind : (uint64_t)(uintptr_t)arena; } arena->dss_prec = chunk_dss_prec_get(); arena->spare = NULL; arena->lg_dirty_mult = arena_lg_dirty_mult_default_get(); arena->purging = false; arena->nactive = 0; arena->ndirty = 0; arena_avail_tree_new(&arena->runs_avail); qr_new(&arena->runs_dirty, rd_link); qr_new(&arena->chunks_cache, cc_link); ql_new(&arena->huge); if (malloc_mutex_init(&arena->huge_mtx)) return (NULL); extent_tree_szad_new(&arena->chunks_szad_cached); extent_tree_ad_new(&arena->chunks_ad_cached); extent_tree_szad_new(&arena->chunks_szad_retained); extent_tree_ad_new(&arena->chunks_ad_retained); if (malloc_mutex_init(&arena->chunks_mtx)) return (NULL); ql_new(&arena->node_cache); if (malloc_mutex_init(&arena->node_cache_mtx)) return (NULL); arena->chunk_hooks = chunk_hooks_default; /* Initialize bins. */ for (i = 0; i < NBINS; i++) { bin = &arena->bins[i]; if (malloc_mutex_init(&bin->lock)) return (NULL); bin->runcur = NULL; arena_run_tree_new(&bin->runs); if (config_stats) memset(&bin->stats, 0, sizeof(malloc_bin_stats_t)); } return (arena); } /* * Calculate bin_info->run_size such that it meets the following constraints: * * *) bin_info->run_size <= arena_maxrun * *) bin_info->nregs <= RUN_MAXREGS * * bin_info->nregs and bin_info->reg0_offset are also calculated here, since * these settings are all interdependent. */ static void bin_info_run_size_calc(arena_bin_info_t *bin_info) { size_t pad_size; size_t try_run_size, perfect_run_size, actual_run_size; uint32_t try_nregs, perfect_nregs, actual_nregs; /* * Determine redzone size based on minimum alignment and minimum * redzone size. Add padding to the end of the run if it is needed to * align the regions. The padding allows each redzone to be half the * minimum alignment; without the padding, each redzone would have to * be twice as large in order to maintain alignment. */ if (config_fill && unlikely(opt_redzone)) { size_t align_min = ZU(1) << (jemalloc_ffs(bin_info->reg_size) - 1); if (align_min <= REDZONE_MINSIZE) { bin_info->redzone_size = REDZONE_MINSIZE; pad_size = 0; } else { bin_info->redzone_size = align_min >> 1; pad_size = bin_info->redzone_size; } } else { bin_info->redzone_size = 0; pad_size = 0; } bin_info->reg_interval = bin_info->reg_size + (bin_info->redzone_size << 1); /* * Compute run size under ideal conditions (no redzones, no limit on run * size). */ try_run_size = PAGE; try_nregs = try_run_size / bin_info->reg_size; do { perfect_run_size = try_run_size; perfect_nregs = try_nregs; try_run_size += PAGE; try_nregs = try_run_size / bin_info->reg_size; } while (perfect_run_size != perfect_nregs * bin_info->reg_size); assert(perfect_nregs <= RUN_MAXREGS); actual_run_size = perfect_run_size; actual_nregs = (actual_run_size - pad_size) / bin_info->reg_interval; /* * Redzones can require enough padding that not even a single region can * fit within the number of pages that would normally be dedicated to a * run for this size class. Increase the run size until at least one * region fits. */ while (actual_nregs == 0) { assert(config_fill && unlikely(opt_redzone)); actual_run_size += PAGE; actual_nregs = (actual_run_size - pad_size) / bin_info->reg_interval; } /* * Make sure that the run will fit within an arena chunk. */ while (actual_run_size > arena_maxrun) { actual_run_size -= PAGE; actual_nregs = (actual_run_size - pad_size) / bin_info->reg_interval; } assert(actual_nregs > 0); assert(actual_run_size == s2u(actual_run_size)); /* Copy final settings. */ bin_info->run_size = actual_run_size; bin_info->nregs = actual_nregs; bin_info->reg0_offset = actual_run_size - (actual_nregs * bin_info->reg_interval) - pad_size + bin_info->redzone_size; if (actual_run_size > small_maxrun) small_maxrun = actual_run_size; assert(bin_info->reg0_offset - bin_info->redzone_size + (bin_info->nregs * bin_info->reg_interval) + pad_size == bin_info->run_size); } static void bin_info_init(void) { arena_bin_info_t *bin_info; #define BIN_INFO_INIT_bin_yes(index, size) \ bin_info = &arena_bin_info[index]; \ bin_info->reg_size = size; \ bin_info_run_size_calc(bin_info); \ bitmap_info_init(&bin_info->bitmap_info, bin_info->nregs); #define BIN_INFO_INIT_bin_no(index, size) #define SC(index, lg_grp, lg_delta, ndelta, bin, lg_delta_lookup) \ BIN_INFO_INIT_bin_##bin(index, (ZU(1)<> LG_PAGE)); if (small_run_tab == NULL) return (true); #define TAB_INIT_bin_yes(index, size) { \ arena_bin_info_t *bin_info = &arena_bin_info[index]; \ small_run_tab[bin_info->run_size >> LG_PAGE] = true; \ } #define TAB_INIT_bin_no(index, size) #define SC(index, lg_grp, lg_delta, ndelta, bin, lg_delta_lookup) \ TAB_INIT_bin_##bin(index, (ZU(1)<= the result * from (2), and will always be correct. */ map_bias = 0; for (i = 0; i < 3; i++) { size_t header_size = offsetof(arena_chunk_t, map_bits) + ((sizeof(arena_chunk_map_bits_t) + sizeof(arena_chunk_map_misc_t)) * (chunk_npages-map_bias)); map_bias = (header_size + PAGE_MASK) >> LG_PAGE; } assert(map_bias > 0); map_misc_offset = offsetof(arena_chunk_t, map_bits) + sizeof(arena_chunk_map_bits_t) * (chunk_npages-map_bias); arena_maxrun = chunksize - (map_bias << LG_PAGE); assert(arena_maxrun > 0); large_maxclass = index2size(size2index(chunksize)-1); if (large_maxclass > arena_maxrun) { /* * For small chunk sizes it's possible for there to be fewer * non-header pages available than are necessary to serve the * size classes just below chunksize. */ large_maxclass = arena_maxrun; } assert(large_maxclass > 0); nlclasses = size2index(large_maxclass) - size2index(SMALL_MAXCLASS); nhclasses = NSIZES - nlclasses - NBINS; bin_info_init(); return (small_run_size_init()); } void arena_prefork(arena_t *arena) { unsigned i; malloc_mutex_prefork(&arena->lock); malloc_mutex_prefork(&arena->huge_mtx); malloc_mutex_prefork(&arena->chunks_mtx); malloc_mutex_prefork(&arena->node_cache_mtx); for (i = 0; i < NBINS; i++) malloc_mutex_prefork(&arena->bins[i].lock); } void arena_postfork_parent(arena_t *arena) { unsigned i; for (i = 0; i < NBINS; i++) malloc_mutex_postfork_parent(&arena->bins[i].lock); malloc_mutex_postfork_parent(&arena->node_cache_mtx); malloc_mutex_postfork_parent(&arena->chunks_mtx); malloc_mutex_postfork_parent(&arena->huge_mtx); malloc_mutex_postfork_parent(&arena->lock); } void arena_postfork_child(arena_t *arena) { unsigned i; for (i = 0; i < NBINS; i++) malloc_mutex_postfork_child(&arena->bins[i].lock); malloc_mutex_postfork_child(&arena->node_cache_mtx); malloc_mutex_postfork_child(&arena->chunks_mtx); malloc_mutex_postfork_child(&arena->huge_mtx); malloc_mutex_postfork_child(&arena->lock); } Index: head/contrib/jemalloc/src/huge.c =================================================================== --- head/contrib/jemalloc/src/huge.c (revision 289899) +++ head/contrib/jemalloc/src/huge.c (revision 289900) @@ -1,433 +1,435 @@ #define JEMALLOC_HUGE_C_ #include "jemalloc/internal/jemalloc_internal.h" /******************************************************************************/ static extent_node_t * huge_node_get(const void *ptr) { extent_node_t *node; node = chunk_lookup(ptr, true); assert(!extent_node_achunk_get(node)); return (node); } static bool huge_node_set(const void *ptr, extent_node_t *node) { assert(extent_node_addr_get(node) == ptr); assert(!extent_node_achunk_get(node)); return (chunk_register(ptr, node)); } static void huge_node_unset(const void *ptr, const extent_node_t *node) { chunk_deregister(ptr, node); } void * huge_malloc(tsd_t *tsd, arena_t *arena, size_t size, bool zero, tcache_t *tcache) { size_t usize; usize = s2u(size); if (usize == 0) { /* size_t overflow. */ return (NULL); } return (huge_palloc(tsd, arena, usize, chunksize, zero, tcache)); } void * huge_palloc(tsd_t *tsd, arena_t *arena, size_t size, size_t alignment, bool zero, tcache_t *tcache) { void *ret; size_t usize; extent_node_t *node; bool is_zeroed; /* Allocate one or more contiguous chunks for this request. */ usize = sa2u(size, alignment); if (unlikely(usize == 0)) return (NULL); assert(usize >= chunksize); /* Allocate an extent node with which to track the chunk. */ node = ipallocztm(tsd, CACHELINE_CEILING(sizeof(extent_node_t)), CACHELINE, false, tcache, true, arena); if (node == NULL) return (NULL); /* * Copy zero into is_zeroed and pass the copy to chunk_alloc(), so that * it is possible to make correct junk/zero fill decisions below. */ is_zeroed = zero; arena = arena_choose(tsd, arena); if (unlikely(arena == NULL) || (ret = arena_chunk_alloc_huge(arena, size, alignment, &is_zeroed)) == NULL) { idalloctm(tsd, node, tcache, true); return (NULL); } extent_node_init(node, arena, ret, size, is_zeroed, true); if (huge_node_set(ret, node)) { arena_chunk_dalloc_huge(arena, ret, size); idalloctm(tsd, node, tcache, true); return (NULL); } /* Insert node into huge. */ malloc_mutex_lock(&arena->huge_mtx); ql_elm_new(node, ql_link); ql_tail_insert(&arena->huge, node, ql_link); malloc_mutex_unlock(&arena->huge_mtx); if (zero || (config_fill && unlikely(opt_zero))) { if (!is_zeroed) memset(ret, 0, size); } else if (config_fill && unlikely(opt_junk_alloc)) memset(ret, 0xa5, size); return (ret); } #ifdef JEMALLOC_JET #undef huge_dalloc_junk #define huge_dalloc_junk JEMALLOC_N(huge_dalloc_junk_impl) #endif static void huge_dalloc_junk(void *ptr, size_t usize) { if (config_fill && have_dss && unlikely(opt_junk_free)) { /* * Only bother junk filling if the chunk isn't about to be * unmapped. */ if (!config_munmap || (have_dss && chunk_in_dss(ptr))) memset(ptr, 0x5a, usize); } } #ifdef JEMALLOC_JET #undef huge_dalloc_junk #define huge_dalloc_junk JEMALLOC_N(huge_dalloc_junk) huge_dalloc_junk_t *huge_dalloc_junk = JEMALLOC_N(huge_dalloc_junk_impl); #endif static void huge_ralloc_no_move_similar(void *ptr, size_t oldsize, size_t usize_min, size_t usize_max, bool zero) { size_t usize, usize_next; extent_node_t *node; arena_t *arena; chunk_hooks_t chunk_hooks = CHUNK_HOOKS_INITIALIZER; - bool zeroed; + bool pre_zeroed, post_zeroed; /* Increase usize to incorporate extra. */ for (usize = usize_min; usize < usize_max && (usize_next = s2u(usize+1)) <= oldsize; usize = usize_next) ; /* Do nothing. */ if (oldsize == usize) return; node = huge_node_get(ptr); arena = extent_node_arena_get(node); + pre_zeroed = extent_node_zeroed_get(node); /* Fill if necessary (shrinking). */ if (oldsize > usize) { size_t sdiff = oldsize - usize; if (config_fill && unlikely(opt_junk_free)) { memset((void *)((uintptr_t)ptr + usize), 0x5a, sdiff); - zeroed = false; + post_zeroed = false; } else { - zeroed = !chunk_purge_wrapper(arena, &chunk_hooks, ptr, - CHUNK_CEILING(oldsize), usize, sdiff); + post_zeroed = !chunk_purge_wrapper(arena, &chunk_hooks, + ptr, CHUNK_CEILING(oldsize), usize, sdiff); } } else - zeroed = true; + post_zeroed = pre_zeroed; malloc_mutex_lock(&arena->huge_mtx); /* Update the size of the huge allocation. */ assert(extent_node_size_get(node) != usize); extent_node_size_set(node, usize); - /* Clear node's zeroed field if zeroing failed above. */ - extent_node_zeroed_set(node, extent_node_zeroed_get(node) && zeroed); + /* Update zeroed. */ + extent_node_zeroed_set(node, post_zeroed); malloc_mutex_unlock(&arena->huge_mtx); arena_chunk_ralloc_huge_similar(arena, ptr, oldsize, usize); /* Fill if necessary (growing). */ if (oldsize < usize) { if (zero || (config_fill && unlikely(opt_zero))) { - if (!zeroed) { + if (!pre_zeroed) { memset((void *)((uintptr_t)ptr + oldsize), 0, usize - oldsize); } } else if (config_fill && unlikely(opt_junk_alloc)) { memset((void *)((uintptr_t)ptr + oldsize), 0xa5, usize - oldsize); } } } static bool huge_ralloc_no_move_shrink(void *ptr, size_t oldsize, size_t usize) { extent_node_t *node; arena_t *arena; chunk_hooks_t chunk_hooks; size_t cdiff; - bool zeroed; + bool pre_zeroed, post_zeroed; node = huge_node_get(ptr); arena = extent_node_arena_get(node); + pre_zeroed = extent_node_zeroed_get(node); chunk_hooks = chunk_hooks_get(arena); assert(oldsize > usize); /* Split excess chunks. */ cdiff = CHUNK_CEILING(oldsize) - CHUNK_CEILING(usize); if (cdiff != 0 && chunk_hooks.split(ptr, CHUNK_CEILING(oldsize), CHUNK_CEILING(usize), cdiff, true, arena->ind)) return (true); if (oldsize > usize) { size_t sdiff = oldsize - usize; if (config_fill && unlikely(opt_junk_free)) { huge_dalloc_junk((void *)((uintptr_t)ptr + usize), sdiff); - zeroed = false; + post_zeroed = false; } else { - zeroed = !chunk_purge_wrapper(arena, &chunk_hooks, + post_zeroed = !chunk_purge_wrapper(arena, &chunk_hooks, CHUNK_ADDR2BASE((uintptr_t)ptr + usize), CHUNK_CEILING(oldsize), CHUNK_ADDR2OFFSET((uintptr_t)ptr + usize), sdiff); } } else - zeroed = true; + post_zeroed = pre_zeroed; malloc_mutex_lock(&arena->huge_mtx); /* Update the size of the huge allocation. */ extent_node_size_set(node, usize); - /* Clear node's zeroed field if zeroing failed above. */ - extent_node_zeroed_set(node, extent_node_zeroed_get(node) && zeroed); + /* Update zeroed. */ + extent_node_zeroed_set(node, post_zeroed); malloc_mutex_unlock(&arena->huge_mtx); /* Zap the excess chunks. */ arena_chunk_ralloc_huge_shrink(arena, ptr, oldsize, usize); return (false); } static bool huge_ralloc_no_move_expand(void *ptr, size_t oldsize, size_t usize, bool zero) { extent_node_t *node; arena_t *arena; bool is_zeroed_subchunk, is_zeroed_chunk; node = huge_node_get(ptr); arena = extent_node_arena_get(node); malloc_mutex_lock(&arena->huge_mtx); is_zeroed_subchunk = extent_node_zeroed_get(node); malloc_mutex_unlock(&arena->huge_mtx); /* * Copy zero into is_zeroed_chunk and pass the copy to chunk_alloc(), so * that it is possible to make correct junk/zero fill decisions below. */ is_zeroed_chunk = zero; if (arena_chunk_ralloc_huge_expand(arena, ptr, oldsize, usize, &is_zeroed_chunk)) return (true); malloc_mutex_lock(&arena->huge_mtx); /* Update the size of the huge allocation. */ extent_node_size_set(node, usize); malloc_mutex_unlock(&arena->huge_mtx); if (zero || (config_fill && unlikely(opt_zero))) { if (!is_zeroed_subchunk) { memset((void *)((uintptr_t)ptr + oldsize), 0, CHUNK_CEILING(oldsize) - oldsize); } if (!is_zeroed_chunk) { memset((void *)((uintptr_t)ptr + CHUNK_CEILING(oldsize)), 0, usize - CHUNK_CEILING(oldsize)); } } else if (config_fill && unlikely(opt_junk_alloc)) { memset((void *)((uintptr_t)ptr + oldsize), 0xa5, usize - oldsize); } return (false); } bool huge_ralloc_no_move(void *ptr, size_t oldsize, size_t usize_min, size_t usize_max, bool zero) { assert(s2u(oldsize) == oldsize); /* Both allocations must be huge to avoid a move. */ if (oldsize < chunksize || usize_max < chunksize) return (true); if (CHUNK_CEILING(usize_max) > CHUNK_CEILING(oldsize)) { /* Attempt to expand the allocation in-place. */ if (!huge_ralloc_no_move_expand(ptr, oldsize, usize_max, zero)) return (false); /* Try again, this time with usize_min. */ if (usize_min < usize_max && CHUNK_CEILING(usize_min) > CHUNK_CEILING(oldsize) && huge_ralloc_no_move_expand(ptr, oldsize, usize_min, zero)) return (false); } /* * Avoid moving the allocation if the existing chunk size accommodates * the new size. */ if (CHUNK_CEILING(oldsize) >= CHUNK_CEILING(usize_min) && CHUNK_CEILING(oldsize) <= CHUNK_CEILING(usize_max)) { huge_ralloc_no_move_similar(ptr, oldsize, usize_min, usize_max, zero); return (false); } /* Attempt to shrink the allocation in-place. */ if (CHUNK_CEILING(oldsize) > CHUNK_CEILING(usize_max)) return (huge_ralloc_no_move_shrink(ptr, oldsize, usize_max)); return (true); } static void * huge_ralloc_move_helper(tsd_t *tsd, arena_t *arena, size_t usize, size_t alignment, bool zero, tcache_t *tcache) { if (alignment <= chunksize) return (huge_malloc(tsd, arena, usize, zero, tcache)); return (huge_palloc(tsd, arena, usize, alignment, zero, tcache)); } void * huge_ralloc(tsd_t *tsd, arena_t *arena, void *ptr, size_t oldsize, size_t usize, size_t alignment, bool zero, tcache_t *tcache) { void *ret; size_t copysize; /* Try to avoid moving the allocation. */ if (!huge_ralloc_no_move(ptr, oldsize, usize, usize, zero)) return (ptr); /* * usize and oldsize are different enough that we need to use a * different size class. In that case, fall back to allocating new * space and copying. */ ret = huge_ralloc_move_helper(tsd, arena, usize, alignment, zero, tcache); if (ret == NULL) return (NULL); copysize = (usize < oldsize) ? usize : oldsize; memcpy(ret, ptr, copysize); isqalloc(tsd, ptr, oldsize, tcache); return (ret); } void huge_dalloc(tsd_t *tsd, void *ptr, tcache_t *tcache) { extent_node_t *node; arena_t *arena; node = huge_node_get(ptr); arena = extent_node_arena_get(node); huge_node_unset(ptr, node); malloc_mutex_lock(&arena->huge_mtx); ql_remove(&arena->huge, node, ql_link); malloc_mutex_unlock(&arena->huge_mtx); huge_dalloc_junk(extent_node_addr_get(node), extent_node_size_get(node)); arena_chunk_dalloc_huge(extent_node_arena_get(node), extent_node_addr_get(node), extent_node_size_get(node)); idalloctm(tsd, node, tcache, true); } arena_t * huge_aalloc(const void *ptr) { return (extent_node_arena_get(huge_node_get(ptr))); } size_t huge_salloc(const void *ptr) { size_t size; extent_node_t *node; arena_t *arena; node = huge_node_get(ptr); arena = extent_node_arena_get(node); malloc_mutex_lock(&arena->huge_mtx); size = extent_node_size_get(node); malloc_mutex_unlock(&arena->huge_mtx); return (size); } prof_tctx_t * huge_prof_tctx_get(const void *ptr) { prof_tctx_t *tctx; extent_node_t *node; arena_t *arena; node = huge_node_get(ptr); arena = extent_node_arena_get(node); malloc_mutex_lock(&arena->huge_mtx); tctx = extent_node_prof_tctx_get(node); malloc_mutex_unlock(&arena->huge_mtx); return (tctx); } void huge_prof_tctx_set(const void *ptr, prof_tctx_t *tctx) { extent_node_t *node; arena_t *arena; node = huge_node_get(ptr); arena = extent_node_arena_get(node); malloc_mutex_lock(&arena->huge_mtx); extent_node_prof_tctx_set(node, tctx); malloc_mutex_unlock(&arena->huge_mtx); } void huge_prof_tctx_reset(const void *ptr) { huge_prof_tctx_set(ptr, (prof_tctx_t *)(uintptr_t)1U); } Index: head/contrib/jemalloc/src/prof.c =================================================================== --- head/contrib/jemalloc/src/prof.c (revision 289899) +++ head/contrib/jemalloc/src/prof.c (revision 289900) @@ -1,2225 +1,2237 @@ #define JEMALLOC_PROF_C_ #include "jemalloc/internal/jemalloc_internal.h" /******************************************************************************/ #ifdef JEMALLOC_PROF_LIBUNWIND #define UNW_LOCAL_ONLY #include #endif #ifdef JEMALLOC_PROF_LIBGCC #include #endif /******************************************************************************/ /* Data. */ bool opt_prof = false; bool opt_prof_active = true; bool opt_prof_thread_active_init = true; size_t opt_lg_prof_sample = LG_PROF_SAMPLE_DEFAULT; ssize_t opt_lg_prof_interval = LG_PROF_INTERVAL_DEFAULT; bool opt_prof_gdump = false; bool opt_prof_final = false; bool opt_prof_leak = false; bool opt_prof_accum = false; char opt_prof_prefix[ /* Minimize memory bloat for non-prof builds. */ #ifdef JEMALLOC_PROF PATH_MAX + #endif 1]; /* * Initialized as opt_prof_active, and accessed via * prof_active_[gs]et{_unlocked,}(). */ bool prof_active; static malloc_mutex_t prof_active_mtx; /* * Initialized as opt_prof_thread_active_init, and accessed via * prof_thread_active_init_[gs]et(). */ static bool prof_thread_active_init; static malloc_mutex_t prof_thread_active_init_mtx; /* * Initialized as opt_prof_gdump, and accessed via * prof_gdump_[gs]et{_unlocked,}(). */ bool prof_gdump_val; static malloc_mutex_t prof_gdump_mtx; uint64_t prof_interval = 0; size_t lg_prof_sample; /* * Table of mutexes that are shared among gctx's. These are leaf locks, so * there is no problem with using them for more than one gctx at the same time. * The primary motivation for this sharing though is that gctx's are ephemeral, * and destroying mutexes causes complications for systems that allocate when * creating/destroying mutexes. */ static malloc_mutex_t *gctx_locks; static unsigned cum_gctxs; /* Atomic counter. */ /* * Table of mutexes that are shared among tdata's. No operations require * holding multiple tdata locks, so there is no problem with using them for more * than one tdata at the same time, even though a gctx lock may be acquired * while holding a tdata lock. */ static malloc_mutex_t *tdata_locks; /* * Global hash of (prof_bt_t *)-->(prof_gctx_t *). This is the master data * structure that knows about all backtraces currently captured. */ static ckh_t bt2gctx; static malloc_mutex_t bt2gctx_mtx; /* * Tree of all extant prof_tdata_t structures, regardless of state, * {attached,detached,expired}. */ static prof_tdata_tree_t tdatas; static malloc_mutex_t tdatas_mtx; static uint64_t next_thr_uid; static malloc_mutex_t next_thr_uid_mtx; static malloc_mutex_t prof_dump_seq_mtx; static uint64_t prof_dump_seq; static uint64_t prof_dump_iseq; static uint64_t prof_dump_mseq; static uint64_t prof_dump_useq; /* * This buffer is rather large for stack allocation, so use a single buffer for * all profile dumps. */ static malloc_mutex_t prof_dump_mtx; static char prof_dump_buf[ /* Minimize memory bloat for non-prof builds. */ #ifdef JEMALLOC_PROF PROF_DUMP_BUFSIZE #else 1 #endif ]; static unsigned prof_dump_buf_end; static int prof_dump_fd; /* Do not dump any profiles until bootstrapping is complete. */ static bool prof_booted = false; /******************************************************************************/ /* * Function prototypes for static functions that are referenced prior to * definition. */ static bool prof_tctx_should_destroy(prof_tctx_t *tctx); static void prof_tctx_destroy(tsd_t *tsd, prof_tctx_t *tctx); static bool prof_tdata_should_destroy(prof_tdata_t *tdata, bool even_if_attached); static void prof_tdata_destroy(tsd_t *tsd, prof_tdata_t *tdata, bool even_if_attached); static char *prof_thread_name_alloc(tsd_t *tsd, const char *thread_name); /******************************************************************************/ /* Red-black trees. */ JEMALLOC_INLINE_C int prof_tctx_comp(const prof_tctx_t *a, const prof_tctx_t *b) { uint64_t a_thr_uid = a->thr_uid; uint64_t b_thr_uid = b->thr_uid; int ret = (a_thr_uid > b_thr_uid) - (a_thr_uid < b_thr_uid); if (ret == 0) { uint64_t a_thr_discrim = a->thr_discrim; uint64_t b_thr_discrim = b->thr_discrim; ret = (a_thr_discrim > b_thr_discrim) - (a_thr_discrim < b_thr_discrim); if (ret == 0) { uint64_t a_tctx_uid = a->tctx_uid; uint64_t b_tctx_uid = b->tctx_uid; ret = (a_tctx_uid > b_tctx_uid) - (a_tctx_uid < b_tctx_uid); } } return (ret); } rb_gen(static UNUSED, tctx_tree_, prof_tctx_tree_t, prof_tctx_t, tctx_link, prof_tctx_comp) JEMALLOC_INLINE_C int prof_gctx_comp(const prof_gctx_t *a, const prof_gctx_t *b) { unsigned a_len = a->bt.len; unsigned b_len = b->bt.len; unsigned comp_len = (a_len < b_len) ? a_len : b_len; int ret = memcmp(a->bt.vec, b->bt.vec, comp_len * sizeof(void *)); if (ret == 0) ret = (a_len > b_len) - (a_len < b_len); return (ret); } rb_gen(static UNUSED, gctx_tree_, prof_gctx_tree_t, prof_gctx_t, dump_link, prof_gctx_comp) JEMALLOC_INLINE_C int prof_tdata_comp(const prof_tdata_t *a, const prof_tdata_t *b) { int ret; uint64_t a_uid = a->thr_uid; uint64_t b_uid = b->thr_uid; ret = ((a_uid > b_uid) - (a_uid < b_uid)); if (ret == 0) { uint64_t a_discrim = a->thr_discrim; uint64_t b_discrim = b->thr_discrim; ret = ((a_discrim > b_discrim) - (a_discrim < b_discrim)); } return (ret); } rb_gen(static UNUSED, tdata_tree_, prof_tdata_tree_t, prof_tdata_t, tdata_link, prof_tdata_comp) /******************************************************************************/ void prof_alloc_rollback(tsd_t *tsd, prof_tctx_t *tctx, bool updated) { prof_tdata_t *tdata; cassert(config_prof); if (updated) { /* * Compute a new sample threshold. This isn't very important in * practice, because this function is rarely executed, so the * potential for sample bias is minimal except in contrived * programs. */ tdata = prof_tdata_get(tsd, true); if (tdata != NULL) prof_sample_threshold_update(tdata); } if ((uintptr_t)tctx > (uintptr_t)1U) { malloc_mutex_lock(tctx->tdata->lock); tctx->prepared = false; if (prof_tctx_should_destroy(tctx)) prof_tctx_destroy(tsd, tctx); else malloc_mutex_unlock(tctx->tdata->lock); } } void prof_malloc_sample_object(const void *ptr, size_t usize, prof_tctx_t *tctx) { prof_tctx_set(ptr, usize, tctx); malloc_mutex_lock(tctx->tdata->lock); tctx->cnts.curobjs++; tctx->cnts.curbytes += usize; if (opt_prof_accum) { tctx->cnts.accumobjs++; tctx->cnts.accumbytes += usize; } tctx->prepared = false; malloc_mutex_unlock(tctx->tdata->lock); } void prof_free_sampled_object(tsd_t *tsd, size_t usize, prof_tctx_t *tctx) { malloc_mutex_lock(tctx->tdata->lock); assert(tctx->cnts.curobjs > 0); assert(tctx->cnts.curbytes >= usize); tctx->cnts.curobjs--; tctx->cnts.curbytes -= usize; if (prof_tctx_should_destroy(tctx)) prof_tctx_destroy(tsd, tctx); else malloc_mutex_unlock(tctx->tdata->lock); } void bt_init(prof_bt_t *bt, void **vec) { cassert(config_prof); bt->vec = vec; bt->len = 0; } JEMALLOC_INLINE_C void prof_enter(tsd_t *tsd, prof_tdata_t *tdata) { cassert(config_prof); assert(tdata == prof_tdata_get(tsd, false)); if (tdata != NULL) { assert(!tdata->enq); tdata->enq = true; } malloc_mutex_lock(&bt2gctx_mtx); } JEMALLOC_INLINE_C void prof_leave(tsd_t *tsd, prof_tdata_t *tdata) { cassert(config_prof); assert(tdata == prof_tdata_get(tsd, false)); malloc_mutex_unlock(&bt2gctx_mtx); if (tdata != NULL) { bool idump, gdump; assert(tdata->enq); tdata->enq = false; idump = tdata->enq_idump; tdata->enq_idump = false; gdump = tdata->enq_gdump; tdata->enq_gdump = false; if (idump) prof_idump(); if (gdump) prof_gdump(); } } #ifdef JEMALLOC_PROF_LIBUNWIND void prof_backtrace(prof_bt_t *bt) { int nframes; cassert(config_prof); assert(bt->len == 0); assert(bt->vec != NULL); nframes = unw_backtrace(bt->vec, PROF_BT_MAX); if (nframes <= 0) return; bt->len = nframes; } #elif (defined(JEMALLOC_PROF_LIBGCC)) static _Unwind_Reason_Code prof_unwind_init_callback(struct _Unwind_Context *context, void *arg) { cassert(config_prof); return (_URC_NO_REASON); } static _Unwind_Reason_Code prof_unwind_callback(struct _Unwind_Context *context, void *arg) { prof_unwind_data_t *data = (prof_unwind_data_t *)arg; void *ip; cassert(config_prof); ip = (void *)_Unwind_GetIP(context); if (ip == NULL) return (_URC_END_OF_STACK); data->bt->vec[data->bt->len] = ip; data->bt->len++; if (data->bt->len == data->max) return (_URC_END_OF_STACK); return (_URC_NO_REASON); } void prof_backtrace(prof_bt_t *bt) { prof_unwind_data_t data = {bt, PROF_BT_MAX}; cassert(config_prof); _Unwind_Backtrace(prof_unwind_callback, &data); } #elif (defined(JEMALLOC_PROF_GCC)) void prof_backtrace(prof_bt_t *bt) { #define BT_FRAME(i) \ if ((i) < PROF_BT_MAX) { \ void *p; \ if (__builtin_frame_address(i) == 0) \ return; \ p = __builtin_return_address(i); \ if (p == NULL) \ return; \ bt->vec[(i)] = p; \ bt->len = (i) + 1; \ } else \ return; cassert(config_prof); BT_FRAME(0) BT_FRAME(1) BT_FRAME(2) BT_FRAME(3) BT_FRAME(4) BT_FRAME(5) BT_FRAME(6) BT_FRAME(7) BT_FRAME(8) BT_FRAME(9) BT_FRAME(10) BT_FRAME(11) BT_FRAME(12) BT_FRAME(13) BT_FRAME(14) BT_FRAME(15) BT_FRAME(16) BT_FRAME(17) BT_FRAME(18) BT_FRAME(19) BT_FRAME(20) BT_FRAME(21) BT_FRAME(22) BT_FRAME(23) BT_FRAME(24) BT_FRAME(25) BT_FRAME(26) BT_FRAME(27) BT_FRAME(28) BT_FRAME(29) BT_FRAME(30) BT_FRAME(31) BT_FRAME(32) BT_FRAME(33) BT_FRAME(34) BT_FRAME(35) BT_FRAME(36) BT_FRAME(37) BT_FRAME(38) BT_FRAME(39) BT_FRAME(40) BT_FRAME(41) BT_FRAME(42) BT_FRAME(43) BT_FRAME(44) BT_FRAME(45) BT_FRAME(46) BT_FRAME(47) BT_FRAME(48) BT_FRAME(49) BT_FRAME(50) BT_FRAME(51) BT_FRAME(52) BT_FRAME(53) BT_FRAME(54) BT_FRAME(55) BT_FRAME(56) BT_FRAME(57) BT_FRAME(58) BT_FRAME(59) BT_FRAME(60) BT_FRAME(61) BT_FRAME(62) BT_FRAME(63) BT_FRAME(64) BT_FRAME(65) BT_FRAME(66) BT_FRAME(67) BT_FRAME(68) BT_FRAME(69) BT_FRAME(70) BT_FRAME(71) BT_FRAME(72) BT_FRAME(73) BT_FRAME(74) BT_FRAME(75) BT_FRAME(76) BT_FRAME(77) BT_FRAME(78) BT_FRAME(79) BT_FRAME(80) BT_FRAME(81) BT_FRAME(82) BT_FRAME(83) BT_FRAME(84) BT_FRAME(85) BT_FRAME(86) BT_FRAME(87) BT_FRAME(88) BT_FRAME(89) BT_FRAME(90) BT_FRAME(91) BT_FRAME(92) BT_FRAME(93) BT_FRAME(94) BT_FRAME(95) BT_FRAME(96) BT_FRAME(97) BT_FRAME(98) BT_FRAME(99) BT_FRAME(100) BT_FRAME(101) BT_FRAME(102) BT_FRAME(103) BT_FRAME(104) BT_FRAME(105) BT_FRAME(106) BT_FRAME(107) BT_FRAME(108) BT_FRAME(109) BT_FRAME(110) BT_FRAME(111) BT_FRAME(112) BT_FRAME(113) BT_FRAME(114) BT_FRAME(115) BT_FRAME(116) BT_FRAME(117) BT_FRAME(118) BT_FRAME(119) BT_FRAME(120) BT_FRAME(121) BT_FRAME(122) BT_FRAME(123) BT_FRAME(124) BT_FRAME(125) BT_FRAME(126) BT_FRAME(127) #undef BT_FRAME } #else void prof_backtrace(prof_bt_t *bt) { cassert(config_prof); not_reached(); } #endif static malloc_mutex_t * prof_gctx_mutex_choose(void) { unsigned ngctxs = atomic_add_u(&cum_gctxs, 1); return (&gctx_locks[(ngctxs - 1) % PROF_NCTX_LOCKS]); } static malloc_mutex_t * prof_tdata_mutex_choose(uint64_t thr_uid) { return (&tdata_locks[thr_uid % PROF_NTDATA_LOCKS]); } static prof_gctx_t * prof_gctx_create(tsd_t *tsd, prof_bt_t *bt) { /* * Create a single allocation that has space for vec of length bt->len. */ prof_gctx_t *gctx = (prof_gctx_t *)iallocztm(tsd, offsetof(prof_gctx_t, vec) + (bt->len * sizeof(void *)), false, tcache_get(tsd, true), true, NULL); if (gctx == NULL) return (NULL); gctx->lock = prof_gctx_mutex_choose(); /* * Set nlimbo to 1, in order to avoid a race condition with * prof_tctx_destroy()/prof_gctx_try_destroy(). */ gctx->nlimbo = 1; tctx_tree_new(&gctx->tctxs); /* Duplicate bt. */ memcpy(gctx->vec, bt->vec, bt->len * sizeof(void *)); gctx->bt.vec = gctx->vec; gctx->bt.len = bt->len; return (gctx); } static void prof_gctx_try_destroy(tsd_t *tsd, prof_tdata_t *tdata_self, prof_gctx_t *gctx, prof_tdata_t *tdata) { cassert(config_prof); /* * Check that gctx is still unused by any thread cache before destroying * it. prof_lookup() increments gctx->nlimbo in order to avoid a race * condition with this function, as does prof_tctx_destroy() in order to * avoid a race between the main body of prof_tctx_destroy() and entry * into this function. */ prof_enter(tsd, tdata_self); malloc_mutex_lock(gctx->lock); assert(gctx->nlimbo != 0); if (tctx_tree_empty(&gctx->tctxs) && gctx->nlimbo == 1) { /* Remove gctx from bt2gctx. */ if (ckh_remove(tsd, &bt2gctx, &gctx->bt, NULL, NULL)) not_reached(); prof_leave(tsd, tdata_self); /* Destroy gctx. */ malloc_mutex_unlock(gctx->lock); idalloctm(tsd, gctx, tcache_get(tsd, false), true); } else { /* * Compensate for increment in prof_tctx_destroy() or * prof_lookup(). */ gctx->nlimbo--; malloc_mutex_unlock(gctx->lock); prof_leave(tsd, tdata_self); } } /* tctx->tdata->lock must be held. */ static bool prof_tctx_should_destroy(prof_tctx_t *tctx) { if (opt_prof_accum) return (false); if (tctx->cnts.curobjs != 0) return (false); if (tctx->prepared) return (false); return (true); } static bool prof_gctx_should_destroy(prof_gctx_t *gctx) { if (opt_prof_accum) return (false); if (!tctx_tree_empty(&gctx->tctxs)) return (false); if (gctx->nlimbo != 0) return (false); return (true); } /* tctx->tdata->lock is held upon entry, and released before return. */ static void prof_tctx_destroy(tsd_t *tsd, prof_tctx_t *tctx) { prof_tdata_t *tdata = tctx->tdata; prof_gctx_t *gctx = tctx->gctx; bool destroy_tdata, destroy_tctx, destroy_gctx; assert(tctx->cnts.curobjs == 0); assert(tctx->cnts.curbytes == 0); assert(!opt_prof_accum); assert(tctx->cnts.accumobjs == 0); assert(tctx->cnts.accumbytes == 0); ckh_remove(tsd, &tdata->bt2tctx, &gctx->bt, NULL, NULL); destroy_tdata = prof_tdata_should_destroy(tdata, false); malloc_mutex_unlock(tdata->lock); malloc_mutex_lock(gctx->lock); switch (tctx->state) { case prof_tctx_state_nominal: tctx_tree_remove(&gctx->tctxs, tctx); destroy_tctx = true; if (prof_gctx_should_destroy(gctx)) { /* * Increment gctx->nlimbo in order to keep another * thread from winning the race to destroy gctx while * this one has gctx->lock dropped. Without this, it * would be possible for another thread to: * * 1) Sample an allocation associated with gctx. * 2) Deallocate the sampled object. * 3) Successfully prof_gctx_try_destroy(gctx). * * The result would be that gctx no longer exists by the * time this thread accesses it in * prof_gctx_try_destroy(). */ gctx->nlimbo++; destroy_gctx = true; } else destroy_gctx = false; break; case prof_tctx_state_dumping: /* * A dumping thread needs tctx to remain valid until dumping * has finished. Change state such that the dumping thread will * complete destruction during a late dump iteration phase. */ tctx->state = prof_tctx_state_purgatory; destroy_tctx = false; destroy_gctx = false; break; default: not_reached(); destroy_tctx = false; destroy_gctx = false; } malloc_mutex_unlock(gctx->lock); if (destroy_gctx) { prof_gctx_try_destroy(tsd, prof_tdata_get(tsd, false), gctx, tdata); } if (destroy_tdata) prof_tdata_destroy(tsd, tdata, false); if (destroy_tctx) idalloctm(tsd, tctx, tcache_get(tsd, false), true); } static bool prof_lookup_global(tsd_t *tsd, prof_bt_t *bt, prof_tdata_t *tdata, void **p_btkey, prof_gctx_t **p_gctx, bool *p_new_gctx) { union { prof_gctx_t *p; void *v; } gctx; union { prof_bt_t *p; void *v; } btkey; bool new_gctx; prof_enter(tsd, tdata); if (ckh_search(&bt2gctx, bt, &btkey.v, &gctx.v)) { /* bt has never been seen before. Insert it. */ gctx.p = prof_gctx_create(tsd, bt); if (gctx.v == NULL) { prof_leave(tsd, tdata); return (true); } btkey.p = &gctx.p->bt; if (ckh_insert(tsd, &bt2gctx, btkey.v, gctx.v)) { /* OOM. */ prof_leave(tsd, tdata); idalloctm(tsd, gctx.v, tcache_get(tsd, false), true); return (true); } new_gctx = true; } else { /* * Increment nlimbo, in order to avoid a race condition with * prof_tctx_destroy()/prof_gctx_try_destroy(). */ malloc_mutex_lock(gctx.p->lock); gctx.p->nlimbo++; malloc_mutex_unlock(gctx.p->lock); new_gctx = false; } prof_leave(tsd, tdata); *p_btkey = btkey.v; *p_gctx = gctx.p; *p_new_gctx = new_gctx; return (false); } prof_tctx_t * prof_lookup(tsd_t *tsd, prof_bt_t *bt) { union { prof_tctx_t *p; void *v; } ret; prof_tdata_t *tdata; bool not_found; cassert(config_prof); tdata = prof_tdata_get(tsd, false); if (tdata == NULL) return (NULL); malloc_mutex_lock(tdata->lock); not_found = ckh_search(&tdata->bt2tctx, bt, NULL, &ret.v); if (!not_found) /* Note double negative! */ ret.p->prepared = true; malloc_mutex_unlock(tdata->lock); if (not_found) { tcache_t *tcache; void *btkey; prof_gctx_t *gctx; bool new_gctx, error; /* * This thread's cache lacks bt. Look for it in the global * cache. */ if (prof_lookup_global(tsd, bt, tdata, &btkey, &gctx, &new_gctx)) return (NULL); /* Link a prof_tctx_t into gctx for this thread. */ tcache = tcache_get(tsd, true); ret.v = iallocztm(tsd, sizeof(prof_tctx_t), false, tcache, true, NULL); if (ret.p == NULL) { if (new_gctx) prof_gctx_try_destroy(tsd, tdata, gctx, tdata); return (NULL); } ret.p->tdata = tdata; ret.p->thr_uid = tdata->thr_uid; ret.p->thr_discrim = tdata->thr_discrim; memset(&ret.p->cnts, 0, sizeof(prof_cnt_t)); ret.p->gctx = gctx; ret.p->tctx_uid = tdata->tctx_uid_next++; ret.p->prepared = true; ret.p->state = prof_tctx_state_initializing; malloc_mutex_lock(tdata->lock); error = ckh_insert(tsd, &tdata->bt2tctx, btkey, ret.v); malloc_mutex_unlock(tdata->lock); if (error) { if (new_gctx) prof_gctx_try_destroy(tsd, tdata, gctx, tdata); idalloctm(tsd, ret.v, tcache, true); return (NULL); } malloc_mutex_lock(gctx->lock); ret.p->state = prof_tctx_state_nominal; tctx_tree_insert(&gctx->tctxs, ret.p); gctx->nlimbo--; malloc_mutex_unlock(gctx->lock); } return (ret.p); } void prof_sample_threshold_update(prof_tdata_t *tdata) { /* * The body of this function is compiled out unless heap profiling is * enabled, so that it is possible to compile jemalloc with floating * point support completely disabled. Avoiding floating point code is * important on memory-constrained systems, but it also enables a * workaround for versions of glibc that don't properly save/restore * floating point registers during dynamic lazy symbol loading (which * internally calls into whatever malloc implementation happens to be * integrated into the application). Note that some compilers (e.g. * gcc 4.8) may use floating point registers for fast memory moves, so * jemalloc must be compiled with such optimizations disabled (e.g. * -mno-sse) in order for the workaround to be complete. */ #ifdef JEMALLOC_PROF uint64_t r; double u; if (!config_prof) return; if (lg_prof_sample == 0) { tdata->bytes_until_sample = 0; return; } /* * Compute sample interval as a geometrically distributed random * variable with mean (2^lg_prof_sample). * * __ __ * | log(u) | 1 * tdata->bytes_until_sample = | -------- |, where p = --------------- * | log(1-p) | lg_prof_sample * 2 * * For more information on the math, see: * * Non-Uniform Random Variate Generation * Luc Devroye * Springer-Verlag, New York, 1986 * pp 500 * (http://luc.devroye.org/rnbookindex.html) */ prng64(r, 53, tdata->prng_state, UINT64_C(6364136223846793005), UINT64_C(1442695040888963407)); u = (double)r * (1.0/9007199254740992.0L); tdata->bytes_until_sample = (uint64_t)(log(u) / log(1.0 - (1.0 / (double)((uint64_t)1U << lg_prof_sample)))) + (uint64_t)1U; #endif } #ifdef JEMALLOC_JET static prof_tdata_t * prof_tdata_count_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata, void *arg) { size_t *tdata_count = (size_t *)arg; (*tdata_count)++; return (NULL); } size_t prof_tdata_count(void) { size_t tdata_count = 0; malloc_mutex_lock(&tdatas_mtx); tdata_tree_iter(&tdatas, NULL, prof_tdata_count_iter, (void *)&tdata_count); malloc_mutex_unlock(&tdatas_mtx); return (tdata_count); } #endif #ifdef JEMALLOC_JET size_t prof_bt_count(void) { size_t bt_count; tsd_t *tsd; prof_tdata_t *tdata; tsd = tsd_fetch(); tdata = prof_tdata_get(tsd, false); if (tdata == NULL) return (0); malloc_mutex_lock(&bt2gctx_mtx); bt_count = ckh_count(&bt2gctx); malloc_mutex_unlock(&bt2gctx_mtx); return (bt_count); } #endif #ifdef JEMALLOC_JET #undef prof_dump_open #define prof_dump_open JEMALLOC_N(prof_dump_open_impl) #endif static int prof_dump_open(bool propagate_err, const char *filename) { int fd; fd = creat(filename, 0644); if (fd == -1 && !propagate_err) { malloc_printf(": creat(\"%s\"), 0644) failed\n", filename); if (opt_abort) abort(); } return (fd); } #ifdef JEMALLOC_JET #undef prof_dump_open #define prof_dump_open JEMALLOC_N(prof_dump_open) prof_dump_open_t *prof_dump_open = JEMALLOC_N(prof_dump_open_impl); #endif static bool prof_dump_flush(bool propagate_err) { bool ret = false; ssize_t err; cassert(config_prof); err = write(prof_dump_fd, prof_dump_buf, prof_dump_buf_end); if (err == -1) { if (!propagate_err) { malloc_write(": write() failed during heap " "profile flush\n"); if (opt_abort) abort(); } ret = true; } prof_dump_buf_end = 0; return (ret); } static bool prof_dump_close(bool propagate_err) { bool ret; assert(prof_dump_fd != -1); ret = prof_dump_flush(propagate_err); close(prof_dump_fd); prof_dump_fd = -1; return (ret); } static bool prof_dump_write(bool propagate_err, const char *s) { unsigned i, slen, n; cassert(config_prof); i = 0; slen = strlen(s); while (i < slen) { /* Flush the buffer if it is full. */ if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) if (prof_dump_flush(propagate_err) && propagate_err) return (true); if (prof_dump_buf_end + slen <= PROF_DUMP_BUFSIZE) { /* Finish writing. */ n = slen - i; } else { /* Write as much of s as will fit. */ n = PROF_DUMP_BUFSIZE - prof_dump_buf_end; } memcpy(&prof_dump_buf[prof_dump_buf_end], &s[i], n); prof_dump_buf_end += n; i += n; } return (false); } JEMALLOC_FORMAT_PRINTF(2, 3) static bool prof_dump_printf(bool propagate_err, const char *format, ...) { bool ret; va_list ap; char buf[PROF_PRINTF_BUFSIZE]; va_start(ap, format); malloc_vsnprintf(buf, sizeof(buf), format, ap); va_end(ap); ret = prof_dump_write(propagate_err, buf); return (ret); } /* tctx->tdata->lock is held. */ static void prof_tctx_merge_tdata(prof_tctx_t *tctx, prof_tdata_t *tdata) { malloc_mutex_lock(tctx->gctx->lock); switch (tctx->state) { case prof_tctx_state_initializing: malloc_mutex_unlock(tctx->gctx->lock); return; case prof_tctx_state_nominal: tctx->state = prof_tctx_state_dumping; malloc_mutex_unlock(tctx->gctx->lock); memcpy(&tctx->dump_cnts, &tctx->cnts, sizeof(prof_cnt_t)); tdata->cnt_summed.curobjs += tctx->dump_cnts.curobjs; tdata->cnt_summed.curbytes += tctx->dump_cnts.curbytes; if (opt_prof_accum) { tdata->cnt_summed.accumobjs += tctx->dump_cnts.accumobjs; tdata->cnt_summed.accumbytes += tctx->dump_cnts.accumbytes; } break; case prof_tctx_state_dumping: case prof_tctx_state_purgatory: not_reached(); } } /* gctx->lock is held. */ static void prof_tctx_merge_gctx(prof_tctx_t *tctx, prof_gctx_t *gctx) { gctx->cnt_summed.curobjs += tctx->dump_cnts.curobjs; gctx->cnt_summed.curbytes += tctx->dump_cnts.curbytes; if (opt_prof_accum) { gctx->cnt_summed.accumobjs += tctx->dump_cnts.accumobjs; gctx->cnt_summed.accumbytes += tctx->dump_cnts.accumbytes; } } /* tctx->gctx is held. */ static prof_tctx_t * prof_tctx_merge_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *arg) { switch (tctx->state) { case prof_tctx_state_nominal: /* New since dumping started; ignore. */ break; case prof_tctx_state_dumping: case prof_tctx_state_purgatory: prof_tctx_merge_gctx(tctx, tctx->gctx); break; default: not_reached(); } return (NULL); } /* gctx->lock is held. */ static prof_tctx_t * prof_tctx_dump_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *arg) { bool propagate_err = *(bool *)arg; - if (prof_dump_printf(propagate_err, - " t%"FMTu64": %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]\n", - tctx->thr_uid, tctx->dump_cnts.curobjs, tctx->dump_cnts.curbytes, - tctx->dump_cnts.accumobjs, tctx->dump_cnts.accumbytes)) - return (tctx); + switch (tctx->state) { + case prof_tctx_state_initializing: + case prof_tctx_state_nominal: + /* Not captured by this dump. */ + break; + case prof_tctx_state_dumping: + case prof_tctx_state_purgatory: + if (prof_dump_printf(propagate_err, + " t%"FMTu64": %"FMTu64": %"FMTu64" [%"FMTu64": " + "%"FMTu64"]\n", tctx->thr_uid, tctx->dump_cnts.curobjs, + tctx->dump_cnts.curbytes, tctx->dump_cnts.accumobjs, + tctx->dump_cnts.accumbytes)) + return (tctx); + break; + default: + not_reached(); + } return (NULL); } /* tctx->gctx is held. */ static prof_tctx_t * prof_tctx_finish_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *arg) { prof_tctx_t *ret; switch (tctx->state) { case prof_tctx_state_nominal: /* New since dumping started; ignore. */ break; case prof_tctx_state_dumping: tctx->state = prof_tctx_state_nominal; break; case prof_tctx_state_purgatory: ret = tctx; goto label_return; default: not_reached(); } ret = NULL; label_return: return (ret); } static void prof_dump_gctx_prep(prof_gctx_t *gctx, prof_gctx_tree_t *gctxs) { cassert(config_prof); malloc_mutex_lock(gctx->lock); /* * Increment nlimbo so that gctx won't go away before dump. * Additionally, link gctx into the dump list so that it is included in * prof_dump()'s second pass. */ gctx->nlimbo++; gctx_tree_insert(gctxs, gctx); memset(&gctx->cnt_summed, 0, sizeof(prof_cnt_t)); malloc_mutex_unlock(gctx->lock); } static prof_gctx_t * prof_gctx_merge_iter(prof_gctx_tree_t *gctxs, prof_gctx_t *gctx, void *arg) { size_t *leak_ngctx = (size_t *)arg; malloc_mutex_lock(gctx->lock); tctx_tree_iter(&gctx->tctxs, NULL, prof_tctx_merge_iter, NULL); if (gctx->cnt_summed.curobjs != 0) (*leak_ngctx)++; malloc_mutex_unlock(gctx->lock); return (NULL); } static void prof_gctx_finish(tsd_t *tsd, prof_gctx_tree_t *gctxs) { prof_tdata_t *tdata = prof_tdata_get(tsd, false); prof_gctx_t *gctx; /* * Standard tree iteration won't work here, because as soon as we * decrement gctx->nlimbo and unlock gctx, another thread can * concurrently destroy it, which will corrupt the tree. Therefore, * tear down the tree one node at a time during iteration. */ while ((gctx = gctx_tree_first(gctxs)) != NULL) { gctx_tree_remove(gctxs, gctx); malloc_mutex_lock(gctx->lock); { prof_tctx_t *next; next = NULL; do { prof_tctx_t *to_destroy = tctx_tree_iter(&gctx->tctxs, next, prof_tctx_finish_iter, NULL); if (to_destroy != NULL) { next = tctx_tree_next(&gctx->tctxs, to_destroy); tctx_tree_remove(&gctx->tctxs, to_destroy); idalloctm(tsd, to_destroy, tcache_get(tsd, false), true); } else next = NULL; } while (next != NULL); } gctx->nlimbo--; if (prof_gctx_should_destroy(gctx)) { gctx->nlimbo++; malloc_mutex_unlock(gctx->lock); prof_gctx_try_destroy(tsd, tdata, gctx, tdata); } else malloc_mutex_unlock(gctx->lock); } } static prof_tdata_t * prof_tdata_merge_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata, void *arg) { prof_cnt_t *cnt_all = (prof_cnt_t *)arg; malloc_mutex_lock(tdata->lock); if (!tdata->expired) { size_t tabind; union { prof_tctx_t *p; void *v; } tctx; tdata->dumping = true; memset(&tdata->cnt_summed, 0, sizeof(prof_cnt_t)); for (tabind = 0; !ckh_iter(&tdata->bt2tctx, &tabind, NULL, &tctx.v);) prof_tctx_merge_tdata(tctx.p, tdata); cnt_all->curobjs += tdata->cnt_summed.curobjs; cnt_all->curbytes += tdata->cnt_summed.curbytes; if (opt_prof_accum) { cnt_all->accumobjs += tdata->cnt_summed.accumobjs; cnt_all->accumbytes += tdata->cnt_summed.accumbytes; } } else tdata->dumping = false; malloc_mutex_unlock(tdata->lock); return (NULL); } static prof_tdata_t * prof_tdata_dump_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata, void *arg) { bool propagate_err = *(bool *)arg; if (!tdata->dumping) return (NULL); if (prof_dump_printf(propagate_err, " t%"FMTu64": %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]%s%s\n", tdata->thr_uid, tdata->cnt_summed.curobjs, tdata->cnt_summed.curbytes, tdata->cnt_summed.accumobjs, tdata->cnt_summed.accumbytes, (tdata->thread_name != NULL) ? " " : "", (tdata->thread_name != NULL) ? tdata->thread_name : "")) return (tdata); return (NULL); } #ifdef JEMALLOC_JET #undef prof_dump_header #define prof_dump_header JEMALLOC_N(prof_dump_header_impl) #endif static bool prof_dump_header(bool propagate_err, const prof_cnt_t *cnt_all) { bool ret; if (prof_dump_printf(propagate_err, "heap_v2/%"FMTu64"\n" " t*: %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]\n", ((uint64_t)1U << lg_prof_sample), cnt_all->curobjs, cnt_all->curbytes, cnt_all->accumobjs, cnt_all->accumbytes)) return (true); malloc_mutex_lock(&tdatas_mtx); ret = (tdata_tree_iter(&tdatas, NULL, prof_tdata_dump_iter, (void *)&propagate_err) != NULL); malloc_mutex_unlock(&tdatas_mtx); return (ret); } #ifdef JEMALLOC_JET #undef prof_dump_header #define prof_dump_header JEMALLOC_N(prof_dump_header) prof_dump_header_t *prof_dump_header = JEMALLOC_N(prof_dump_header_impl); #endif /* gctx->lock is held. */ static bool prof_dump_gctx(bool propagate_err, prof_gctx_t *gctx, const prof_bt_t *bt, prof_gctx_tree_t *gctxs) { bool ret; unsigned i; cassert(config_prof); /* Avoid dumping such gctx's that have no useful data. */ if ((!opt_prof_accum && gctx->cnt_summed.curobjs == 0) || (opt_prof_accum && gctx->cnt_summed.accumobjs == 0)) { assert(gctx->cnt_summed.curobjs == 0); assert(gctx->cnt_summed.curbytes == 0); assert(gctx->cnt_summed.accumobjs == 0); assert(gctx->cnt_summed.accumbytes == 0); ret = false; goto label_return; } if (prof_dump_printf(propagate_err, "@")) { ret = true; goto label_return; } for (i = 0; i < bt->len; i++) { if (prof_dump_printf(propagate_err, " %#"FMTxPTR, (uintptr_t)bt->vec[i])) { ret = true; goto label_return; } } if (prof_dump_printf(propagate_err, "\n" " t*: %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]\n", gctx->cnt_summed.curobjs, gctx->cnt_summed.curbytes, gctx->cnt_summed.accumobjs, gctx->cnt_summed.accumbytes)) { ret = true; goto label_return; } if (tctx_tree_iter(&gctx->tctxs, NULL, prof_tctx_dump_iter, (void *)&propagate_err) != NULL) { ret = true; goto label_return; } ret = false; label_return: return (ret); } JEMALLOC_FORMAT_PRINTF(1, 2) static int prof_open_maps(const char *format, ...) { int mfd; va_list ap; char filename[PATH_MAX + 1]; va_start(ap, format); malloc_vsnprintf(filename, sizeof(filename), format, ap); va_end(ap); mfd = open(filename, O_RDONLY); return (mfd); } static bool prof_dump_maps(bool propagate_err) { bool ret; int mfd; cassert(config_prof); #ifdef __FreeBSD__ mfd = prof_open_maps("/proc/curproc/map"); #else { int pid = getpid(); mfd = prof_open_maps("/proc/%d/task/%d/maps", pid, pid); if (mfd == -1) mfd = prof_open_maps("/proc/%d/maps", pid); } #endif if (mfd != -1) { ssize_t nread; if (prof_dump_write(propagate_err, "\nMAPPED_LIBRARIES:\n") && propagate_err) { ret = true; goto label_return; } nread = 0; do { prof_dump_buf_end += nread; if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) { /* Make space in prof_dump_buf before read(). */ if (prof_dump_flush(propagate_err) && propagate_err) { ret = true; goto label_return; } } nread = read(mfd, &prof_dump_buf[prof_dump_buf_end], PROF_DUMP_BUFSIZE - prof_dump_buf_end); } while (nread > 0); } else { ret = true; goto label_return; } ret = false; label_return: if (mfd != -1) close(mfd); return (ret); } static void prof_leakcheck(const prof_cnt_t *cnt_all, size_t leak_ngctx, const char *filename) { if (cnt_all->curbytes != 0) { malloc_printf(": Leak summary: %"FMTu64" byte%s, %" FMTu64" object%s, %zu context%s\n", cnt_all->curbytes, (cnt_all->curbytes != 1) ? "s" : "", cnt_all->curobjs, (cnt_all->curobjs != 1) ? "s" : "", leak_ngctx, (leak_ngctx != 1) ? "s" : ""); malloc_printf( ": Run jeprof on \"%s\" for leak detail\n", filename); } } static prof_gctx_t * prof_gctx_dump_iter(prof_gctx_tree_t *gctxs, prof_gctx_t *gctx, void *arg) { prof_gctx_t *ret; bool propagate_err = *(bool *)arg; malloc_mutex_lock(gctx->lock); if (prof_dump_gctx(propagate_err, gctx, &gctx->bt, gctxs)) { ret = gctx; goto label_return; } ret = NULL; label_return: malloc_mutex_unlock(gctx->lock); return (ret); } static bool prof_dump(tsd_t *tsd, bool propagate_err, const char *filename, bool leakcheck) { prof_tdata_t *tdata; prof_cnt_t cnt_all; size_t tabind; union { prof_gctx_t *p; void *v; } gctx; size_t leak_ngctx; prof_gctx_tree_t gctxs; cassert(config_prof); tdata = prof_tdata_get(tsd, true); if (tdata == NULL) return (true); malloc_mutex_lock(&prof_dump_mtx); prof_enter(tsd, tdata); /* * Put gctx's in limbo and clear their counters in preparation for * summing. */ gctx_tree_new(&gctxs); for (tabind = 0; !ckh_iter(&bt2gctx, &tabind, NULL, &gctx.v);) prof_dump_gctx_prep(gctx.p, &gctxs); /* * Iterate over tdatas, and for the non-expired ones snapshot their tctx * stats and merge them into the associated gctx's. */ memset(&cnt_all, 0, sizeof(prof_cnt_t)); malloc_mutex_lock(&tdatas_mtx); tdata_tree_iter(&tdatas, NULL, prof_tdata_merge_iter, (void *)&cnt_all); malloc_mutex_unlock(&tdatas_mtx); /* Merge tctx stats into gctx's. */ leak_ngctx = 0; gctx_tree_iter(&gctxs, NULL, prof_gctx_merge_iter, (void *)&leak_ngctx); prof_leave(tsd, tdata); /* Create dump file. */ if ((prof_dump_fd = prof_dump_open(propagate_err, filename)) == -1) goto label_open_close_error; /* Dump profile header. */ if (prof_dump_header(propagate_err, &cnt_all)) goto label_write_error; /* Dump per gctx profile stats. */ if (gctx_tree_iter(&gctxs, NULL, prof_gctx_dump_iter, (void *)&propagate_err) != NULL) goto label_write_error; /* Dump /proc//maps if possible. */ if (prof_dump_maps(propagate_err)) goto label_write_error; if (prof_dump_close(propagate_err)) goto label_open_close_error; prof_gctx_finish(tsd, &gctxs); malloc_mutex_unlock(&prof_dump_mtx); if (leakcheck) prof_leakcheck(&cnt_all, leak_ngctx, filename); return (false); label_write_error: prof_dump_close(propagate_err); label_open_close_error: prof_gctx_finish(tsd, &gctxs); malloc_mutex_unlock(&prof_dump_mtx); return (true); } #define DUMP_FILENAME_BUFSIZE (PATH_MAX + 1) #define VSEQ_INVALID UINT64_C(0xffffffffffffffff) static void prof_dump_filename(char *filename, char v, uint64_t vseq) { cassert(config_prof); if (vseq != VSEQ_INVALID) { /* "...v.heap" */ malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE, "%s.%d.%"FMTu64".%c%"FMTu64".heap", opt_prof_prefix, (int)getpid(), prof_dump_seq, v, vseq); } else { /* "....heap" */ malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE, "%s.%d.%"FMTu64".%c.heap", opt_prof_prefix, (int)getpid(), prof_dump_seq, v); } prof_dump_seq++; } static void prof_fdump(void) { tsd_t *tsd; char filename[DUMP_FILENAME_BUFSIZE]; cassert(config_prof); assert(opt_prof_final); assert(opt_prof_prefix[0] != '\0'); if (!prof_booted) return; tsd = tsd_fetch(); malloc_mutex_lock(&prof_dump_seq_mtx); prof_dump_filename(filename, 'f', VSEQ_INVALID); malloc_mutex_unlock(&prof_dump_seq_mtx); prof_dump(tsd, false, filename, opt_prof_leak); } void prof_idump(void) { tsd_t *tsd; prof_tdata_t *tdata; cassert(config_prof); if (!prof_booted) return; tsd = tsd_fetch(); tdata = prof_tdata_get(tsd, false); if (tdata == NULL) return; if (tdata->enq) { tdata->enq_idump = true; return; } if (opt_prof_prefix[0] != '\0') { char filename[PATH_MAX + 1]; malloc_mutex_lock(&prof_dump_seq_mtx); prof_dump_filename(filename, 'i', prof_dump_iseq); prof_dump_iseq++; malloc_mutex_unlock(&prof_dump_seq_mtx); prof_dump(tsd, false, filename, false); } } bool prof_mdump(const char *filename) { tsd_t *tsd; char filename_buf[DUMP_FILENAME_BUFSIZE]; cassert(config_prof); if (!opt_prof || !prof_booted) return (true); tsd = tsd_fetch(); if (filename == NULL) { /* No filename specified, so automatically generate one. */ if (opt_prof_prefix[0] == '\0') return (true); malloc_mutex_lock(&prof_dump_seq_mtx); prof_dump_filename(filename_buf, 'm', prof_dump_mseq); prof_dump_mseq++; malloc_mutex_unlock(&prof_dump_seq_mtx); filename = filename_buf; } return (prof_dump(tsd, true, filename, false)); } void prof_gdump(void) { tsd_t *tsd; prof_tdata_t *tdata; cassert(config_prof); if (!prof_booted) return; tsd = tsd_fetch(); tdata = prof_tdata_get(tsd, false); if (tdata == NULL) return; if (tdata->enq) { tdata->enq_gdump = true; return; } if (opt_prof_prefix[0] != '\0') { char filename[DUMP_FILENAME_BUFSIZE]; malloc_mutex_lock(&prof_dump_seq_mtx); prof_dump_filename(filename, 'u', prof_dump_useq); prof_dump_useq++; malloc_mutex_unlock(&prof_dump_seq_mtx); prof_dump(tsd, false, filename, false); } } static void prof_bt_hash(const void *key, size_t r_hash[2]) { prof_bt_t *bt = (prof_bt_t *)key; cassert(config_prof); hash(bt->vec, bt->len * sizeof(void *), 0x94122f33U, r_hash); } static bool prof_bt_keycomp(const void *k1, const void *k2) { const prof_bt_t *bt1 = (prof_bt_t *)k1; const prof_bt_t *bt2 = (prof_bt_t *)k2; cassert(config_prof); if (bt1->len != bt2->len) return (false); return (memcmp(bt1->vec, bt2->vec, bt1->len * sizeof(void *)) == 0); } JEMALLOC_INLINE_C uint64_t prof_thr_uid_alloc(void) { uint64_t thr_uid; malloc_mutex_lock(&next_thr_uid_mtx); thr_uid = next_thr_uid; next_thr_uid++; malloc_mutex_unlock(&next_thr_uid_mtx); return (thr_uid); } static prof_tdata_t * prof_tdata_init_impl(tsd_t *tsd, uint64_t thr_uid, uint64_t thr_discrim, char *thread_name, bool active) { prof_tdata_t *tdata; tcache_t *tcache; cassert(config_prof); /* Initialize an empty cache for this thread. */ tcache = tcache_get(tsd, true); tdata = (prof_tdata_t *)iallocztm(tsd, sizeof(prof_tdata_t), false, tcache, true, NULL); if (tdata == NULL) return (NULL); tdata->lock = prof_tdata_mutex_choose(thr_uid); tdata->thr_uid = thr_uid; tdata->thr_discrim = thr_discrim; tdata->thread_name = thread_name; tdata->attached = true; tdata->expired = false; tdata->tctx_uid_next = 0; if (ckh_new(tsd, &tdata->bt2tctx, PROF_CKH_MINITEMS, prof_bt_hash, prof_bt_keycomp)) { idalloctm(tsd, tdata, tcache, true); return (NULL); } tdata->prng_state = (uint64_t)(uintptr_t)tdata; prof_sample_threshold_update(tdata); tdata->enq = false; tdata->enq_idump = false; tdata->enq_gdump = false; tdata->dumping = false; tdata->active = active; malloc_mutex_lock(&tdatas_mtx); tdata_tree_insert(&tdatas, tdata); malloc_mutex_unlock(&tdatas_mtx); return (tdata); } prof_tdata_t * prof_tdata_init(tsd_t *tsd) { return (prof_tdata_init_impl(tsd, prof_thr_uid_alloc(), 0, NULL, prof_thread_active_init_get())); } /* tdata->lock must be held. */ static bool prof_tdata_should_destroy(prof_tdata_t *tdata, bool even_if_attached) { if (tdata->attached && !even_if_attached) return (false); if (ckh_count(&tdata->bt2tctx) != 0) return (false); return (true); } /* tdatas_mtx must be held. */ static void prof_tdata_destroy_locked(tsd_t *tsd, prof_tdata_t *tdata, bool even_if_attached) { tcache_t *tcache; assert(prof_tdata_should_destroy(tdata, even_if_attached)); assert(tsd_prof_tdata_get(tsd) != tdata); tdata_tree_remove(&tdatas, tdata); tcache = tcache_get(tsd, false); if (tdata->thread_name != NULL) idalloctm(tsd, tdata->thread_name, tcache, true); ckh_delete(tsd, &tdata->bt2tctx); idalloctm(tsd, tdata, tcache, true); } static void prof_tdata_destroy(tsd_t *tsd, prof_tdata_t *tdata, bool even_if_attached) { malloc_mutex_lock(&tdatas_mtx); prof_tdata_destroy_locked(tsd, tdata, even_if_attached); malloc_mutex_unlock(&tdatas_mtx); } static void prof_tdata_detach(tsd_t *tsd, prof_tdata_t *tdata) { bool destroy_tdata; malloc_mutex_lock(tdata->lock); if (tdata->attached) { destroy_tdata = prof_tdata_should_destroy(tdata, true); /* * Only detach if !destroy_tdata, because detaching would allow * another thread to win the race to destroy tdata. */ if (!destroy_tdata) tdata->attached = false; tsd_prof_tdata_set(tsd, NULL); } else destroy_tdata = false; malloc_mutex_unlock(tdata->lock); if (destroy_tdata) prof_tdata_destroy(tsd, tdata, true); } prof_tdata_t * prof_tdata_reinit(tsd_t *tsd, prof_tdata_t *tdata) { uint64_t thr_uid = tdata->thr_uid; uint64_t thr_discrim = tdata->thr_discrim + 1; char *thread_name = (tdata->thread_name != NULL) ? prof_thread_name_alloc(tsd, tdata->thread_name) : NULL; bool active = tdata->active; prof_tdata_detach(tsd, tdata); return (prof_tdata_init_impl(tsd, thr_uid, thr_discrim, thread_name, active)); } static bool prof_tdata_expire(prof_tdata_t *tdata) { bool destroy_tdata; malloc_mutex_lock(tdata->lock); if (!tdata->expired) { tdata->expired = true; destroy_tdata = tdata->attached ? false : prof_tdata_should_destroy(tdata, false); } else destroy_tdata = false; malloc_mutex_unlock(tdata->lock); return (destroy_tdata); } static prof_tdata_t * prof_tdata_reset_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata, void *arg) { return (prof_tdata_expire(tdata) ? tdata : NULL); } void prof_reset(tsd_t *tsd, size_t lg_sample) { prof_tdata_t *next; assert(lg_sample < (sizeof(uint64_t) << 3)); malloc_mutex_lock(&prof_dump_mtx); malloc_mutex_lock(&tdatas_mtx); lg_prof_sample = lg_sample; next = NULL; do { prof_tdata_t *to_destroy = tdata_tree_iter(&tdatas, next, prof_tdata_reset_iter, NULL); if (to_destroy != NULL) { next = tdata_tree_next(&tdatas, to_destroy); prof_tdata_destroy_locked(tsd, to_destroy, false); } else next = NULL; } while (next != NULL); malloc_mutex_unlock(&tdatas_mtx); malloc_mutex_unlock(&prof_dump_mtx); } void prof_tdata_cleanup(tsd_t *tsd) { prof_tdata_t *tdata; if (!config_prof) return; tdata = tsd_prof_tdata_get(tsd); if (tdata != NULL) prof_tdata_detach(tsd, tdata); } bool prof_active_get(void) { bool prof_active_current; malloc_mutex_lock(&prof_active_mtx); prof_active_current = prof_active; malloc_mutex_unlock(&prof_active_mtx); return (prof_active_current); } bool prof_active_set(bool active) { bool prof_active_old; malloc_mutex_lock(&prof_active_mtx); prof_active_old = prof_active; prof_active = active; malloc_mutex_unlock(&prof_active_mtx); return (prof_active_old); } const char * prof_thread_name_get(void) { tsd_t *tsd; prof_tdata_t *tdata; tsd = tsd_fetch(); tdata = prof_tdata_get(tsd, true); if (tdata == NULL) return (""); return (tdata->thread_name != NULL ? tdata->thread_name : ""); } static char * prof_thread_name_alloc(tsd_t *tsd, const char *thread_name) { char *ret; size_t size; if (thread_name == NULL) return (NULL); size = strlen(thread_name) + 1; if (size == 1) return (""); ret = iallocztm(tsd, size, false, tcache_get(tsd, true), true, NULL); if (ret == NULL) return (NULL); memcpy(ret, thread_name, size); return (ret); } int prof_thread_name_set(tsd_t *tsd, const char *thread_name) { prof_tdata_t *tdata; unsigned i; char *s; tdata = prof_tdata_get(tsd, true); if (tdata == NULL) return (EAGAIN); /* Validate input. */ if (thread_name == NULL) return (EFAULT); for (i = 0; thread_name[i] != '\0'; i++) { char c = thread_name[i]; if (!isgraph(c) && !isblank(c)) return (EFAULT); } s = prof_thread_name_alloc(tsd, thread_name); if (s == NULL) return (EAGAIN); if (tdata->thread_name != NULL) { idalloctm(tsd, tdata->thread_name, tcache_get(tsd, false), true); tdata->thread_name = NULL; } if (strlen(s) > 0) tdata->thread_name = s; return (0); } bool prof_thread_active_get(void) { tsd_t *tsd; prof_tdata_t *tdata; tsd = tsd_fetch(); tdata = prof_tdata_get(tsd, true); if (tdata == NULL) return (false); return (tdata->active); } bool prof_thread_active_set(bool active) { tsd_t *tsd; prof_tdata_t *tdata; tsd = tsd_fetch(); tdata = prof_tdata_get(tsd, true); if (tdata == NULL) return (true); tdata->active = active; return (false); } bool prof_thread_active_init_get(void) { bool active_init; malloc_mutex_lock(&prof_thread_active_init_mtx); active_init = prof_thread_active_init; malloc_mutex_unlock(&prof_thread_active_init_mtx); return (active_init); } bool prof_thread_active_init_set(bool active_init) { bool active_init_old; malloc_mutex_lock(&prof_thread_active_init_mtx); active_init_old = prof_thread_active_init; prof_thread_active_init = active_init; malloc_mutex_unlock(&prof_thread_active_init_mtx); return (active_init_old); } bool prof_gdump_get(void) { bool prof_gdump_current; malloc_mutex_lock(&prof_gdump_mtx); prof_gdump_current = prof_gdump_val; malloc_mutex_unlock(&prof_gdump_mtx); return (prof_gdump_current); } bool prof_gdump_set(bool gdump) { bool prof_gdump_old; malloc_mutex_lock(&prof_gdump_mtx); prof_gdump_old = prof_gdump_val; prof_gdump_val = gdump; malloc_mutex_unlock(&prof_gdump_mtx); return (prof_gdump_old); } void prof_boot0(void) { cassert(config_prof); memcpy(opt_prof_prefix, PROF_PREFIX_DEFAULT, sizeof(PROF_PREFIX_DEFAULT)); } void prof_boot1(void) { cassert(config_prof); /* * opt_prof must be in its final state before any arenas are * initialized, so this function must be executed early. */ if (opt_prof_leak && !opt_prof) { /* * Enable opt_prof, but in such a way that profiles are never * automatically dumped. */ opt_prof = true; opt_prof_gdump = false; } else if (opt_prof) { if (opt_lg_prof_interval >= 0) { prof_interval = (((uint64_t)1U) << opt_lg_prof_interval); } } } bool prof_boot2(void) { cassert(config_prof); if (opt_prof) { tsd_t *tsd; unsigned i; lg_prof_sample = opt_lg_prof_sample; prof_active = opt_prof_active; if (malloc_mutex_init(&prof_active_mtx)) return (true); prof_gdump_val = opt_prof_gdump; if (malloc_mutex_init(&prof_gdump_mtx)) return (true); prof_thread_active_init = opt_prof_thread_active_init; if (malloc_mutex_init(&prof_thread_active_init_mtx)) return (true); tsd = tsd_fetch(); if (ckh_new(tsd, &bt2gctx, PROF_CKH_MINITEMS, prof_bt_hash, prof_bt_keycomp)) return (true); if (malloc_mutex_init(&bt2gctx_mtx)) return (true); tdata_tree_new(&tdatas); if (malloc_mutex_init(&tdatas_mtx)) return (true); next_thr_uid = 0; if (malloc_mutex_init(&next_thr_uid_mtx)) return (true); if (malloc_mutex_init(&prof_dump_seq_mtx)) return (true); if (malloc_mutex_init(&prof_dump_mtx)) return (true); if (opt_prof_final && opt_prof_prefix[0] != '\0' && atexit(prof_fdump) != 0) { malloc_write(": Error in atexit()\n"); if (opt_abort) abort(); } gctx_locks = (malloc_mutex_t *)base_alloc(PROF_NCTX_LOCKS * sizeof(malloc_mutex_t)); if (gctx_locks == NULL) return (true); for (i = 0; i < PROF_NCTX_LOCKS; i++) { if (malloc_mutex_init(&gctx_locks[i])) return (true); } tdata_locks = (malloc_mutex_t *)base_alloc(PROF_NTDATA_LOCKS * sizeof(malloc_mutex_t)); if (tdata_locks == NULL) return (true); for (i = 0; i < PROF_NTDATA_LOCKS; i++) { if (malloc_mutex_init(&tdata_locks[i])) return (true); } } #ifdef JEMALLOC_PROF_LIBGCC /* * Cause the backtracing machinery to allocate its internal state * before enabling profiling. */ _Unwind_Backtrace(prof_unwind_init_callback, NULL); #endif prof_booted = true; return (false); } void prof_prefork(void) { if (opt_prof) { unsigned i; malloc_mutex_prefork(&tdatas_mtx); malloc_mutex_prefork(&bt2gctx_mtx); malloc_mutex_prefork(&next_thr_uid_mtx); malloc_mutex_prefork(&prof_dump_seq_mtx); for (i = 0; i < PROF_NCTX_LOCKS; i++) malloc_mutex_prefork(&gctx_locks[i]); for (i = 0; i < PROF_NTDATA_LOCKS; i++) malloc_mutex_prefork(&tdata_locks[i]); } } void prof_postfork_parent(void) { if (opt_prof) { unsigned i; for (i = 0; i < PROF_NTDATA_LOCKS; i++) malloc_mutex_postfork_parent(&tdata_locks[i]); for (i = 0; i < PROF_NCTX_LOCKS; i++) malloc_mutex_postfork_parent(&gctx_locks[i]); malloc_mutex_postfork_parent(&prof_dump_seq_mtx); malloc_mutex_postfork_parent(&next_thr_uid_mtx); malloc_mutex_postfork_parent(&bt2gctx_mtx); malloc_mutex_postfork_parent(&tdatas_mtx); } } void prof_postfork_child(void) { if (opt_prof) { unsigned i; for (i = 0; i < PROF_NTDATA_LOCKS; i++) malloc_mutex_postfork_child(&tdata_locks[i]); for (i = 0; i < PROF_NCTX_LOCKS; i++) malloc_mutex_postfork_child(&gctx_locks[i]); malloc_mutex_postfork_child(&prof_dump_seq_mtx); malloc_mutex_postfork_child(&next_thr_uid_mtx); malloc_mutex_postfork_child(&bt2gctx_mtx); malloc_mutex_postfork_child(&tdatas_mtx); } } /******************************************************************************/ Index: head/contrib/jemalloc/src/tsd.c =================================================================== --- head/contrib/jemalloc/src/tsd.c (revision 289899) +++ head/contrib/jemalloc/src/tsd.c (revision 289900) @@ -1,190 +1,193 @@ #define JEMALLOC_TSD_C_ #include "jemalloc/internal/jemalloc_internal.h" /******************************************************************************/ /* Data. */ static unsigned ncleanups; static malloc_tsd_cleanup_t cleanups[MALLOC_TSD_CLEANUPS_MAX]; malloc_tsd_data(, , tsd_t, TSD_INITIALIZER) /******************************************************************************/ void * malloc_tsd_malloc(size_t size) { return (a0malloc(CACHELINE_CEILING(size))); } void malloc_tsd_dalloc(void *wrapper) { a0dalloc(wrapper); } void malloc_tsd_no_cleanup(void *arg) { not_reached(); } #if defined(JEMALLOC_MALLOC_THREAD_CLEANUP) || defined(_WIN32) #ifndef _WIN32 JEMALLOC_EXPORT #endif void _malloc_thread_cleanup(void) { bool pending[MALLOC_TSD_CLEANUPS_MAX], again; unsigned i; for (i = 0; i < ncleanups; i++) pending[i] = true; do { again = false; for (i = 0; i < ncleanups; i++) { if (pending[i]) { pending[i] = cleanups[i](); if (pending[i]) again = true; } } } while (again); } #endif void malloc_tsd_cleanup_register(bool (*f)(void)) { assert(ncleanups < MALLOC_TSD_CLEANUPS_MAX); cleanups[ncleanups] = f; ncleanups++; } void tsd_cleanup(void *arg) { tsd_t *tsd = (tsd_t *)arg; switch (tsd->state) { + case tsd_state_uninitialized: + /* Do nothing. */ + break; case tsd_state_nominal: #define O(n, t) \ n##_cleanup(tsd); MALLOC_TSD #undef O tsd->state = tsd_state_purgatory; tsd_set(tsd); break; case tsd_state_purgatory: /* * The previous time this destructor was called, we set the * state to tsd_state_purgatory so that other destructors * wouldn't cause re-creation of the tsd. This time, do * nothing, and do not request another callback. */ break; case tsd_state_reincarnated: /* * Another destructor deallocated memory after this destructor * was called. Reset state to tsd_state_purgatory and request * another callback. */ tsd->state = tsd_state_purgatory; tsd_set(tsd); break; default: not_reached(); } } bool malloc_tsd_boot0(void) { ncleanups = 0; if (tsd_boot0()) return (true); *tsd_arenas_cache_bypassp_get(tsd_fetch()) = true; return (false); } void malloc_tsd_boot1(void) { tsd_boot1(); *tsd_arenas_cache_bypassp_get(tsd_fetch()) = false; } #ifdef _WIN32 static BOOL WINAPI _tls_callback(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved) { switch (fdwReason) { #ifdef JEMALLOC_LAZY_LOCK case DLL_THREAD_ATTACH: isthreaded = true; break; #endif case DLL_THREAD_DETACH: _malloc_thread_cleanup(); break; default: break; } return (true); } #ifdef _MSC_VER # ifdef _M_IX86 # pragma comment(linker, "/INCLUDE:__tls_used") # else # pragma comment(linker, "/INCLUDE:_tls_used") # endif # pragma section(".CRT$XLY",long,read) #endif JEMALLOC_SECTION(".CRT$XLY") JEMALLOC_ATTR(used) static BOOL (WINAPI *const tls_callback)(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved) = _tls_callback; #endif #if (!defined(JEMALLOC_MALLOC_THREAD_CLEANUP) && !defined(JEMALLOC_TLS) && \ !defined(_WIN32)) void * tsd_init_check_recursion(tsd_init_head_t *head, tsd_init_block_t *block) { pthread_t self = pthread_self(); tsd_init_block_t *iter; /* Check whether this thread has already inserted into the list. */ malloc_mutex_lock(&head->lock); ql_foreach(iter, &head->blocks, link) { if (iter->thread == self) { malloc_mutex_unlock(&head->lock); return (iter->data); } } /* Insert block into list. */ ql_elm_new(block, link); block->thread = self; ql_tail_insert(&head->blocks, block, link); malloc_mutex_unlock(&head->lock); return (NULL); } void tsd_init_finish(tsd_init_head_t *head, tsd_init_block_t *block) { malloc_mutex_lock(&head->lock); ql_remove(&head->blocks, block, link); malloc_mutex_unlock(&head->lock); } #endif