Page MenuHomeFreeBSD

kib (Konstantin Belousov)
User

Projects

User Details

User Since
May 16 2014, 7:35 PM (248 w, 2 d)

Recent Activity

Yesterday

kib added a comment to D19226: i386 PAE: avoid atomic for pte_store()..

It looks like this is assuming that the old value is not valid, but that's not always true. For example, pmap_promote_pde() simply overwrites the old PDE.

I don't really see how pmap_kextract() is safe wrt torn writes.

Sun, Feb 17, 8:33 PM
kib added inline comments to D19225: amd64: cleanup pmap_init_pat()..
Sun, Feb 17, 8:30 PM
kib retitled D19226: i386 PAE: avoid atomic for pte_store(). from i386: avoid atomic for pte_store(). to i386 PAE: avoid atomic for pte_store()..
Sun, Feb 17, 7:48 PM
kib added reviewers for D19226: i386 PAE: avoid atomic for pte_store().: alc, markj.
Sun, Feb 17, 7:48 PM
kib created D19226: i386 PAE: avoid atomic for pte_store()..
Sun, Feb 17, 7:48 PM
kib created D19225: amd64: cleanup pmap_init_pat()..
Sun, Feb 17, 7:41 PM
kib updated the diff for D19211: pkru(3).

Alan' notes.

Sun, Feb 17, 6:26 PM
kib accepted D19224: Move a racy assertion in filt_pipewrite()..

Why cannot the same happen for read side ?

Sun, Feb 17, 5:43 PM
kib updated the diff for D19211: pkru(3).

More Mark' notes.

Sun, Feb 17, 5:31 PM
kib committed rS344230: MFC r343723:.
MFC r343723:
Sun, Feb 17, 10:02 AM
kib committed rS344229: MFC r343723:.
MFC r343723:
Sun, Feb 17, 9:58 AM
kib committed rS344228: MFC r343966:.
MFC r343966:
Sun, Feb 17, 9:56 AM

Sat, Feb 16

kib accepted D19214: Fix refcount leak in SGX Linux compat ioctl handlers..
Sat, Feb 16, 8:26 PM
kib accepted D19215: Get rid of a redundant flag variable..
Sat, Feb 16, 8:23 PM
kib added inline comments to D19211: pkru(3).
Sat, Feb 16, 8:15 PM
kib updated the diff for D19211: pkru(3).

Editing.

Sat, Feb 16, 8:13 PM
kib updated the diff for D19211: pkru(3).

Grammar.

Sat, Feb 16, 3:47 PM
kib added inline comments to D19211: pkru(3).
Sat, Feb 16, 3:44 PM
kib created D19211: pkru(3).
Sat, Feb 16, 1:36 PM
kib added a comment to D18893: Add support for Intel userspace protection keys feature on Skylake Xeons..

The code changes look ok to me. Do you plan to upload man page changes here too?

Sat, Feb 16, 7:26 AM

Fri, Feb 15

kib accepted D18423: Prototype WITH_PIE knob.
Fri, Feb 15, 7:06 PM
kib added a comment to D18423: Prototype WITH_PIE knob.

I believe this is ready to commit now (disabled by default); commit message:

Do static binaries work ? I suspect they do not, or they work by a chance. Most scary is init(8) of course, but other static binaries esp. rescue(8) are also questionable.

Fri, Feb 15, 3:10 PM
kib added inline comments to D18028: Summary: Add support for Intel Speed Shift.
Fri, Feb 15, 1:28 PM
kib committed rS344158: MFC r343859:.
MFC r343859:
Fri, Feb 15, 11:36 AM
kib committed rS344157: MFC r343855, r343859:.
MFC r343855, r343859:
Fri, Feb 15, 11:34 AM
kib committed rS344156: MFC r343899:.
MFC r343899:
Fri, Feb 15, 11:28 AM
kib committed rS344155: MFC r343897, r343898:.
MFC r343897, r343898:
Fri, Feb 15, 11:27 AM
kib committed rS344154: MFC r343899:.
MFC r343899:
Fri, Feb 15, 11:20 AM
kib committed rS344153: MFC r343897, r343898:.
MFC r343897, r343898:
Fri, Feb 15, 11:19 AM
kib committed rS344152: MFC r343891:.
MFC r343891:
Fri, Feb 15, 11:13 AM
kib committed rS344149: MFC r343617, r343618:.
MFC r343617, r343618:
Fri, Feb 15, 9:49 AM

Thu, Feb 14

kib committed rS344128: Make anon clustering more compatible..
Make anon clustering more compatible.
Thu, Feb 14, 3:46 PM
kib closed D19194: Make anon clustering more compatible..
Thu, Feb 14, 3:46 PM
kib committed rS344122: MFC r343850:.
MFC r343850:
Thu, Feb 14, 2:51 PM
kib committed rS344121: Enable enabling ASLR on non-x86 architectures..
Enable enabling ASLR on non-x86 architectures.
Thu, Feb 14, 2:45 PM
kib updated the diff for D19072: Fix initial exec TLS mode for dynamically loaded shared objects..

Provide implementations for all arches.

Thu, Feb 14, 2:17 PM
kib updated the diff for D19194: Make anon clustering more compatible..

s/-/=/

Thu, Feb 14, 2:15 PM
kib committed rS344120: Unify i386 and amd64 getcontextx.c, and use ifuncs while there..
Unify i386 and amd64 getcontextx.c, and use ifuncs while there.
Thu, Feb 14, 2:02 PM
kib committed rS344119: x86 __vdso_gettc(): use machine/cpufunc.h function for CPUID..
x86 __vdso_gettc(): use machine/cpufunc.h function for CPUID.
Thu, Feb 14, 1:59 PM
kib committed rS344118: Provide userspace versions of do_cpuid() and cpuid_count() on i386..
Provide userspace versions of do_cpuid() and cpuid_count() on i386.
Thu, Feb 14, 1:53 PM
kib created D19194: Make anon clustering more compatible..
Thu, Feb 14, 11:19 AM

Tue, Feb 12

kib committed rS344053: Fix PAE modules build on i386..
Fix PAE modules build on i386.
Tue, Feb 12, 4:56 PM

Mon, Feb 11

kib committed rS344011: MFC r339877-r339879,r343564-r343566,r343580,r343754:.
MFC r339877-r339879,r343564-r343566,r343580,r343754:
Mon, Feb 11, 3:02 PM
kib committed rS343997: MFC r343890:.
MFC r343890:
Mon, Feb 11, 7:09 AM

Sun, Feb 10

kib committed rS343966: struct xswdev on amd64 requires compat32 shims after ino64..
struct xswdev on amd64 requires compat32 shims after ino64.
Sun, Feb 10, 7:01 PM
kib closed D5603: ASLR.
Sun, Feb 10, 5:20 PM
kib committed rS343964: Implement Address Space Layout Randomization (ASLR).
Implement Address Space Layout Randomization (ASLR)
Sun, Feb 10, 5:20 PM
kib committed rS343948: MFC r343724:.
MFC r343724:
Sun, Feb 10, 5:42 AM
kib committed rS343947: MFC r343724:.
MFC r343724:
Sun, Feb 10, 5:40 AM

Sat, Feb 9

kib updated the diff for D19072: Fix initial exec TLS mode for dynamically loaded shared objects..

Fix quite a bug in the loading of depended objects.
Restructure to avoid coding current thread offset calculation twice.

Sat, Feb 9, 11:42 AM
kib committed rS343929: i386: honor kern.elf32.read_exec for ommap(2) and break(2), as already.
i386: honor kern.elf32.read_exec for ommap(2) and break(2), as already
Sat, Feb 9, 3:57 AM
kib committed rS343928: Normalize the declaration of i386_read_exec variable..
Normalize the declaration of i386_read_exec variable.
Sat, Feb 9, 3:52 AM
kib added a comment to D19122: mdmfs: Fix many bugs in automatic md(4) creation..

The buffer is not statically allocated, it is statically sized (in the revision summary).

Sat, Feb 9, 3:27 AM
kib updated the diff for D18893: Add support for Intel userspace protection keys feature on Skylake Xeons..

Rename rangeset_get() to rangeset_lookup().
Remove one more _NOFREE.
Clear dst after failed copy.

Sat, Feb 9, 2:51 AM
kib added inline comments to D18893: Add support for Intel userspace protection keys feature on Skylake Xeons..
Sat, Feb 9, 2:49 AM

Fri, Feb 8

kib committed rS343899: Un null_vptocnp(), cache vp->v_mount and use it for null_nodeget() call..
Un null_vptocnp(), cache vp->v_mount and use it for null_nodeget() call.
Fri, Feb 8, 8:20 AM
kib committed rS343898: Before using VTONULL(), check that the covered vnode belongs to nullfs..
Before using VTONULL(), check that the covered vnode belongs to nullfs.
Fri, Feb 8, 8:17 AM
kib committed rS343897: Some style for nullfs_mount(). Also use bool type for isvnunlocked..
Some style for nullfs_mount(). Also use bool type for isvnunlocked.
Fri, Feb 8, 8:15 AM
kib committed rS343896: Correctly return exit status from the exited process..
Correctly return exit status from the exited process.
Fri, Feb 8, 7:36 AM
kib committed rS343891: Fix renameat(2) for CAPABILITIES kernels..
Fix renameat(2) for CAPABILITIES kernels.
Fri, Feb 8, 4:18 AM
kib closed D19096: Fix renameat(2) for CAPABILITIES kernels..
Fri, Feb 8, 4:18 AM
kib added a comment to D19096: Fix renameat(2) for CAPABILITIES kernels..
In D19096#408759, @ngie wrote:

Thanks so much!

@emaste: MFC note: this breaks struct namei layout. Do we want the merge ?

Should this field be at the end of the struct to avoid breaking ABI?

Fri, Feb 8, 4:17 AM
kib committed rS343890: do_execve(): lock vnode when needed..
do_execve(): lock vnode when needed.
Fri, Feb 8, 4:07 AM
kib committed rS343888: MFC r343808:.
MFC r343808:
Fri, Feb 8, 3:29 AM
kib committed rS343887: MFC r343808:.
MFC r343808:
Fri, Feb 8, 3:25 AM

Thu, Feb 7

kib added a comment to D19063: mlx5en: add pfil ethernet hook.

The reason to have this in the driver is for performance. The plan for pfil is to eventually just use a pointer to a memory blob for filtering. The current fake mbuf stuff is a step in that direction.

By passing just a pointer from inside the driver, filtering and dropping floods of unwanted traffic can become incredibly cheap. For each packet examined and dropped, we avoid the cache misses inherent in allocating an mbuf and filling it in (at least 2 cachelines just for the mbuf), as well as the memory copy that most drivers do for small packets (likely 2 more cachelines). For larger packets, or drivers that do not employ the copy-for-small packets strategy, instead of the copy, we avoid the virtual to physical (or worse, IOMMU) translation for new rx buffers. Avoiding non-IO cache misses is especially beneficial on Intel Xeon w/DDIO, when recently received traffic may still be resident in the DDIO cache ways.

Thu, Feb 7, 7:36 PM
kib edited reviewers for D19098: Parallel zfs mounting, added: avg; removed: kib.
Thu, Feb 7, 10:34 AM
kib added a comment to D19063: mlx5en: add pfil ethernet hook.

I have a generic question about the patch. Why this needs to be done inside the driver, and why cannot pfil hooks be applied e.g. in ether_input() ? It is strange that all drivers would need this patch.

Thu, Feb 7, 5:07 AM
kib committed rS343859: Add comment noting that the strange spelling of GenuineIntel is for reason..
Add comment noting that the strange spelling of GenuineIntel is for reason.
Thu, Feb 7, 3:58 AM
kib created D19096: Fix renameat(2) for CAPABILITIES kernels..
Thu, Feb 7, 3:53 AM
kib committed rS343855: Use ifunc to select the barrier instruction for RDTSC..
Use ifunc to select the barrier instruction for RDTSC.
Thu, Feb 7, 2:56 AM
kib committed rS343852: Port sysctl kern.elf32.read_exec from amd64 to i386..
Port sysctl kern.elf32.read_exec from amd64 to i386.
Thu, Feb 7, 2:17 AM
kib committed rS343851: Fix resume on i386 PAE..
Fix resume on i386 PAE.
Thu, Feb 7, 2:10 AM
kib committed rS343850: contigmalloc: handle M_EXEC..
contigmalloc: handle M_EXEC.
Thu, Feb 7, 2:00 AM
kib closed D19092: contigmalloc: handle M_EXEC.
Thu, Feb 7, 2:00 AM
kib committed rS343849: MFC r343757:.
MFC r343757:
Thu, Feb 7, 1:55 AM
kib committed rS343848: MFC r343757:.
MFC r343757:
Thu, Feb 7, 1:54 AM
kib committed rS343847: MFC r343629:.
MFC r343629:
Thu, Feb 7, 1:52 AM
kib committed rS343846: MFC r343628:.
MFC r343628:
Thu, Feb 7, 1:50 AM

Wed, Feb 6

kib created D19092: contigmalloc: handle M_EXEC.
Wed, Feb 6, 8:20 AM

Tue, Feb 5

kib committed rS343809: Make it possible to override PAE mode on boot..
Make it possible to override PAE mode on boot.
Tue, Feb 5, 8:09 PM
kib committed rS343808: Remove pointless initial value for i386 vm.pmap.pat_works sysctl definition..
Remove pointless initial value for i386 vm.pmap.pat_works sysctl definition.
Tue, Feb 5, 8:02 PM
kib committed rS343782: MFC r343780:.
MFC r343780:
Tue, Feb 5, 5:54 PM
kib committed rS343781: MFC r343780:.
MFC r343780:
Tue, Feb 5, 5:52 PM
kib committed rS343780: amd64: clear callee-preserved registers on syscall exit..
amd64: clear callee-preserved registers on syscall exit.
Tue, Feb 5, 5:49 PM
kib accepted D19083: extattr_list_vp: Narrow locked section somewhat.
Tue, Feb 5, 4:20 AM
kib accepted D19082: extattr_list_vp: Only take shared vnode lock.
Tue, Feb 5, 2:08 AM

Mon, Feb 4

kib committed rS343757: Update CPUID bits definitions and CPU identification based on changes.
Update CPUID bits definitions and CPU identification based on changes
Mon, Feb 4, 11:58 PM
kib committed rS343754: Fixes for very early use of the pthread_mutex_* and libthr malloc..
Fixes for very early use of the pthread_mutex_* and libthr malloc.
Mon, Feb 4, 9:16 PM
kib accepted D19074: Use NDLT to get number of LDTs.
Mon, Feb 4, 12:58 PM
kib added inline comments to D19063: mlx5en: add pfil ethernet hook.
Mon, Feb 4, 9:02 AM

Sun, Feb 3

kib added inline comments to D19058: posix_spawn(3): handle potential signal issues with vfork.
Sun, Feb 3, 11:41 PM
kib added a comment to D19072: Fix initial exec TLS mode for dynamically loaded shared objects..

Should we encourage lld to revive the https://reviews.llvm.org/D33041 patch?

Sun, Feb 3, 10:25 PM
kib committed rS343724: Do not call PHOLD() while owning the allproc_lock sx..
Do not call PHOLD() while owning the allproc_lock sx.
Sun, Feb 3, 9:32 PM
kib committed rS343723: i386: Do not ever store to other-CPU counter64 slot..
i386: Do not ever store to other-CPU counter64 slot.
Sun, Feb 3, 9:29 PM
kib added inline comments to D19058: posix_spawn(3): handle potential signal issues with vfork.
Sun, Feb 3, 9:24 PM
kib created D19072: Fix initial exec TLS mode for dynamically loaded shared objects..
Sun, Feb 3, 9:11 PM
kib added inline comments to D19058: posix_spawn(3): handle potential signal issues with vfork.
Sun, Feb 3, 6:17 PM
kib committed rS343695: MFC r343607:.
MFC r343607:
Sun, Feb 3, 12:46 AM
kib committed rS343694: MFC r343484:.
MFC r343484:
Sun, Feb 3, 12:44 AM