Page MenuHomeFreeBSD

markj (Mark Johnston)
User

Projects (8)

User Details

User Since
Mar 12 2014, 1:00 AM (497 w, 1 d)

Recent Activity

Mon, Sep 18

markj added a comment to D41897: KASAN: Add KASAN_UMA_NEVERUSED redzone value.

Note that the change as-is isn't complete: after a slab is allocated, a per-CPU bucket will be populated with some items from the cache, and kasan_mark_item_invalid() will overwrite the state bytes, setting them back to KASAN_UMA_FREED even though they are still conceptually still in the NEVERUSED state.

Mon, Sep 18, 10:49 AM
markj added a comment to D41897: KASAN: Add KASAN_UMA_NEVERUSED redzone value.

@markj I don't know enough about the UMA code to be 100% certain that this is correct, but I think it's right -- we only call this from keg_alloc_slab, which is assigning never-before-used pages to a zone, right?

Mon, Sep 18, 10:46 AM

Sun, Sep 17

markj accepted D41634: arm64 pmap: allocate leaf page table page for wired userspace superpages.
Sun, Sep 17, 11:23 AM
markj accepted D41633: riscv pmap: allocate leaf page table page for wired userspace superpages.
Sun, Sep 17, 11:22 AM
markj accepted D41635: i386 pmap: allocate leaf page table page for wired userspace superpages.
Sun, Sep 17, 11:21 AM
markj committed rGaa3bcaad5107: amd64: Add a leaf PTP when pmap_enter(psind=1) creates a wired mapping (authored by bojan.novkovic_fer.hr).
amd64: Add a leaf PTP when pmap_enter(psind=1) creates a wired mapping
Sun, Sep 17, 10:38 AM
markj committed rGf7c733e4fe5d: amd64: Convert a cheap DIAGNOSTIC check to a KASSERT (authored by markj).
amd64: Convert a cheap DIAGNOSTIC check to a KASSERT
Sun, Sep 17, 10:38 AM
markj committed rG8f26ed01bd74: mlock tests: Add a regression test (authored by markj).
mlock tests: Add a regression test
Sun, Sep 17, 10:38 AM
markj closed D41132: amd64 pmap: allocate leaf page table page for wired userspace 2M pages.
Sun, Sep 17, 10:37 AM
markj added inline comments to D41132: amd64 pmap: allocate leaf page table page for wired userspace 2M pages.
Sun, Sep 17, 10:24 AM

Sat, Sep 16

markj requested review of D41884: unix: Fix a lock order reveral.
Sat, Sep 16, 9:46 AM
markj requested review of D41883: hdac: Defer interrupt allocation in hdac_attach().
Sat, Sep 16, 9:45 AM
markj committed rGd643925a79ca: unix tests: Use the right type for sendmsg(2) return values (authored by markj).
unix tests: Use the right type for sendmsg(2) return values
Sat, Sep 16, 9:33 AM
markj committed rG38426b32e106: socket tests: Add a regression test for MSG_WAITALL (authored by markj).
socket tests: Add a regression test for MSG_WAITALL
Sat, Sep 16, 9:29 AM
markj closed D41783: sockets: re-check socket state after call to pr_rcvd().
Sat, Sep 16, 9:28 AM · committers, tests
markj committed rGf44102411e1b: sockets: re-check socket state after call to pr_rcvd() (authored by becker.greg_att.net).
sockets: re-check socket state after call to pr_rcvd()
Sat, Sep 16, 9:28 AM
markj added a comment to D41783: sockets: re-check socket state after call to pr_rcvd().

I've been able to trigger an sbwait() assertion failure in the receive path when running KTLS tests, but only when running tests in parallel in QEMU. I believe the problem is the following: KTLS introduces a state for received data wherein it is being decrypted but is not yet queued in the socket buffer. In this case (sb_tlsdcc != 0), we can end up calling sbwait() with CANTRECVMORE set because that sleep/wakeup mechanism is used to signal completion of decryption operations.

Sat, Sep 16, 9:23 AM · committers, tests

Fri, Sep 15

markj accepted D41876: kinst: do not look for a function epilogue.

Approved.

Fri, Sep 15, 3:33 PM
markj added a comment to D41876: kinst: do not look for a function epilogue.

arm64 doesn't need any changes?

Fri, Sep 15, 3:14 PM
markj added a comment to D40262: Allow setting a per-Jail fallback ABI brand..

The overall idea seems ok to me for what it's worth, my comments are about cosmetic issues.

Fri, Sep 15, 3:12 PM · Jails
markj added inline comments to D41787: Reimplement rangelocks.
Fri, Sep 15, 11:22 AM
markj committed rGf799833f4fce: gve: Make LRO work for jumbo packets (authored by shailend_google.com).
gve: Make LRO work for jumbo packets
Fri, Sep 15, 8:50 AM

Thu, Sep 14

markj committed rG04ada3cc2b53: gve: Make LRO work for jumbo packets (authored by shailend_google.com).
gve: Make LRO work for jumbo packets
Thu, Sep 14, 7:51 AM
markj committed rGffaa91929f2d: qat: Intel 4xxx Series driver API extension (authored by piotrx.kasierski_intel.com).
qat: Intel 4xxx Series driver API extension
Thu, Sep 14, 7:51 AM

Wed, Sep 13

markj committed rGc265e6217951: qat: Intel 4xxx Series driver API extension (authored by piotrx.kasierski_intel.com).
qat: Intel 4xxx Series driver API extension
Wed, Sep 13, 9:15 AM

Tue, Sep 12

markj added inline comments to D41305: ossl: Add AES-GCM support for NEON-enabled armv7.
Tue, Sep 12, 8:21 AM
markj updated the diff for D41305: ossl: Add AES-GCM support for NEON-enabled armv7.

Fix the in-kernel build.

Tue, Sep 12, 8:20 AM

Sat, Sep 9

markj committed rG425c8e509df1: scandir.3: Fix several typos (authored by trond.endrestol_ximalas.info).
scandir.3: Fix several typos
Sat, Sep 9, 5:23 PM
markj accepted D41344: radix_trie: have vm_radix use pctrie code.

Are there any more issues to address before I commit this?

Sat, Sep 9, 4:48 PM
markj added a comment to D41787: Reimplement rangelocks.

So far I mostly just looked at the cheat locks.

Sat, Sep 9, 4:43 PM
markj accepted D41783: sockets: re-check socket state after call to pr_rcvd().

Looks good to me. Thanks for working on this.

Sat, Sep 9, 3:51 PM · committers, tests
markj committed rGd0a5da5c1cf9: scandir.3: Fix several typos (authored by trond.endrestol_ximalas.info).
scandir.3: Fix several typos
Sat, Sep 9, 3:49 PM

Fri, Sep 8

markj added a comment to D41778: bhyve: always generate ACPI tables.

The bhyve manual page needs to be updated as well.

Fri, Sep 8, 2:06 PM · bhyve
markj committed rG8ded9cd3d554: scandir.3: Fix several typos (authored by trond.endrestol_ximalas.info).
scandir.3: Fix several typos
Fri, Sep 8, 1:40 PM

Thu, Sep 7

markj committed rG5f62584a9adb: gve: Make LRO work for jumbo packets (authored by shailend_google.com).
gve: Make LRO work for jumbo packets
Thu, Sep 7, 1:40 PM
markj closed D41754: gve: Make LRO work for jumbo packets.
Thu, Sep 7, 1:39 PM
markj added inline comments to D41739: bhyve: Add PCI mappings for arm64.
Thu, Sep 7, 1:26 PM

Wed, Sep 6

markj added inline comments to D41344: radix_trie: have vm_radix use pctrie code.
Wed, Sep 6, 4:44 PM
markj added inline comments to D41739: bhyve: Add PCI mappings for arm64.
Wed, Sep 6, 4:35 PM
markj updated the diff for D41739: bhyve: Add PCI mappings for arm64.

Address review comments.

Wed, Sep 6, 4:30 PM
markj added a comment to D41745: dtrace: Add the 'oformat' libdtrace option..

Perhaps its own man page is warranted in order to avoid polluting dtrace(1)?

Wed, Sep 6, 2:46 PM · DTrace
markj updated the diff for D41740: bhyve: Use vm_raise_msi() instead of vm_lapic_msi().

Restore PCI b/s/f tuple parameters.

Wed, Sep 6, 2:15 PM
markj updated the diff for D41752: libvmmapi: Make vm_raise_msi() a common function.

Restore PCI b/s/f tuple parameters.

Wed, Sep 6, 2:15 PM
markj committed rG22cf89c93888: qat: Intel 4xxx Series driver API extension (authored by piotrx.kasierski_intel.com).
qat: Intel 4xxx Series driver API extension
Wed, Sep 6, 2:02 PM
markj committed rG8527bb2aee6c: dtrace: Fix a kernel panic in printm() (authored by domagoj.stolfa_gmail.com).
dtrace: Fix a kernel panic in printm()
Wed, Sep 6, 2:02 PM
markj closed D41613: qat: Intel® 4xxx Series driver API extention.
Wed, Sep 6, 2:01 PM
markj closed D41722: dtrace: Fix a kernel panic in printm()..
Wed, Sep 6, 2:01 PM
markj accepted D41754: gve: Make LRO work for jumbo packets.

The main diff looks ok to me, I just don't understand the alignment change.

Wed, Sep 6, 1:44 PM
markj added a comment to D41344: radix_trie: have vm_radix use pctrie code.

Please make sure to bump __FreeBSD_version in sys/param.h in this commit. The layout change would affect some out-of-tree kernel modules, requiring them to be recompiled.

Wed, Sep 6, 1:36 PM
markj requested review of D41753: bhyve: Push option parsing down into bhyverun_machdep.c.
Wed, Sep 6, 12:32 AM
markj added a comment to D41740: bhyve: Use vm_raise_msi() instead of vm_lapic_msi().
In D41740#951159, @jhb wrote:

Hmm, looking back in this series, I only seem vm_raise_msi added for arm64, but not for x86?

Wed, Sep 6, 12:31 AM
markj requested review of D41752: libvmmapi: Make vm_raise_msi() a common function.
Wed, Sep 6, 12:30 AM

Tue, Sep 5

markj added a comment to D41742: arm64: Connect bhyve and libvmmapi to the build.

I still have a couple of WIP patches: one to update the bhyve man page a bit, and another to update option parsing and usage strings. There are quite a few further TODOs: gdb support, PCI passthrough support, ACPI support. I've started working on the gdb stub in the background.

Tue, Sep 5, 6:45 PM
markj updated the diff for D41742: arm64: Connect bhyve and libvmmapi to the build.

Update the src.conf.5 description for WITH_BHYVE.

Tue, Sep 5, 6:38 PM
markj added a comment to D40551: bhyve: Move some more amd64-specific drivers to their own subdir.

Will you combine all of the move-amd64-files into one commit?

Tue, Sep 5, 6:32 PM
markj requested review of D41742: arm64: Connect bhyve and libvmmapi to the build.
Tue, Sep 5, 6:19 PM
markj requested review of D41740: bhyve: Use vm_raise_msi() instead of vm_lapic_msi().
Tue, Sep 5, 6:19 PM
markj requested review of D41741: bhyve: Partially disable INT#x support in virtio for arm64.
Tue, Sep 5, 6:19 PM
markj requested review of D41739: bhyve: Add PCI mappings for arm64.
Tue, Sep 5, 6:18 PM
markj requested review of D41738: bhyve: Do not compile PCI passthrough support on arm64.
Tue, Sep 5, 6:18 PM
markj accepted D41727: libc: Suppress format checks on printf() / scanf() tests..
Tue, Sep 5, 1:55 PM
markj updated the diff for D41006: bhyve: Add bhyverun and vmexit handlers for arm64.

Make vmexit_hyp() a stub. Translation faults are handled in the kernel.

Tue, Sep 5, 1:45 PM
markj added a comment to D39951: kgdb: Handle stoppcbs compat.

@markj are you wanting to commit this, or do you mind if I just grab it when I push the other one + bump PORTREVISION?

Tue, Sep 5, 1:35 PM

Mon, Sep 4

markj accepted D41722: dtrace: Fix a kernel panic in printm()..
Mon, Sep 4, 9:29 PM
markj added inline comments to D41722: dtrace: Fix a kernel panic in printm()..
Mon, Sep 4, 9:15 PM
markj added inline comments to D41721: Cross-mountpoint copy_file_range() for ZFS.
Mon, Sep 4, 8:31 PM · ZFS
markj committed rG34e146632e62: geli tests: Add a regression test for geli setkey -J (authored by markj).
geli tests: Add a regression test for geli setkey -J
Mon, Sep 4, 2:12 PM
markj committed rG134d36676e94: geli: Fix geli setkey -J for detached providers (authored by Arjan de Vet <freebsd@devet.org>).
geli: Fix geli setkey -J for detached providers
Mon, Sep 4, 2:12 PM
markj accepted D41656: virtio_random: Pipeline fetching the data.
Mon, Sep 4, 2:07 PM
markj added inline comments to D41344: radix_trie: have vm_radix use pctrie code.
Mon, Sep 4, 1:31 PM
markj accepted D41467: syslogd: Enter capability mode.
Mon, Sep 4, 1:23 PM
markj accepted D41703: syslogd: Use anonymous union in struct filed.
Mon, Sep 4, 1:19 PM
markj accepted D41699: riscv: Save gp in the trapframe in both modes.
Mon, Sep 4, 1:14 PM
markj accepted D41700: riscv: Print stval in dump_regs for fatal exceptions.
Mon, Sep 4, 1:13 PM
markj accepted D41702: riscv: Don't print zero offsets for register addresses.
Mon, Sep 4, 1:10 PM
markj committed rG472fce62bb53: geli tests: Add a regression test for geli setkey -J (authored by markj).
geli tests: Add a regression test for geli setkey -J
Mon, Sep 4, 12:56 PM
markj committed rGb785ac18c7d3: geli: Fix geli setkey -J for detached providers (authored by Arjan de Vet <freebsd@devet.org>).
geli: Fix geli setkey -J for detached providers
Mon, Sep 4, 12:56 PM

Fri, Sep 1

markj committed rGb7000cadfbf3: scandir.3: Fix several typos (authored by trond.endrestol_ximalas.info).
scandir.3: Fix several typos
Fri, Sep 1, 8:58 PM
markj accepted D41686: libc: Fix fmemopen(3) prototype in fopen(3) man page..
Fri, Sep 1, 8:38 PM
markj accepted D41641: timerfd: Namespace pollution adjustments.
Fri, Sep 1, 8:20 PM

Thu, Aug 31

markj added inline comments to D41677: arm64: initialize pcb in the TBI/PAC/etc. fault case.
Thu, Aug 31, 8:41 PM
markj accepted D41677: arm64: initialize pcb in the TBI/PAC/etc. fault case.

Seems ok to me, but:

Thu, Aug 31, 8:30 PM
markj added inline comments to D37428: Import the kernel parts of bhyve/arm64.
Thu, Aug 31, 6:14 PM
markj accepted D41462: intrng: switch from MAXCPU to mp_ncpus.
Thu, Aug 31, 5:55 PM
markj added inline comments to D41673: lib/libc/amd64/string: add timingsafe_bcmp(3) scalar, baseline implementations.
Thu, Aug 31, 5:50 PM
markj added a comment to D41330: netlink: fix adding an interface route.

@melifaro does this change seem correct to you?

Thu, Aug 31, 5:47 PM
markj added inline comments to D41656: virtio_random: Pipeline fetching the data.
Thu, Aug 31, 5:40 PM
markj accepted D41467: syslogd: Enter capability mode.
Thu, Aug 31, 2:48 PM
markj accepted D41463: syslogd: Create syslogd libcasper service.
Thu, Aug 31, 2:47 PM
markj added inline comments to D41465: syslogd: Log messages using libcasper.
Thu, Aug 31, 2:44 PM
markj accepted D41671: armv8_crypto: fix recursive fpu_kern_enter call.
Thu, Aug 31, 2:27 PM
markj added inline comments to D41464: syslogd: Refresh configuration using libcasper.
Thu, Aug 31, 2:24 PM
markj committed rG77740df27c10: syslogd: Add some basic regression tests (authored by jfree).
syslogd: Add some basic regression tests
Thu, Aug 31, 1:56 PM
markj committed rGb5f0f20e9bd1: intr: merge interrupt table uses of MAXCOMLEN into INTRNAME_LEN (authored by ehem_freebsd_m5p.com).
intr: merge interrupt table uses of MAXCOMLEN into INTRNAME_LEN
Thu, Aug 31, 1:56 PM
markj added inline comments to D41656: virtio_random: Pipeline fetching the data.
Thu, Aug 31, 1:54 PM
markj accepted D41654: savecore: fsync after writing the core file.
Thu, Aug 31, 1:40 PM
markj added a comment to D41641: timerfd: Namespace pollution adjustments.

Include <sys/time.h> instead of <sys/timespec.h>. This causes intentional namespace pollution that mimics Linux.

Thu, Aug 31, 1:38 PM
markj accepted D41640: timerfd: Relocate 32-bit compat code.
Thu, Aug 31, 1:35 PM
markj accepted D41600: timerfd: Define a locking regime.
Thu, Aug 31, 1:32 PM

Tue, Aug 29

markj accepted D41589: libcrypto: Refactor Makefile.asm so it can be run outside of buildenv.
Tue, Aug 29, 1:43 PM