Page MenuHomeFreeBSD

brooks (Brooks Davis)
User

Projects (3)

User Details

User Since
Jul 1 2014, 6:02 PM (646 w, 2 d)

Recent Activity

Today

brooks accepted D61774: freebsd32: Fix the ffclock estimate conversions.
Fri, Nov 20, 3:20 PM
brooks committed rG3f10dc83c790: swapoff: release vnode held since swapon.
Fri, Nov 20, 10:27 AM
brooks committed rGd6f99f67507d: swapoff: don't leak vnode pointer to device.
Fri, Nov 20, 10:26 AM
brooks committed rGf30c296f200e: swapoff: release vnode held since swapon.
Fri, Nov 20, 9:31 AM
brooks committed rG87599597b02e: swapoff: don't leak vnode pointer to device.
Fri, Nov 20, 9:30 AM
brooks accepted D61774: freebsd32: Fix the ffclock estimate conversions.

I was pondering if there was an engineering practice that let types prevent this error and realized there is. If we'd habitually split these thunks into e.g., void ffe2ffe32(const struct ffclock_estimate*, struct ffclock_estimate32*) and vs versa then const would have been on our side.

Fri, Nov 20, 9:16 AM

Wed, Nov 18

brooks committed R11:80c7dbf9a8d0: devel/llvm23: 23.1.6 release.
Wed, Nov 18, 7:23 PM
brooks added a comment to D61531: New port: devel/llvm-apple..

I assume this was bootstrapped from devel/llvm21? If so, when committing, please follow my usual practice when adding LLVM ports of copying the directory unmodified and then committing the actual port on top and hooking it up. This aids git's rename detection.

It was actually bootstrapped from llvm-devel. Maybe I should have bootstrapped it from llvm21, I’m not sure.
Anyway, I don’t have a commit bit anymore.

Wed, Nov 18, 3:56 PM
brooks accepted D61675: syscalls.master: Add a capsicum note about sync(2).
Wed, Nov 18, 12:16 PM
brooks added a comment to D61675: syscalls.master: Add a capsicum note about sync(2).

Seems reasonable. I do wonder if there are any side channels related for forced I/O, but I doubt they are interesting vs e.g., things you can do with CPU load.

Wed, Nov 18, 12:16 PM

Tue, Nov 17

brooks updated subscribers of D61661: [RFC] Introduce Solaris-style Door IPC subsystem (preview).

Wow, this is a lot of work. In CheriBSD, I suspect this will be quite interesting for our work on co-processes (applications consisting of multiple processes in a single address space). I know @jhb has suggested this might be a better set of interfaces than some of the ones we've developed.

Tue, Nov 17, 2:45 PM
brooks committed R11:dae9009308e8: net-im/pidgin-sipe: update to 1.25.0.
Tue, Nov 17, 1:01 PM
brooks committed R11:3aecfc549942: lang/opensycl: allow a range of LLVM versions.
Tue, Nov 17, 1:01 PM
brooks committed R11:25d7f5884e4b: japanese/Wnn7-lib: DEPRECATE.
Tue, Nov 17, 1:01 PM
brooks committed R11:94dce73068a9: devel/llvm12: thin the port further.
Tue, Nov 17, 12:47 PM
brooks committed R11:479e29396ceb: devel/ikos: mark DEPRECATED.
Tue, Nov 17, 12:47 PM
brooks committed R11:e3b935bfa66a: editors/hexer: mark DEPRECATED.
Tue, Nov 17, 12:47 PM
brooks committed R11:2e6a84eecb38: devel/llvm13: mark DEPRECATED.
Tue, Nov 17, 12:47 PM
brooks committed R11:67cdb99bb1b4: devel/spirv-llvm-translator: drop support for LLVM 12 13 14.
Tue, Nov 17, 12:47 PM

Mon, Nov 16

brooks committed rG766061eea071: modstat: check errno before calling feature_present.
Mon, Nov 16, 6:13 PM
brooks committed rG9d6b8de8c225: modstat: allow compat support to fail in tests.
Mon, Nov 16, 6:13 PM
brooks committed rG3b321d5d6136: modstat: don't copyout while holding a lock.
Mon, Nov 16, 6:13 PM
brooks committed rG776493a19330: modstat: avoid copying from freed memory.
Mon, Nov 16, 6:13 PM
brooks committed rGec213d186be9: kern_module.c: add kern_ implementations.
Mon, Nov 16, 6:13 PM
brooks committed rGa04f2d46dda2: extattr_list*: avoid panic with absurd buffer lengths.
Mon, Nov 16, 6:13 PM
brooks committed rG37d71d6f1c6a: bsd.sys.mk: suppress some clang warnings for C++ for >= 19 (authored by arichardson).
Mon, Nov 16, 1:22 PM
brooks added a comment to D61531: New port: devel/llvm-apple..

Seems like a useful thing to have around.

Mon, Nov 16, 11:49 AM
brooks committed rGed8dc5e715ff: modstat: check errno before calling feature_present.
Mon, Nov 16, 10:46 AM
brooks committed rG4a9770b9256c: modstat: allow compat support to fail in tests.
Mon, Nov 16, 10:46 AM
brooks committed rG044b90b0f78f: modstat: don't copyout while holding a lock.
Mon, Nov 16, 10:46 AM
brooks committed rG80fed5b6b5bd: modstat: avoid copying from freed memory.
Mon, Nov 16, 10:46 AM
brooks committed rG5c7d799c53cc: kern_module.c: add kern_ implementations.
Mon, Nov 16, 10:46 AM
brooks committed rG0967b36fe783: extattr_list*: avoid panic with absurd buffer lengths.
Mon, Nov 16, 10:46 AM
brooks committed rG933f5f616a8a: swapoff: release vnode held since swapon.
Mon, Nov 16, 9:59 AM
brooks closed D61465: swapoff: release vnode held since swapon.
Mon, Nov 16, 9:59 AM

Fri, Nov 13

brooks committed rG143da26632a9: kern_revoke: remove always UIO_USERSPACE pathseg argument.
Fri, Nov 13, 3:39 PM
brooks added a comment to D61465: swapoff: release vnode held since swapon.

After I pushed D61464 llm-postcr reanalyzed this file and gave a higher quality report that reflects our conversation.

POSTCR-9506d8e0b1788068cbd4c2cbc984cb6ee0dcb0ab-1: pre-existing: swapoff of a                 
GEOM-backed swap device never releases the devfs vnode reference taken by swapon
Fri, Nov 13, 10:32 AM
brooks committed rGacd270c8c654: modstat: check errno before calling feature_present.
Fri, Nov 13, 10:27 AM
brooks committed rG70cf77eaefa8: aio_cancel(2): add kern_aio_cancel.
Fri, Nov 13, 10:21 AM
brooks committed rG9506d8e0b178: swapoff: don't leak vnode pointer to device.
Fri, Nov 13, 10:07 AM
brooks closed D61464: swapoff: don't leak vnode pointer to device.
Fri, Nov 13, 10:07 AM
brooks committed rG118067f51c11: extattr_list*: avoid panic with absurd buffer lengths.
Fri, Nov 13, 10:07 AM
brooks committed rGef726c645300: modstat: allow compat support to fail in tests.
Fri, Nov 13, 10:07 AM
brooks closed D61461: extattr_list*: avoid panic with absurd buffer lengths.
Fri, Nov 13, 10:07 AM
brooks closed D61460: modstat: allow compat support to fail in tests.
Fri, Nov 13, 10:07 AM
brooks updated the summary of D61465: swapoff: release vnode held since swapon.
Fri, Nov 13, 10:05 AM
brooks updated the diff for D61465: swapoff: release vnode held since swapon.

Remove comment about swapgeom_close being called by biodone.

Fri, Nov 13, 10:05 AM

Thu, Nov 12

brooks updated the diff for D61460: modstat: allow compat support to fail in tests.

Make it an error if modstat fails with a v2 struct when compat_freebsd13 is present

Thu, Nov 12, 8:17 PM
brooks updated the diff for D61465: swapoff: release vnode held since swapon.

spelling

Thu, Nov 12, 2:49 PM
brooks added inline comments to D61464: swapoff: don't leak vnode pointer to device.
Thu, Nov 12, 2:19 PM
brooks updated the diff for D61464: swapoff: don't leak vnode pointer to device.

Host the vrele out of the swap device lock region

Thu, Nov 12, 2:19 PM
brooks updated the summary of D61464: swapoff: don't leak vnode pointer to device.
Thu, Nov 12, 2:18 PM
brooks added a comment to D61465: swapoff: release vnode held since swapon.

postcr report

POSTCR-965170da4548176cb0ec4ee65e01248af37f8824-2: pre-existing:                              
swapgeom_close() never releases the sw_vp reference that kern_swapon() keeps for              
GEOM swap devices
Thu, Nov 12, 1:55 PM
brooks added a comment to D61464: swapoff: don't leak vnode pointer to device.

postcr report:

POSTCR-965170da4548176cb0ec4ee65e01248af37f8824-1: pre-existing: kern_swapoff()               
never vrele()s the vnode returned by namei(), leaking a usecount on every call                
where the lookup succeeds
Thu, Nov 12, 1:54 PM
brooks requested review of D61465: swapoff: release vnode held since swapon.
Thu, Nov 12, 1:54 PM
brooks requested review of D61464: swapoff: don't leak vnode pointer to device.
Thu, Nov 12, 1:53 PM
brooks added a comment to D61461: extattr_list*: avoid panic with absurd buffer lengths.

llm-postcr says:

POSTCR-f20c712f167acd692240350179d4d98d33971375-1: pre-existing:                       
extattr_list_{fd,file,link} with nbytes > SSIZE_MAX makes uio_resid negative,          
which gets past the IOSIZE_MAX check and trips the uiomove() KASSERT (panic            
triggerable by an unprivileged user on INVARIANTS kernels)
Thu, Nov 12, 1:22 PM
brooks requested review of D61461: extattr_list*: avoid panic with absurd buffer lengths.
Thu, Nov 12, 1:19 PM
brooks added a comment to D61460: modstat: allow compat support to fail in tests.

llm-postcr correctly reported that the v1 tests would now fail after 1ebf46ecb627 (thought it didn't realize they would pass on armv7 and powerpc). It also didn't report that v2 tests would fail on a kernel without COMPAT_FREEBSD13. I've update both to allow support to not be present.

Thu, Nov 12, 1:17 PM
brooks requested review of D61460: modstat: allow compat support to fail in tests.
Thu, Nov 12, 1:13 PM
brooks committed rGb75412fea1da: kern_undelete: remove unused pathseg argument.
Thu, Nov 12, 1:00 PM
brooks committed rGd60162816083: freebsd14_[gs]groups: add kern_ and user_ implementations.
Thu, Nov 12, 11:02 AM
brooks committed rG30d759067be9: pdwait(2): add user_pdwait.
Thu, Nov 12, 11:02 AM
brooks committed rG9f47314ebdb6: pdrfork(2): add kern_pdrfork.
Thu, Nov 12, 11:02 AM
brooks committed rG965170da4548: swap{on,off}(2): add/expose kern_ implementations.
Thu, Nov 12, 11:02 AM
brooks committed rGb1482e09830e: SysV SHM: make kern_shmat static.
Thu, Nov 12, 11:02 AM
brooks closed D61432: modstat: don't copyout while holding a lock.
Thu, Nov 12, 10:24 AM
brooks committed rG1ebf46ecb627: modstat: don't copyout while holding a lock.
Thu, Nov 12, 10:24 AM
brooks committed rG622365caf3bd: ntp_gettime(2): add kern_ntp_gettime.
Thu, Nov 12, 10:24 AM
brooks committed rG7273d80ba18c: vfs_syscalls.c: add kern_ and user_ implementations.
Thu, Nov 12, 10:24 AM
brooks committed rG87d8a910d30f: nmount(2): add kern_nmount.
Thu, Nov 12, 10:24 AM
brooks committed rG7ca5de287e58: __getcwd(2), __getrealpath(2); add kern_ implementations.
Thu, Nov 12, 10:24 AM
brooks committed rGf20c712f167a: extattr*(2): add kern_ and user_ implementaions.
Thu, Nov 12, 10:24 AM

Wed, Nov 11

brooks updated the diff for D61432: modstat: don't copyout while holding a lock.

freebsd32_modstat: Don't unlock twice, do free statbuf

Wed, Nov 11, 3:12 PM
brooks added a comment to D61432: modstat: don't copyout while holding a lock.

Thanks for the review. That error I made probably argues for further consolidation, but I'll hold off for the moment.

Wed, Nov 11, 3:11 PM
brooks committed rG431cc94c0395: user_connectat: correctly pass fd through.
Wed, Nov 11, 2:58 PM
brooks committed rG6e87733f73ff: __acl_*(2): add kern_ implementations.
Wed, Nov 11, 1:21 PM
brooks committed rG65e94d4621c8: uipc_syscalls.c: add kern_ and user_ implementations.
Wed, Nov 11, 1:21 PM
brooks committed rG0e100820c4bb: uipc_shm.c: add kern_ implementations.
Wed, Nov 11, 1:21 PM
brooks committed rG1b2f03fe94ef: pipe2(2): add kern_pipe2.
Wed, Nov 11, 1:21 PM
brooks committed rG22dd3a0df230: pdgetpid(2): add user_pdgetpid.
Wed, Nov 11, 1:21 PM
brooks added a comment to D61432: modstat: don't copyout while holding a lock.

The second LLM review convinced me to fix this more properly.

Wed, Nov 11, 12:25 PM
brooks updated the summary of D61432: modstat: don't copyout while holding a lock.
Wed, Nov 11, 12:22 PM
brooks requested review of D61432: modstat: don't copyout while holding a lock.
Wed, Nov 11, 12:21 PM
brooks committed rG1eada4295600: getrandom(2): add kern_getrandom.
Wed, Nov 11, 9:59 AM
brooks committed rG9acec4c86370: sys_generic.c: add kern_ and user_ implementations.
Wed, Nov 11, 9:59 AM
brooks committed rG974320334844: sys_capability.c: add kern_ and user_ implementations.
Wed, Nov 11, 9:59 AM
brooks committed rG1c17a0983ac3: profil(2): add kern_ implemenation.
Wed, Nov 11, 9:59 AM
brooks committed rGe5100cd50a05: sched_*(2): add user_ implementations.
Wed, Nov 11, 9:59 AM
brooks closed D61398: modstat: avoid copying from freed memory.
Wed, Nov 11, 9:58 AM
brooks committed rG7f8c2056db1e: modstat: avoid copying from freed memory.
Wed, Nov 11, 9:58 AM
brooks added a comment to D61398: modstat: avoid copying from freed memory.

I think this is okay. It might not be a good idea to hold this global lock across copyin/copyout. If someone runs kldstat or similar and the kernel needs to fault in the user output buffer, then we could block under memory pressure with that lock held, which could further hurt responsiveness of the system.

Wed, Nov 11, 9:36 AM

Tue, Nov 10

brooks updated the diff for D61398: modstat: avoid copying from freed memory.
  • Handle freebsd32_modstat
  • Use the same pattern for the lookup failure
Tue, Nov 10, 6:32 PM
brooks updated the summary of D61398: modstat: avoid copying from freed memory.
Tue, Nov 10, 4:39 PM
brooks updated the diff for D61398: modstat: avoid copying from freed memory.

use goto out pattern

Tue, Nov 10, 4:38 PM
brooks committed rG3d101bbff45e: uuidgen(2): add user_ implementation.
Tue, Nov 10, 3:55 PM
brooks committed rG22f93edf2014: kern_time.c: add kern_ and user_ implementations.
Tue, Nov 10, 3:55 PM
brooks committed rG698ba43e0ec7: sig*(2): add kern_ and user_ implementations.
Tue, Nov 10, 3:55 PM
brooks committed rGa81038c0dc02: thr_set_name(2): add kern_ implementation.
Tue, Nov 10, 3:55 PM
brooks committed rG23abe1880071: __sysctl(2): add kern_ implementation.
Tue, Nov 10, 3:55 PM