Page MenuHomeFreeBSD

markj (Mark Johnston)
User

Projects (9)

User Details

User Since
Mar 12 2014, 1:00 AM (569 w, 5 d)

Recent Activity

Today

markj added inline comments to D47917: sound: Allocate vchans on-demand.
Mon, Feb 10, 7:00 PM
markj added a comment to D48899: ktr: Avoid assertion failures from an unlocked STAILQ_EMPTY check.

That said, shouldn't we add the new check to TAILQ_EMPTY() as well, and add TAILQ_EMPTY_ATOMIC() etc.?

Most probably, yes. I'll be off next week, but can take care of that when coming back (not preventing you from doing it in the meantime if you prefer).

Mon, Feb 10, 4:42 PM
markj updated the diff for D48899: ktr: Avoid assertion failures from an unlocked STAILQ_EMPTY check.

Add *_EMPTY_ATOMIC variants for other list types.

Mon, Feb 10, 4:41 PM
markj updated the diff for D48270: vmm: Consolidate code which manages guest memory regions.

Clarify locking in vm_cleanup(), add assertions for the !destroy case.

Mon, Feb 10, 4:33 PM
markj added inline comments to D48270: vmm: Consolidate code which manages guest memory regions.
Mon, Feb 10, 4:29 PM
markj requested review of D48911: pf: Log the intended action when a NAT rule matches a packet.
Mon, Feb 10, 3:53 PM
markj added a comment to D48908: bhyve: Avoid holding /dev/pci open unnecessarily.
In D48908#1115607, @jhb wrote:

Would it be simpler to just avoid using pcifd at all?

Mon, Feb 10, 3:44 PM
markj added inline comments to D48755: linuxkpi: Add `woken_wake_function()` and `wait_woken()` .
Mon, Feb 10, 3:19 PM
markj requested review of D48908: bhyve: Avoid holding /dev/pci open unnecessarily.
Mon, Feb 10, 3:18 AM
markj updated the diff for D48270: vmm: Consolidate code which manages guest memory regions.

Restore vm_gla2gpa_nofault() on arm64 and riscv. Fix an apparent bug in the
former where the return value was not being checked.

Mon, Feb 10, 2:12 AM

Yesterday

markj added a comment to D48899: ktr: Avoid assertion failures from an unlocked STAILQ_EMPTY check.

Thanks. I think it's worthwhile to keep the added check, which can't be guaranteed to work when an atomic access is necessary without resorting to more drastic measures.

Certainly, I'd rather have an explicit variant for atomic checks anyway.

Sun, Feb 9, 3:57 PM
markj added a comment to D48899: ktr: Avoid assertion failures from an unlocked STAILQ_EMPTY check.

Thanks. I think it's worthwhile to keep the added check, which can't be guaranteed to work when an atomic access is necessary without resorting to more drastic measures.

Sun, Feb 9, 3:56 PM
markj added a comment to D48523: linuxkpi: Introduce a properly typed jiffies.
In D48523#1114719, @kib wrote:
In D48523#1111731, @kib wrote:

I think pretty much every use of msecs_to_jiffies() in mlx5 needs to be fixed. Every reference to jiffies needs to be audited too, at least.

Ok I will do the pass.

This diff + D48878 passed internal NVidia verification for mlx5.

Sun, Feb 9, 3:46 PM
markj accepted D48873: Fix remaining zgrep(1) wrapper script regressions.
Sun, Feb 9, 3:24 PM
markj added a comment to D48903: mtw: Fix mtw_mcu_radio().
In D48903#1115325, @jsm wrote:

good catch! @jsm what do you think?

Acutally it looks like I mixed up some functions here, the radio cmd in OpenBSD is different (8 here and 20 in OpenBSD) as Mark noted. And we should also probably add a disable radio like here if (disable) {

		/* disable radio */
		error = mtw_mcu_radio(sc, 0x30, 0x1);
		usbd_delay_ms(sc->sc_udev, 10);

}
https://github.com/openbsd/src/blob/2347e6edcd5e06bc97a8ea95077c54fc76d2e634/sys/dev/usb/if_mtw.c#L3344C1-L3348C3
Thanks

Sun, Feb 9, 3:14 PM

Sat, Feb 8

markj accepted D48588: pctrie: add parent pointer to nodes.
Sat, Feb 8, 8:36 PM
markj requested review of D48903: mtw: Fix mtw_mcu_radio().
Sat, Feb 8, 7:14 PM
markj added inline comments to D48588: pctrie: add parent pointer to nodes.
Sat, Feb 8, 5:45 PM
markj requested review of D48899: ktr: Avoid assertion failures from an unlocked STAILQ_EMPTY check.
Sat, Feb 8, 4:20 PM
markj committed rGc2de0116c801: vmm tests: Check for /dev/vmmctl to decide if vmm is initialized (authored by markj).
vmm tests: Check for /dev/vmmctl to decide if vmm is initialized
Sat, Feb 8, 4:06 PM
markj committed rG400ce6248be9: fibs_multibind_test: Explicitly cast the checksum value (authored by markj).
fibs_multibind_test: Explicitly cast the checksum value
Sat, Feb 8, 4:06 PM
markj accepted D48863: sockets: make pr_aio_queue the default method.
Sat, Feb 8, 4:04 PM
markj added a comment to D48588: pctrie: add parent pointer to nodes.

I read through most of the updated routines and find the new version to be a fair bit easier to read.

Sat, Feb 8, 3:59 PM

Fri, Feb 7

markj accepted D48895: mtw(4) remove misplaced DEBUG_FLAGS and add missing man link to if_mtw.

I already fixed the man page link (was doing some bugzilla triage today), but the makefile change looks good to me!

Fri, Feb 7, 8:18 PM · wireless
markj committed rGf468366ecc8c: mtw.4: Fix a couple of nits (authored by markj).
mtw.4: Fix a couple of nits
Fri, Feb 7, 7:51 PM
markj committed rGd3b1dcb408e2: acpi_iicbus: install the address space handler by default (authored by ahmadkhalifa570_gmail.com).
acpi_iicbus: install the address space handler by default
Fri, Feb 7, 6:44 PM
markj committed rG546b910adb88: savecore: add a test for LOG_PERROR (authored by stephane.rochoy_stormshield.eu).
savecore: add a test for LOG_PERROR
Fri, Feb 7, 2:48 PM
markj committed rGdff74bbdd060: savecore: add missing call to cap_openlog when in capabilities mode (authored by stephane.rochoy_stormshield.eu).
savecore: add missing call to cap_openlog when in capabilities mode
Fri, Feb 7, 2:48 PM
markj committed rG794283f94b82: thread: Add a missing include of asan.h (authored by markj).
thread: Add a missing include of asan.h
Fri, Feb 7, 2:48 PM
markj committed rG0db4588bbe6e: thread: Simplify sanitizer integration with thread creation (authored by markj).
thread: Simplify sanitizer integration with thread creation
Fri, Feb 7, 2:48 PM
markj committed rG703c6df05260: ure(4): Add support for ELECOM EDC-QUA3C (authored by FUKAUMI Naoki <naoki@radxa.com>).
ure(4): Add support for ELECOM EDC-QUA3C
Fri, Feb 7, 2:48 PM
markj committed rGbd8c494343ae: dtrace: Use size_t instead of uintptr_t to represent buffer offsets (authored by markj).
dtrace: Use size_t instead of uintptr_t to represent buffer offsets
Fri, Feb 7, 2:48 PM
markj committed rGaf5c7b110f47: bhyve: Implement the libslirp notify callback (authored by markj).
bhyve: Implement the libslirp notify callback
Fri, Feb 7, 2:48 PM
markj committed rGf34a651bc988: bhyve: Use a non-blocking read in slirp_recv() (authored by markj).
bhyve: Use a non-blocking read in slirp_recv()
Fri, Feb 7, 2:48 PM
markj committed rG028323d61649: setfib.2: Note that the number of FIBs can be adjusted after boot (authored by markj).
setfib.2: Note that the number of FIBs can be adjusted after boot
Fri, Feb 7, 2:47 PM

Thu, Feb 6

markj committed rG23a28fe7776f: libvmmapi: Fix auto-loading of vmm.ko (authored by markj).
libvmmapi: Fix auto-loading of vmm.ko
Thu, Feb 6, 4:28 PM
markj committed rG4cfdfd34472b: RELNOTES: Add an entry for the new bind_all_fibs tunables (authored by markj).
RELNOTES: Add an entry for the new bind_all_fibs tunables
Thu, Feb 6, 4:28 PM
markj committed rG3fa552149885: tools: Add a small program to demonstrate FIB handling in bind(2) (authored by markj).
tools: Add a small program to demonstrate FIB handling in bind(2)
Thu, Feb 6, 4:28 PM
markj committed rG7a66b3008693: pf: Stop using net_epoch to synchronize access to eth rules (authored by markj).
pf: Stop using net_epoch to synchronize access to eth rules
Thu, Feb 6, 4:28 PM
markj committed rGca94f92c23fd: inpcb: Move the definition of struct inpcblbgroup to in_pcb_var.h (authored by markj).
inpcb: Move the definition of struct inpcblbgroup to in_pcb_var.h
Thu, Feb 6, 4:28 PM
markj committed rGee951eb59f21: socket: Add an option to retrieve a socket's FIB number (authored by markj).
socket: Add an option to retrieve a socket's FIB number
Thu, Feb 6, 4:28 PM
markj committed rG7034563f8ef3: tests: Add some FIB multibind test cases (authored by markj).
tests: Add some FIB multibind test cases
Thu, Feb 6, 4:28 PM
markj committed rG4009a98fe80b: rawip: Add a bind_all_fibs sysctl (authored by markj).
rawip: Add a bind_all_fibs sysctl
Thu, Feb 6, 4:28 PM
markj closed D48797: libvmmapi: Fix auto-loading of vmm.ko.
Thu, Feb 6, 4:28 PM
markj committed rGcaccbaef8e26: socket: Move SO_SETFIB handling to protocol layers (authored by markj).
socket: Move SO_SETFIB handling to protocol layers
Thu, Feb 6, 4:28 PM
markj closed D48822: pf: Stop using net_epoch to synchronize access to eth rules.
Thu, Feb 6, 4:28 PM
markj closed D48834: socket: Add an option to retrieve a socket's FIB number.
Thu, Feb 6, 4:28 PM
markj closed D48707: rawip: Add a bind_all_fibs mode.
Thu, Feb 6, 4:28 PM
markj committed rG08e638c089ab: udp: Add a sysctl to modify listening socket FIB inheritance (authored by markj).
udp: Add a sysctl to modify listening socket FIB inheritance
Thu, Feb 6, 4:28 PM
markj committed rG5dc99e9bb985: tcp: Add a sysctl to modify listening socket FIB inheritance (authored by markj).
tcp: Add a sysctl to modify listening socket FIB inheritance
Thu, Feb 6, 4:28 PM
markj closed D48666: socket: Move SO_SETFIB handling to protocol layers.
Thu, Feb 6, 4:28 PM
markj closed D48665: tests: Add some FIB multibind test cases.
Thu, Feb 6, 4:28 PM
markj committed rGda806e8db685: inpcb: Add FIB-aware inpcb lookup (authored by markj).
inpcb: Add FIB-aware inpcb lookup
Thu, Feb 6, 4:28 PM
markj closed D48664: udp: Add a sysctl to control binding.
Thu, Feb 6, 4:28 PM
markj closed D48663: tcp: Add a sysctl to affect listening socket behaviour.
Thu, Feb 6, 4:28 PM
markj committed rGbbd0084baf75: inpcb: Add a flags parameter to in_pcbbind() (authored by markj).
inpcb: Add a flags parameter to in_pcbbind()
Thu, Feb 6, 4:28 PM
markj closed D48662: inpcb: Add FIB-aware inpcb lookup.
Thu, Feb 6, 4:28 PM
markj committed rG9a4131629bb3: inpcb: Imbue in(6)_pcblookup_local() with a FIB parameter (authored by markj).
inpcb: Imbue in(6)_pcblookup_local() with a FIB parameter
Thu, Feb 6, 4:28 PM
markj closed D48661: inpcb: Add a flags parameter to in_pcbbind().
Thu, Feb 6, 4:28 PM
markj closed D48660: inpcb: Imbue in(6)_pcblookup_local() with a FIB parameter.
Thu, Feb 6, 4:27 PM

Tue, Feb 4

markj added a comment to D48834: socket: Add an option to retrieve a socket's FIB number.

If we move the fib to pcb would we be able to preserve this API? Of course it would start fail on sockets where FIB doesn't make any sense. But would we be able to preserve for meaningful sockets?

Tue, Feb 4, 7:09 PM
markj committed R11:7b3028c92e9d: security/sssd2: Fix two bugs in the find_uid patch (authored by markj).
security/sssd2: Fix two bugs in the find_uid patch
Tue, Feb 4, 6:00 PM
markj requested review of D48835: tools: Add a small program to demonstrate FIB handling in bind(2).
Tue, Feb 4, 5:13 PM
markj requested review of D48834: socket: Add an option to retrieve a socket's FIB number.
Tue, Feb 4, 5:11 PM
markj abandoned D48833: socket: Move SO_SETFIB handling to protocol layers.

Wrong diff, sorry for the noise.

Tue, Feb 4, 5:11 PM
markj requested review of D48833: socket: Move SO_SETFIB handling to protocol layers.
Tue, Feb 4, 5:11 PM
markj updated the diff for D48797: libvmmapi: Fix auto-loading of vmm.ko.

Fix a silly mistake in the legacy vm_create().

Tue, Feb 4, 3:31 PM

Mon, Feb 3

markj added inline comments to D45409: vm_phys: reduce touching of page->pool fields.
Mon, Feb 3, 10:04 PM
markj added inline comments to D45409: vm_phys: reduce touching of page->pool fields.
Mon, Feb 3, 10:00 PM
markj added inline comments to D45409: vm_phys: reduce touching of page->pool fields.
Mon, Feb 3, 9:32 PM
markj requested review of D48822: pf: Stop using net_epoch to synchronize access to eth rules.
Mon, Feb 3, 7:30 PM
markj added inline comments to D48808: riscv vmm: various fixes in APLIC.
Mon, Feb 3, 6:23 PM
markj added inline comments to D48808: riscv vmm: various fixes in APLIC.
Mon, Feb 3, 2:58 PM
markj accepted D48807: bhyve/riscv: fix HSM.
Mon, Feb 3, 2:55 PM
markj accepted D48811: bhyve: Sleep for a short period after VM_EXITCODE_DEBUG exits.
Mon, Feb 3, 2:45 PM
markj added a comment to D48588: pctrie: add parent pointer to nodes.

With this patch {F108995447} from @alc

Mon, Feb 3, 2:44 PM

Sun, Feb 2

markj added a comment to D48797: libvmmapi: Fix auto-loading of vmm.ko.

I don't really like the auto-loading, but that's how it used to work, and changing it in 15.0 would be a POLA violation.

Sun, Feb 2, 9:51 PM
markj requested review of D48797: libvmmapi: Fix auto-loading of vmm.ko.
Sun, Feb 2, 9:45 PM

Sat, Feb 1

markj added inline comments to D48709: sockets: refactor solisten() and pr_listen.
Sat, Feb 1, 5:20 PM

Fri, Jan 31

markj added inline comments to D48709: sockets: refactor solisten() and pr_listen.
Fri, Jan 31, 9:55 PM
markj committed rGc55f457df678: ip: Defer checks for an unspecified dstaddr until after pfil hooks (authored by markj).
ip: Defer checks for an unspecified dstaddr until after pfil hooks
Fri, Jan 31, 7:18 PM
markj committed rGafc38f1f23ac: sysctl: Add a regression test which runs sysctl -a (authored by markj).
sysctl: Add a regression test which runs sysctl -a
Fri, Jan 31, 3:23 PM
markj closed D48659: sysctl: Add a regression test which runs sysctl -a.
Fri, Jan 31, 3:23 PM
markj added inline comments to D48709: sockets: refactor solisten() and pr_listen.
Fri, Jan 31, 2:41 PM

Thu, Jan 30

markj accepted D48729: netstat: restore printing the "default" keyword, provide -nn option.
Thu, Jan 30, 6:36 PM
markj added inline comments to D48709: sockets: refactor solisten() and pr_listen.
Thu, Jan 30, 6:35 PM
markj added a comment to D48729: netstat: restore printing the "default" keyword, provide -nn option.

Some test cases need to be updated (again). See commits 81b076e43aa63ac255996093233ab3560a23977e and e090646d6f5a4a6848ecd4bcb1f2db498ea3b3e2.

Thu, Jan 30, 5:06 PM
markj added inline comments to D48709: sockets: refactor solisten() and pr_listen.
Thu, Jan 30, 5:05 PM
markj added inline comments to D48587: xinstall: Validate numeric uids/gids.
Thu, Jan 30, 4:54 PM
markj updated the diff for D48523: linuxkpi: Introduce a properly typed jiffies.

Update round_jiffies() as well.

Thu, Jan 30, 4:22 PM
markj added a comment to D48523: linuxkpi: Introduce a properly typed jiffies.
In D48523#1111725, @kib wrote:

When testing the new patch, I encountered the same issue where the mce0 device disappeared,
accompanied by the following error in the dmesg log.

Thank you for testing. It seems that mlx5 (and probably mlx4) assume that jiffies is an int in some places, e.g., wait_func(). I am willing to work on converting these to unsigned long where appropriate. Presumably this would ease maintenance of shared code, but I am not the maintainer of these drivers - how would you like to handle it?

Indeed I tried the following but it was not enough at least.

Thu, Jan 30, 4:21 PM
markj committed rGd45326bfeab7: hwpstate_amd: add amd pstate for zen5 (authored by aokblast).
hwpstate_amd: add amd pstate for zen5
Thu, Jan 30, 4:09 PM
markj committed rG2839e32a597e: hwpstate_amd: add amd pstate for zen4 (authored by aokblast).
hwpstate_amd: add amd pstate for zen4
Thu, Jan 30, 4:09 PM
markj committed rG3d30774f0056: pf: Force logging if pf_create_state() fails (authored by markj).
pf: Force logging if pf_create_state() fails
Thu, Jan 30, 4:09 PM
markj committed rGb4da6ed899d7: dtrace tests: Fix the ATF config variable name (authored by markj).
dtrace tests: Fix the ATF config variable name
Thu, Jan 30, 4:09 PM
markj committed rG9c60e4e4e721: dtrace tests: Avoid hard-coding paths to required programs (authored by markj).
dtrace tests: Avoid hard-coding paths to required programs
Thu, Jan 30, 4:09 PM
markj committed rG89073e0ad444: dtrace: Fix the definition of pc_t (authored by markj).
dtrace: Fix the definition of pc_t
Thu, Jan 30, 4:09 PM
markj committed rG18a39dedbdd3: dtrace: Build systrace_freebsd32 only if COMPAT_FREEBSD32 is configured (authored by markj).
dtrace: Build systrace_freebsd32 only if COMPAT_FREEBSD32 is configured
Thu, Jan 30, 4:09 PM
markj committed rG2e0d595651b4: dtrace/arm64: Simplify variable declarations in the invop handler (authored by markj).
dtrace/arm64: Simplify variable declarations in the invop handler
Thu, Jan 30, 4:09 PM
markj committed rGd603868ddcd9: libdtrace: Use C99 designated initializers for dt_idops_t (authored by markj).
libdtrace: Use C99 designated initializers for dt_idops_t
Thu, Jan 30, 4:09 PM