On Wed, Jun 08, 2022 at 06:50:18AM +0200, Anton Lindqvist wrote:
> On Sun, May 01, 2022 at 04:17:34PM +0000, Visa Hankala wrote:
> > On Sat, Apr 30, 2022 at 09:40:24AM +0200, Anton Lindqvist wrote:
> > > On Sun, Mar 13, 2022 at 04:17:07PM +0100, Mark Kettenis wrote:
> >
bulk build on octeon.ports.openbsd.org
started on Sun May 29 16:05:09 UTC 2022
finished at Sun Jun 5 17:45:36 UTC 2022
lasted 08D01h40m
done with kern.version=OpenBSD 7.1-current (GENERIC.MP) #948: Sat May 28
21:07:21 MDT 2022
built packages:8758
May 29:2056
May 30:962
May 31:525
Jun 1:475
Jun
On Mon, Jun 06, 2022 at 06:47:32AM +1000, David Gwynne wrote:
> On Sun, Jun 05, 2022 at 03:57:39PM +0000, Visa Hankala wrote:
> > On Sun, Jun 05, 2022 at 12:27:32PM +0200, Martin Pieuchot wrote:
> > > On 05/06/22(Sun) 05:20, Visa Hankala wrote:
> > > > E
On Sun, Jun 05, 2022 at 12:27:32PM +0200, Martin Pieuchot wrote:
> On 05/06/22(Sun) 05:20, Visa Hankala wrote:
> > Encountered the following panic:
> >
> > panic: kernel diagnostic assertion "(p->p_flag & P_TIMEOUT) == 0" failed:
> > file "/usr
Encountered the following panic:
panic: kernel diagnostic assertion "(p->p_flag & P_TIMEOUT) == 0" failed: file
"/usr/src/sys/kern/kern_synch.c", line 373
Stopped at db_enter+0x10: popq%rbp
TIDPIDUID PRFLAGS PFLAGS CPU COMMAND
423109 57118 55 0x3
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/06/01 08:21:50
Modified files:
regress/sys/kern/flock: Makefile flock.c
Log message:
Add a regress test for the end of lock range ambiguity bug.
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/06/01 08:18:43
Modified files:
sys/kern : vfs_lockf.c
Log message:
Fix ambiguity with lock range end
When the user requests a lock range that ends at LLONG_MAX, replace
the end point with the
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/06/01 08:16:28
Modified files:
sys/kern : vfs_lockf.c
Log message:
Fix lock range start when l_whence == SEEK_END and l_len < 0.
OK anton@
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/05/30 08:06:16
Modified files:
sys/kern : sys_pipe.c
Log message:
Replace selwakeup() with KNOTE() in pipe event activation.
Recommit the reverted change selectively so that only pipes are
bulk build on octeon.ports.openbsd.org
started on Sun May 22 12:27:55 UTC 2022
finished at Sun May 29 15:05:00 UTC 2022
lasted 08D02h37m
done with kern.version=OpenBSD 7.1-current (GENERIC.MP) #940: Sat May 21
21:03:52 MDT 2022
built packages:8745
May 22:2348
May 23:833
May 24:499
May 25:614
On Tue, Feb 08, 2022 at 09:12:11AM +, Visa Hankala wrote:
> Now that poll(2) is based on kqueue, the old, non-MP-safe poll backend
> is not used any longer. Event sources can call KNOTE() directly instead
> of selwakeup().
>
> This diff does the KNOTE() conversion for pi
bulk build on octeon.ports.openbsd.org
started on Sun May 15 08:25:45 UTC 2022
finished at Sun May 22 06:21:20 UTC 2022
lasted 07D21h55m
done with kern.version=OpenBSD 7.1-current (GENERIC.MP) #932: Sat May 14
18:56:38 MDT 2022
built packages:8739
May 15:2553
May 16:678
May 17:629
May 18:606
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/05/21 22:52:17
Modified files:
sys/nfs: nfs_socket.c
Log message:
Lock kernel in nfsrv_rcv() because NFS subsystem is not MP-safe yet.
Tested in snaps for a week.
OK bluhm@
bulk build on octeon.ports.openbsd.org
started on Fri May 6 15:33:54 UTC 2022
finished at Fri May 13 16:07:42 UTC 2022
lasted 08D00h33m
done with kern.version=OpenBSD 7.1-current (GENERIC.MP) #921: Fri May 6
04:30:57 MDT 2022
built packages:8749
May 6:2116
May 7:862
May 8:500
May 9:457
May
The NFS subsystem is not MP-safe yet. Take this into account
in the NFS server socket upcall by locking the kernel.
This might help with the NFS server hanging that was seen recently
as a result of the now-reverted selwakeup()-to-KNOTE() conversion.
Unfortunately, I have not been able to confirm
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/05/12 07:33:00
Modified files:
sys/kern : kern_event.c
Log message:
kqueue: Fix race condition in knote_remove()
Always fetch the knlist array pointer at the start of every iteration
in
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/05/09 08:49:55
Modified files:
sys/kern : sys_pipe.c uipc_socket.c uipc_socket2.c
Log message:
Revert "Replace selwakeup() with KNOTE() in pipe and socket event activation."
The commit caused
Qwinci created this revision.
Herald added subscribers: usaxena95, kadircet, arphaman.
Herald added a project: All.
Qwinci requested review of this revision.
Herald added projects: clang, clang-tools-extra.
Herald added a subscriber: cfe-commits.
Repository:
rG LLVM Github Monorepo
On Sat, May 07, 2022 at 07:47:12PM +1000, Joel Sing wrote:
> After upgrading an OpenBSD arm64 (pine64+ board) Go builder to 7.1, the
> following has been triggered twice (once every couple of days):
>
> panic: kernel diagnostic assertion "kn->kn_kq == kq" failed: file
>
On Sat, May 07, 2022 at 07:47:12PM +1000, Joel Sing wrote:
> After upgrading an OpenBSD arm64 (pine64+ board) Go builder to 7.1, the
> following has been triggered twice (once every couple of days):
>
> panic: kernel diagnostic assertion "kn->kn_kq == kq" failed: file
>
Remove unused struct fileops field fo_poll and callbacks.
(After this, VOP_POLL() is next in line for removal.)
OK?
Index: sys/dev/pci/drm/drm_linux.c
===
RCS file: src/sys/dev/pci/drm/drm_linux.c,v
retrieving revision 1.92
diff -u
Qwinci created this revision.
Herald added subscribers: usaxena95, kadircet, arphaman.
Herald added a project: All.
Qwinci requested review of this revision.
Herald added projects: clang, clang-tools-extra.
Herald added a subscriber: cfe-commits.
Repository:
rG LLVM Github Monorepo
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/05/06 07:12:16
Modified files:
sys/kern : kern_event.c
Log message:
Replace selwakeup() with KNOTE() in kqueue event activation.
The deferred activation can now run in an MP-safe task queue.
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/05/06 07:09:41
Modified files:
sys/kern : sys_pipe.c uipc_socket.c uipc_socket2.c
Log message:
Replace selwakeup() with KNOTE() in pipe and socket event activation.
OK mpi@
On Sat, Apr 30, 2022 at 09:40:24AM +0200, Anton Lindqvist wrote:
> On Sun, Mar 13, 2022 at 04:17:07PM +0100, Mark Kettenis wrote:
> > > Date: Fri, 11 Mar 2022 07:53:13 +0100
> > > From: Anton Lindqvist
> > >
> > > On Tue, Mar 08, 2022 at 01:44:47PM +,
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/04/30 08:44:04
Modified files:
share/man/man9 : refcnt_init.9
sys/kern : kern_synch.c
Log message:
Enforce proper memory ordering in refcnt_rele() and refcnt_finalize()
Make
-user.c
diff -N regress/sys/kern/kqueue/kqueue-user.c
--- /dev/null 1 Jan 1970 00:00:00 -
+++ regress/sys/kern/kqueue/kqueue-user.c 30 Apr 2022 13:33:12 -
@@ -0,0 +1,189 @@
+/* $OpenBSD$ */
+
+/*
+ * Copyright (c) 2022 Visa Hankala
+ *
+ * Permission to use, copy, modify
On Tue, Feb 08, 2022 at 09:12:11AM +, Visa Hankala wrote:
> Now that poll(2) is based on kqueue, the old, non-MP-safe poll backend
> is not used any longer. Event sources can call KNOTE() directly instead
> of selwakeup().
>
> This diff does the KNOTE() conversion for pi
On Thu, Apr 21, 2022 at 11:10:40PM +0200, Mark Kettenis wrote:
> > Date: Thu, 21 Apr 2022 22:17:31 +0200
> > From: Alexander Bluhm
> >
> > On Mon, Apr 18, 2022 at 08:33:06AM +, Visa Hankala wrote:
> > > I think the sanest solution is to add
On Mon, Mar 14, 2022 at 04:14:47PM +, Visa Hankala wrote:
> On Mon, Mar 14, 2022 at 02:01:07AM -0700, Philip Guenther wrote:
> > On Mon, Mar 14, 2022 at 12:47 AM Visa Hankala wrote:
> >
> > > On Sun, Mar 13, 2022 at 06:26:19PM -0700, Philip Guenther wrote:
> >
CVSROOT:/cvs
Module name:www
Changes by: v...@cvs.openbsd.org2022/04/15 09:55:54
Modified files:
. : 71.html
Log message:
mips64 package count
bulk build on octeon.ports.openbsd.org
started on Fri Apr 8 15:58:37 UTC 2022
finished at Fri Apr 15 15:35:46 UTC 2022
lasted 07D23h37m
done with kern.version=OpenBSD 7.1 (GENERIC.MP) #903: Thu Apr 7 20:59:31 MDT
2022
built packages:8782
Apr 8:2083
Apr 9:1089
Apr 10:503
Apr 11:652
Apr 12:830
raid.c. I
> added
> them.
>
> Comments or OK ?
Please move the declarations of `s' next to the other variable
declarations at the starts of the functions. With that, OK visa@
> Index: dev/softraid.c
> ===
> RCS f
On Mon, Apr 11, 2022 at 07:14:46PM +0200, Martin Pieuchot wrote:
> Now that poll(2) & select(2) use the kqueue backend under the hood we
> can start retiring the old machinery.
>
> The diff below does not touch driver definitions, however it :
>
> - kills selrecord() & doselwakeup()
>
> - make
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/04/11 08:30:05
Modified files:
sys/dev/gpio : gpio.c
Log message:
gpio: Add missing device_unref() calls
device_lookup() returns a reference that the caller has to release.
OK jca@
Make gpio(4) release the device reference that device_lookup() takes.
OK?
Index: dev/gpio/gpio.c
===
RCS file: src/sys/dev/gpio/gpio.c,v
retrieving revision 1.16
diff -u -p -r1.16 gpio.c
--- dev/gpio/gpio.c 6 Apr 2022 18:59:28
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/04/10 07:23:14
Modified files:
sys/arch/mips64/mips64: ipifuncs.c
Log message:
mips64: Enforce memory ordering on SMP rendezvous
Make sure that each CPU sees the initiator's pre-action state when
On Thu, Apr 07, 2022 at 07:55:11PM +0200, Alexander Bluhm wrote:
> On Wed, Mar 23, 2022 at 06:13:27PM +0100, Alexander Bluhm wrote:
> > In my opinion tracepoints give insight at minimal cost. It is worth
> > it to have it in GENERIC to make it easy to use.
>
> After release I want to revive the
bulk build on octeon.ports.openbsd.org
started on Sat Mar 26 11:57:09 UTC 2022
finished at Sat Apr 2 08:50:09 UTC 2022
lasted 07D20h53m
done with kern.version=OpenBSD 7.1-beta (GENERIC.MP) #889: Fri Mar 25 20:10:55
MDT 2022
built packages:8766
Mar 26:2431
Mar 27:741
Mar 28:534
Mar 29:479
Mar
bulk build on octeon.ports.openbsd.org
started on Thu Mar 10 09:53:44 UTC 2022
finished at Sat Mar 26 09:24:06 UTC 2022
lasted 16D23h30m
done with kern.version=OpenBSD 7.1-beta (GENERIC.MP) #880: Wed Mar 9 16:52:25
MST 2022
built packages:8758
Mar 10:2468
Mar 11:639
Mar 12:621
Mar 13:585
Mar
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/20 23:45:52
Modified files:
sys/sys: atomic.h
Log message:
Remove data dependency barrier from atomic_load_* functions
This makes the atomic_load_* functions relaxed in terms of memory
Linux uses pseudo-attribute __read_mostly to tag variables that are read
frequently but written seldom. These variables are grouped together
at link time. This improves cache usage by reducing false sharing.
__read_mostly has been adopted by FreeBSD, DragonFly BSD and NetBSD.
The former two also
On Sat, Mar 19, 2022 at 12:10:11AM +0100, Alexander Bluhm wrote:
> On Thu, Mar 17, 2022 at 07:25:27AM +0000, Visa Hankala wrote:
> > On Thu, Mar 17, 2022 at 12:42:13AM +0100, Alexander Bluhm wrote:
> > > I would like to use btrace to debug refernce counting. The idea
> >
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/18 08:45:39
Modified files:
sys/kern : kern_resource.c
sys/sys: resourcevar.h
Log message:
Use the refcnt API with struct plimit.
OK bluhm@ dlg@
On Thu, Mar 17, 2022 at 06:16:51PM +0100, Alexander Bluhm wrote:
> On Thu, Mar 17, 2022 at 07:25:27AM +0000, Visa Hankala wrote:
> > On Thu, Mar 17, 2022 at 12:42:13AM +0100, Alexander Bluhm wrote:
> > > I would like to use btrace to debug refernce counting. The idea
> >
Use the refcnt API with struct plimit.
OK?
Index: kern/kern_resource.c
===
RCS file: src/sys/kern/kern_resource.c,v
retrieving revision 1.71
diff -u -p -r1.71 kern_resource.c
--- kern/kern_resource.c8 Feb 2021 10:51:01 -
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/17 08:23:34
Modified files:
sys/nfs: nfs_socket.c
sys/kern : kern_fork.c kern_prot.c
sys/sys: ucred.h
Log message:
Use the refcnt API with struct ucred.
OK
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/17 08:22:03
Modified files:
sys/net: bpf.c bpfdesc.h
Log message:
Use the refcnt API in bpf.
OK sashan@ bluhm@
On Thu, Mar 17, 2022 at 12:42:13AM +0100, Alexander Bluhm wrote:
> I would like to use btrace to debug refernce counting. The idea
> is to a a tracepoint for every type of refcnt we have. When it
> changes, print the actual object, the current counter and the change
> value.
> Do we want that
On Wed, Mar 16, 2022 at 11:09:12PM +0100, Alexander Bluhm wrote:
> On Tue, Mar 15, 2022 at 09:15:34AM +0000, Visa Hankala wrote:
> > However, some DEC Alpha CPUs have their data caches divided into cache
> > banks to improve bandwidth. These cache banks are relatively
> > in
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/16 10:17:47
Modified files:
sys/kern : kern_event.c
Log message:
Remove an unneeded include.
Use the refcnt API with struct ucred.
OK?
Index: nfs/nfs_socket.c
===
RCS file: src/sys/nfs/nfs_socket.c,v
retrieving revision 1.139
diff -u -p -r1.139 nfs_socket.c
--- nfs/nfs_socket.c22 Feb 2022 01:15:02 - 1.139
+++
Use the refcnt API in bpf.
OK?
Index: net/bpf.c
===
RCS file: src/sys/net/bpf.c,v
retrieving revision 1.215
diff -u -p -r1.215 bpf.c
--- net/bpf.c 15 Feb 2022 08:43:50 - 1.215
+++ net/bpf.c 16 Mar 2022 15:42:05 -
@@
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/16 08:38:43
Modified files:
sys/kern : kern_event.c
sys/sys: eventvar.h
Log message:
Use the refcnt API in kqueue.
OK dlg@ bluhm@
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/16 08:13:01
Modified files:
share/man/man9 : refcnt_init.9
sys/kern : kern_synch.c
sys/sys: refcnt.h
Log message:
Add refcnt_shared() and refcnt_read()
Make kqueue use the refcnt API.
OK?
Index: sys/kern/kern_event.c
===
RCS file: src/sys/kern/kern_event.c,v
retrieving revision 1.183
diff -u -p -r1.183 kern_event.c
--- sys/kern/kern_event.c 22 Feb 2022 01:15:01 -
On Tue, Mar 15, 2022 at 09:11:30AM +, Visa Hankala wrote:
> This patch adds a function for getting a snapshot of a reference
> counter. This will let code like crcopy() access the value in an
> API-observing way.
Here is a revised version.
Based on input from dlg@, the patch
This removes the data dependency consumer barrier from the atomic_load_*
functions. I think the intent was to keep these functions relaxed in
terms of CPU memory order.
This makes these functions more agreeable in code that assertions may
use, such as the suggested refcnt_read().
Removing the
This patch adds a function for getting a snapshot of a reference
counter. This will let code like crcopy() access the value in an
API-observing way.
OK?
Index: share/man/man9/refcnt_init.9
===
RCS file:
On Mon, Mar 14, 2022 at 02:01:07AM -0700, Philip Guenther wrote:
> On Mon, Mar 14, 2022 at 12:47 AM Visa Hankala wrote:
>
> > On Sun, Mar 13, 2022 at 06:26:19PM -0700, Philip Guenther wrote:
> > > On Sun, Mar 13, 2022 at 10:27 AM Visa Hankala wrote:
> > >
> &g
On Sun, Mar 13, 2022 at 06:26:19PM -0700, Philip Guenther wrote:
> On Sun, Mar 13, 2022 at 10:27 AM Visa Hankala wrote:
>
> > On Sun, Mar 13, 2022 at 04:29:44PM +0100, Mark Kettenis wrote:
> >
> ...
>
> > > Under what circumstances does memory ordering
On Sun, Mar 13, 2022 at 04:29:44PM +0100, Mark Kettenis wrote:
> > Date: Sun, 13 Mar 2022 12:47:13 +
> > From: Visa Hankala
> >
> > This makes the refcnt implementation issue memory barriers when
> > releasing references, splitting memory activity cleanly int
This makes the refcnt implementation issue memory barriers when
releasing references, splitting memory activity cleanly into preceding
and succeeding stages around refcnt 1->0 transition.
I think the while loop could be optimized a little by re-reading
r->r_refs just after sleep_finish(). That
This is a slightly tweaked version of the FRELE() memory barrier patch.
The barriers aim to provide clearer memory access behaviour around the
f_count 1->0 transition. With the barriers, the transition splits the
memory activity into preceding and succeeding stages that do not
overlap.
OK?
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/11 07:42:08
Modified files:
share/man/man9 : cond_init.9
Log message:
Fix formatting of cond_signal().
On Fri, Mar 11, 2022 at 11:51:31AM +0100, Alexander Bluhm wrote:
> On Fri, Mar 11, 2022 at 05:32:11AM +0000, Visa Hankala wrote:
> > On Thu, Mar 10, 2022 at 07:17:43PM +0100, Alexander Bluhm wrote:
> > > On Thu, Mar 10, 2022 at 04:39:49PM +0100, Alexander Bluhm wrote:
> >
On Thu, Mar 10, 2022 at 07:17:43PM +0100, Alexander Bluhm wrote:
> On Thu, Mar 10, 2022 at 04:39:49PM +0100, Alexander Bluhm wrote:
> > > Below is a patch that shows how to accomplish release semantics with
> > > the file and refcnt APIs. (The added memory barriers ensure that the
> > > CPU
On Thu, Mar 10, 2022 at 10:45:47AM +, Laurence Tratt wrote:
> On Thu, Mar 10, 2022 at 09:05:54AM +0000, Visa Hankala wrote:
>
> Hello Visa,
>
> > In general, atomic_* functions have not provided implicit memory
> > barriers on OpenBSD.
>
> I've used atomi
ery weak and does not seem useful in any of the use cases
in the patch. However, the barrier does not appear to make things worse
in terms of correctness. Except maybe in assertions where they cause
subtle side effects.
However, the patch looks good.
OK visa@
> Index: sys/dev/pci/if_iwm.c
> ==
bulk build on octeon.ports.openbsd.org
started on Wed Mar 2 16:47:19 UTC 2022
finished at Wed Mar 9 16:21:10 UTC 2022
lasted 07D23h33m
done with kern.version=OpenBSD 7.1-beta (GENERIC.MP) #870: Tue Mar 1 18:30:28
MST 2022
built packages:8747
Mar 2:1912
Mar 3:1183
Mar 4:503
Mar 5:544
Mar 6:772
On Wed, Mar 09, 2022 at 08:40:50AM +0100, Jeremie Courreges-Anglas wrote:
> On Tue, Mar 08 2022, Visa Hankala wrote:
> > This patch adds kernel stack trace saving for riscv64, for the benefit
> > of dt(4) and witness(4).
>
> Nice!
>
> > The unwinder is slow be
This patch adds kernel stack trace saving for riscv64, for the benefit
of dt(4) and witness(4). The unwinder is slow because of the symbol
lookup, but this can be tweaked later.
The limit variable prevents the unwinder from using user-controllable
register values. The limit has to reflect the
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/08 09:13:08
Modified files:
sys/dev/fdt: if_cad.c
Log message:
Make cad(4) handle rxr ioctl.
On Tue, Mar 08, 2022 at 08:04:36AM +0100, Anton Lindqvist wrote:
> On Mon, Mar 07, 2022 at 07:36:35AM +0000, Visa Hankala wrote:
> > I still think that checking TXFF and using the same code for both
> > SBSA and true PL011 UARTs would be the best choice. This would avoid
> >
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/03/07 06:02:53
Modified files:
sys/dev/fdt: if_cad.c
Log message:
Prevent deadlock in cad_down()
Introduce an rwlock to serialize cad(4) ioctl operations so that
cad_down() can release NET_LOCK()
On Wed, Mar 02, 2022 at 07:25:36AM +0100, Anton Lindqvist wrote:
> On Tue, Mar 01, 2022 at 01:41:10PM +0000, Visa Hankala wrote:
> > On Tue, Mar 01, 2022 at 07:08:51AM +0100, Anton Lindqvist wrote:
> > > On Sun, Feb 27, 2022 at 05:33:17PM +0100, Mark Kettenis wrote:
> >
bulk build on octeon.ports.openbsd.org
started on Mon Feb 21 17:16:58 UTC 2022
finished at Wed Mar 2 07:33:37 UTC 2022
lasted 09D14h16m
done with kern.version=OpenBSD 7.1-beta (GENERIC.MP) #857: Sun Feb 20 19:35:55
MST 2022
built packages:9037
Feb 21:1808
Feb 22:1006
Feb 23:647
Feb 24:475
Feb
On Wed, Mar 02, 2022 at 12:11:30PM +0100, Mark Kettenis wrote:
> > Date: Wed, 2 Mar 2022 10:53:54 +
> > From: Visa Hankala
> >
> > Holding NET_LOCK() while invoking a taskq barrier can result in
> > a deadlock because the lock can prevent the taskq f
Holding NET_LOCK() while invoking a taskq barrier can result in
a deadlock because the lock can prevent the taskq from making progress.
Avoid this problem in cad_down() by serializing the driver's ioctl
handler with an rwlock so that NET_LOCK() can be released temporarily.
The patch additionally
On Tue, Mar 01, 2022 at 07:08:51AM +0100, Anton Lindqvist wrote:
> On Sun, Feb 27, 2022 at 05:33:17PM +0100, Mark Kettenis wrote:
> > > Date: Sun, 27 Feb 2022 16:01:25 +0100
> > > From: Anton Lindqvist
> > >
> > > r1p5 is revision 3 with a 32 bytes FIFO. My only concern is that the
> > >
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/28 08:49:57
Modified files:
sys/arch/mips64/mips64: trap.c
Log message:
Remove unneeded symbol name lookup.
On Sun, Feb 27, 2022 at 09:56:38AM +0100, Anton Lindqvist wrote:
> On Sun, Feb 27, 2022 at 06:19:02AM +0000, Visa Hankala wrote:
> > On Sat, Feb 26, 2022 at 08:40:25AM +0100, Anton Lindqvist wrote:
> > > Hi,
> > > This enables fifo support in pluart(4). While here, I c
On Sat, Feb 26, 2022 at 08:40:25AM +0100, Anton Lindqvist wrote:
> Hi,
> This enables fifo support in pluart(4). While here, I changed the
> attachment output to look more like com(4). Tested on my rpi3b which has
> a 16 byte fifo.
>
> Comments? OK?
>
> diff --git sys/dev/ic/pluart.c
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/25 06:51:02
Modified files:
sys/arch/arm64/arm64: machdep.c
sys/arch/arm64/conf: GENERIC RAMDISK
Log message:
Enable cduart(4) on arm64.
OK kettenis@
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/24 07:19:10
Modified files:
sys/arch/riscv64/include: frame.h
sys/arch/riscv64/riscv64: vm_machdep.c
Log message:
Fix kernel stack alignment on riscv64
Pad trapframe and switchframe
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/24 07:16:53
Modified files:
sys/arch/riscv64/riscv64: vm_machdep.c
Log message:
Reserve room for holding curcpu pointer in u-area
The tp register contains the curcpu pointer in kernel mode. The
create calls. if the vnode has been made bad/dead after
> all the sleeping, it returns with ENXIO.
>
> this appears to fix the issue. i had a test program that would trigger
> the KASSERT in a matter of seconds and it's been running for an hour
> now.
>
> here's the diff.
The di
So far this patch has worked fine on my test system.
On Tue, Feb 22, 2022 at 06:41:31PM +, Visa Hankala wrote:
> Index: arch/riscv64/include/frame.h
> ===
> RCS file: src/sys/arch/riscv64/include/frame.h,v
> retriev
On Tue, Feb 22, 2022 at 07:04:23PM +0100, Mark Kettenis wrote:
> > Date: Tue, 22 Feb 2022 17:45:26 +
> > From: Visa Hankala
> >
> > On Tue, Feb 22, 2022 at 06:13:54PM +0100, Mark Kettenis wrote:
> > > > Date: Tue, 22 Feb 2022 16:59:24 +
> > >
On Tue, Feb 22, 2022 at 06:13:54PM +0100, Mark Kettenis wrote:
> > Date: Tue, 22 Feb 2022 16:59:24 +
> > From: Visa Hankala
> >
> > On Tue, Feb 22, 2022 at 05:31:31PM +0100, Mark Kettenis wrote:
> > > > Date: Tue, 22 Feb 2022 15:58
On Tue, Feb 22, 2022 at 05:31:31PM +0100, Mark Kettenis wrote:
> > Date: Tue, 22 Feb 2022 15:58:31 +
> > From: Visa Hankala
> >
> > The standard RISC-V calling convention says that the stack pointer
> > should be 16-byte aligned.
> >
> > The patch
The standard RISC-V calling convention says that the stack pointer
should be 16-byte aligned.
The patch below corrects the alignment of the kernel stack in context
switching and exception handling.
OK?
The diff reveals that curcpu is stored in an unnamed spot near the upper
end of u-area when
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/22 06:34:23
Modified files:
sys/arch/powerpc64/dev: ipmi_opal.c kexec.c
sys/arch/riscv64/riscv64: syscall.c
Log message:
Remove extra ;
No functional change.
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/22 00:47:46
Modified files:
sys/arch/riscv64/riscv64: cpuswitch.S vm_machdep.c
Log message:
Clear frame pointer in cpu_fork() on riscv64
This ensures the chain of call frames is terminated
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/22 00:46:04
Modified files:
sys/arch/riscv64/riscv64: db_trace.c
Log message:
ddb: Improve stack unwinding on riscv64
Add logic to navigate across trapframes. Stop unwinding if the frame
pointer is
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/22 00:34:06
Modified files:
sys/dev: audio.c
Log message:
Make audio(4) event filters MP-safe
Add f_modify and f_process callbacks, and initialize the knote lists
to use audio_lock. This
When setting up the stack for a (kernel) thread on riscv64, zero the
frame pointer in switchframe so that the frame chain terminates cleanly.
At the moment, the unwinding of kernel thread stacks stops to "bad frame
pointer" error.
OK?
Index: arch/riscv64/riscv64/cpuswitch.S
On riscv64, ddb's stack unwinder performs poorly. The main problem is
that the exception handlers use a frame structure (trapframe) that
differs from the typical call frame.
The following patch does several adjustments, including:
* Detect and handle exception frames. (Alternatively, the
bulk build on octeon.ports.openbsd.org
started on Sat Feb 12 15:10:36 UTC 2022
finished at Mon Feb 21 09:13:10 UTC 2022
lasted 09D18h02m
done with kern.version=OpenBSD 7.0-current (GENERIC.MP) #848: Fri Feb 11
18:31:13 MST 2022
built packages:9018
Feb 12:2144
Feb 13:894
Feb 14:279
Feb 15:413
CVSROOT:/cvs
Module name:src
Changes by: v...@cvs.openbsd.org2022/02/19 01:33:28
Modified files:
distrib/armv7/ramdisk: install.md list
Log message:
Use installboot(8) in install.md of armv7.
OK kettenis@ deraadt@
201 - 300 of 2253 matches
Mail list logo