On 2011/07/08 20:51, Henning Brauer wrote:
> CVSROOT: /cvs
> Module name: src
> Changes by: henn...@cvs.openbsd.org 2011/07/08 20:51:18
>
> Modified files:
> usr.sbin/bgpd : rde.c session.c
>
> Log message:
> remove that rlimit code, rc.d and login classes do it much betterer thes
On Sat, Jul 09, 2011 at 12:56:12AM -0400, Ted Unangst wrote:
> Top said I didn't have much memory free. Where did it all go?
>
> Oh, here it is:
>
> Memory: Real: 391M/2730M act/tot Free: 178M Cache: 1711M Swap: 0K/4097M
I like the idea, but I think "Cache" is ot the ricght name. How about
BCa
Top said I didn't have much memory free. Where did it all go?
Oh, here it is:
Memory: Real: 391M/2730M act/tot Free: 178M Cache: 1711M Swap: 0K/4097M
Index: machine.c
===
RCS file: /home/tedu/cvs/src/usr.bin/top/machine.c,v
retriev
So, there is actually another bug in that chunk of code. This diff
fixes them:
- Read the register from the correct location: HCSPARAMS is a
capability register.
- Construct the resulting mask correctly by adding
parenthesis where needed: '|' takes precedence over '?'
so currently the expr
On Fri, Jul 08, 2011 at 04:22:55PM +0200, David Coppa wrote:
> Hi,
>
> Isn't this wrong?
It may be wrong indeed. I don't think it matters before rthreads are
ready.
Anyways don't change it without having someone run
a full ports test build first.
>
> $ cat /usr/X11R6/lib/pkgconfig/x11.pc
>
On Fri, Jul 8, 2011 at 9:50 PM, Nicholas Marriott
wrote:
> Move TEST.sh to regress/bin/test/ and remove bin/test/TEST.*.
>
> ok?
Makes sense to me. ok matthew@
Move TEST.sh to regress/bin/test/ and remove bin/test/TEST.*.
ok?
Index: regress/bin/test/Makefile
--- /dev/null
+++ regress/bin/test/Makefile
@@ -0,0 +1,8 @@
+# $OpenBSD$
+
+REGRESS_TARGETS= run_test
+
+run_test:
+ @test=/bin/test /bin/sh TEST.sh >/dev/null
+
+.include
Index: regre
Changes as suggested by tedu@.Also the cas() diff is in.
Index: files.linux
===
RCS file: /cvs/src/sys/compat/linux/files.linux,v
retrieving revision 1.16
diff -u -p -r1.16 files.linux
--- files.linux 9 Jul 2011 00:10:52 - 1
Diff below simplifies umass(4) slightly: instead of attaching a two
target bus and reserving one target for the "host" (which doesn't do
anything), just attach a one target bus and let the SCSI subsystem
know that no target is reserved for the adapter.
Basically, this just changes the dmesg line f
Please check that SD cards still work with this diff.
Index: sdmmc_scsi.c
===
RCS file: /cvs/src/sys/dev/sdmmc/sdmmc_scsi.c,v
retrieving revision 1.28
diff -u -p -r1.28 sdmmc_scsi.c
--- sdmmc_scsi.c9 Jul 2011 00:39:29 -
On Fri, Jul 08, 2011 at 08:39:34PM -0400, Ted Unangst wrote:
> On Sat, Jul 09, 2011, Paul Irofti wrote:
> > Mostly for people asking for my secret diffs.
> >
> > This is the cas diff...
>
> > +int
> > +tstohz(const struct timespec *ts)
> > +{
>
> > +int
> > +itimespecfix(struct timespec *ts)
> >
On Sat, Jul 09, 2011, Paul Irofti wrote:
> Mostly for people asking for my secret diffs.
>
> This is the cas diff...
> +int
> +tstohz(const struct timespec *ts)
> +{
> +int
> +itimespecfix(struct timespec *ts)
> +{
These functions can't go here with these names. That's too generic.
Add futex p
Mostly for people asking for my secret diffs.
This is the cas diff...
Index: i386/locore.s
===
RCS file: /cvs/src/sys/arch/i386/i386/locore.s,v
retrieving revision 1.137
diff -u -p -r1.137 locore.s
--- i386/locore.s 5 Jul 2011
On Fri, Jul 08, 2011 at 04:34:36PM -0700, Matthew Dempsky wrote:
> autoconf doesn't allow you to attach multiple child devices to the
> same device with different attach arg types. sdmmc(4) tries to attach
> both scsibus(4) and SDIO devices (currently just the disabled sbt(4)
> bluetooth adapter).
autoconf doesn't allow you to attach multiple child devices to the
same device with different attach arg types. sdmmc(4) tries to attach
both scsibus(4) and SDIO devices (currently just the disabled sbt(4)
bluetooth adapter). The way it does this is by creating a
sdmmc_attach_args structure that
On Fri, Jul 08, 2011 at 11:18:59PM +0200, Thomas Gerlach wrote:
> however, there's still the following problem:
> if wol is enabled via /etc/hostname.xl0 (by adding the wol keyword),
> it's not possible to log into the system, since the error messages "xl0:
> transmission error: ff" and "xl0: comma
If we have successfully found and read in the disk block we expect
to hold the OpenBSD disklabel, then what is there must be treated
as the disklabel. If it is currently invalid (e.g. all zeros) we
still want to write the new label in this MD location, and thus
should pass the spoofed label back up
On 07/08/2011 02:56 PM, Stefan Sperling wrote:
> On Mon, Jun 27, 2011 at 09:31:26AM +, Thomas Gerlach wrote:
>> Tobias Ulmer tmux.org> writes:
>>
>>> I've lost track which patches need to be applied or not, but once a
>>> complete patch appears, I'm willing to test it on a machine that has
>>>
Mark kettenis found an issue with the cleaner on the previous diff I sent out.
This diff fixes his issue, replaces the earlier diff.
Previous comments apply, please test in lots of places.
Index: kern/kern_sysctl.c
===
On Fri, 8 Jul 2011, Marco Peereboom wrote:
> On Fri, Jul 08, 2011 at 09:39:26PM +0200, Piotr Durlej wrote:
> > Reading RAID 1 volumes can be slow because of the way reads are
> > interleaved:
> >
> > # dd if=/dev/rsd0c of=/dev/null bs=1m count=256
> > 256+0 records in
> > 256+0 records out
> > 2
forgot to mention, that gets us rid of the best switch statement ever:
case SIOCSIFADDR:
switch (ifa->ifa_addr->sa_family) {
#ifdef NETATALK
case AF_APPLETALK:
/* Nothing to do. */
break;
#endif /* NETATALK */
On Fri, Jul 08, 2011 at 09:39:26PM +0200, Piotr Durlej wrote:
> Reading RAID 1 volumes can be slow because of the way reads are
> interleaved:
>
> # dd if=/dev/rsd0c of=/dev/null bs=1m count=256
> 256+0 records in
> 256+0 records out
> 268435456 bytes transferred in 1.841 secs (145780529 bytes/se
Reading RAID 1 volumes can be slow because of the way reads are
interleaved:
# dd if=/dev/rsd0c of=/dev/null bs=1m count=256
256+0 records in
256+0 records out
268435456 bytes transferred in 1.841 secs (145780529 bytes/sec)
# dd if=/dev/rsd1c of=/dev/null bs=1m count=256
256+0 records in
256+0 re
Similar to O_CLOEXEC, F_DUPFD_CLOEXEC allows you to dup a file
descriptor and atomically have FD_CLOEXEC set. (Also POSIX Issue 7.)
ok?
Index: sys/fcntl.h
===
RCS file: /home/mdempsky/anoncvs/cvs/src/sys/sys/fcntl.h,v
retrieving rev
A bit of confusion crept in somewhere along the line and LABELSECTOR was being
added
to the hfspartition offset, which caused the OpenBSD disklabel to be read from
sector
1 on the disk, rather than sector 0. To compensate the hfspartition offset was
decremented to make the location come out as be
if we sleep during the copyout, we don't want weird stuff to happen to
our fds. and in the 99.99% case where you neither sleep nor contend on
the lock, no change.
Index: sys_pipe.c
===
RCS file: /home/tedu/cvs/src/sys/kern/sys_pipe.c
On Fri, Jul 08, 2011 at 04:51:47PM +0100, Nicholas Marriott wrote:
> Except it isn't sizeof(b) or '\0'.
Ooops, right. It's counting in units of ints.
> This actually passes the right length to mbstowcs as well:
This looks good to me.
> Index: wcsdup.3
> =
I think this is wrong. WHat do you expect this line to do?
> + user = name;
I think modifying user might be better left outside.
Also:
- You don't need endpwent() for getpwnam() so you can remove that.
- One of those silly _XFOOBAR defines hides the strsep definition. I
guess they can ju
On Fri, Jul 08, 2011 at 04:23:14PM +0100, Nicholas Marriott wrote:
> On Fri, Jul 08, 2011 at 10:31:19AM +0200, Stefan Sperling wrote:
> > On Fri, Jul 08, 2011 at 06:21:17AM +0100, Nicholas Marriott wrote:
> > > EL_BIND and EL_SETTC must have their argument lists end in NULL.
> > >
> > > ok?
> > >
On Fri, Jul 08, 2011 at 07:43:04PM +0300, Paul Irofti wrote:
>
> /usr/src/usr.bin/ktrace/ktrace.c: In function 'main':
> /usr/src/usr.bin/ktrace/ktrace.c:66: error: 'KTRFAC_STRUCT' undeclared
> (first use in this function)
> /usr/src/usr.bin/ktrace/ktrace.c:66: error: (Each undeclared identifier
On Fri, Jul 08, 2011 at 03:44:49AM -0600, Theo de Raadt wrote:
> If you do try the "full trim" diff from Ted, you should not use
> softdep. The softdep is making some assumptions and taking some
> rather nasty actions when ffs_blkfree returns early.
>
> Test softdep, but not with trim.
> Test tri
On Fri, Jul 08, 2011, Paul Irofti wrote:
>> There is a macro named p_emul, now, so no more variables of that name.
>> See below.
>
> Ah, I really don't like this. Its confusing.
Don't worry, guenther convinced me we don't need this.
On Fri, Jul 08, 2011 at 02:23:22PM +0200, Otto Moerbeek wrote:
> On Tue, Jul 05, 2011 at 07:09:38PM +0200, Otto Moerbeek wrote:
>
> > On Tue, Jul 05, 2011 at 06:17:53PM +0200, Otto Moerbeek wrote:
> >
> > > On Tue, Jul 05, 2011 at 12:08:26PM -0400, Ted Unangst wrote:
> > >
> > > > On Tue, Jul 05
I like this a lot, it reads good to me and works great on i386. Sorry
for not testing sooner :-/.
I'd like to see struct termios and sigaction too :-).
On Fri, Jul 08, 2011 at 02:23:22PM +0200, Otto Moerbeek wrote:
> On Tue, Jul 05, 2011 at 07:09:38PM +0200, Otto Moerbeek wrote:
>
> > On Tue, J
On Fri, Jul 08, 2011 at 04:40:16AM -0400, Ted Unangst wrote:
> On Fri, Jul 08, 2011, Paul Irofti wrote:
> > On Fri, Jul 08, 2011 at 01:35:18AM -0400, Ted Unangst wrote:
> >> I don't think we're ever going to support different threads of a process
> >> running with different emulations, so move p_em
Except it isn't sizeof(b) or '\0'. This actually passes the right length
to mbstowcs as well:
Index: wcsdup.3
===
RCS file: /cvs/src/lib/libc/string/wcsdup.3,v
retrieving revision 1.2
diff -u -p -r1.2 wcsdup.3
--- wcsdup.35 Jul 20
Currently dkcsumattach() walks the alldevs list whilst performing
operations that will sleep. This is rather bad if the list happens to be
modified (i.e. a device detachs) whilst it is sleeping.
The following diff resolves this issue. The process is very similar to
that used for softraid - we main
On Fri, Jul 08, 2011 at 10:31:19AM +0200, Stefan Sperling wrote:
> On Fri, Jul 08, 2011 at 06:21:17AM +0100, Nicholas Marriott wrote:
> > EL_BIND and EL_SETTC must have their argument lists end in NULL.
> >
> > ok?
> >
>
> To make sure I understood correctly, it crashes in ct_decode_argv()
> bec
On Fri, Jul 08, 2011 at 10:18:59AM +0200, Stefan Sperling wrote:
> On Fri, Jul 08, 2011 at 03:01:45AM +0100, Nicholas Marriott wrote:
> > Index: wcsdup.3
> > ===
> > RCS file: /cvs/src/lib/libc/string/wcsdup.3,v
> > retrieving revision
Hi,
Isn't this wrong?
$ cat /usr/X11R6/lib/pkgconfig/x11.pc
prefix=/usr/X11R6
exec_prefix=${prefix}
libdir=${exec_prefix}/lib
includedir=${prefix}/include
xthreadlib=-lpthread
Name: X11
Description: X Library
Version: 1.4.3
Requires: xproto kbproto
Requires.private: xcb >= 1.1.92
Cflags: -I${i
On Mon, Jun 27, 2011 at 09:31:26AM +, Thomas Gerlach wrote:
> Tobias Ulmer tmux.org> writes:
>
> >
> > I've lost track which patches need to be applied or not, but once a
> > complete patch appears, I'm willing to test it on a machine that has
> > three different xl's.
> >
> >
>
>
> hi t
On Tue, Jul 05, 2011 at 07:09:38PM +0200, Otto Moerbeek wrote:
> On Tue, Jul 05, 2011 at 06:17:53PM +0200, Otto Moerbeek wrote:
>
> > On Tue, Jul 05, 2011 at 12:08:26PM -0400, Ted Unangst wrote:
> >
> > > On Tue, Jul 05, 2011, Otto Moerbeek wrote:
> > > > Hi,
> > > >
> > > > this adds decoding
On Sun, 26 Jun 2011, Pascal Stumpf wrote:
This is my first try at getting support for terminating '+' in find(1)'s
-exec statement to work, as required by POSIX. Code shamelessly
guttenberged from NetBSD, with some minor modifications.
FWIW, it's been working nicely here for some time now.
20 éxitos / Cae la noche / Agua de lluvia
/ Bienvenido al club / Al fin solos/ Vuela conmigo
/ Contigo a la distancia / Corazón de bolero
/ La magia de Dyango / Amante Gaviota / Colección
I recall someone (maybe oga and/or djm?) being interested in having
O_CLOEXEC support. While here, I noticed POSIX defines O_DIRECTORY,
which is pretty trivial to implement as well.
Tested that these both work as expected on directory and non-directory
files.
ok?
Index: sys/fcntl.h
===
I missed one flag: AT_REMOVEDIR. This makes unlinkat() behave the
same as rmdir(), and the diff below changes sys_rmdir() to call
dounlinkat() using it.
Here's how to read this diff and be convinced it's correct:
1. First read dounlinkat() by itself, and convince yourself that if
flag ==
On 2011/07/08 03:51, Matthew Dempsky wrote:
> On Fri, Jul 8, 2011 at 3:45 AM, Stuart Henderson wrote:
> > On 2011/07/08 01:53, Matthew Dempsky wrote:
> >> Now that all of the new AT_* flags are supported, it's okay to expose
> >> openat(2), etc. in libc.
> >>
> >> This will need at least a minor l
No diff -N, but I assume you were removing it. Good.
COMPAT_09 is now extra officially dead. Only one function actually
used, moved to linux compat.
Index: compat/linux/linux_misc.c
===
RCS file: /home/tedu/cvs/src/sys/compat/linux/linux_misc.c,v
retrieving revision 1.70
diff -u -p -r1
On Fri, Jul 8, 2011 at 3:45 AM, Stuart Henderson wrote:
> On 2011/07/08 01:53, Matthew Dempsky wrote:
>> Now that all of the new AT_* flags are supported, it's okay to expose
>> openat(2), etc. in libc.
>>
>> This will need at least a minor libc bump (not included below). I
>> know martynas@ has
On 2011/07/08 01:53, Matthew Dempsky wrote:
> Now that all of the new AT_* flags are supported, it's okay to expose
> openat(2), etc. in libc.
>
> This will need at least a minor libc bump (not included below). I
> know martynas@ has a major bump planned for libc, so I figure it makes
> sense to
If you do try the "full trim" diff from Ted, you should not use
softdep. The softdep is making some assumptions and taking some
rather nasty actions when ffs_blkfree returns early.
Test softdep, but not with trim.
Test trim, but not with softdep :)
Now that all of the new AT_* flags are supported, it's okay to expose
openat(2), etc. in libc.
This will need at least a minor libc bump (not included below). I
know martynas@ has a major bump planned for libc, so I figure it makes
sense to either go before that or ride the same bump. Opinions
w
On Fri, Jul 08, 2011, Paul Irofti wrote:
> On Fri, Jul 08, 2011 at 01:35:18AM -0400, Ted Unangst wrote:
>> I don't think we're ever going to support different threads of a process
>> running with different emulations, so move p_emul and p_emuldata to
>> struct process. This uses the p__pgid kinfo_
On Fri, Jul 08, 2011 at 06:21:17AM +0100, Nicholas Marriott wrote:
> EL_BIND and EL_SETTC must have their argument lists end in NULL.
>
> ok?
>
To make sure I understood correctly, it crashes in ct_decode_argv()
because it ends up passing a garbage pointer to mbstowcs(), right?
If so, ok.
If no
On 2011/07/07 19:30, Dale Rahn wrote:
> At c2k11 I managed to revive the powerpc (macppc/socppc) interrupt
> rewrite from a few years ago. It had been backed out because of problems
> attributed to other previously fixed bugs. At some prodding the dust
> was blown off the diff and it is working on
On Fri, Jul 08, 2011 at 03:01:45AM +0100, Nicholas Marriott wrote:
> Index: wcsdup.3
> ===
> RCS file: /cvs/src/lib/libc/string/wcsdup.3,v
> retrieving revision 1.2
> diff -u -p -r1.2 wcsdup.3
> --- wcsdup.3 5 Jul 2011 19:01:31 -
On Fri, Jul 08, 2011 at 01:35:18AM -0400, Ted Unangst wrote:
> I don't think we're ever going to support different threads of a process
> running with different emulations, so move p_emul and p_emuldata to
> struct process. This uses the p__pgid kinfo_proc hack to allow p_emul
> to remain as a mac
58 matches
Mail list logo