On 8 October 2015 at 11:48, Martin Pieuchot wrote:
> Recent NFS-related rtisvalid(9) regressions turns out to be related
> to the use of DOWN RTF_CLONED route entries. Such entries are DOWN
> because they are cloned from a DOWN RTF_CLONING entry.
>
> While investigating this
When guenther@ switched isatty(3) to F_ISATTY, he forgot ttyname(3).
With this, simple callers of ttyname(3) like tty(1) and who(1) no
longer need pledge("tty").
Index: lib/libc/gen/ttyname.c
===
RCS file:
On Mon, Oct 12, 2015 at 11:47:08AM +0200, Martin Pieuchot wrote:
> Why not using an "enum ieee80211_edca_ac" while here? Plus does it make
> sense to pass a constant argument to iwm_tx()?
Agreed and filed as "future work" ;-)
> Anyway I'm fine with this.
Thanks, committed.
Index: bsd.port.mk.5
===
RCS file: /cvs/src/share/man/man5/bsd.port.mk.5,v
retrieving revision 1.426
diff -u -p -u -r1.426 bsd.port.mk.5
--- bsd.port.mk.5 23 Sep 2015 01:38:36 - 1.426
+++ bsd.port.mk.5 12 Oct 2015
On 2015/10/12 10:57, Remco wrote:
> Index: bsd.port.mk.5
> ===
> RCS file: /cvs/src/share/man/man5/bsd.port.mk.5,v
> retrieving revision 1.426
> diff -u -p -u -r1.426 bsd.port.mk.5
> --- bsd.port.mk.5 23 Sep 2015 01:38:36 -
On Sun, Oct 11, 2015 at 09:02:27PM +0200, Mark Kettenis wrote:
> Using dma_alloc() here would be wrong. Drivers should use bus_dmamem_alloc().
Thanks, noted.
Do you think iwm should use a buffer allocated with bus_dmamem_alloc()
here, or stick with the mbuf?
On 11/10/15(Sun) 14:23, Stefan Sperling wrote:
> The index into sc->txq used by iwm_tx() is only initialized for
> management frames. For data frames, iwm picks a garbage index off
> the stack which is not initialized unless a mgmt frame was processed
> first in the same loop.
>
> This could end
On 10/10/15(Sat) 20:02, Alexandr Nedvedicky wrote:
> Hello,
>
> Patch fixes two small nits related to source node table in PF (a.k.a.
> pf_src_tree_tracking).
>
> The first issue comes to `global` argument of pf_insert_src_node(). It is
> always 0 everywhere in source code. The `global` is
I noticed a number of "fuser(PID): syscall 37" entries in my dmesg.
If 'sflg' is set (i.e., the -k or -s options were specified),
fuser might call kill(2) (see fuser.c: 176), otherwise "stdio rpath"
is apparently sufficient for fuser.
Index: usr.bin/fstat/fstat.c
> Actually, plain old printf should be OK in ping.c since buffering
> is disabled for ping -f. If you want to keep dprintf(), I think
> we can lose the setbuf() call. Whatever you decide, it would
> be nice to make ping6.c match.
No, disagree strongly.
ping is doing this inside a signal
Hi,
I was wondering if it would be possible to allow the override the
definition of MACHINE_ARCH /__MACHINE_ARCH in amd64/param.h by wrapping
them around ifndef statement.
Index: sys/arch/amd64/include/param.h
===
RCS file:
> I was wondering if it would be possible to allow the override the
> definition of MACHINE_ARCH /__MACHINE_ARCH in amd64/param.h by wrapping
> them around ifndef statement.
>
> Index: sys/arch/amd64/include/param.h
> ===
> RCS file:
Ah, I didn't realize that pinger() was still called via a signal
handler in ping. It looks like ping6 is better in this regard.
- todd
> Ah, I didn't realize that pinger() was still called via a signal
> handler in ping. It looks like ping6 is better in this regard.
ping and ping6 need to be merged, as happened to traceroute.
First step: make all ping6's options match ping options. Rename
ping6 options with wild abandon if
On 12/10/2015 16:30, Theo de Raadt wrote:
> You have failed to explain the usage case; I don't believe there is
> any scenario where people benefit from more architecture divergence.
As an example, in pkgsrc, CPUs architectures are grouped under a common
alias so they can be addressed once
Hello,
The updated patch addresses additional nit found by mpi:
> > Here can't you also change:
> >
> > if ((*sn)->rule.ptr != NULL)
> > (*sn)->rule.ptr->src_nodes++;
> >
> > into:
> >
> > (*sn)->rule.ptr->src_nodes++;
> >
> > I don't know enough to
On 12/10/2015 17:11, Theo de Raadt wrote:
> You want 10 layers up of high level software, to redefine fields in
> a system .h file -- for an ecosystem that isn't part of OpenBSD at
> all.
OpenBSD is the exception to the rule within the framework, there is no
specific changes needed to the said
> Date: Mon, 12 Oct 2015 11:16:27 +0200
> From: Stefan Sperling
>
> On Sun, Oct 11, 2015 at 09:02:27PM +0200, Mark Kettenis wrote:
> > Using dma_alloc() here would be wrong. Drivers should use
> > bus_dmamem_alloc().
>
> Thanks, noted.
>
> Do you think iwm should use a buffer
On 02/10/15(Fri) 16:03, Martin Pieuchot wrote:
> I'd like to change the behavior of arptfree() to stop checking the
> reference counter of the route:
>
> - to keep the logic for reference counting rout entries simple,
>
> - and to align it with the NDP behavior (see nd6_free()).
>
>
> With
On 12/10/15(Mon) 14:59, Mike Belopuhov wrote:
> On Mon, Oct 12, 2015 at 14:06 +0200, Martin Pieuchot wrote:
> > It's currently possible to add a RTF_LOCAL route on an interface with a
> > DOWN link state (em(4) to not say which one).
> >
> > Diff below prevents a race and make sure such routes
RTF_LOCAL are always UP so this is safe. ok?
Index: netinet/ip_input.c
===
RCS file: /cvs/src/sys/netinet/ip_input.c,v
retrieving revision 1.256
diff -u -p -r1.256 ip_input.c
--- netinet/ip_input.c 25 Sep 2015 09:51:20 -
On Mon, Oct 12, 2015 at 14:06 +0200, Martin Pieuchot wrote:
> It's currently possible to add a RTF_LOCAL route on an interface with a
> DOWN link state (em(4) to not say which one).
>
> Diff below prevents a race and make sure such routes will be UP until
> if_link_state_change() is executed.
>
On Mon, Oct 12, 2015 at 14:08 +0200, Martin Pieuchot wrote:
> RTF_LOCAL are always UP so this is safe. ok?
>
Makes sense to me after the "RTF_LOCAL and link state" diff.
OK mikeb
> Index: netinet/ip_input.c
> ===
> RCS file:
On 10 October 2015 at 20:02, Alexandr Nedvedicky
wrote:
> Hello,
>
> Patch fixes two small nits related to source node table in PF (a.k.a.
> pf_src_tree_tracking).
>
> The first issue comes to `global` argument of pf_insert_src_node(). It is
> always 0 everywhere
It's currently possible to add a RTF_LOCAL route on an interface with a
DOWN link state (em(4) to not say which one).
Diff below prevents a race and make sure such routes will be UP until
if_link_state_change() is executed.
ok?
Index: net/route.c
On Mon, Oct 12, 2015 at 11:13 +0200, Martin Pieuchot wrote:
> On 10/10/15(Sat) 20:02, Alexandr Nedvedicky wrote:
> > Hello,
> >
> > Patch fixes two small nits related to source node table in PF (a.k.a.
> > pf_src_tree_tracking).
> >
> > The first issue comes to `global` argument of
> When guenther@ switched isatty(3) to F_ISATTY, he forgot ttyname(3).
That was a change I made.
> With this, simple callers of ttyname(3) like tty(1) and who(1) no
> longer need pledge("tty").
That is correct, these programs could do without the ability to set
modes on the tty. I think this
On Sun, 11 Oct 2015 01:26:49 +0200, Tobias Stoeckmann wrote:
> If we limit support of substitutions to s/.//, then it is easily
> transformed into code, e.g. p++. No need for any form of regexp
> support, at all... and we are GNU patch compatible. ;)
OK millert@
- todd
On Sat, 10 Oct 2015 16:35:02 -0700, Philip Guenther wrote:
> Some isfoo(char) usages crept back into ftp
OK.
- todd
Actually, plain old printf should be OK in ping.c since buffering
is disabled for ping -f. If you want to keep dprintf(), I think
we can lose the setbuf() call. Whatever you decide, it would
be nice to make ping6.c match.
- todd
When scanning for is*() function uses with signed chars, I found that
lex(1) uses an unecessary #define for unsigned char. The below diff
removes it and fixes an undefined is*() usage or two as well.
Index: ecs.c
===
RCS file:
Our tree uses lex pretty much as-is from upstream, so this refactoring
isn't the way to go.
If you find an actual bug however.. fix them, using the established
practice.
> When scanning for is*() function uses with signed chars, I found that
> lex(1) uses an unecessary #define for unsigned char.
On Sat, 10 Oct 2015 16:24:38 -0700, Philip Guenther wrote:
> As a general rule, programs should not use errno values as an exit status.
>
> Compare "sleep 10001" w/ and w/o this diff.
OK.
- todd
I think this is the correct approach. I've only taken a brief look
so far but I think you should make write_lines() static in its
declaration to match the prototype.
- todd
Philip Guenther wrote:
> On Sun, Oct 11, 2015 at 8:19 AM, Ted Unangst wrote:
> > agreed, but why not return 1? i don't want to have to slap 2>/dev/null
> > around
> > all my sleep calls now.
>
> So don't. Why would you need to?
doesn't err() print a message?
Went to build the example in man 4 pf and believe it needs a diff
Index: pf.4
===
RCS file: /cvs/src/share/man/man4/pf.4,v
retrieving revision 1.85
diff -u -p -r1.85 pf.4
--- pf.418 Apr 2014 21:42:04 - 1.85
+++ pf.4
Sevan Janiyan wrote:
>
>
> On 12/10/2015 16:30, Theo de Raadt wrote:
> > You have failed to explain the usage case; I don't believe there is
> > any scenario where people benefit from more architecture divergence.
>
> As an example, in pkgsrc, CPUs architectures are grouped under a common
>
Hi Ted,
Ted Unangst wrote on Mon, Oct 12, 2015 at 01:09:22PM -0400:
> Philip Guenther wrote:
>> On Sun, Oct 11, 2015 at 8:19 AM, Ted Unangst wrote:
>>> agreed, but why not return 1? i don't want to have to slap 2>/dev/null
>>> around all my sleep calls now.
>> So don't.
On Mon, Oct 12, 2015 at 01:49:54PM +0200, Martin Pieuchot wrote:
> Turns out that our stack does a very bad job at tracking which routes
> should be used or not based on the RTF_UP flag. So I'm not going to
> change that.
>
> Nonetheless I'd like to use rtisvalid(9) for the existing route
ping
On Sun, Sep 27, 2015 at 09:52:16AM -0500, Matthew Martin wrote:
> The following outputs only $ with a newline.
>
> { i=0; while [ $i -lt 89 ]; do printf $i\\0; : $(( i++ )); done; } |
> xargs -0 -I $ printf '%s\n' $
>
> The problem is two fold. The first is that strnsubst gives up
On 12/10/2015 18:32, Ted Unangst wrote:
> just don't include or look at param.h.
Looking into doing that now.
Sevan
Small fix to tcpdump nfs packet printing. Crash occurs when
printing nfs request filename of malformed packet.
parsefn() now passes snapend to filename print function
(attempting to print packet data so NULL isn't right). Also
check return value from fn_printn() and handle truncation.
Behavior
On Fri, Oct 02, 2015 at 04:03:46PM +0200, Martin Pieuchot wrote:
> I'd like to change the behavior of arptfree() to stop checking the
> reference counter of the route:
>
> - to keep the logic for reference counting rout entries simple,
>
> - and to align it with the NDP behavior (see
Hello,
Richard Procter came back to me in private email with one more nit to fix:
we can get rid of
if (sn->rule.ptr != NULL)
test condition in pfioctl() function as well.
The relevant snippet looks as follows:
2188 p = psn->psn_src_nodes;
2189
On Mon, 12 Oct 2015 13:09:22 -0400, "Ted Unangst" wrote:
> Philip Guenther wrote:
> > On Sun, Oct 11, 2015 at 8:19 AM, Ted Unangst wrote:
> > > agreed, but why not return 1? i don't want to have to slap 2>/dev/null ar
> ound
> > > all my sleep calls now.
> >
> > So don't.
Instead of just seeing if a line in a dired buffer is longer than 0
before marking it for deletion, check if a filename has the potential
to be extracted from a line. ok?
-lum
Index: dired.c
===
RCS file:
While trying to make a fresh install with the from the Oct 12th miniroot58.fs,
while installing the sets I get a bunch of:
tar: Unable to set file uid/gid of ./blah/blah: Operation not permitted
tar: Unable to set file uid/gid of ./...: Operation not permitted
tar: Unable to set file uid/gid of
Many will have observed that pledge(2) usage is being pushed into the
source tree at a very rapid pace.
I'd like if everyone looks in their dmesg logs for pledge errors. But
please don't immediately mail a report! Instead, look for if someone
else reports an error in the same command. If noone
48 matches
Mail list logo