Sudo runs on more systems thsan just OpenBSD and so has a lot of
configure goo and defines as a result. There's really no point in
removing that. Any changes made to the sudo in OpenBSD just makes
updates harder.
The alloc functions implement integer overflow checks that are not
present on most
Currently, if you have a host netgroup with an unresolvable hostname,
the file system will not be exported to any of the netgroup members.
This is because we don't zero out grp before reusing it when a host
is unresolvable.
- todd
Index: sbin/mountd/mountd.c
On Fri, 08 Aug 2014 16:38:11 -0400, Jared Yanovich wrote:
I cannot forward to a socket on the remote host (No forward host name.).
Looks correct, but we should also add to the regress tests.
- todd
The following diff adds names for unknown devices on an MSI Z97 PC
MATE motherboard. Premuably other Z97-based boards will be similar.
- todd
OpenBSD 5.6 (GENERIC.MP) #312: Wed Jul 30 12:01:37 MDT 2014
dera...@amd64.openbsd.org:/usr/src/sys/arch/amd64/compile/GENERIC.MP
real mem =
On Wed, 13 Aug 2014 00:46:25 -0400, Jared Yanovich wrote:
It might be desired to change some of the usage strings as well.
The actual usage is too convoluted to describe in the SYNOPSIS.
What you have is OK but I think what is really needed is a
subsection devoted to port forwarding.
- todd
There are two unnamed USB hubs (one attached to each EHCI root hub).
From the data sheet I would expect these to be 0x8000 and 0x8008
but my board actually has 0x8001 and 0x8009.
This is purely cosmetic and nothing currently uses these defines.
- todd
Index: sys/dev/usb/usbdevs
Years ago I bought a Promise TX2plus PCI card which is a two channel
SATAII board w/ one ATA channel that uses a PDC20775. It wasn't
supported at the time so it went on a shelf. The other day I tried
to get it working and found that while it is recognized, I get a
solid hang after fd0 if a drive
On Mon, 08 Sep 2014 02:28:42 +1000, Jonathan Gray wrote:
I'd be fine with that one going in as well.
Are there any reasons not to add it? I don't see a portable alternative
here as brought up by Mark in that thread, and the only if it's
supported on the majority of UNIX-ike operating
On Thu, 11 Sep 2014 22:03:04 -0700, William Orr wrote:
I'm resubmitting this patch since the source tree was locked last time I
submitted. Any thoughts?
I think we've discussed this one to death already. It looks fine
to me.
- todd
No objection here. I'd go a step farther and nuke all the FOO_END
macros as well. They add nothing and just make code harder to read.
- todd
On Fri, 12 Sep 2014 13:50:51 -0700, Philip Guenther wrote:
On Fri, 12 Sep 2014, Todd C. Miller wrote:
No objection here. I'd go a step farther and nuke all the FOO_END
macros as well. They add nothing and just make code harder to read.
Better?
Looks good. OK millert@
- todd
On Fri, 12 Sep 2014 21:09:45 -0700, Doug Hogan wrote:
Index: lib/libevent/event-internal.h
===
RCS file: /cvs/src/lib/libevent/event-internal.h,v
retrieving revision 1.6
diff -u -p -r1.6 event-internal.h
---
I have no objection to this but I don't think the System-V setpgrp()
API belongs in compat-43. We can just move it to gen/setpgrp.c.
Like Ted says, we should ready the source tree first by using
setpgid(). However, all the uses of setpgrp() in the tree are the
equivalent of:
setpgrp(0,
On Wed, 17 Sep 2014 13:51:37 +0200, =?utf-8?Q?J=C3=A9r=C3=A9mie_Courr=C3=A8ges-
Anglas?= wrote:
However I don't think that changing our setpgrp definition would bring
much (any?) benefit. The mismatch here between SysV and BSD is known
since a long time, and I bet that a bunch of stuff in
On Wed, 17 Sep 2014 15:14:19 +0200, Mark Kettenis wrote:
Note that the SysV version of setpgrp is marked as an XSI extension in
the combined POSIX and X/Open specification. As such it isn't
actually part of POSIX and isn't needed for POSIX compliance.
Good point. As far as POSIX goes,
EHOSTUNREACH, EOVERFLOW and ECANCELED are now part of POSIX so we
should adjust sys/errno.h accordingly.
- todd
Index: /sys/sys/errno.h
===
RCS file: /home/cvs/openbsd/src/sys/sys/errno.h,v
retrieving revision 1.22
diff -u -r1.22
The following diff should fix it.
- todd
Index: sys/miscfs/fifofs/fifo_vnops.c
===
RCS file: /home/cvs/openbsd/src/sys/miscfs/fifofs/fifo_vnops.c,v
retrieving revision 1.41
diff -u -r1.41 fifo_vnops.c
---
Two comments inline, otherwise OK.
- todd
On Thu, 09 Oct 2014 14:52:50 +1000, David Gwynne wrote:
Index: net/rcmd.c
===
RCS file: /cvs/src/lib/libc/net/rcmd.c,v
retrieving revision 1.56
diff -u -p -r1.56 rcmd.c
---
You should #include time.h for the nanosleep prototype.
Otherwise OK.
- todd
On Thu, 09 Oct 2014 15:22:29 +1000, David Gwynne wrote:
deraadt points out i suck at grep.
ok?
Index: termios/tcsendbreak.c
===
RCS file:
Here is a better diff that passes the newly-updated regress test.
It does two extra things:
1) causes POLLHUP to be returned in revents on EOF
2) clears the EOF condition on read so when another writer
connects we don't still have an EOF pending
Ultimately we should investigate using a
On Fri, 10 Oct 2014 15:17:55 +1000, David Gwynne wrote:
theo made me look at where i was using POLLHUP, so i read your
comment and the poll manpage again.
OK millert@
- todd
On Tue, 14 Oct 2014 22:08:20 +0200, Jonas 'Sortie' Termansen wrote:
Unfortunately, OS X doesn't have clock_gettime, so the portable version will
have to add back a times call as a fallback, or perhaps use gettimeofday (but
this doesn't have the proper time-doesn't-go-backwards semantics).
You
Since this came up in another thread. Trivial implementations of
CLOCK_VIRTUAL and CLOCK_PROF, modeled after what FreeBSD does.
- todd
Index: sys/sys/_time.h
===
RCS file: /home/cvs/openbsd/src/sys/sys/_time.h,v
retrieving
On Wed, 15 Oct 2014 20:36:33 +0200, Mark Kettenis wrote:
Hmm, looking at the FreeBSD man page... isn't CLOCK_PROF the same
thing as CLOCK_PROCESS_CPUTIME_ID?
I was wondering that too, but I get different results as they are
not calculated the same way. CLOCK_PROCESS_CPUTIME_ID always yields
a
On Wed, 15 Oct 2014 20:18:10 +0200, Mark Kettenis wrote:
Shouldn't this do a tuagg() on all the threads of the process like we
do for getrusage? Otherwise the CLOCK_VIRTUAL and CLOCK_PROF clocks
will only be updated upon a context switch.
Probably. I wasn't 100% sure going the rusage route
On Wed, 15 Oct 2014 21:04:43 +0200, Mark Kettenis wrote:
Oh, and while FreeBSD seems to implement CLOCK_VIRTUAL and CLOCK_PROF
as per-process, Solaris implements them as per-thread (but doesn't
document them). And on Solaris CLOCK_PROF is just an alias for
CLOCK_THREAD_CPUTIME_ID.
On Wed, 15 Oct 2014 21:53:47 +0200, Alexandre Ratchov wrote:
On Wed, Oct 15, 2014 at 11:37:26AM -0600, Todd C. Miller wrote:
Since this came up in another thread. Trivial implementations of
CLOCK_VIRTUAL and CLOCK_PROF, modeled after what FreeBSD does.
out of curiousity, what program
On Wed, 15 Oct 2014 21:50:44 -0700, Philip Guenther wrote:
IMO we should just delete CLOCK_VIRTUAL from sys/_time.h and clock_gettime(2)
Easy enough.
- todd
Index: sys/sys/_time.h
===
RCS file:
That's a fun one. OK millert@
- todd
This looks correct and matches what svc_run() does.
- todd
I'm not convinced we actually want to do this, but your wrapper is
missing something like this:
mode = ~ACCESSPERMS;
- todd
On Tue, 04 Nov 2014 14:27:39 +, Dimitris Papastamos wrote:
It seems to me that we should not free `pfd' at this point. The saved
max poll fd is not reset to 0 and I do not see any guarantees that `pfd'
will point to valid memory after calling free() here.
Other code that follows the
On Wed, 12 Nov 2014 09:40:17 +0100, Martin Natano wrote:
Looks good in general but I have a few comments inline.
- todd
Index: common/main.c
===
RCS file: /cvs/src/usr.bin/vi/common/main.c,v
retrieving revision 1.23
diff -u
===
RCS file: libexec/spamd/gdcopy.c
diff -N libexec/spamd/gdcopy.c
--- /dev/null 1 Jan 1970 00:00:00 -
+++ libexec/spamd/gdcopy.c 22 Apr 2013 20:03:45 -
@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2013 Todd C. Miller todd.mil...@courtesan.com
+ *
+ * Permission to use, copy, modify
On Tue, 20 Aug 2013 20:36:38 -0700, William Orr wrote:
Theo pointed out that it would be better to change whitecount to an int,
so as to match the call to configure_pf().
Since trapcount is logically similar, and uses the same iterator
variable in freeaddrlists(), I changed that to an int
Looks good but I think this idiom used in hangman:
pos = (double) random() / (RAND_MAX + 1.0) * (double) Dict_size;
Can be replaced with:
pos = arc4random_uniform(Dict_size);
so long as Dict_size = UINT32_MAX, which should always be the case.
A similar idiom is used by monop (compare
Assuming we want to make this a non-fatal error the following should
do.
- todd
Index: usr.sbin/user/user.c
===
RCS file: /home/cvs/openbsd/src/usr.sbin/user/user.c,v
retrieving revision 1.95
diff -u -r1.95 user.c
---
I changed my mind and decided it is better to just move the chown
and chmod out of copydotfiles() and add an explicit check for skeldir
set to the empty string. Much as I would like to prettify the
user.c code it is a losing battle so here is a minimal diff.
- todd
Index: usr.sbin/user/user.c
On Mon, 30 Sep 2013 19:26:20 -0700, Constantine A. Murenin wrote:
Whereas it remains to be seen what kind of bug I'm facing here
(Google reveals I'm not alone), it would appear that changes
introduced in 5.4-current would no longer cause spamd to report
such situation, because the 0 that
Setting a flag seems more straighforward than (ab)using timeouts
for this. psignal() is a rather complicated function but I don't
see any problems offhand with running it from userret().
- todd
That looks reasonable to me. I can't think of a reason to not
return the proper errno values, especially as things like ENOMEM
are already documented for execve(2).
- todd
It might be worth updating the in-tree perl to the latest, version
5.18.1. This is not something I have time for myself, however.
- todd
Your diff looks correct to me. In the old code, we had this loop:
for (lim = mp + *mp; mp lim;)
if (*mp++ *np++)
return 1;
return 0;
Which was replaced by:
lim = mp + *mp;
return (memcmp(mp, np, *lim) 0);
which is not equivalent since lim points to
On Wed, 11 Dec 2013 17:54:06 +0100, =?utf-8?Q?J=C3=A9r=C3=A9mie_Courr=C3=A8ges-
Anglas?= wrote:
As reported by on bugs@[1]. I had a longer diff but got no reply so
let's try to get this in first.
OK but please just change the unsigned long to u_int32_t. There's
no need for this to be any
On Tue, 10 Dec 2013 14:49:26 -0700, Theo de Raadt wrote:
Noticed TIOCGSID in that list. Don't think that is a 4.3 compat
ioctl. Rather a System V compat ioctl.
Well, it's implemented in compat/common/tty_43.c. If we want to save
it, we need to move it to kern/tty.c:ttioctl(). Do
On Thu, 12 Dec 2013 22:03:53 +, Christian Weisgerber wrote:
make build, release, and xenocara are fine.
Full ports build done and remaining fallout fixed.
Any okays?
OK millert@
- todd
On Thu, 12 Dec 2013 23:20:55 +0100, Christian Weisgerber wrote:
Earlier today I found myself patching struct winsize to ttysize in
a port, because I had been confused by sys/ioctl.h into thinking
OpenBSD had the latter instead of the former. Ugh.
Heh, the same thing happened to me with sudo.
On Sun, 22 Dec 2013 19:08:53 -0500, Ted Unangst wrote:
1. bigger is better. make a link to sha512.
2. simplify md5.c. there's really only two modes.
OK. program_mode was originally used for more things but now it
just for the getopt string and usage.
3. add a note to sha256.1 about
On Mon, 23 Dec 2013 10:12:23 +0100, =?utf-8?Q?J=C3=A9r=C3=A9mie_Courr=C3=A8ges-
Anglas?= wrote:
This should now be:
static const char *optstr[2] = {
bcpqrs:tx,
- bcpqrs:tx,
- bcpqrs:tx,
- a:bco:pqrs:tx,
a:bco:pqrs:tx
Note that yacc skeleton is a special case since there is no configure
goo to rely on. With portable ssh we have configure checks available.
- todd
This is nit picking but it bugs me to see the conditional repeated
like that. Perhaps instead of:
if (nextbyte tarindex + 1 = targsize)
return (-1);
if (tarindex + 1 targsize)
target[tarindex+1] = nextbyte;
Something more like:
if (tarindex + 1 targsize) {
Since both sfi and sfo refer to the same underlying descriptor it
is a bad idea to fclose() both of them. In this case, it is not a
big deal as sfo has already been flushed. There's really no need
for two streams so I committed a change to use a single one, opened
with r+, and added the
On Fri, 17 Jan 2014 21:49:53 +0100, Tobias Stoeckmann wrote:
lpd wants to verify that it doesn't open a symbolic link, checking with
lstat(), then open()ing the file. The only reason I can see that the
code does not simply use O_NOFOLLOW is a different return value if
it encounters a symlink
Perhaps something like this? Only compile-tested.
- todd
Index: usr.sbin/lpr/lpd/printjob.c
===
RCS file: /home/cvs/openbsd/src/usr.sbin/lpr/lpd/printjob.c,v
retrieving revision 1.49
diff -u -r1.49 printjob.c
---
On Tue, 21 Jan 2014 10:44:00 +0100, Otto Moerbeek wrote:
And here's the man page diff, our ctime and asctime actually do not
ever return NULL, while posix allows that.
Isn't it worth documenting that ctime and asctime are allowed to
return NULL, even though they do not on OpenBSD?
- todd
On Wed, 22 Jan 2014 08:48:29 +1300, Philip Guenther wrote:
I think this is a doc clarity/consistency issue. The manpage says:
readpassphrase() takes the following optional flags:
If the flags are optional, then you can leave them all out, which in
this case means you use zero.
It
On Wed, 22 Jan 2014 10:26:47 +0100, Otto Moerbeek wrote:
yeah, I first had that and then deleted it.
OK.
- todd
I'd prefer something like this using strtoll(). We can't easily
use strtonum() here since there can be anything past the digits.
The lexer is in charge of matching whitespace and space and newlines
are explicit in the yacc grammar.
For numbers larger than INT_MAX the lexer now returns BIGNUM
On Mon, 17 Feb 2014 15:25:12 -0500, Ted Unangst wrote:
It's only been 25 years. I think we can depend on prototypes and
const now.
Is there really a reason to break KR compatibility in the generated
code? What problem are you solving?
- todd
On Mon, 17 Feb 2014 17:30:41 -0500, Ted Unangst wrote:
I don't see how KR compat helps us. None of the headers in
/usr/include are KR anymore. How would one compile the generated output?
You're assuming that the resulting code will be built on OpenBSD.
That's a bad assumption. The code
On Tue, 18 Feb 2014 16:38:07 -0500, Ted Unangst wrote:
True, I phrased that poorly. What I'm assuming is that the code will
be built by a compiler that supports the const keyword. Or in other
words, if you're using a 25 year old cross compiler, I don't think
it's unreasonable to expect you to
On Tue, 18 Feb 2014 23:05:23 +0100, Marc Espie wrote:
Come on Jim, it's dead.
I'm a doctor, not a programmer! Err, wait...
- todd
On Tue, 18 Feb 2014 18:49:21 -0700, Theo de Raadt wrote:
The *.y file is the source code. the .c and .h files it generates are not
the source code, but an intermediate language.
To stetch this to the limit, we should be shipping cpp-processed files, for
maximum portability.
NO WAY! .y
On Sat, 18 Jun 2011 00:42:30 EDT, Ted Unangst wrote:
lint does not understand the %m format, leading to false positives. The
too few format args is one of the important ones, IMO, so we should try to
make it accurate.
That seems reasonable but I think you want to treat %m like %%.
That way
The code looks fine to me, though you don't need to include stdint.h.
It is basically identical to what is in FreeBSD too.
In the man page example:
if (p = wcsdup(Lfoobar), p == NULL) {
I think it is best to either write this as:
if ((p = wcsdup(Lfoobar)) == NULL) {
or just split it
On Tue, 05 Jul 2011 20:03:14 BST, Nicholas Marriott wrote:
I think they would be better as separate pages but I'm not volunteering
to do it, at least not right now ;-). FreeBSD and NetBSD look to have a
few split out.
Me neither :-) This would be a easy project for someone with a
little bit
They've all been normalized to be /usr/share/tabset already, except
for one that was missed (I've altered the upstream maintainer).
Also remove blank lines while we are at it.
- todd
Index: share/termtypes/Makefile
===
RCS file:
Since cgetnext() iterates over the file it can just pass the entire
record in to getent() so getent() doesn't need to re-parse the file
again to find it. This speeds up cap_mkdb when building termcap.db.
Other potential speedups include useing stdio in getent() and to
using pfp instead of
Using stdio in getent() results in a minor wallclock speedup but
halves the system time on my test machine. As a bonus we can reuse
pfp in tcgetnext().
- todd
Index: lib/libc/gen/getcap.c
===
RCS file:
On Thu, 07 Jul 2011 12:22:28 +0200, Stefan Sperling wrote:
The proper way to fix this is to add conversion from/to the various
character sets into unicode code points within mbrtowc() and wcrtomb()
in citrus_none.c. This wouldn't need anything special like iconv.
A couple of statically
This time for terminfo. These are the same basic changes I've made
to getcap.c but applied to getinfo.c.
Before:
$ /usr/bin/time cap_mkdb -i -f obj/terminfo obj/terminfo.src
6.57 real 5.51 user 0.93 sys
After:
$ /usr/bin/time cap_mkdb -i -f obj/terminfo obj/terminfo.src
On Sun, 31 Jul 2011 11:51:19 EDT, Loganaden Velvindron wrote:
Shouldn't we account for errors when fclose() is called?
Yes, since fclose() calls fflush() first which may do the equivalent
of fwrite(). You need to check the return value of fclose()
to make sure everything actually got written.
On Wed, 17 Aug 2011 21:33:36 PDT, Philip Guenther wrote:
This diff adds to touch(1) support for the -d option specified by POSIX
2008 that permits specifying subsecond timestamps, as well as nanosecond
support in the -r option, and a general simplification over the main loop.
OK millert@
On Wed, 17 Aug 2011 18:42:45 PDT, Philip Guenther wrote:
The diff below adds support to various utilities to preserve timestamps to
the nanosecond. Most of them already preserve down to microseconds and
this just makes them preserve the nanoseconds part too. That might seem
pointless but
On Sun, 15 Jan 2012 22:30:31 MST, Theo de Raadt wrote:
you forgot https_proxy and no_proxy...
however, im against this change since it allows a user to redirect a progra
m
they need privileges to use to an arbitrary proxy they specify, something
there is no good mitigation against.
Why do we care if the user exists? Ideally, you want the code to
behave more or less the same whether the user is real or not.
Otherwise, a remote attacker can guess valid usernames by timing a
login attempt.
For safety's sake, it makes sense to reject a username with a '/'
in it since the
On Tue, 13 Nov 2012 11:02:00 +0100, Alexander Hall wrote:
Since I switched to SMTPD I noticed a few cron emails being marked as
spam by spamassassin, largely caused by the From: and To: headers not
containing a domain part.
Sendmail will add the domain if the always_add_domain feature is
On Thu, 06 Dec 2012 10:55:50 +0100, Alexander Hall wrote:
If, between the internal grep'ing and the printout, a process has
disappeared, we currently get an empty line and pgrep will return
nonzero.
Wouldn't it be better to just defer the printing of the delimiter
under we get the args? Then
On Sun, 09 Dec 2012 00:04:57 +0100, Alexander Hall wrote:
This diff follows your line, but allows *action to return match, error
or nomatch, handled appropriately by the main loop. Also make the same
changes to killact().
I have yet to make it fail my tests. Makes sense? OK?
Looks good to
On Wed, 12 Dec 2012 01:50:46 +0100, Alexander Hall wrote:
Hm, reading the diff again, I felt grepact() got a bit messy. This
cleans it up a bit. Rest should be the same.
OK.
- todd
Alternately, we can simply convert the while loop to a do-while.
As far as I can tell that loop is always entered with either
uio-uio_resid 0 or cc 0.
- todd
Index: sys/kern/tty_pty.c
===
RCS file: /cvs/src/sys/kern/tty_pty.c,v
On Fri, 28 Dec 2012 11:50:49 -0200, Gleydson Soares wrote:
following diff tweak malloc(9) to use M_NOWAIT instead of M_DONTWAIT.
this will make the code more consistent since that M_DONTWAIT is a flag defin
ed for mbuf(9),
so we shouldn't to confuse.
no functional change since that
On Mon, 04 Mar 2013 16:19:56 GMT, Stuart Henderson wrote:
Hmm. Looking at the number of bytes in the error message which matches
what's seen on anoncvs.usa and differs to my server... Todd, does the
copy in your chroot pre-date this commit?
The cvs binary may not have had that fix on
The documentation is correct, that line should be:
set_loginclass(runas_pw ? runas_pw : sudo_user.pw);
This was fixed some time ago upstream. I'll commit the fix.
- todd
On Sat, 23 Mar 2013 06:55:40 -0400, Ted Unangst wrote:
this hides more kernel pointers in the kinfo proc struct and
introduces a backdoor for the kmem group. also hoist the permission
test up out of the loops.
Why should we have a backdoor for the group kmem? There are several
programs
On Sat, 23 Mar 2013 06:37:10 -0400, Ted Unangst wrote:
same as in top.
OK millert@
- todd
On Fri, 10 May 2013 17:27:26 +0300, Arto Jonsson wrote:
Here's an updated diff. Compared to the previous diff '-' is now handled
as stdin. From the freebsd version I noticed that the previous diff also
had useless exit() call which I removed. Comments?
The FreeBSD version has other
On Tue, 07 May 2013 19:10:44 +0300, Arto Jonsson wrote:
While writing an email vi(1) crashed with segmentation fault.
When ^W (WERASE) is hit in insert mode it's possible that the line
buffer is accessed out of bounds. If 'max' == 0 and 'tp-cno' == 1 the
'tp-cno' value is first reduced by
On Sun, 12 May 2013 20:57:50 +0300, Arto Jonsson wrote:
Thanks for the review. Here's an updated diff. If anyone's curious this
bug is about one month short of being 20 years old.
Committed, thanks.
- todd
On Fri, 10 May 2013 23:58:23 +0200, =?utf-8?Q?J=C3=A9r=C3=A9mie_Courr=C3=A8ges-
Anglas?= wrote:
+ switch (numbering_properties[section].type) {
+ case number_all:
+ /*
+* Doing this for number_all only is disputable, but
+
I've taken your diff and merged some useful bits from FreeBSD.
Specifically, the use of getline() and multibyte support for the
-d option.
I also made the functions non-static (though I don't think that is
such a big deal) and tweaked the manual to treat '-' like cat(1)
as per j...@wxcvbn.org's
On Wed, 15 May 2013 16:16:53 +0300, Arto Jonsson wrote:
I asked stsp@ about the multibyte support yesterday and it was his
opinion that it's not currently needed.
Seemed like it might be useful for the future but I suppose we can
add things like this when we have better multibyte support.
On Mon, 20 May 2013 12:43:19 +0300, Arto Jonsson wrote:
Updated diff. I removed the int width handling and modified the
separator printing based on your comment.
That looks good to me.
- todd
On Tue, 07 Dec 2010 23:58:39 PST, patrick keshishian wrote:
The patch below resets the HMAP-soff (screen offset of line)
to 1 if the number of lines HMAP-lno spans has changed and is
now less than HMAP-soff. This essentially forces the entire
line to be redrawn at the top of the screen,
On Tue, 18 Nov 2014 16:41:52 +0100, Tobias Stoeckmann wrote:
Not in this case. We use the output of fgets.
Good catch, OK millert@
- todd
On Tue, 18 Nov 2014 15:19:50 +0100, Martin Pieuchot wrote:
Here's the diff inline to facilitate review:
Thank you, I was having trouble navigating the web version to find
a unified diff.
The diff looks good to me. I took a stab at using fstatat() instead
of the existing lstat() to avoid the
I was considering doing something like this was but was unsure about
interactions with FTS_LOGICAL (and thus FTS_NOCHDIR). I suppose
since fts_name is effectively d_name this is OK.
I also prefer passing dirp directly so the caller doesn't need to
use dirfd itself but that's not a big deal.
-
On Tue, 18 Nov 2014 20:51:56 -0700, Anthony J. Bentley wrote:
Martin Natano writes:
The system util.h header is shadowed by the eponymous perl header. This
specific errors exists since December 2012. I strongly suspect that no
one uses the perl extension of vi - Let's remove it!
I'm
On Wed, 19 Nov 2014 20:30:04 -0800, Philip Guenther wrote:
Committed, which leads to the next diff: use O_CLOEXEC on all internal
fds, O_DIRECTORY when opening a directory other than ., and drop the 3rd
argument to open() as unnecessary when O_CREAT isn't used.
Looks good to me. OK
On Wed, 19 Nov 2014 22:32:59 -0500, Ted Unangst wrote:
I don't think this qualifies as a bug. Using the wrong client is user
error. fingerd should not be responsible for filtering anything you
shouldn't send it.
OK millert@ (who wonders how many people even know what a TIP is these days)
-
1 - 100 of 1493 matches
Mail list logo