Aaron Bannert [EMAIL PROTECTED] writes:
On Fri, Mar 01, 2002 at 12:34:49PM -0500, Jeff Trawick wrote:
printf(%d %d %d %d\n,
offsetof(struct s1,b),
offsetof(struct s2,b),
offsetof(struct s3,b),
offsetof(struct s4,b));
s3
. Try this, for
example:
Jeff calling Cliff and Aaron: apr_shm_baseaddr_get() returns
addresses which aren't 64-bit aligned. That is broken. End of story.
--
Jeff Trawick | [EMAIL PROTECTED] | PGP public key at web site:
http://www.geocities.com/SiliconValley/Park/9289/
Born
Cliff Woolley [EMAIL PROTECTED] writes:
On 1 Mar 2002, Jeff Trawick wrote:
Jeff calling Cliff and Aaron: apr_shm_baseaddr_get() returns
addresses which aren't 64-bit aligned. That is broken. End of story.
Yeah, your last message cleared a lot up. Thanks. But I believe we're
all
would have done it
--
Jeff Trawick | [EMAIL PROTECTED] | PGP public key at web site:
http://www.geocities.com/SiliconValley/Park/9289/
Born in Roswell... married an alien...
() prior to the mmap()?
+*/
+
nbytes = sizeof(new_m-realsize);
status = apr_file_read(file, (void *)(new_m-realsize),
nbytes);
Perhaps whoever wrote this could take a look at the section of code
where I added that comment?
--
Jeff
Cliff Woolley [EMAIL PROTECTED] writes:
On Fri, 1 Mar 2002, Cliff Woolley wrote:
Does this look right? (attached to avoid line wrapping)
Hmm, no, it doesn't. :-/ Scratch that. Back to the drawing board.
What was wrong with it?
--
Jeff Trawick | [EMAIL PROTECTED] | PGP public key
' on libapr.so and httpd show all the
necessary functions have been redefined ( open() - open64(), stat()
-stat64(), mmap() - mmap64(), etc )
/dale
Index: apr/configure.in
Any reason this didn't get committed?
--
Jeff Trawick | [EMAIL PROTECTED] | PGP public key at web site:
http
.
Since MAX() isn't standard by any means, if we defined it we would
become the bad guys w.r.t. namespace protection if we defined it.
--
Jeff Trawick | [EMAIL PROTECTED] | PGP public key at web site:
http://www.geocities.com/SiliconValley/Park/9289/
Born in Roswell... married
be
instructive.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
looks to me that it may be needed on Win32.)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
(**rwlock));
+(*rwlock)-pool = pool;
committed...
Thanks!
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
on?
Was there a '\n' at the end of the file?
Do you have a patch?
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
it the same everywhere
+ */
+if (r-finfo.filetype == APR_REG
+r-filename[strlen(r-filename) - 1] == '/') {
+r-finfo.filetype = 0; /* forget what we learned */
+}
}
if (r-finfo.filetype == APR_REG) {
--
Jeff Trawick | [EMAIL PROTECTED]
Born
passed to Apache's configure and down to other
configures.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
(not APR_ENOTIMPL) it can
then try to retrieve the time via apr_stat() and verify that it was
set correctly.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
. This should be fixed now.
Do you know of any others?
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
-time per call handles your issues nicely. It might be
nice to have a flag that says to just set everything to the one time
so that we don't waste a syscall trying to preserve the times we don't
think we're supposed to modify.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
[EMAIL PROTECTED] writes:
wrowe 02/03/14 14:21:38
Modified:test testfile.c
Log:
Win32 doesn't support remove-open-file semantics.
a needlessly-friendly way of saying that I forgot to close the damn
file :)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell
, I hear, let's not make those poor [Windows|Netware|IDEfan] users
set up and configure perl, it's too hard(SM).
Phoey.
so do we get to ditch awk some day?
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
to APR so that
these priorities are irrelevant (they should only be used when we
don't have specific OS knowledge). Even if we think we ave the
priorities the same between 2.0 and 1.3 it isn't really the same if we
don't bring forward the cases where the priorities aren't even used.
--
Jeff Trawick
Jim Jagielski [EMAIL PROTECTED] writes:
Jeff Trawick wrote:
Jim Jagielski [EMAIL PROTECTED] writes:
Are people happy with the priority order of the accept mutex?
Right now it's flock - sysvsem - fcntl - pthread.
I think it should be pthread - sysvsem - fcntl - flock, which
with a cracked case waiting for me somewhere. Or an iMac that
came out puke green by mistake. (No, ssh to moof is not the same :) )
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
and our bug
databases will be even a little easier to look at, then I am strongly
in favor of keeping the -g as a default. If somebody is worried about
that amount of space then let them strip the executable files.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
it
is best to override the default and use 1:1.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
(or just call
+ * apr_set_socket_vars() to do it */
(*new)-remote_addr-port = ntohs((*new)-remote_addr-sa.sin.sin_port);
apr_pool_cleanup_register((*new)-cntxt, (void *)(*new),
socket_cleanup, apr_pool_cleanup_null);
--
Jeff Trawick | [EMAIL PROTECTED]
Born
it (and update my patches for OS/2 and
Win32) :)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
locks. They now work for me on
FreeBSD 3.4. Try it at your leisure :)
As far as the ENOSPC failure, I would assume that it is not a software
problem until/unless we get better doc. SysV sems work fine for me on
FreeBSD 3.4.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
/install/bin: No such file or
directory
../apr/apr-config: /home/regress/regress/install/bin: No such file or
directory
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
).
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
by user id remm; no set was available for
apache so the system returned ENOSPC
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
is a new lock mechanism
suddenly so high in the priority list? Aren't we throwing away a lot
of past experience (1.3 and 2.0)?
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Aaron Bannert [EMAIL PROTECTED] writes:
On Thu, Apr 04, 2002 at 03:39:41PM -0500, Jeff Trawick wrote:
sem_open(/ApR.whatever, O_CREAT, 0644,1) returns ENOSYS on Linux
(2.2.12 kernel). I'll try to add an autoconf check for that.
Beyond the issue mentioned above, why is a new lock
Jim Jagielski [EMAIL PROTECTED] writes:
Jeff Trawick wrote:
sem_open(/ApR.whatever, O_CREAT, 0644,1) returns ENOSYS on Linux
(2.2.12 kernel). I'll try to add an autoconf check for that.
Yeah... right now almost all the APR checks are just for existance
and not implementation... we
Aaron Bannert [EMAIL PROTECTED] writes:
On Thu, Apr 04, 2002 at 04:04:46PM -0500, Jeff Trawick wrote:
You should hard-code posix semaphores for Darwin in apr_hints.m4. We
should not be playing around with the lock choice and affecting
multiple systems when we get specific platform
Jeff Trawick [EMAIL PROTECTED] writes:
Also, we don't know how these things work in practice. What if (wild
hypothesis) Solaris has it but the default kernel config allows only a
handful of them? Then we start getting PRs from a bunch of the people
who did --disable-threads on Solaris
Ryan Morgan [EMAIL PROTECTED] writes:
On Fri, Apr 05, 2002 at 04:42:33PM -0500, Jeff Trawick wrote:
Ryan Morgan [EMAIL PROTECTED] writes:
The cleanup of the unix networking code a while back has broken
apr_sockaddr_port_get when 0 is specifed as the port to bind
errno == EINTR);
+if (sock-timeout rv *len) {
+sock-netmask |= APR_INCOMPLETE_WRITE;
+}
}
}
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
./buildconf before ./configure.
I build successfully on Tru64 (hudson) all the time using that
mechanism (though I don't start with a tarball, so your verification
will be helpful).
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Jim Jagielski [EMAIL PROTECTED] writes:
At present, it appears that libtool 1.4.2 is not compatible with
at least 2 platforms that Apache 2.0 runs on: Darwin (OS X) and AIX.
FYI... Apache 2.0 and libtool 1.4.2 and AIX get along reasonably
well. Pre-1.4.?, on the other hand :)
--
Jeff
Brian Pane [EMAIL PROTECTED] writes:
Jeff Trawick wrote:
+if (sock-netmask APR_INCOMPLETE_WRITE) {
+sock-netmask = ~APR_INCOMPLETE_WRITE;
+goto do_select;
+}
+
do {
rv = sendfile(sock-socketdes, /* socket */
file-filedes
attempting the next write.
go Brian
Obtained from: Jeff Trawick
mostly just a hint :)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
the whole issue. What problems would that present?
Thanks,
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
on the apr_mmap_t when we create
an mmap bucket?
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Cliff Woolley [EMAIL PROTECTED] writes:
On 15 Apr 2002, Jeff Trawick wrote:
What happens if we kill the cleanup on the apr_mmap_t when we create
an mmap bucket?
That would work [and would be the preferable solution as far as I'm
concerned], but there's currently no API to do
the binary.
Any idea what we have to change to make that happen? That requires
shipping all the .o .lo .a .la, right?
BTW, are you talking about pre-compiled binaries provided by
apache.org or pre-compiled binaries available in RPM/DEB/etc format?
binaries provided by apache.org
--
Jeff Trawick
);
...
}
I saw code like this:
while (apr_file_read(file, buffer, len) != APR_SUCCESS)
{
...
}
which one is better to use?
I would suggest the second one. What if you hit an I/O error? I
always did
while (!feof() !ferror())
with stdio.
--
Jeff
Stas Bekman [EMAIL PROTECTED] writes:
without this patch, while(!apr_file_eof(...)) spins for unbuffered
fh. ...
Committed... Thanks!
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
;
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Stas Bekman [EMAIL PROTECTED] writes:
Jeff Trawick wrote:
Stas Bekman [EMAIL PROTECTED] writes:
2. What's the idiomatic apr read till eof? From grepping the source
code, I see apr_file_eof is hardly ever used. Is something wrong with:
while (!apr_file_eof(fp
read the relevant text in Stevens'
APUE?
After reading it, my understanding is that Dave's change shouldn't
hurt any platforms and is definitely going to help some platforms.
Just my 2cents; you'll do want you want.
Jeff Trawick wrote:
This patch looks reasonable to me. Any comments from
will
be available as soon as it accepts a new connection. You could add a
poll for data available before the apr_recv() in the server.
More easily you could set a timeout on the socket, but then we already
test that with client.c I believe.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married
this be changed please? I had to so that I could compile my sources.
oops :)
fixed now
Thanks
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
the
segment of code in sa_common.c (~ line 420) where we do a strspn to see
if we are dealing with dotted-quad notation?
fair enough (+0.9)... we can zap the configure test and always handle
that part of processing...
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
the problems are mostly due to bad build tool setup and we'd be
better off trying to detect common problems and issuing a message that
would help them fix it.)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
for that or for a --disable-atomic switch which could be
used to alleviate any problems if they happen.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
-specific
implementations, but we can not rely on Linux to help us here. So,
remove this.
Great, thanks. There is a stale comment at line 96 which could probably
be deleted, and also this one:
Thanks, committed!
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
that we haven't mistakenly optimized
global mutexes on the current platform.
If folks could verify that the above info holds across releases (e.g.,
Linux 2.0.x, Linux 2.4.x, Solaris 2.6, etc.), then at least one of us
would feel more comfortable about making use of the information.
--
Jeff Trawick
Greg Stein [EMAIL PROTECTED] writes:
+1. Nuke the damned thing.
yep
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
?
If somebody has a patch, I'm willing to try it on AIX and HP-UX.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Jeff Trawick [EMAIL PROTECTED] writes:
The simple answer is to require libtool 1.4.
Two concerns:
1) libtool 1.4.2 fails on HP-UX. One of the Apache tarballs was
created with libtool 1.4.2 but it didn't work on HP-UX. Maybe
1.4.x works on HP-UX.
I was unable to recreate any
Jeff Trawick [EMAIL PROTECTED] writes:
Greg Stein [EMAIL PROTECTED] writes:
libtool 1.3 has problems with intra-library dependencies. This is making
some of the dependency stuff in apr(-util) a bit more complicated than it
needs to be.
The simple answer is to require libtool 1.4
-2.0-powerpc-ibm-aix4.3.3.0.README) in the
parent directory.
I need to look into this further.
(I first patched binbuild.sh to work-around current breakage with the
BinaryDistribution layout. I'm 90% sure that binbuild.sh didn't have
this problem before your patch, but I'll double check.)
--
Jeff
Jeff Trawick [EMAIL PROTECTED] writes:
I'm getting a weird problem with an Apache binary build:
Binary image successfully created...
exec(): 0509-036 Cannot load program ./bindist/bin/httpd because of the
following errors:
0509-150 Dependent module libaprutil.so could
catch the signal and reap status
in
+ * the handler to avoid zombies
+ */
+if ((signo == SIGCHLD) (func == SIG_IGN)) {
+act.sa_handler = avoid_zombies;
}
#endif
if (sigaction(signo, act, oact) 0)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married
have a timeout on the socket */
...disable incomplete reads
}
Also, we don't turn on the incomplete read flag until we get certain
feedback on I/O. It isn't enabled just because a certain option has
been set.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
as it can be done
without impacting the performance of the other, more heavily-used formats
in a horrendous way.
+1
The API needs to be changed to use const char * if at all possible.
This isn't the first time this has come up.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Does somebody want to break the tie? (so far two alternate opinions
of the same code, both indicating that there is a problem)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Jeff Trawick [EMAIL PROTECTED] writes:
[EMAIL PROTECTED] writes:
brianp 02/05/23 18:07:47
Modified:strmatch apr_strmatch.c
Log:
Switched to unsigned chars in the searching functions to avoid
problems when input strings contain character values 127
Reported
Brian Pane [EMAIL PROTECTED] writes:
There's only one operation in the whole algorithm that
really needs an unsigned char. I'll switch back to signed
chars, with a cast to unsigned only in the case where the
char value is used as an array index.
excellent!
--
Jeff Trawick | [EMAIL
for you with this patch? I never got to the bottom
of what was failing on AIX, but I suspected a build order problem that
caused it to fail if one of the support libraries didn't already exist
(yeah, that is pretty vague :( ). It didn't seem like a
libtool-on-AIX problem.
--
Jeff Trawick | [EMAIL
submitted a patch for that one about 2 months ago).
if no Win32 committer speaks up/commits soon, I'll do it myself,
albeit with no test-compile
Thanks for persisting!
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
William A. Rowe, Jr. [EMAIL PROTECTED] writes:
That's a placeholder for an decision not made (I believe we apr_pcalloc,
so it isn't a huge issue between = 0 or omitting it entirely.)
Oh right :) (To most people, it just looks like the same bug as with
the flags line right above it.)
--
Jeff
figuring out what to do if
mod_auth_digest won't initialize.
(clear as mud?)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
:04:34 - 1.8
+++ proc_mutex.h26 Jun 2002 16:29:52 -
@@ -142,7 +142,7 @@
#if !APR_HAVE_UNION_SEMUN defined(APR_HAS_SYSVSEM_SERIALIZE)
union semun {
-long val;
+int val;
struct semid_ds *buf;
unsigned short *array;
};
--
Jeff Trawick | [EMAIL
systems).
BTW, we were able to get Apache (2.0.39) to work on the HP-UX / Itanium
architecture without any other problems. Great work !!.
So I guess sizeof(long) is 8 bytes on 64-bit HP-UX? sizeof(long) is
still 4 on 64-bit AIX.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married
Jim Jagielski [EMAIL PROTECTED] writes:
Jeff Trawick wrote:
MATHIHALLI,MADHUSUDAN (HP-Cupertino,ex1) [EMAIL PROTECTED] writes:
On platforms where the semun union is not available, the APR defines the
semun structure - but I believe the first parameter should be a int val
Jeff Trawick [EMAIL PROTECTED] writes:
The header file defines in apr_private.h are busted and APR won't
build. Here is a glaring example:
/* Define if you have the string.h header file. */
/* #undef HAVE_STRING_H */
config.cache seems to have the right value:
ac_cv_header_string_h
food deficit)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Jeff Trawick [EMAIL PROTECTED] writes:
Justin Erenkrantz [EMAIL PROTECTED] writes:
I tried installing local copies of GNU m4 and autoconf 2.13, but the
problem persists. I even tried specifying /usr/local/bin/bash instead
of /bin/sh in configure. Any other ideas to narrow down what
Garrett Rooney [EMAIL PROTECTED] writes:
On Thu, Jun 27, 2002 at 06:17:41PM -0400, Jeff Trawick wrote:
talk about PRs coming our way for people upgrading FreeBSD... first,
old builds are suddenly unreliable, and new builds are impossible
without switching to GNU sed
Jeff Trawick [EMAIL PROTECTED] writes:
The header file defines in apr_private.h are busted and APR won't
build. Here is a glaring example:
Summary of ths solution:
We picked up a bad sed from 4.6-STABLE, which broke this. We then
picked up a subsequent fix, and we build again on daedalus
each of
the make files (probably not a fatal condition):
config.status: creating Makefile
mv: Makefile: set owner/group (was: 1121/0): Operation not permitted
make works, so I overreacted about all the messages.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
usec!
stderr:
thread_cond_timedwait test failed : [20507] The timeout specified has expired
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Jeff Trawick [EMAIL PROTECTED] writes:
I dunno what the cause is yet, but I'm getting testlock failures for
the last day or so.
Tru64:
thread_cond_timedwait Tests
Initializing the first apr_thread_mutex_t OK
Initializing the apr_thread_cond_t
;
+#endif /* HAVE_POLL */
}
apr_status_t apr_send(apr_socket_t *sock, const char *buf, apr_size_t *len)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
.
This works for me with some testing (timeouts on read and write work
for me).
Can we remove the #ifdef's by just using apr_poll here?
I'd rather we not, since that introduces a fair amount of extra
overhead.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
... this is an important path within APR... if we can use the
most efficient mechanism without much extra maintenance then we
should...
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
and less
speedy by inspection than using poll() directly is going to be
vetoed.
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
*/
extern int randbyte(void); /* from the truerand library */
Thanks,
Victor
--
Victor J. Orlikowski | The Wall is Down, But the Threat Remains!
==
[EMAIL PROTECTED] | [EMAIL PROTECTED] | [EMAIL PROTECTED]
--
Jeff
)) {
aprset[i].events |= APR_POLLERR;
no, this is POLLPRI
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
`ap_http_header_filter':
http_protocol.c:1561: warning: passing arg 1 of `apr_is_empty_table' from
incompatible pointer type
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
sdbm__delpair
sdbm__duppair
sdbm__putpair
sdbm__getpair
apr:
(none)
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
to the
previous or next year,
that year is used instead. (TZ)
FreeBSD:
%Gis replaced by a year as a decimal number with century. This year
is the one that contains the greater part of the week (Monday as
the first day of the week).
--
Jeff Trawick
document which format strings we support (limit it to
ANSI for portability)
alternatively:
http://www.freebsd.org/cgi/cvsweb.cgi/src/lib/libc/stdtime/strftime.c
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Jim Jagielski [EMAIL PROTECTED] writes:
Jeff Trawick wrote:
FreeBSD:
%Gis replaced by a year as a decimal number with century. This
year
is the one that contains the greater part of the week (Monday as
the first day of the week).
Do we need
in these cases.
Doesn't your code assume that errno gets cleared by the iconv routine?
Is that really true, or does errno need to be cleared before iconv
calls?
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
Ed Holyat [EMAIL PROTECTED] writes:
!DOCTYPE HTML PUBLIC -//W3C//DTD HTML 3.2//EN
Can you post in plain text please?
Thanks!
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
[EMAIL PROTECTED] writes:
Added: buildapu-iconv.m4
our CVS repository is now infected with the carriage-return virus
--
Jeff Trawick | [EMAIL PROTECTED]
Born in Roswell... married an alien...
301 - 400 of 1573 matches
Mail list logo