Re: Intent to TR 2.2.12
HEAD on httpd-2.2 passes the perl framework tests and looks good. Planning on tagging/rolling later on today assuming nothing pops up, so please test beforehand :)
RE: Intent to TR 2.2.12
-Original Message- From: Jim Jagielski Sent: Montag, 20. Juli 2009 13:29 To: dev@httpd.apache.org Subject: Re: Intent to TR 2.2.12 HEAD on httpd-2.2 passes the perl framework tests and looks good. Planning on tagging/rolling later on today assuming nothing pops up, so please test beforehand :) What about the dup3 / accept4 and so on detection issue in APR? Do we want to see a fixed APR release before or do we live with this issue in 2.2.12? Regards Rüdiger
Re: Intent to TR 2.2.12
On Jul 20, 2009, at 7:47 AM, Plüm, Rüdiger, VF-Group wrote: -Original Message- From: Jim Jagielski Sent: Montag, 20. Juli 2009 13:29 To: dev@httpd.apache.org Subject: Re: Intent to TR 2.2.12 HEAD on httpd-2.2 passes the perl framework tests and looks good. Planning on tagging/rolling later on today assuming nothing pops up, so please test beforehand :) What about the dup3 / accept4 and so on detection issue in APR? Do we want to see a fixed APR release before or do we live with this issue in 2.2.12? I get the impression that we won't be seeing a new APR release anytime soon, due to the concern on whether this is an APR issue or an OS related one. However, instead of waiting for a full APR release, it would be nice to maybe tag an interim version of APR and bundle *that* with 2.2.12... CCing d...@apr
Re: Intent to TR 2.2.12
all, Jim Jagielski schrieb: HEAD on httpd-2.2 passes the perl framework tests and looks good. Planning on tagging/rolling later on today assuming nothing pops up, so please test beforehand :) would be really great if I could get some votes on the gen_test_char change - it doesnt alter code for any other platform, but only makes it possible to decouple gen_test_char from APR with a define so I'm able to build a native version of it when cross-compiling: http://people.apache.org/~fuankg/diffs/gen_test_char.c.diff if nobody objects I would like to start in around one hour, and prepare our NetWare build system for this where I set it inactive by default unless I get the votes for gen_test_char changes. Günter.
Re: Intent to TR 2.2.12
Guenter Knauf wrote: would be really great if I could get some votes on the gen_test_char change - it doesnt alter code for any other platform, but only makes it possible to decouple gen_test_char from APR with a define so I'm able to build a native version of it when cross-compiling: http://people.apache.org/~fuankg/diffs/gen_test_char.c.diff I see there is a WANT_WIN32_OS2 symbol as well which seems unrelated to the CROSS_COMPILE symbol, can you confirm whether you need both? Regards, Graham -- smime.p7s Description: S/MIME Cryptographic Signature
Re: Intent to TR 2.2.12
On Jul 20, 2009, at 12:56 PM, Graham Leggett wrote: Guenter Knauf wrote: would be really great if I could get some votes on the gen_test_char change - it doesnt alter code for any other platform, but only makes it possible to decouple gen_test_char from APR with a define so I'm able to build a native version of it when cross-compiling: http://people.apache.org/~fuankg/diffs/gen_test_char.c.diff I see there is a WANT_WIN32_OS2 symbol as well which seems unrelated to the CROSS_COMPILE symbol, can you confirm whether you need both? +1
Re: Intent to TR 2.2.12
Hi Graham, Graham Leggett schrieb: Guenter Knauf wrote: would be really great if I could get some votes on the gen_test_char change - it doesnt alter code for any other platform, but only makes it possible to decouple gen_test_char from APR with a define so I'm able to build a native version of it when cross-compiling: http://people.apache.org/~fuankg/diffs/gen_test_char.c.diff I see there is a WANT_WIN32_OS2 symbol as well which seems unrelated to the CROSS_COMPILE symbol, can you confirm whether you need both? yes, for two reasons: 1. I had problems with undefining WIN32 with the CodeWarrior Win32 compiler, so the WIN32 (now WANT_WIN32_OS2) part snapped in unwanted. 2. I also thought of a Win32 cross compile where I want to have the WANT_WIN32_OS2 ifdef'd part in so that I can build the right gen_test_char for Win32 platform which runs native on Linux (of course in this case also a -DWIN32 would have done). Günter.
Re: Intent to TR 2.2.12
Hi, Graham Leggett schrieb: I see there is a WANT_WIN32_OS2 symbol as well which seems unrelated to the CROSS_COMPILE symbol, can you confirm whether you need both? probably the name was not good - I was also thinking of something like NEED_ENHANCED_ESCAPES or so ...; if someone has a better idea please tell me ... in the #else part of CROSS_COMPILE there I set WANT_WIN32_OS2: #if defined(WIN32) || defined(OS2) #define WANT_WIN32_OS2 #endif Günter.
Re: Intent to TR 2.2.12
On Jul 20, 2009, at 1:48 PM, Guenter Knauf wrote: Hi, Graham Leggett schrieb: I see there is a WANT_WIN32_OS2 symbol as well which seems unrelated to the CROSS_COMPILE symbol, can you confirm whether you need both? probably the name was not good - I was also thinking of something like NEED_ENHANCED_ESCAPES or so ...; if someone has a better idea please tell me ... in the #else part of CROSS_COMPILE there I set WANT_WIN32_OS2: #if defined(WIN32) || defined(OS2) #define WANT_WIN32_OS2 #endif Looks safe to me...
Re: Intent to TR 2.2.12
On Jul 20, 2009, at 1:23 PM, Guenter Knauf wrote: Hi Graham, Graham Leggett schrieb: Guenter Knauf wrote: would be really great if I could get some votes on the gen_test_char change - it doesnt alter code for any other platform, but only makes it possible to decouple gen_test_char from APR with a define so I'm able to build a native version of it when cross-compiling: http://people.apache.org/~fuankg/diffs/gen_test_char.c.diff I see there is a WANT_WIN32_OS2 symbol as well which seems unrelated to the CROSS_COMPILE symbol, can you confirm whether you need both? yes, for two reasons: 1. I had problems with undefining WIN32 with the CodeWarrior Win32 compiler, so the WIN32 (now WANT_WIN32_OS2) part snapped in unwanted. 2. I also thought of a Win32 cross compile where I want to have the WANT_WIN32_OS2 ifdef'd part in so that I can build the right gen_test_char for Win32 platform which runs native on Linux (of course in this case also a -DWIN32 would have done). +1...
Re: Intent to TR 2.2.12
Jim Jagielski wrote: However, instead of waiting for a full APR release, it would be nice to maybe tag an interim version of APR and bundle *that* with 2.2.12... No, it would not, httpd will not become responsible for APR's releases unless the APR project is folded and httpd project votes to accept the responsibility for this code. So... -1 on any APR fork in an httpd release (and my feelings are similar on PCRE or expat forks, and for very similar reasons).
Re: Intent to TR 2.2.12
Nick Kew wrote on 2009-07-19 00:04:59: Just been reviewing it with the testcase Bob found. I'm not able to reproduce the problem on this platform because Sun CC sets the non-matches to 0, so it all works. But the problem is clear. This throws up a non-serious problem with the patch: testing for 0. Wouldn't a better test be rm_eo == rm_so, meaning null match? I think you are right. The tests for 0 are part of the original code so I was just keeping them but testing for re-have_match first. Patching trunk based on the above. Will propose for backport if noone disputes my amendment to the patch. I've seen you added both tests in your patch so we are good anyway. +1 (and thanks for getting this fixed in trunk, I didn't had the time to look at this over the weekend). cheers... -- Lars Eilebrecht l...@eilebrecht.net
Re: Intent to TR 2.2.12
Nick Kew wrote: Patching trunk based on the above. Will propose for backport if noone disputes my amendment to the patch. Done in r795445. -- Nick Kew
Re: Intent to TR 2.2.12
Lars Eilebrecht wrote: Jim Jagielski wrote: Over the weekend I'll be doing some final things with the intent to tag and roll 2.2.12 on Monday... I just realized that I still have one patch for 2.2.12 which fixes an SSI-related bug causing a segfault when handling regex back-references (see attachment). Just been reviewing it with the testcase Bob found. I'm not able to reproduce the problem on this platform because Sun CC sets the non-matches to 0, so it all works. But the problem is clear. This throws up a non-serious problem with the patch: testing for 0. Wouldn't a better test be rm_eo == rm_so, meaning null match? Patching trunk based on the above. Will propose for backport if noone disputes my amendment to the patch. -- Nick Kew
Re: Intent to TR 2.2.12
Jim Jagielski wrote:
Over the weekend I'll be doing some final things with the intent
to tag and roll 2.2.12 on Monday...
I just realized that I still have one patch for 2.2.12 which fixes an
SSI-related bug causing a segfault when handling regex back-references
(see attachment).
I didn't propose it yet for inclusion in 2.2.12 as I didn't had the
chance to fix this in trunk yet. The code/api in trunk changed
and I don't know if this bug actually exists in trunk.
I don't know if I will have the time to do this over the weekend.
ciao...
--
Lars Eilebrecht
l...@eilebrecht.net
--- mod_include.c.orig 2008-12-17 14:27:41.0 +
+++ mod_include.c 2009-02-27 15:39:22.0 +
@@ -158,6 +158,7 @@
const char *rexp;
apr_size_t nsub;
ap_regmatch_t match[AP_MAX_REG_MATCH];
+int have_match;
} backref_t;
typedef struct {
@@ -664,6 +665,11 @@
return NULL;
}
else {
+if (!re-have_match ||
+ re-match[idx].rm_so 0 || re-match[idx].rm_eo 0) {
+return NULL;
+}
+
if (re-nsub idx || idx = AP_MAX_REG_MATCH) {
ap_log_rerror(APLOG_MARK, APLOG_WARNING, 0, r,
regex capture $% APR_SIZE_T_FMT
@@ -672,10 +678,6 @@
return NULL;
}
-if (re-match[idx].rm_so 0 || re-match[idx].rm_eo 0) {
-return NULL;
-}
-
val = apr_pstrmemdup(ctx-dpool, re-source + re-match[idx].rm_so,
re-match[idx].rm_eo - re-match[idx].rm_so);
}
@@ -923,7 +925,6 @@
{
ap_regex_t *compiled;
backref_t *re = ctx-intern-re;
-int rc;
compiled = ap_pregcomp(ctx-dpool, rexp, AP_REG_EXTENDED);
if (!compiled) {
@@ -939,10 +940,11 @@
re-source = apr_pstrdup(ctx-pool, string);
re-rexp = apr_pstrdup(ctx-pool, rexp);
re-nsub = compiled-re_nsub;
-rc = !ap_regexec(compiled, string, AP_MAX_REG_MATCH, re-match, 0);
+re-have_match = !ap_regexec(compiled, string, AP_MAX_REG_MATCH,
+ re-match, 0);
ap_pregfree(ctx-dpool, compiled);
-return rc;
+return re-have_match;
}
static int get_ptoken(include_ctx_t *ctx, const char **parse, token_t *token, token_t *previous)
RE: Intent to TR 2.2.12
-Original Message- From: Lars Eilebrecht Sent: Freitag, 17. Juli 2009 15:49 To: dev@httpd.apache.org Subject: Re: Intent to TR 2.2.12 Jim Jagielski wrote: Over the weekend I'll be doing some final things with the intent to tag and roll 2.2.12 on Monday... I just realized that I still have one patch for 2.2.12 which fixes an SSI-related bug causing a segfault when handling regex back-references (see attachment). I didn't propose it yet for inclusion in 2.2.12 as I didn't had the chance to fix this in trunk yet. The code/api in trunk changed and I don't know if this bug actually exists in trunk. IMHO trunk should have the same problem as the code looks similar. I don't know if I will have the time to do this over the weekend. If you have time (fix in trunk, backport proposal) I will have a look at the proposal and vote on it to get it in. Regards Rüdiger
Re: Intent to TR 2.2.12
Plüm, Rüdiger, VF-Group wrote: If you have time (fix in trunk, backport proposal) I will have a look at the proposal and vote on it to get it in. +1. Segfault sounds serious enough to prioritise! -- Nick Kew
Re: Intent to TR 2.2.12
On Jul 17, 2009, at 9:49 AM, Lars Eilebrecht wrote: Jim Jagielski wrote: Over the weekend I'll be doing some final things with the intent to tag and roll 2.2.12 on Monday... I just realized that I still have one patch for 2.2.12 which fixes an SSI-related bug causing a segfault when handling regex back-references (see attachment). I didn't propose it yet for inclusion in 2.2.12 as I didn't had the chance to fix this in trunk yet. The code/api in trunk changed and I don't know if this bug actually exists in trunk. I don't know if I will have the time to do this over the weekend. ciao... -- Lars Eilebrecht l...@eilebrecht.net mod_include_backref.patch I'll look and review... Most likely we'll have enough others to see this thru :)
Re: Intent to TR 2.2.12
2009/7/17 Nick Kew n...@webthing.com: I've a faint recollection of someone raising this issue, but a quick google didn't find it. Do you have a test-case that provokes the bug you're fixing? http://markmail.org/message/jlc7t5edsjujbe37 ;-) Bob
