Bug #69820 [Ver->Csd]: CLI crashes at start with various extensions enabled

From: Date: Sun, 14 Jun 2015 11:12:25 +0000
Subject: Bug #69820 [Ver->Csd]: CLI crashes at start with various extensions enabled
References: 1  Groups: php.bugs 
Request: Send a blank email to php-bugs+get-193469@lists.php.net to get a copy of this message
Edit report at https://bugs.php.net/bug.php?id=69820&edit=1

 ID:                 69820
 Updated by:         ab@php.net
 Reported by:        maikgreubel at gmail dot com
 Summary:            CLI crashes at start with various extensions enabled
-Status:             Verified
+Status:             Closed
 Type:               Bug
 Package:            Reproducible crash
 Operating System:   Windows 7 64bit
 PHP Version:        7.0.0alpha1
 Assigned To:        ab
 Block user comment: N
 Private report:     N

 New Comment:

Thanks for checking the alpha1 and for the good backtrace :)

Can be closed then.

Thanks.


Previous Comments:
------------------------------------------------------------------------
[2015-06-14 10:50:59] maikgreubel at gmail dot com

Tested the snapshot build and et'voila, crash does not appear.

Thank you, nice done.

------------------------------------------------------------------------
[2015-06-14 10:43:41] ab@php.net

@maikgreubel, or alternatively - just check this snapshot http://windows.php.net/downloads/snaps/ostc/69814/
(built from the latest master).

Thanks.

------------------------------------------------------------------------
[2015-06-14 10:31:03] maikgreubel at gmail dot com

Going back to qa-build binary and debugging.

+		&p->val	0x000000000417bf80 {value={lval=112 dval=5.533535233422e-322#DEN
counted=0x0000000000000070 {refcount=...} ...} ...}	_zval_struct *
		apply_func	0x000007fede71c480 {php7ts.dll!zend_startup_module_zval(_zval_struct *)}	int
(_zval_struct *) *
+		ht	0x000007fedeecdb00 {php7ts.dll!_zend_array module_registry} {gc={refcount=1 u={v={type=7
'\a' flags=...} ...} } ...}	_zend_array *
+		ht->arData	0x00000000060a2800 {val={value={lval=68645888 dval=3.391557498907e-316#DEN
counted=0x0000000004177400 {...} ...} ...} ...}	_Bucket *
		ht->nNumUsed	33	unsigned int
		idx	8	unsigned int
+		p	0x000000000417bf80 {val={value={lval=112 dval=5.533535233422e-322#DEN
counted=0x0000000000000070 {refcount=...} ...} ...} ...}	_Bucket *
+		p->val	{value={lval=112 dval=5.533535233422e-322#DEN counted=0x0000000000000070 {refcount=???
u={v={type=??? ...} ...} } ...} ...}	_zval_struct
		result	0	int

Due to optimization I can not add conditional breakpoint at for-loop iteration of HashTable in
zend_hash_apply to find out, which module entry causes the problem. I try to investigate the scope
where the module_registry HashTable is filled with entries and what happens between initialization
of HT and the module activation.

------------------------------------------------------------------------
[2015-06-14 10:25:40] ab@php.net

@maikgreubel, yeah, see bug #69823. My insight about ext/hash was correct. I think we're done
here. Please fetch the latest master to check this.

Thanks.

------------------------------------------------------------------------
[2015-06-14 10:01:32] ab@php.net

Thanks for the followup. Yeah, --enable-pgi is a part of the PGO build scenario, not needed if you
don't do it. 

But something strange happens. Yesterday I was able to reproduce your initial report using 

x64\Release_TS\php.exe -n -d extension_dir=x64\Release_TS -d extension=php_fileinfo.dll -d
extension=php_mbstring.dll -d extension=php_pdo_mysql.dll -d extension=php_pdo_sqlite.dll

But this doesn't crash when I've fetched and rebuilt today. Not doing PGO build as well.
But that's why i guess it's not an optimization issue, still quite strange. Maybe you can
repro it with this configure line?

cscript /nologo configure.js  "--enable-snapshot-build" "--enable-debug-pack"
"--enable-com-dotnet=shared"

That picks up literally everything. But of course, maybe that's just PGO, so probably were
worth to check it (but need setup for scenarios). Also, it might be a vc14 issue as it's still
RC. Whereby I seem have not received any vc14 updates over the night.

Thanks.

------------------------------------------------------------------------


The remainder of the comments for this report are too long. To view
the rest of the comments, please view the bug report online at

    https://bugs.php.net/bug.php?id=69820


--
Edit this bug report at https://bugs.php.net/bug.php?id=69820&edit=1


Thread (14 messages)

« previous php.bugs (#193469) next »