Subject: bin/12283: Apache LoadModudule crashes httpd - dlopen()/dlclose() broken ?
To: None <gnats-bugs@gnats.netbsd.org>
From: Jaromír <jdolecek@NetBSD.org>
List: netbsd-bugs
Date: 02/25/2001 18:26:22
>Number:         12283
>Category:       bin
>Synopsis:       Apache LoadModudule crashes httpd - dlopen()/dlclose() broken ?
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    bin-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Feb 25 09:25:00 PST 2001
>Closed-Date:
>Last-Modified:
>Originator:     Jaromír Dolecek
>Release:        NetBSD-20010224
>Organization:
	N/A
>Environment:
	
System: NetBSD saruman.ics.muni.cz 1.5S NetBSD 1.5S (SARUMAN) #21: Sat Feb 24 16:48:12 CET 2001 dolecek@saruman.ics.muni.cz:/usr/home/dolecek/soft/netbsd/sys/arch/i386/compile/SARUMAN i386
Architecture: i386
Machine: i386
>Description:
	When I recently updated userland (I used some mid-February userland),
	httpd started crashing when any LoadModule directive would be
	used. I recompiled both httpd and modules in question couple
	of times, so this should not be any versioning problem.	Unfortunately,
	I also upgraded httpd in between, so this MIGHT be caused
	by some version mismatch.
>How-To-Repeat:
	on -current machine, LoadModule e.g. PHP4 module and see httpd
	crash
>Fix:
	Don't know. I can provide stack backtrace upon request. Repeatably,
	the program crashes within _rtld_symlook_obj(), obj->buckets
	points to incorrect memory. When I look at obj contents,
	couple of other structure members point to invalid (unmapped) memory.
	This leads me to think this is really ld.elf_so issue. This MIGHT
	be triggered by latest kernel changes (I have ~up-to-date kernel, too).
>Release-Note:
>Audit-Trail:
>Unformatted: