87054 - pthreads and boehm problem on non-(i386 and glibc) system

Status: NEW

(NSPR :: NSPR, defect, P3)

Description

[John Mark Vandenberg]

mozilla/nsprpub/pr/src/pthreads/ptthread.c uses GetStackTop more liberally than 
it implements it.

#if defined(GC_LEAK_DETECTOR) && (__GLIBC__ >= 2) && defined(__i386__)
...
static void* GetStackTop()
{
...
}
#endif

#if defined(GC_LEAK_DETECTOR)
         ts->stackTop    = GetStackTop();
...
#endif

Comment 1

[John Mark Vandenberg]

Attachment: Patch to restrict usage of GetStackTop

Comment 2

[Wan-Teh Chang]

If you want to use the Boehm GC leak detector on your
platform, you must define the GetStackTop() function
for your platform.

Comment 3

[Nick Blievers]

Attachment: nsprpub/pr/src/pthreads/ptthread.c - Add GetStackTop() for IRIX

This is a first cut (eg I need to test properly) at a GetStackTop() IRIX. I've
made some assumptions and really need to test them.

Comment 4

[Nick Blievers]

Added myself to cc-list

Comment 5

[Nick Blievers]

Attachment: Bah, above without returning 1 all the time

Comment 6

[Nick Blievers]

The only problem with this is related to which compiler is used. eg:
% /usr/local/gcc-2.95.3/bin/gcc -v
Reading specs from
/usr/local/gcc-2.95.3/lib/gcc-lib/mips-sgi-irix6.5/2.95.3/specs
gcc version 2.95.3 20010315 (release)
% /usr/local/gcc-2.95.3/bin/gcc -o stacktrace stack.c
test_stacktrace.c -lexc
% ./stacktrace
__exc_unwind: Error: pc(0x100013d8) not in any .debug_frame table
 0x000000000001 <unknown function>
% /usr/local/gcc-2.95.3/bin/gcc -g -o stacktrace stack.c
test_stacktrace.c -lexc
% ./stacktrace
 0x0000100014bc backtrace
 0x000010001504 cfunc
 0x00001000154c bfunc
 0x00001000159c afunc
 0x000010001290 main

% gcc -v
Reading specs from /usr/freeware/lib/gcc-lib/mips-sgi-irix6.5/3.0.4/specs
Configured with: ../configure --prefix=/usr/freeware
--enable-version-specific-runtime-libs --disable-shared
--enable-threads --enable-haifa
Thread model: single
gcc version 3.0.4
% gcc -o stacktrace stack.c test_stacktrace.c -lexc
% ./stacktrace
__exc_unwind: Error: pc(0x10001498) not in any .debug_frame table
% gcc -g -o stacktrace stack.c test_stacktrace.c -lexc
% ./stacktrace
 0x000010001578 backtrace
 0x0000100015c0 cfunc
 0x000010001608 bfunc
 0x000010001658 afunc
 0x000010001320 main

% cc -version
MIPSpro Compilers: Version 7.3.1.2m
% cc -o stacktrace stack.c test_stacktrace.c -lexc
stack.c:
test_stacktrace.c:
% ./stacktrace
 0x000010000f94 backtrace
 0x000010000fd4 cfunc
 0x000010001014 bfunc
 0x00001000105c afunc
 0x000010000e50 main
% cc -g -o stacktrace stack.c test_stacktrace.c -lexc
stack.c:
test_stacktrace.c:
% ./stacktrace
 0x000010000fb4 backtrace
 0x000010000ff4 cfunc
 0x000010001034 bfunc
 0x00001000107c afunc
 0x000010000e70 main

% cc -Ofast -o stacktrace stack.c test_stacktrace.c -lexc
stack.c:
test_stacktrace.c:
% ./stacktrace
 0x000010001688 backtrace
 0x0000100010e0 main


gcc on IRIX does not support -rdynamic. So, its of limited use with gcc. Of
course MipsPro works fine, even at high levels of optimisation.

Comment 7

[Nick Blievers]

Attachment: gc/boehm/gcconfig.h - patch to allow IRIX to use generic push regs

Comment 8

[Nick Blievers]

Attachment: nsprpub/pr/src/Makefile.in - link with libexc if IRIX and using gc

Comment 9

[Christophe Ravel]

The target milestone is already released. Resetting target milestone.

Comment 10

[Philip Chee]

Closing old IRIX bugs.

Comment 11

[BugBot [:suhaib / :marco/ :calixte]]

The bug assignee didn't login in Bugzilla in the last 7 months, so the assignee is being reset.