[ltt-dev] Insight into crash on powerpc with ust-0.11

[ltt-dev] Insight into crash on powerpc with ust-0.11

[Jason Wessel]

I built and linked a PPC application with -lust it gets a segmentation fault before it ever gets to main().

Breakpoint 1 at 0x10001b04: file src/main.c, line 196.
(gdb) c
Continuing.

Program received signal SIGSEGV, Segmentation fault.
0x0fd97b5c in marker_probe_register (channel=0xfdae034 "metadata", name=0xfdae050 "core_marker_format", format=0x0, probe=0xfd9f6d4 <ltt_vtrace>,
    probe_private=0x1002e070) at marker.c:799
799                     trace_mark(metadata, core_marker_id,
(gdb) bt
#0  0x0fd97b5c in marker_probe_register (channel=0xfdae034 "metadata", name=0xfdae050 "core_marker_format", format=0x0, probe=0xfd9f6d4 <ltt_vtrace>,
    probe_private=0x1002e070) at marker.c:799
#1  0x0fd9a5f8 in ltt_marker_connect (channel=0xfdae034 "metadata", mname=0xfdae050 "core_marker_format", pname=<value optimized out>) at marker-control.c:187
#2  0x0fd9a740 in init_marker_control () at marker-control.c:416
#3  0x0fdabc50 in __do_global_ctors_aux () from /ord-gp11/jwessel/42/lemon_p2020/export/dist/usr/lib/libust.so.0
#4  0x0fd94a84 in _init () from /ord-gp11/jwessel/42/lemon_p2020/export/dist/usr/lib/libust.so.0
#5  0x0ffceaf8 in call_init (l=0x48000a70, argc=1, argv=0xbfe2cc34, env=0xbfe2cc3c) at dl-init.c:70
#6  0x0ffcec60 in _dl_init (main_map=0xfff08e8, argc=1, argv=0xbfe2cc34, env=0xbfe2cc3c) at dl-init.c:134
#7  0x0ffd67d4 in _start () from /ord-gp11/jwessel/42/lemon_p2020/export/dist/lib/ld.so.1
Backtrace stopped: frame did not save the PC
(gdb)

With some disassembly:
(gdb)  disas /m $pc-16,$pc+16
Dump of assembler code from 0xfd97b4c to 0xfd97b6c:
797                     entry->event_id = ret;
   0x0fd97b4c <marker_probe_register+1888>:     clrlwi  r8,r28,16
   0x0fd97b50 <marker_probe_register+1892>:     sth     r8,54(r31)

798                     ret = 0;
799                     trace_mark(metadata, core_marker_id,
   0x0fd97b54 <marker_probe_register+1896>:     lis     r3,3
   0x0fd97b58 <marker_probe_register+1900>:     ori     r3,r3,12120
=> 0x0fd97b5c <marker_probe_register+1904>:     lbz     r0,12(r3)
   0x0fd97b60 <marker_probe_register+1908>:     li      r24,1
   0x0fd97b64 <marker_probe_register+1912>:     cmpwi   cr7,r0,0
   0x0fd97b68 <marker_probe_register+1916>:     beq+    cr7,0xfd97bf0 <marker_probe_register+2052>
   0x0fd97b6c <marker_probe_register+1920>:     li      r0,4
   0x0fd97b70 <marker_probe_register+1924>:     li      r4,0
   0x0fd97b74 <marker_probe_register+1928>:     stw     r0,8(r1)
   0x0fd97b78 <marker_probe_register+1932>:     addi    r5,r1,24
   0x0fd97b7c <marker_probe_register+1936>:     stw     r0,12(r1)

I am assuming there is something wrong with the trace_mark macro but I had not dug into how it actually works, and thought I might ask here first if anyone had any ideas.

Thanks,
Jason.

[ltt-dev] Insight into crash on powerpc with ust-0.11

[Mathieu Desnoyers]

* Jason Wessel (jason.wessel at windriver.com) wrote:
> I built and linked a PPC application with -lust it gets a segmentation
> fault before it ever gets to main().
> 

Hi Jason,

Before we dig further, what version of UST are you using ? Can you try
with the version in the git HEAD at http://lttng.org/ust ?

Thanks!

Mathieu

> Breakpoint 1 at 0x10001b04: file src/main.c, line 196.
> (gdb) c
> Continuing.
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x0fd97b5c in marker_probe_register (channel=0xfdae034 "metadata", name=0xfdae050 "core_marker_format", format=0x0, probe=0xfd9f6d4 <ltt_vtrace>,
>     probe_private=0x1002e070) at marker.c:799
> 799                     trace_mark(metadata, core_marker_id,
> (gdb) bt
> #0  0x0fd97b5c in marker_probe_register (channel=0xfdae034 "metadata", name=0xfdae050 "core_marker_format", format=0x0, probe=0xfd9f6d4 <ltt_vtrace>,
>     probe_private=0x1002e070) at marker.c:799
> #1  0x0fd9a5f8 in ltt_marker_connect (channel=0xfdae034 "metadata", mname=0xfdae050 "core_marker_format", pname=<value optimized out>) at marker-control.c:187
> #2  0x0fd9a740 in init_marker_control () at marker-control.c:416
> #3  0x0fdabc50 in __do_global_ctors_aux () from /ord-gp11/jwessel/42/lemon_p2020/export/dist/usr/lib/libust.so.0
> #4  0x0fd94a84 in _init () from /ord-gp11/jwessel/42/lemon_p2020/export/dist/usr/lib/libust.so.0
> #5  0x0ffceaf8 in call_init (l=0x48000a70, argc=1, argv=0xbfe2cc34, env=0xbfe2cc3c) at dl-init.c:70
> #6  0x0ffcec60 in _dl_init (main_map=0xfff08e8, argc=1, argv=0xbfe2cc34, env=0xbfe2cc3c) at dl-init.c:134
> #7  0x0ffd67d4 in _start () from /ord-gp11/jwessel/42/lemon_p2020/export/dist/lib/ld.so.1
> Backtrace stopped: frame did not save the PC
> (gdb)
> 
> With some disassembly:
> (gdb)  disas /m $pc-16,$pc+16
> Dump of assembler code from 0xfd97b4c to 0xfd97b6c:
> 797                     entry->event_id = ret;
>    0x0fd97b4c <marker_probe_register+1888>:     clrlwi  r8,r28,16
>    0x0fd97b50 <marker_probe_register+1892>:     sth     r8,54(r31)
> 
> 798                     ret = 0;
> 799                     trace_mark(metadata, core_marker_id,
>    0x0fd97b54 <marker_probe_register+1896>:     lis     r3,3
>    0x0fd97b58 <marker_probe_register+1900>:     ori     r3,r3,12120
> => 0x0fd97b5c <marker_probe_register+1904>:     lbz     r0,12(r3)
>    0x0fd97b60 <marker_probe_register+1908>:     li      r24,1
>    0x0fd97b64 <marker_probe_register+1912>:     cmpwi   cr7,r0,0
>    0x0fd97b68 <marker_probe_register+1916>:     beq+    cr7,0xfd97bf0 <marker_probe_register+2052>
>    0x0fd97b6c <marker_probe_register+1920>:     li      r0,4
>    0x0fd97b70 <marker_probe_register+1924>:     li      r4,0
>    0x0fd97b74 <marker_probe_register+1928>:     stw     r0,8(r1)
>    0x0fd97b78 <marker_probe_register+1932>:     addi    r5,r1,24
>    0x0fd97b7c <marker_probe_register+1936>:     stw     r0,12(r1)
> 
> I am assuming there is something wrong with the trace_mark macro but I had not dug into how it actually works, and thought I might ask here first if anyone had any ideas.
> 
> Thanks,
> Jason.
> 
> _______________________________________________
> ltt-dev mailing list
> ltt-dev at lists.casi.polymtl.ca
> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
> 

-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com