[Pkg-chromium-maint] Bug#680566: chromium: SIGSEGV in fopen64 at content/browser/zygote_main_linux.cc:840 on armel

[Jonathan Nieder]

shawn wrote:
> On Fri, 2012-07-06 at 17:57 -0500, Jonathan Nieder wrote: 

>> Does 0x... work any better?  (The leading "0x" means this is a
>> hexadecimal address.)
>
> oh wow, i feel like an idiot now, i just copy/pasted the address....
>
> (gdb)  disassemble 0x2c2e0a4c
> Dump of assembler code for function fopen64(char const*, char const*):
>    0x2c2e09d4 <+0>: push {r4, r5, lr}
>    0x2c2e09d8 <+4>: ldr r3, [pc, #316] ; 0x2c2e0b1c <fopen64(char const*, char const*)+328>

The disassembly you sent before started with

   0x022e09d4 <+0>: push {r4, r5, lr}
   0x022e09d8 <+4>: ldr r3, [pc, #316] ; 0x22e0b1c <fopen64(char const*,

so the difference is conveniently 0x0a000000.

That means we were at

   0x022e0a4c <+120>: mov r3, #0

(initializing the index for comparing the path to "/dev/urandom").
Um.  Or somewhere nearby, I guess --- the addresses seem to be off
a little.