GTTF: dump registers on crash on 64-bit Linux

GTTF: dump registers on crash on 64-bit Linux

BUG=none

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=176110

[jar (doing other things), 2013-01-04 20:57:36]

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix.cc
File base/debug/stack_trace_posix.cc (right):

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix....
base/debug/stack_trace_posix.cc:199: { "  r8: ",
context->uc_mcontext.gregs[REG_R8] },
Question: Are all these registers held intact during this execution of the
initializer?  Are you guaranteed that they won't be changed?

[Paweł Hajdan Jr., 2013-01-04 21:53:22]

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix.cc
File base/debug/stack_trace_posix.cc (right):

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix....
base/debug/stack_trace_posix.cc:199: { "  r8: ",
context->uc_mcontext.gregs[REG_R8] },
On 2013/01/04 20:57:36, jar wrote:
> Question: Are all these registers held intact during this execution of the
> initializer?  Are you guaranteed that they won't be changed?

See http://pubs.opengroup.org/onlinepubs/007904975/functions/sigaction.html .
The context we get as an argument is the context from time the signal was
caught, not the current context, which could have been obtained with just
getcontext(2).

[jar (doing other things), 2013-01-09 19:04:25]

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix.cc
File base/debug/stack_trace_posix.cc (right):

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix....
base/debug/stack_trace_posix.cc:131: OutputPointer(trace[i], handler);
Why did you change the order of this in the output?  Does anyone use and parse
the current output?

[Paweł Hajdan Jr., 2013-01-09 19:08:25]

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix.cc
File base/debug/stack_trace_posix.cc (right):

https://codereview.chromium.org/11745003/diff/1/base/debug/stack_trace_posix....
base/debug/stack_trace_posix.cc:131: OutputPointer(trace[i], handler);
On 2013/01/09 19:04:25, jar wrote:
> Why did you change the order of this in the output?  Does anyone use and parse
> the current output?

I'm not aware of anyone parsing this output, it's not designed for parsing, it's
not intended for parsing. The new order makes stack addresses aligned in the
output, possibly making it easier to see some regularities.

[jar (doing other things), 2013-01-10 04:55:05]

lgtm