[ASan] share code at the beginning/end of error reports

[glider1, 2012-08-10 12:03:50]

https://codereview.appspot.com/6446107/diff/4001/asan_report.cc
File asan_report.cc (right):

https://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode48
asan_report.cc:48: for (uptr i = 0; i < byte_num; i++) {
++i is more stylistically preferable, but looks like we're not using it at all
in ASan codebase.

https://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode63
asan_report.cc:63: PrintBytes("  ", (uptr*)(aligned_shadow-4*kWordSize));
How about writing a loop here?

https://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode63
asan_report.cc:63: PrintBytes("  ", (uptr*)(aligned_shadow-4*kWordSize));
Please also fix the spacing between the operators.

https://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode218
asan_report.cc:218: AsanPrintf("AddressSanitizer: while reporting a bug found
another one."
It's better to use AsanReport() here.

https://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode220
asan_report.cc:220: SleepForSeconds(5);
I think we should sleep for |sleep_before_dying| seconds here.
Keep in mind that Clang may replace the current while(1) loop with a call to
_exit(1), so we should either enforce it or let the user to control the sleep
time.

[samsonov, 2012-08-10 14:34:58]

http://codereview.appspot.com/6446107/diff/4001/asan_report.cc
File asan_report.cc (right):

http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode48
asan_report.cc:48: for (uptr i = 0; i < byte_num; i++) {
On 2012/08/10 12:03:50, glider1 wrote:
> ++i is more stylistically preferable, but looks like we're not using it at all
> in ASan codebase.

Yeah, "i++" vs. "++i" = 27:0. I'll leave it as is for consistency.

http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode63
asan_report.cc:63: PrintBytes("  ", (uptr*)(aligned_shadow-4*kWordSize));
On 2012/08/10 12:03:50, glider1 wrote:
> Please also fix the spacing between the operators.

Done.

http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode63
asan_report.cc:63: PrintBytes("  ", (uptr*)(aligned_shadow-4*kWordSize));
On 2012/08/10 12:03:50, glider1 wrote:
> How about writing a loop here?

Done.

http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode218
asan_report.cc:218: AsanPrintf("AddressSanitizer: while reporting a bug found
another one."
On 2012/08/10 12:03:50, glider1 wrote:
> It's better to use AsanReport() here.

Done.

http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode220
asan_report.cc:220: SleepForSeconds(5);
On 2012/08/10 12:03:50, glider1 wrote:
> I think we should sleep for |sleep_before_dying| seconds here.
> Keep in mind that Clang may replace the current while(1) loop with a call to
> _exit(1), so we should either enforce it or let the user to control the sleep
> time.

Done.

[samsonov, 2012-08-10 15:14:25]

On 2012/08/10 14:34:58, samsonov wrote:
> http://codereview.appspot.com/6446107/diff/4001/asan_report.cc
> File asan_report.cc (right):
> 
> http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode48
> asan_report.cc:48: for (uptr i = 0; i < byte_num; i++) {
> On 2012/08/10 12:03:50, glider1 wrote:
> > ++i is more stylistically preferable, but looks like we're not using it at
all
> > in ASan codebase.
> 
> Yeah, "i++" vs. "++i" = 27:0. I'll leave it as is for consistency.
> 
> http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode63
> asan_report.cc:63: PrintBytes("  ", (uptr*)(aligned_shadow-4*kWordSize));
> On 2012/08/10 12:03:50, glider1 wrote:
> > Please also fix the spacing between the operators.
> 
> Done.
> 
> http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode63
> asan_report.cc:63: PrintBytes("  ", (uptr*)(aligned_shadow-4*kWordSize));
> On 2012/08/10 12:03:50, glider1 wrote:
> > How about writing a loop here?
> 
> Done.
> 
> http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode218
> asan_report.cc:218: AsanPrintf("AddressSanitizer: while reporting a bug found
> another one."
> On 2012/08/10 12:03:50, glider1 wrote:
> > It's better to use AsanReport() here.
> 
> Done.
> 
> http://codereview.appspot.com/6446107/diff/4001/asan_report.cc#newcode220
> asan_report.cc:220: SleepForSeconds(5);
> On 2012/08/10 12:03:50, glider1 wrote:
> > I think we should sleep for |sleep_before_dying| seconds here.
> > Keep in mind that Clang may replace the current while(1) loop with a call to
> > _exit(1), so we should either enforce it or let the user to control the
sleep
> > time.
> 
> Done.

r161666