Mailing List Archive

On 22 Jan 2013 at 10:56, Grant wrote:

> google-chrome suffers intermittent crashes on x86 unless I enable
> softmode. Is there any other option to keep it running?

can you get some details on the nature of crashes? any logs perhaps?
if softmode fixes it then it's probably MPROTECT related, did you
already disable it on the binary?

>> google-chrome suffers intermittent crashes on x86 unless I enable
>> softmode. Is there any other option to keep it running?
>
> can you get some details on the nature of crashes? any logs perhaps?
> if softmode fixes it then it's probably MPROTECT related, did you
> already disable it on the binary?

MPROTECT is actually already disabled:

# paxctl -v /opt/google/chrome/chrome
PaX control v0.7
Copyright 2004,2005,2006,2007,2009,2010,2011,2012 PaX Team
<pageexec@freemail.hu>
- PaX flags: -----m-x-e-- [/opt/google/chrome/chrome]
MPROTECT is disabled
RANDEXEC is disabled
EMUTRAMP is disabled

Here's the segfault from dmesg:

chrome[13338]: segfault at bbadbeef ip 14d81f2b sp 5ec2e1b0 error 6 in
chrome[136f4000+4203000]
grsec: Segmentation fault occurred at bbadbeef in
/opt/google/chrome/chrome[chrome:13338] uid/euid:1009/1009
gid/egid:1009/1009, parent /opt/google/chrome/chrome[chrome:12981]
uid/euid:1009/1009 gid/egid:1009/1009

- Grant

On 22 Jan 2013 at 19:44, Grant wrote:

> >> google-chrome suffers intermittent crashes on x86 unless I enable
> >> softmode. Is there any other option to keep it running?
> >
> > can you get some details on the nature of crashes? any logs perhaps?
> > if softmode fixes it then it's probably MPROTECT related, did you
> > already disable it on the binary?
>
> MPROTECT is actually already disabled:

ok, in that case can you report this

> chrome[13338]: segfault at bbadbeef ip 14d81f2b sp 5ec2e1b0 error 6 in chrome[136f4000+4203000]

to the chrome folks? i'm pretty sure that 'bbadbeef' is a magic constant
somewhere in their userland memory management code and signifies some
corruption (use-after-free perhaps?) and they'd probably like to know
how to reproduce this.

On 23/01/13 08:17, PaX Team wrote:
> On 22 Jan 2013 at 19:44, Grant wrote:
>
>>>> google-chrome suffers intermittent crashes on x86 unless I enable
>>>> softmode. Is there any other option to keep it running?
>>>
>>> can you get some details on the nature of crashes? any logs perhaps?
>>> if softmode fixes it then it's probably MPROTECT related, did you
>>> already disable it on the binary?
>>
>> MPROTECT is actually already disabled:
>
> ok, in that case can you report this
>
>> chrome[13338]: segfault at bbadbeef ip 14d81f2b sp 5ec2e1b0 error 6 in chrome[136f4000+4203000]
>
> to the chrome folks? i'm pretty sure that 'bbadbeef' is a magic constant
> somewhere in their userland memory management code and signifies some
> corruption (use-after-free perhaps?) and they'd probably like to know
> how to reproduce this.
>
>
webkit¿?

>> >> google-chrome suffers intermittent crashes on x86 unless I enable
>> >> softmode. Is there any other option to keep it running?
>> >
>> > can you get some details on the nature of crashes? any logs perhaps?
>> > if softmode fixes it then it's probably MPROTECT related, did you
>> > already disable it on the binary?
>>
>> MPROTECT is actually already disabled:
>
> ok, in that case can you report this
>
>> chrome[13338]: segfault at bbadbeef ip 14d81f2b sp 5ec2e1b0 error 6 in chrome[136f4000+4203000]
>
> to the chrome folks? i'm pretty sure that 'bbadbeef' is a magic constant
> somewhere in their userland memory management code and signifies some
> corruption (use-after-free perhaps?) and they'd probably like to know
> how to reproduce this.

OK, so no Gentoo bug, just a chromium bug?

- Grant

On 1/23/13 11:28 AM, Grant wrote:
> OK, so no Gentoo bug, just a chromium bug?

Most likely. Please report it and paste link here.

Also see
<http://www.chromium.org/for-testers/bug-reporting-guidelines/reporting-crash-bug>

Paweł

>> OK, so no Gentoo bug, just a chromium bug?
>
> Most likely. Please report it and paste link here.

I'm sorry for the delay with this. Here's the bug:

http://code.google.com/p/chromium/issues/detail?id=175316

- Grant