Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 6 Aug 2009 12:20:08 -0500 (CDT)
From:      "Sean C. Farley" <scf@FreeBSD.org>
To:        Juergen Lock <nox@jelal.kn-bremen.de>
Cc:        freebsd-emulation@FreeBSD.org
Subject:   Re: nspluginwrapper patch for testing
Message-ID:  <alpine.BSF.2.00.0908061209240.1622@thor.farley.org>
In-Reply-To: <20090806170526.GA10971@triton.kn-bremen.de>
References:  <20090628082701.GA34665@triton.kn-bremen.de> <200908061418.36989.tijl@ulyssis.org> <alpine.BSF.2.00.0908060807160.4271@thor.farley.org> <200908061710.39039.tijl@ulyssis.org> <alpine.BSF.2.00.0908061106150.1521@thor.farley.org> <20090806170526.GA10971@triton.kn-bremen.de>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 6 Aug 2009, Juergen Lock wrote:

> On Thu, Aug 06, 2009 at 11:14:22AM -0500, Sean C. Farley wrote:
>> On Thu, 6 Aug 2009, Tijl Coosemans wrote:
>>> On Thursday 06 August 2009 15:15:10 Sean C. Farley wrote:
>>>> On Thu, 6 Aug 2009, Tijl Coosemans wrote:
>>>>> On Thursday 06 August 2009 12:14:10 Boris Samorodov wrote:
>>>>>> On Thu, 6 Aug 2009 10:59:21 +0200 Tijl Coosemans wrote:
>>>>>>> Can this be committed before the 8.0 ports freeze?
>>>>>>
>>>>>> According to the list this solution appears to be helpful,
>>>>>> so the answer is "yes" (or rather "should").
>>>>>>
>>>>>> Can you provide a commit log?
>>>>>
>>>>> Limit the stack size for plugins. Linux glibc threading behaves
>>>>> slightly differently with a smaller stack size and the Flash plugins
>>>>> rely on this behaviour.
>>>>>
>>>>> Suggested by:   dchagin
>>>>> Patch by:       nox
>>>>
>>>> Also, it helps for running Google Earth.  Until I limited the stack
>>>> size for it, it would crash if I zoomed in too close to the Earth.
>>>> It still crashes for me when exiting Google Earth, but I am able to
>>>> live with that.  This is when running on i386 7-STABLE with the
>>>> Nvidia driver v185.18.29.
>>>
>>> You mean a similar patch should applied to the google-earth port?
>>
>> I am not sure.  It helps me, but I do not know if my case is 
>> peculiar.  I do not have this problem on an amd64 7-STABLE (same 
>> revision as above) using the radeonhd driver.  However, on the amd64 
>> system, the RV610 chip does not have hardware acceleration at this 
>> time.  Can anyone replicate my problem?
>
> Already committed. :)

Thank you.

> (As Linux has a smaller stack by default iiuc its unlikely to hurt, 
> and it _seems_ to help against the hangs/deadlocks I got which I used 
> to work around by forcing googleearth onto a single cpu too.)

I am glad it helps.  I wonder what other applications would benefit from 
this and if there is a more global way to set the default for Linux 
applications to a stack size of 32MB.

Sean
-- 
scf@FreeBSD.org



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.00.0908061209240.1622>