[gccsdk] Dynamic libraries vs pthreads

Peter Naulls peter at chocky.org
Sun Aug 24 20:38:16 BST 2008


Lee Noar wrote:
> Peter Naulls wrote:
>> Lee Noar wrote:
>>> Peter Naulls wrote:
>>>> Lee Noar wrote:
>>>>
>>>>> OK, once you've updated the AB, I'll try to build and run it. It'll 
>>>>> have to be the Risc PC version though. Whether all the libraries 
>>>>> will fit in the RMA is a different matter...
>>>>
>>>> Done.  Increasing the wimpslot (say to 100MB, which you can't do on a
>>>> RiscPC) helps a little, but it still appears to run out of memory. So
>>>> this suggests that the allocations are going wrong somehow.
>>>
>>> I've hit my first problem :-) I've attached the last-failure and 
>>> config.log files for libpopt0.
>>
>> I don't know if you noticed the fix I put in a few days ago for this.
> 
> Yes, thanks and I've managed to build it. From a fresh Boot with about 
> 6MB in the RMA to start, it runs out of library memory after about 39 
> are loaded - see attached SOMStatus output file to see how far it gets 
> and the resulting stderr1 output. However, I've also seen it crash after 
> only 22 libraries are loaded, see attached stderr2. Does the latter one 
> look familiar? How many libraries load before it fails for you?

No, this is different.  I'm currently chasing two issues; one which
is an intermittent call to strcpy with NULL, due to some path issues,
and a later failure to load a security library (which isn't a fatal
error).  The earlier crash I referred to happens much later than tese
(right now I have the security thing abort), but all of these are
sometime after what you're talking about.   Of course, it's
possible I missed out some part of the patches I put in, too.

> Does it 
> fail in pthread_yield_return by any chance?

Well, I certainly don't have enough information to rule out
the later crash there.  Anyway, thanks for looking as this;
people are not often interested in looking at such hard
problems :-|






More information about the gcc mailing list