On 04/09/2014 03:19 AM, Nicolas Richard wrote:
> [I'm cc-ing Bastien because he had a similar problem IIRC]
> 
> Hi David,
> 
>> Still dumps core.  Looks pretty much the same to me:
>> #0  0x40022424 in __kernel_vsyscall ()
>> #1  0x416b60c6 in raise (sig=sig@entry=6)
>>     at ../nptl/sysdeps/unix/sysv/linux/pt-raise.c:37
> 
> You could try Daniel's suggestion to bug#17184, i.e. compile with
> -DSYSTEM_PURESIZE_EXTRA=10000000
> 
> Perhaps it is time to increase BASE_PURESIZE ? FWIW I applied this patch
> to my local branch :
> 

Probably a good idea, assuming the core size increase is natural and
doesn't reflect some deeper problem.

Aside: why do we even bother with the malloc-on-pure-overflow logic? We
can't GC once we've overflown pure storage (although that could be made
to work), so an Emacs in that condition isn't very useful. Wouldn't it
be simpler just to abort when we overflow pure storage?