[Gc] sparc patch

Petr Krajča krajcap at inf.upol.cz
Thu Oct 30 09:25:16 PST 2008


On Sun, 26 Oct 2008 01:32:09 -0400, Hans Boehm <Hans.Boehm at hp.com> wrote:

> On Wed, 22 Oct 2008, Petr Kraj?a wrote:
>
>> Btw. I'm using GC on SPARC even with THREAD_LOCAL_ALLOC option turned on
>> and I think it may be turned on by default, but I don't know how to test
>> if it works correctly.
>
> I would expect that it works correctly.  You should see an appreciable
> performance difference for a test program that just allocates and drops
> objects in a tight loop, especially in multiple threads.  Turning
> this on replaces GC_malloc with a completely different implementation.
> Since it doesn't crash, you're probably OK.  And there's no fundamental
> reason it shouldn't work on sparc linux, aside from possibly the very
> intermittent bug some others are reporting on X86.

I've tested this with my applications---they run significantly faster and  
without any unexpected crash. Hence, I'm attaching a patch that enables  
THREAD_LOCAL_SUPPORT on sparc-linux.

Petr
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gc-sparc-thread.patch
Type: application/octet-stream
Size: 529 bytes
Desc: not available
Url : http://napali.hpl.hp.com/pipermail/gc/attachments/20081030/63c75c3c/gc-sparc-thread.obj


More information about the Gc mailing list