Re: [patch,7.3] Fix JIT clang-lli gdb-7.3 regression

[Paul Pluzhnikov <ppluzhnikov@google.com>]

On Wed, Jul 6, 2011 at 9:32 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> On Wed, 06 Jul 2011 18:07:04 +0200, Joel Brobecker wrote:
>> > That is on unload+reload of the JIT engine.  OTOH (a) this case is probably
>> > not faced by normal users and (b) it needs more work for better performance
>> > and (c) in this case maybe the overall multi-JIT rework would be better.
>>
>> So, would the following approach be acceptable?
>>   (1) HEAD: Apply Paul's patch, then look at solutions for
>>       the remaining failures/regressions...
>
> If Paul is going to implement the multi-JITer one as he was talking about.

I am not sure there is an actual use case for that, and I was not planning
on working on that.

> Otherwise I would be for update by me of the very last patch of mine which
> PASSed unload + reload as 739571cc3151651f49f7171cfd98275d983bfaaa^ did.

Fixing the unload/reload case would (I expect) be fairly simple without
reverting to a global -- just need to clear jit_inferior_data on unload.

>>   (2) BRANCH: Apply Paul's patch; I think we might want to apply
>>       the testcase part as well, even if we still need to figure out
>>       why it doesn't work for Jan...

Updated patch should work for Jan...

Thanks,
-- 
Paul Pluzhnikov