llvm dev - Nov 2011 - [LLVMdev] LLVM: segfault while modifying a thread-local global variable

OK, that's my fault. I'm running jitted code on linux x86.

2011/11/9 Duncan Sands <baldrick at free.fr>
> Hi Semion,
>
>
>  In my case the LLVM is built with assertions, but this doesn't help.
From
>> your
>> answer I conclude that thread-local globals are fully implemented in
x86
>> JIT.
>>
>
> you didn't mention the JIT.  You didn't give any details at all! 
That's
> why I
> was only able to make a generic suggestion.  Don't read too much into
it.
>
> Ciao, Duncan.
>
>  So
>
>> this is strange...
>>
>> 2011/11/9 Duncan Sands <baldrick at free.fr <mailto:baldrick at
free.fr>>
>>
>>
>>    Hi Semion,
>>
>>     > Really strange situation: when I modify a thread-local
>> GlobalVariable I get a
>>     > segfault. It seems to be a bug, doesn't it? Can u suggest
me some
>> solution?
>>
>>    build LLVM with assertions enabled.  Then you may well get an
>> assertion failure
>>    with a helpful error message rather than an obscure crash.
>>
>>    Ciao, Duncan.
>>    ______________________________**_________________
>>    LLVM Developers mailing list
>>    LLVMdev at cs.uiuc.edu <mailto:LLVMdev at cs.uiuc.edu>
>> http://llvm.cs.uiuc.edu
>>   
http://lists.cs.uiuc.edu/**mailman/listinfo/llvmdev<http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111109/402cacab/attachment.html>

Which version of LLVM? At any rate I don't recall TLS working for the JIT.
It may take some work.

-eric


On Nov 9, 2011, at 4:46 AM, Semion Prihodko wrote:
> OK, that's my fault. I'm running jitted code on linux x86.
> 
> 2011/11/9 Duncan Sands <baldrick at free.fr>
> Hi Semion,
> 
> 
> In my case the LLVM is built with assertions, but this doesn't help.
From your
> answer I conclude that thread-local globals are fully implemented in x86
JIT.
> 
> you didn't mention the JIT.  You didn't give any details at all! 
That's why I
> was only able to make a generic suggestion.  Don't read too much into
it.
> 
> Ciao, Duncan.
> 
>  So
> this is strange...
> 
> 2011/11/9 Duncan Sands <baldrick at free.fr <mailto:baldrick at
free.fr>>
> 
> 
>    Hi Semion,
> 
>     > Really strange situation: when I modify a thread-local
GlobalVariable I get a
>     > segfault. It seems to be a bug, doesn't it? Can u suggest me
some solution?
> 
>    build LLVM with assertions enabled.  Then you may well get an assertion
failure
>    with a helpful error message rather than an obscure crash.
> 
>    Ciao, Duncan.
>    _______________________________________________
>    LLVM Developers mailing list
>    LLVMdev at cs.uiuc.edu <mailto:LLVMdev at cs.uiuc.edu>
http://llvm.cs.uiuc.edu
>    http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
> 
> 
> 
> 
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111109/518df029/attachment.html>

I use LLVM 2.9. By the way, I have another question. Is landingpad
instruction is already implemented in JIT or I should use the old approach
with llvm.eh instrinsics? Thanks.

2011/11/9 Eric Christopher <echristo at apple.com>
> Which version of LLVM? At any rate I don't recall TLS working for the
JIT.
> It may take some work.
>
> -eric
>
>
> On Nov 9, 2011, at 4:46 AM, Semion Prihodko wrote:
>
> OK, that's my fault. I'm running jitted code on linux x86.
>
> 2011/11/9 Duncan Sands <baldrick at free.fr>
>
>> Hi Semion,
>>
>>
>>  In my case the LLVM is built with assertions, but this doesn't
help.
>>> From your
>>> answer I conclude that thread-local globals are fully implemented
in x86
>>> JIT.
>>>
>>
>> you didn't mention the JIT.  You didn't give any details at
all!  That's
>> why I
>> was only able to make a generic suggestion.  Don't read too much
into it.
>>
>> Ciao, Duncan.
>>
>>  So
>>
>>> this is strange...
>>>
>>> 2011/11/9 Duncan Sands <baldrick at free.fr <mailto:baldrick
at free.fr>>
>>>
>>>
>>>    Hi Semion,
>>>
>>>     > Really strange situation: when I modify a thread-local
>>> GlobalVariable I get a
>>>     > segfault. It seems to be a bug, doesn't it? Can u
suggest me some
>>> solution?
>>>
>>>    build LLVM with assertions enabled.  Then you may well get an
>>> assertion failure
>>>    with a helpful error message rather than an obscure crash.
>>>
>>>    Ciao, Duncan.
>>>    ______________________________**_________________
>>>    LLVM Developers mailing list
>>>    LLVMdev at cs.uiuc.edu <mailto:LLVMdev at cs.uiuc.edu>
>>> http://llvm.cs.uiuc.edu
>>>   
http://lists.cs.uiuc.edu/**mailman/listinfo/llvmdev<http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev>
>>>
>>>
>>>
>>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20111109/029e8e65/attachment.html>