[Mesa-dev] [PATCH] gallivm: Do not use NoFramePointerElim with LLVM 3.7

[Roland Scheidegger]

Am 28.05.2015 um 14:28 schrieb Jose Fonseca:
> For the record, we also feel the pain, and I did look into the LLVM C
> API for disassembly, but the problem is that we only get the function
> start address -- we don't know where it finishes --, and the LLVM C API
> doesn't expose enough capabilities to describe jump/return instructions,
> which we rely upon to guess where a function ends.
> 
> Another approach would be to get the functions start-stop addresses
> separately -- via the DelegatingJITMemoryManager somehow.
> 
> 
> 
> BTW, this LLVM breakage has nothing to do with the disassembly. It just
> happens to be  on the same file.  It would have been broken even without
> it.
> 
> In fact, doesn't this mean that we'll stumble into
> http://llvm.org/PR21435 again?
> We must disable frame pointer elimination *somehow*, given upstream LLVM
> didn't bother fixing the bug.


Oh indeed. I'm not sure if this still happens it seems the decision to
eliminate frame pointer is now per function.
http://permalink.gmane.org/gmane.comp.compilers.llvm.cvs/248931

Not sure though if this means that bug got fixed as a side effect or
not, if not we sure need to do something about it.

Roland

> 
> 
> Jose
> 
> 
> On 28/05/15 12:56, Roland Scheidegger wrote:
>> I use this sometimes for debugging things in llvmpipe, so I don't think
>> that's a good option. Without it getting the actual assembly of the
>> shaders would be quite annoying.
>> It is unfortunately true that (due to the heavy use of the unstable C++
>> API) it breaks very often, but as long as it isn't possible to do that
>> with llvm's C API it seems unavoidable.
>>
>> Roland
>>
>> Am 28.05.2015 um 12:01 schrieb Marek Olšák:
>>> The disassemble function does more harm than good and is often the
>>> most often broken function after an LLVM update. Shouldn't we remove
>>> it to make our lives easier?
>>>
>>> Marek
>>>
>>>
>>> On Wed, May 27, 2015 at 7:27 AM, Vinson Lee <vlee at freedesktop.org>
>>> wrote:
>>>> TargetOptions::NoFramePointerElim was removed in llvm-3.7.0svn r238244
>>>> "Remove NoFramePointerElim and NoFramePointerElimOverride from
>>>> TargetOptions and remove ExecutionEngine's dependence on CodeGen. NFC."
>>>>
>>>> Signed-off-by: Vinson Lee <vlee at freedesktop.org>
>>>> ---
>>>>   src/gallium/auxiliary/gallivm/lp_bld_debug.cpp | 2 ++
>>>>   src/gallium/auxiliary/gallivm/lp_bld_misc.cpp  | 2 ++
>>>>   2 files changed, 4 insertions(+)
>>>>
>>>> diff --git a/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
>>>> b/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
>>>> index be3e834..76c302f 100644
>>>> --- a/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
>>>> +++ b/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
>>>> @@ -277,8 +277,10 @@ disassemble(const void* func, llvm::raw_ostream
>>>> & Out)
>>>>      options.StackAlignmentOverride = 4;
>>>>   #endif
>>>>   #if defined(DEBUG) || defined(PROFILE)
>>>> +#if HAVE_LLVM < 0x0307
>>>>      options.NoFramePointerElim = true;
>>>>   #endif
>>>> +#endif
>>>>      OwningPtr<TargetMachine> TM(T->createTargetMachine(Triple,
>>>> sys::getHostCPUName(), "", options));
>>>>
>>>>      /*
>>>> diff --git a/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
>>>> b/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
>>>> index 5e8a634..ffed9e6 100644
>>>> --- a/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
>>>> +++ b/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
>>>> @@ -439,8 +439,10 @@
>>>> lp_build_create_jit_compiler_for_module(LLVMExecutionEngineRef *OutJIT,
>>>>   #if HAVE_LLVM < 0x0304
>>>>      options.NoFramePointerElimNonLeaf = true;
>>>>   #endif
>>>> +#if HAVE_LLVM < 0x0307
>>>>      options.NoFramePointerElim = true;
>>>>   #endif
>>>> +#endif
>>>>
>>>>      builder.setEngineKind(EngineKind::JIT)
>>>>             .setErrorStr(&Error)
>>>> -- 
>>>> 2.4.1
>>>>
>>>> _______________________________________________
>>>> mesa-dev mailing list
>>>> mesa-dev at lists.freedesktop.org
>>>> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.freedesktop.org_mailman_listinfo_mesa-2Ddev&d=BQIGaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=Vjtt0vs_iqoI31UfJxBl7yv9I2FeiaeAYgMTLKRBc_I&m=oJN3RQI6auvaC8t4GPlTvSe8h-WZYhBL7xSRhbwFgDI&s=Ht4F7y_dMUmK34b_5vMdCKunj7WvDKU8-8HWca89Ei0&e=
>>>>
>>> _______________________________________________
>>> mesa-dev mailing list
>>> mesa-dev at lists.freedesktop.org
>>> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.freedesktop.org_mailman_listinfo_mesa-2Ddev&d=BQIGaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=Vjtt0vs_iqoI31UfJxBl7yv9I2FeiaeAYgMTLKRBc_I&m=oJN3RQI6auvaC8t4GPlTvSe8h-WZYhBL7xSRhbwFgDI&s=Ht4F7y_dMUmK34b_5vMdCKunj7WvDKU8-8HWca89Ei0&e=
>>>
>>>
>>
>> _______________________________________________
>> mesa-dev mailing list
>> mesa-dev at lists.freedesktop.org
>> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.freedesktop.org_mailman_listinfo_mesa-2Ddev&d=BQIGaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=zfmBZnnVGHeYde45pMKNnVyzeaZbdIqVLprmZCM2zzE&m=j--Khc7uwEjgb39x21uUdm65cn47ORER1E2LLAosy28&s=KOd3_0hp1Z6nppy1NmIibOIP05Q_sl7IY3G7yoFrt9M&e=
>>
>>
>