[llvm-commits] [llvm] r118422 - in /llvm/trunk: lib/Target/ARM/ARMAsmPrinter.cpp lib/Target/ARM/ARMBuildAttrs.h test/CodeGen/ARM/2010-10-19-mc-elf-objheader.ll

Jason Kim jasonwkim at google.com
Mon Nov 8 21:24:31 PST 2010


On Mon, Nov 8, 2010 at 3:41 PM, Evan Cheng <evan.cheng at apple.com> wrote:
>
> On Nov 8, 2010, at 10:41 AM, Jason Kim wrote:
>
>> On Mon, Nov 8, 2010 at 10:37 AM, Anton Korobeynikov
>> <anton at korobeynikov.info> wrote:
>>> Hi Jason,
>>>
>>>> +    if (CPUString != "generic") {
>>>> +      if (CPUString == "cortex-a8") {
>>> This looks pretty gross. Is it possible to generalize stuff somehow?
>>> E.g. via the mapping of CPU names => supported attributes.
>>
>> I agree its very gross. I am going to be iterating over this repeatedly.
>> I don't yet have a clear understanding of the space of cpu/attributes yet.
>> As I get more familiarized with how GNU/as does things, I will be
>> refactoring these.
>>
> it's also wrong. What about other cpu's which are cortex-a8 variants and later? Do we have to get this patch in now? Can you table it until you have figured out how to match these to subtarget attributes?

Hi Evan, and Anton

Thanks for review. This was meant as an incremental patch to get a
corner subcase working. The attributes that it emitted was "correct"
at least according to GNU/as for the specific subcase.

I suspect that the a substantially complete table that encodes the
various switches and their correspondence to .ARM.attributes  might be
somewhat nontrivial - as the
"correctness criteria" is largely dependent upon the quirks of GNU/as
(Documentation from ARM does not help much here )-:
Not to mention, this is for ELF only - Does MachO care for these at all?

In such cases, what ever I commit in an incremental manner will either
be substantially incomplete (i.e. looks like dead code), or
substantially messy looking, or both.

As it turns out, there was a (trivial) bug in the test I committed,
(but it was enough to cause build break on darwin) so it was rolled
back :-( so its a moot point.

I'll refactor and email updated patch to commits for review.

Thanks again
-jason

>
> Evan
>
>>
>>>
>>> --
>>> With best regards, Anton Korobeynikov
>>> Faculty of Mathematics and Mechanics, Saint Petersburg State University
>>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>




More information about the llvm-commits mailing list