[Mesa-dev] Mesa (master): gallium/util: remove the fake format helpers for bptc and etc2

[Roland Scheidegger]

Am 24.11.2015 um 16:08 schrieb Roland Scheidegger:
> Am 24.11.2015 um 16:04 schrieb Ilia Mirkin:
>> On Tue, Nov 24, 2015 at 1:37 AM, Michel Dänzer <michel at daenzer.net> wrote:
>>> On 24.11.2015 01:21, Ilia Mirkin wrote:
>>>> Module: Mesa
>>>> Branch: master
>>>> Commit: 1c7d0a6aa4f5cb38af7e281e1e5437cd1a20f781
>>>> URL:    https://urldefense.proofpoint.com/v2/url?u=http-3A__cgit.freedesktop.org_mesa_mesa_commit_-3Fid-3D1c7d0a6aa4f5cb38af7e281e1e5437cd1a20f781&d=BQIGaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=Vjtt0vs_iqoI31UfJxBl7yv9I2FeiaeAYgMTLKRBc_I&m=XZNFHsl4lFAxxg4-uHH5HyanjvlMvLVTdJeebqtJNNk&s=at-4ysFGpZbYjQkXGULUBbK_sc5mmCztZ2hJGhsczvg&e= 
>>>>
>>>> Author: Ilia Mirkin <imirkin at alum.mit.edu>
>>>> Date:   Sat Nov 21 10:44:00 2015 -0500
>>>>
>>>> gallium/util: remove the fake format helpers for bptc and etc2
>>>>
>>>> This was a silly hack that kept growing and growing. Instead, just write
>>>> NULLs for those functions. No need to have helpers that just assert(0)
>>>> when you call them.
>>>>
>>>> Signed-off-by: Ilia Mirkin <imirkin at alum.mit.edu>
>>>> Reviewed-by: Marek Olšák <marek.olsak at amd.com>
>>>> Reviewed-by: Roland Scheidegger <sroland at vmware.com>
>>>
>>> This change broke make check, specifically llvmpipe's lp_test_format:
>>
>> Hmmm... odd. Calling any of those should have ended up in asserts
>> anyways. I did realize that I forgot to add ASTC to the
>> no-softpipe/llvmpipe list, so I'll do that, but those were all added
>> in the next commit. Will also investigate what I mucked up here,
>> thanks for the heads up.
>>
> 
> If you don't beat me to it, I'm going to remove testing of these formats
> in lp_test_format. Though I'm curious why it didn't assert before as
> well, as you said it can't have really worked. The difference though is
> now it will assert when generating the jit code rather than when the jit
> code is called.
> 

Actually lp_test_format only tests formats for which there are test
cases for in u_format_tests.c, and for these unsupported formats there
aren't any. It does, however, generate jit fetch for all formats. Hence
why it didn't crash before and does now - the generated jit fetch
function is never called.
Speaking of that, some formats which we do claim support for in llvmpipe
are actually missing test cases as well - a quick look suggests etc1,
rgtc, but possibly more.
Patch coming...

Roland