[FFmpeg-devel] [PATCH] libavcodec/hevcdsp: port SIMD idct functions from 32-bit.

Reimar Döffinger Reimar.Doeffinger at gmx.de
Thu Feb 11 22:06:38 EET 2021


Hi Martin!

> On 10 Feb 2021, at 22:53, Martin Storsjö <martin at martin.st> wrote:
> 
>>>>> +.macro idct_16x16 bitdepth
>>>>> +function ff_hevc_idct_16x16_\bitdepth\()_neon, export=1
>>>>> +//r0 - coeffs
>>>>> +        mov             x15, lr
>>>>> +
>>>> Binutils doesn't recognize "lr" as alias for x30
>>> It didn’t have an issue in the Debian unstable VM?
>>> That seems like the kind of workaround where it would be
>>> better to leave a comment with more info, if you know
>>> what exactly is affected.
>> 
>> Binutils 2.28 doesn't recognize "lr" while 2.30 does, it seems.
>> 
>> FWIW, all the existing aarch64 assembly just uses "x30" to refer to this register, none of it uses "lr".
> 
> Do you want to follow up on this patch? IIRC changing it to use "x30" instead of "lr" was the only blocker from my point of view (and the add_residual patch that goes on top of it was mostly fine as well)?

Sorry, I forgot about that comment when I sent the last revision.
Josh has been doing some polishing of these patches, so unless I hear
otherwise I’ll assume he’s volunteering to do these minor fixes
(thanks in advance), otherwise we just end up stepping on each other’s toes.
But I am around in principle and will if necessary help out getting it merged.

Best regards,
Reimar


More information about the ffmpeg-devel mailing list