On Tue, Feb 27, 2024, at 12:35, Ricardo Ribalda wrote: > On Tue, 27 Feb 2024 at 12:17, Hans Verkuil <hverkuil@xxxxxxxxx> wrote: >> >> Ricardo, >> >> First of all, note the typo in theo subject line: vcodedc -> vcodec. >> >> There is also a similar (but not identical!) patch from Arnd: >> >> https://patchwork.linuxtv.org/project/linux-media/patch/20240224121059.1806691-1-arnd@xxxxxxxxxx/ >> >> That patch and yours share the change to common/mtk_vcodec_fw_vpu.c but otherwise >> they are different, which is a bit odd. >> >> Can you take a look at Arnd's patch and see if you need to incorporate his changes >> into your patch? > > We went separate paths :), I tried to make everything const (and > therefore the remoteproc changes) and he removed the const. I had the same patch 3 originally but there was still a warning or a cast from const to non-const pointer, so I went with the simpler approach that I posted. Without that regression, your patch would be nicer, but I think the version you sent has the same regression that I ran into. > His patch looks good to me. Shall I resend the series without this > patch or you can ignore 3/3 and take 1 and 2? I also sent your patches 1 and 2 with minor differences in whitespace. Both of these are already in linux-next, so I think you don't have to resend anything here. I sent a whole lot of -Wcast-function-type-strict warning fixes to address all randconfig builds on arm64, arm and x86, all but three are now merged. Aside from the mediatek vcodec, the only missing ones are https://lore.kernel.org/all/20240213095631.454543-1-arnd@xxxxxxxxxx/ https://lore.kernel.org/all/20240213100238.456912-1-arnd@xxxxxxxxxx/ I also had one patch for s390 but none of the other architectures. I think once Hans applies the vcodec patch, I'll resend the last two patches together with a patch to enable the warning by default. Arnd
