Thanks for the quick reply. Decoding works without issues for me too. I did not change the CMA size or used s5p_mfc.mem parameter. However, according to the Marek, the default 8M should be enough for 3 instances of H264 encoders/decoders. My test was encoding a 30 seconds 720p clip, so I thought memory should not be a big issue; also it’s working w/o these patches applied, so I think CMA size is enough. Nevertheless, I will try setting them, but I would feel better if someone else would try encoding too.
Cheers, Marian > On 15 Mar. 2017, at 10:19 pm, Javier Martinez Canillas <jav...@dowhile0.org> > wrote: > > Hello Marian, > > On Wed, Mar 15, 2017 at 8:36 AM, Marian Mihailescu > <mihailesc...@gmail.com> wrote: >> Hi, >> >> After testing these patches, encoding using MFC fails when requesting >> buffers for capture (it works for output) with ENOMEM (it complains it >> cannot allocate memory on bank1). >> Did anyone else test encoding? >> > > I've not tested encoding, but I did test decoding and it works for me > with Shuah's patch to increase the CMA memory [0]. Did you test with > that one as well? > > Also you could try using the 5p_mfc.mem kernel param as explained in > the commit message of "media: s5p-mfc: Add support for probe-time > preallocated block based allocator". > > [0]: https://patchwork.kernel.org/patch/9596737/ > >> Thanks, >> Marian >> > > Best regards, > Javier
