[FFmpeg-devel] [PATCH] Fix compilation on ARM with android gcc 4.7

Reimar Döffinger Reimar.Doeffinger at gmx.de
Sat Mar 16 19:50:52 CET 2013


On Sat, Mar 16, 2013 at 07:39:25AM -0700, Ronald S. Bultje wrote:
> Hi,
> 
> On Sat, Mar 16, 2013 at 5:37 AM, Reimar Döffinger
> <Reimar.Doeffinger at gmx.de> wrote:
> > With the current code it fails due to running out
> > of registers.
> > So code the store offsets manually into the assembler
> > instead.
> > Note: untested since I did not actually get anything
> > to run yet.
> > ---
> >  libavcodec/arm/dca.h |   74 ++++++++++++++++++++++++--------------------------
> >  1 file changed, 36 insertions(+), 38 deletions(-)
> 
> Copying the file into a running Android emulator and then "adb -s emu
> ffmpeg -i dcafatefile -f pcm_s16le output" (and then testing that the
> output is, I guess, float-equal to some reference generated on the
> host) may work, if you can't run it on the actual phone.

First, I am working on a build with NEON, which last time I checked
the emulator can't run.
Next, the DCA decoder is not bit-exact, so this checking would be
a bit of a pain.
And also running it on the phone is actually easier than running on
the emulator.
The bigger issue is that gdbserver seems unusable because it causes
every attempt by ffmpeg to print something to raise a SIGTTOU.
I'd prefer it by far for someone else to test it, but if not I'll
probably try to run the tests on a BeagleBoard - which should be
fine for this, but the lack of actually working GLES drivers make
it useless for anything else I want to do. What a mess.

Reimar


More information about the ffmpeg-devel mailing list