[FFmpeg-devel] r9017 breaks WMA decoding on Intel Macs
Fri Jun 8 23:27:48 CEST 2007
Guillaume POIRIER wrote:
> On 6/3/07, Michael Niedermayer <michaelni at gmx.at> wrote:
>> On Sun, Jun 03, 2007 at 09:37:37AM -0500, Graham Booker wrote:
>>> Arg!! Forgot about that. Well, I have another idea now, although it
>>> is a bit more hack like, but it seems to work.
>>> I noticed that the linux gas (newer gas really), upon seeing a (value
>>> operator "missing value"), assumes the "missing value" evaluates to
>>> 0. So, 123+(..) is changed to 123+0(...). The Mactel gas (older
>>> one) seems to assume that the evaluation of the operator is 0 (not
>>> the whole expression btw) meaning it evaluates to 0(..). So, what
>>> about offest+1*%number. The newer gas assumes offset+1*0(...) in the
>>> case of no offset in the %number, and the older gas assumes offset+0
>>> (...) in the same case. For both, if the %number contains an offset,
>>> then these evaluate to offset1+1*offset2(%register).
>>> More ugly, yes, but from what I can tell, this seems to work everywhere.
>> patch ok, if it does work ...
> Works here, patch applied.
Shouldn't all other usage of num+(%reg) should be converted to this syntax ?
Otherwise this will likely break depending of the phase of moon (or
version of gcc, optimisation flags, ...).
More information about the ffmpeg-devel