Bug#253568: Bug #253568 (16-bit sound broken on HPPA, and PPC with -ansi)
On Tue, Aug 10, 2004 at 11:21:54AM +0900, GOTO Masanori wrote:
> At Mon, 9 Aug 2004 00:59:16 +0100, Stuart Brady wrote:
[snip]
> > The problem is that <linux/soundcard.h> doesn't test for __hppa__ and
> > __powerpc__ when defining AFMT_S16_NE, so it ends up being defined as
> > AFMT_S16_LE instead of AFMT_S16_BE. Result: samples are effectively
> > byte-swapped.
[snip]
> We're now base freeze period, so I would like to keep the minimum
> modification for lkh and libc6. I would like to know what package is
> this bug affected?
The problem is in lkh. I can't say how many packages are affected.
It's nowhere near as bad as I had feared, though. Indeed, the bulk of
the audio-on-big-endian problem appears to be with individual packages.
Of the packages I've looked at so far:
* mikmod, timidity and madplay are broken, due to this bug.
* esound, freecraft, rplay, nas and s3mod are broken, regardless of
this bug. (They won't work on sparc, powerpc or m68k, either.)
* arts, koules, libao, soundtracker, sox, xmms and xmp are unaffected.
My best guess is that there are around 400-500 packages that use
/dev/dsp directly, and perhaps 50-100 packages affected by the bug.
I could be miles out - I don't really know.
> BTW, "possible patch" is fine for me. Is this already tested?
WFM, but I will ask for help with testing on the parisc-linux list.
> And do you have intension to clean up upstream kernel headers? It
> seems fine for me, so please get submit to upstream and get review.
I'll be pushing the larger patch for upstream. I'll see what they think
about the workaround -- I can't see there being any problem, though.
Cheers,
--
Stuart Brady
Reply to: