Hello,
On 14-May-00 Randy Dunlap wrote:
> Hi Nemosoft,
>
> You are the only person who I recall having made patches to the
> audio driver recently (for Philips microphones ?).
>
> Is Tom complaining about your patch here?
> If so, does this still work for your devices?
No, it will break them.
> Any comments before I apply it (if you are quick enough)?
>
> Thanks,
> Randy
>
>
> Thomas Sailer wrote:
>>
>> This patch against the current backport (please also apply to 2.3)
>> fixes a problem with "short writes" and mostly undoes earlier changes
>> to find_format which are conceptually wrong
Can you be more specific on what's wrong with find_format? Maybe I
overlooked, but it seems the only place to insert the samplerate.
> and broke just about every
>> audio application with the Dallas chip (likely also with the philips
>> chip, > but I didn't try)
No, it makes the Philips cams work :(
>>
>> Tom
>>
>> --- audio.c.orig Sat May 13 14:45:11 2000
>> +++ audio.c Sat May 13 17:30:08 2000
>> @@ -3,7 +3,7 @@
>> /*
>> * audio.c -- USB Audio Class driver
>> *
>> - * Copyright (C) 1999
>> + * Copyright (C) 1999, 2000
>> * Alan Cox ([EMAIL PROTECTED])
>> * Thomas Sailer ([EMAIL PROTECTED])
>> *
>> @@ -59,6 +59,19 @@
>> * 1999-12-20: Fix bad bug in conversion to per interface probing.
>> * disconnect was called multiple times for the audio
>> device,
>> * leading to a premature freeing of the audio structures
>> + * 2000-05-13: I don't remember who changed the find_format routine,
>> + * but the change was completely broken for the Dallas
>> + * chip. Anyway taking sampling rate into account in
>> find_format
>> + * is bad and should not be done unless there are devices
>> with
>> + * completely broken audio descriptors. Unless someone
>> shows
>> + * me such a descriptor, I will not allow find_format to
>> + * take the sampling rate into account.
Stop! You _have_ to take sampling rate into account. Matching only on the
samplesize (8 vs. 16) bit and stereo/mono will not work; the routine as it
was would simply take the first alternate interface that would match these 2
parameters, and disregard samplerate altogether.
Anyway, any tool I tried would report it couldn't set the samplerate to the
desired one, always reporting the first match (8000 Hz).
>> + * Also, the former find_format made:
>> + * - mpg123 play mono instead of stereo
>> + * - sox completely fail for wav's with sample rates <
>> 44.1kHz
>> + * for the Dallas chip.
I'm not sure what this 'Dallas' chip is, but to me it sounds like it
seriously broken, or its logic works different than I anticipated. To the
mono/stereo problem: maybe I made a mistake, dunno.
Anyway, if you have any suggestion as how to _set_ the samplerate properly,
I'm open for suggestions. Turning things back after being in the kernel
for more than a month is a bit weird...
Attached are the full descriptor dumps of both Philips cams.
- Nemosoft
-----------------------------------------------------------------------------
Try SorceryNet! One of the best IRC-networks around! irc.sorcery.net:9000
URL: never IRC: nemosoft IscaBBS (bbs.isca.uiowa.edu): Nemosoft
>> Never mind the daylight <<
pcvc680-dump
pca645-dump
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
