[Prev][Next][Index][Thread]

Re: Linux 1.99q12, probably 2.0 as well, bug



>>>>> "Adam" == Adam J Thornton <adam@tucson.princeton.edu> writes:
In article <4quqc4$2ot@cnn.Princeton.EDU> adam@tucson.princeton.edu (Adam J. Thornton) writes:


    Adam> If you have sound support built as a module, and you're
    Adam> running kerneld to demand-load modules, Executor will come
    Adam> up only with "pretend" sound unless the sound module is
    Adam> already loaded.

Hmmm... that's not good.

    Adam> I suspect there's a check that asks "is sound available"?
    Adam> Instead, it might be better to try to play a sound--say the
    Adam> "machine on bong"--and see if that succeded.  If it did,
    Adam> then you have sound, if not then you don't.  That gives
    Adam> kerneld a chance to load the sound module into kernel space
    Adam> if it can.

But we don't do that.  We open /dev/dsp and then make ioctls and if
any of them fail we decide you don't have sound.  From my perspective
it looks like there's a problem with demand-loading sound.

I will not be able to take a look at this until after E2 is out, but
anyone who is real curious could probably use the system call tracing
program (strace?) and see where the failure is happening.

    Adam> Adam -- adam@phoenix.princeton.edu | Viva HEGGA! | Save the
    Adam> choad! | 64,928 | Fnord "Double integral is also the shape
    Adam> of lovers curled asleep":Pynchon | Linux Thanks for letting
    Adam> me rearrange the chemicals in your head. | Team OS/2 You can
    Adam> have my PGP passphrase when you pry it from my cold, dead
    Adam> brain.

--Cliff
ctm@ardi.com


Follow-Ups: References: