From mboxrd@z Thu Jan 1 00:00:00 1970 Message-Id: <200008030742.IAA26688@hyperion.valhalla.net> Date: Thu, 03 Aug 2000 08:42:25 +0100 Subject: Re: Updated DMASOUND patches From: "Iain Sandoe" To: "Michel Dänzer" CC: linuxppc-dev@lists.linuxppc.org, Benjamin Herrenschmidt Mime-version: 1.0 Content-type: text/plain; charset="ISO-8859-1" Sender: owner-linuxppc-dev@lists.linuxppc.org List-Id: Thu, Aug 3, 2000, Michel Dänzer wrote: > Iain Sandoe wrote: > >> 2.4.0-test5 >> >> This is much more flaky **with identical code** there is no crashing (yet) >> - but a strong tendency for almost any activity to cut off the sound. >> opening the mixer control is a really good way of doing it. Once the panel >> is open you can restart the sound and all is hunky-dory... but... > > I've noticed that 2.4.0-test4 is much more sluggish than any 2.2 kernel I've > used. Turning on DMA for the HD (yep, didn't do it automatically) helps a bit, > but not totally. According to DMESG DMA *is* on for both SCSI & IDE (IDE doesn't enter into this really for me - all my disks are SCSI). How could I tell if dmesg were telling fibs? >> I guess I'm a little suspicious of the new wait-queue stuff (that's really >> the only major change to the sound side). > > There's a whole new kernel around the sound driver ;) Fair point... I guess we do interact with the fs stuff, and IRQs and... and... :-) Iain. ** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/