Date: Thu, 19 Jun 2003 12:08:38 -0400 (EDT) From: Andrew Gallatin <gallatin@cs.duke.edu> To: Josh Tolbert <hemi@just.puresimplicity.net> Cc: freebsd-alpha@freebsd.org Subject: Re: [hemi@just.puresimplicity.net: Re: Unusual sound problems] Message-ID: <16113.57350.330775.726477@grasshopper.cs.duke.edu> In-Reply-To: <20030619155159.GB41810@just.puresimplicity.net> References: <20030619155159.GB41810@just.puresimplicity.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Josh Tolbert writes: > Hi Andrew, > > Looks like ATA things are interrupting on 5 along with the sound card. Should I try putting the sound card in the other ISA slot? Would that make any difference? > > atapci0: interrupting at CIA irq 5 > ata1: interrupting at ISA irq 15 > sbc0: interrupting at ISA irq 5 > If possible, yes. I think the problem is that the audio interrupt handler is MPSAFE (meaning that it does its own locking), while the ata driver is not MPSAFE (meaning that it shares the "Giant" lock with a lot of the rest of the kernel). Since there is a non-MPSAFE driver sharing an IRQ, the MPSAFE driver needs to wait for the Giant kernel lock. Since a lot of the kernel uses the Giant lock, that could translate to very high interrupt latency. If you're not using any IDE disks, CDROMs, etc, the best move would be to just remove the ata driver from your kernel. Drew PS: Please try to break your lines at 72 cols or so.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?16113.57350.330775.726477>