Date: Mon, 20 Jan 2003 15:29:11 -0500 (EST) From: Kenneth Culver <culverk@yumyumyum.org> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: Nate Lawson <nate@root.org>, Trish Lynch <trish@bsdunix.net>, <freebsd-current@FreeBSD.ORG> Subject: Re: FreeBSD panic with umass Message-ID: <20030120152601.E14910-100000@alpha.yumyumyum.org> In-Reply-To: <200301202009.h0KK9120088439@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 20 Jan 2003, Matthew Dillon wrote: > :Hmm, good stuff, but shouldn't something be committed anyway? I mean if it > :causes a panic just by plugging in the device that's totally unacceptable. > :I'll provide a backtrace of the crash on my computer tomorrow I suppose (I > :won't be home until then) and let people know if that's what's causing my > :crash. > : > :Ken > > Yes, but it isn't quite that easy. I did fix the incorrect sense > code issue with UMASS, but that's only one of the potentially many > problems that could occur. It would probably also help (give us more > deterministic panics / errors) if the read_capacity structure were > at least bzero'd by CAM/SCSI. > > But half the problem is with the USB devices themselves. The device > firmware for many of these devices, especially the Sony, was written > by idiots. The entire USB specification was written by idiots IMHO. > For example, the Sony will respond with garbage, and no error whatsoever, > to just about any page inquiry command you send it. The Sony doesn't > even return reasonable data for the code pages that the USB spec requires! > > Ultimately this means that the best we can do is to try to ensure that > garbage data doesn't result in a system panic. That's a fairly tall > order for such a low level subsystem. > Yeah, I suppose I understand your point. I could probably at least take a look at doing some modifications here and there to avoid crashes at least. Even with everything you said, it's still the case that we shouldn't allow just plugging in a device to cause a panic. Maybe we should just not try to attach a device we don't know about? Meaning no wildcard matches in the USB code for the probe/attach. That would solve the problem since we wouldn't be attaching anything we don't know about. Then people could submit pr's or whatever with the device ID's for their devices, and the quirks that device needs in order to work. Ken To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030120152601.E14910-100000>