From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pavel Machek Subject: Re: [RFC][PATCH -mm 3/3] PM: Disable _request_firmware before hibernation/suspend Date: Mon, 28 May 2007 13:15:54 +0200 Message-ID: <20070528111554.GF18807@elf.ucw.cz> References: <200705272229.21263.rjw@sisk.pl> <20070527220412.GC22687@srcf.ucam.org> <1180304166.3131.66.camel@lov.localdomain> <200705280943.54312.rjw@sisk.pl> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <200705280943.54312.rjw@sisk.pl> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: linux-pm-bounces@lists.linux-foundation.org Errors-To: linux-pm-bounces@lists.linux-foundation.org To: "Rafael J. Wysocki" Cc: Matthew Garrett , Kay Sievers , LKML , pm list List-Id: linux-pm@vger.kernel.org Hi! > > > > What exactly is the problem we see here? The timeout of the firmware loader? > > > > What goes wrong with frozen userspace, usually there is only a netlink > > > > message sent from the kernel, which should be received and handled > > > > just fine when userspace is running again. > > > > > > Driver calls request_firmware in the resume method. The userspace helper > > > can't be run because it's been frozen, so the firmware never gets loaded > > > and the call times out. The driver then fails to resume. While all this > > > is happening, the rest of the kernel is blocking on that resume method. > > > The firmware can be loaded once userspace has been started again, but by > > > that time the driver has given up. > > > > Seems, that's just the broken synchronous firmware loading interface > > with the useless timeout handling. The nowait version of the same loader > > doesn't time out, and should not have that problem. The sync version > > should be removed from the kernel, it just causes all sorts of problems > > since it exists. > > > > Userspace should handle the async request just fine when it comes back > > running, regardless of the time it was submitted. > > Okay, so the solution is to convert the drivers to use > request_firmware_nowait() instead of request_firmware() in their .resume() > routines. You'll just get deadlock at different level (and more rare). Imagine disk with its firmware on NFS and NFS with its firmware on disk. (Or maybe firmware loader doing find /, including both disk and NFS). Just don't call request_firmware_* from .resume(). -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html