On Mon, 7 Mar 2005 12:17:29 -0500 (EST), Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > On Mon, 7 Mar 2005, Dmitry Torokhov wrote: > > > > > This locking should be moved to the driver core, if possible. > > > > > > My thought exactly! Would it suffice to have the PM suspend and resume > > > routines get a read lock on the bus subsystem's rwsem? > > > > > > > Ugh, wasn't it just recently killed as it was prevnting discovery of > > new child devices upon resume? > > I don't know. Can you send a pointer to the messages in the email > archives? > > But drivers shouldn't try to discover new devices during resume; they > should wait until the system is fully up and functional. Or if they do > insist on discovery during resume, they should wait until the system is up > before trying to register the new devices. > > The situation is no different from probe(), because the rwsem is locked at > that time too. > I am sorry, I muxed up subsystem's rwsem and dpm_sem. Still issue is prety much the same. http://marc.theaimsgroup.com/?l=linux-kernel&m=109755205112315&w=2 http://linux.bkbits.net:8080/linux-2.5/cset@4186a502_-ZT7ruVgfy3hZZHYSc6UA?nav=index.html|src/.|src/drivers|src/drivers/base|src/drivers/base/power|related/drivers/base/power/main.c http://marc.theaimsgroup.com/?l=linux-kernel&m=109944258827464&w=2 -- Dmitry
