Oliver Neukum <oliver@xxxxxxxxxx> writes: > Am Donnerstag, 2. Februar 2012, 09:48:06 schrieb Bjørn Mork: >> Hmm, looking at the code, and realizing that the reason we're taking >> wlock here only is that the driver used to take the combined read/write >> lock, I don't think it's necessary at all. desc->count is only modified >> by wdm_open() and wdm_release(), both holding the global wdm_mutex at >> the time. The only place desc->count is tested without holding the >> wdm_mutex lock is in recover_from_urb_loss() which is called from >> wdm_resume() and wdm_post_reset(). I don't think that is a problem >> though, as both require holding the device lock which will protect them >> from competing against each other. >> >> So in short: I don't think the wlock offers any extra protection in >> wdm_open and wdm_release and I would like to just drop it if you too >> think that's OK. > > What is to guard pre/post_reset against open() ? Currently nothing, I guess. The patch adding WDM_RESETTING should have included a test for that flag in open(). Will fix that and resend unless you have other suggestions. Bjørn -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
