On Fri, 2015-01-16 at 10:13 +0100, Bjørn Mork wrote: > Oliver Neukum <oneukum@xxxxxxx> writes: > > > On Thu, 2015-01-15 at 17:52 +0100, Bjørn Mork wrote: > >> Oliver Neukum <oneukum@xxxxxxx> writes: > >> > >> > Hi, > >> > > >> > I think I found a race between pre_reset() and the bottom half. > >> > In wdm_rxwork() WDM_RESETTING is not checked for. That means that > >> > > >> > wdm_pre_reset() cannot reliable end IO. What do you think? > >> > >> Yes, it certainly looks that way. There are too many flags meaning > >> "end IO"... > > > > So you think I should consolidate the flags rather add a check? > > If possible. At least that was my initial thought after a quick look. > But I don't know how feasible it is. I can imagine that we might need > to separate the flags we want to clear. At least the ones that can be > set at the same time, if any. At a closer look it looks to me that this is not easily doable, as the error returns and the response to poll() must be different. Regards Oliver -- 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
