Oliver Neukum wrote: > > Hi, > > the attached patch fixes an SMP race in open and adds interruptible locking > to read and write. > > The locking is a new feature, but IMHO very sensible. Allowing only exclusive > open but concurrent reads and writes is not in the same spirit. > In the original code these functions were at least SMP safe, which the patch > restores and it adds some further locking as some drivers are buggy and the > lost latent functionality is very small. (If you absolutly must do so drop > the lock yourself.) > > The behaviour of open is _not_ changed. Regardless of whether you want > exclusive open or not, the situation as is is a bug. > > Please tell me whether you have fundamental problems with this patch or find > bugs. > > Regards > Oliver I still think it is very bad practice to fix what is essentially a driver bug, by adding a restrictive hack to a higher layer. How about just sending an email to the authors of the respective drivers letting them know about the bug. The bugs will be fixed in no time, and everybody will be happy. -justin
