On Wednesday 01 November 2000 00:31, Alan Cox wrote: > > 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 sem still belongs in drivers. To start with they have to lock versus > ioctl and other events anyway. Most drivers are safe here (I did the radio > cards a long time ago) FWIW I tend to agree with the you and the others on this position. Coarse grain locking at too high a level can often lead to many obscure problems down the line, particularly in deadlock situations. Locking should be done at the lowest level possible to keep it simple and reliable. In this case, that means in the drivers themselves. Also, why introduce errors in higher performance boards (ones that have multiple DMA channels and run on systems that have multiple cpus/threads) if it can be avoided? I'm not sure how many drivers there are available, but it seems that the majority of the video cards use the btXXX version. Fixing that first would appear to handle a large percentage of the systems in use. Regards, mike (travis@xxxxxxx)
