Messages in this thread | | | From | Jonathan Corbet <> | Subject | [PATCH/RFC] Fasync BKL removal | Date | Sat, 7 Feb 2009 13:06:53 -0700 |
| |
With any luck at all, this is my final attempt at removing the BKL from the fasync() path and providing proper protection for struct file->f_flags. This week's episode includes these patches:
1) Rename f_ep_lock to f_lock and move it out of CONFIG_EPOLL. Epoll remains the only user, but the lock is now available for others.
2) Use f_lock to protect f_flags. This provides better protection to the flags than we've had before and allows the removal of the BKL in places where its only purpose was protecting accesses to this field.
3) Move FASYNC bit handling into ->fasync(): this is how changes to that bit and calls to f_op->fasync() are kept atomic in the absence of the BKL. Almost every FASYNC implementation uses fasync_helper(), so the actual bit manipulation is done there. The one exception (for sockets) has been fixed up. At this point, there is no more BKL in these paths.
4) Rationalize fasync_helper() return value handling. Some drivers mapped positive return values from fasync_helper() onto zero, most others did not bother. This optional cleanup patch moves that mapping into the VFS code, making things consistent and enabling the removal of some 50 lines of code.
Parts 3 and 4 are essentially the same as the last time around.
Once again, this code works for me and, if there's any regressions here, the LTP test suite cannot find them.
Thanks to all who have looked at this interminable stream of patches...
jon
| |