Messages in this thread | | | Date | Wed, 2 Jun 1999 01:54:57 -0400 (EDT) | From | Alexander Viro <> | Subject | [PATCH][CFT] procfs cleanup - phase 1. |
| |
Folks, I'm putting the first version of procfs cleanup on ftp.math.psu.edu/pub/viro/proc-patch-5.gz. Patch is against 2.3.4.
Contents: * /proc/net/* stuff mostly made dynamic. No more explicit struct proc_dir_entry in networking code. * Closed an oopsable bug in drivers/net/bmac.c - it forgot to release the /proc/net/bmac entry. * system-wide stuff in fs/proc/array.c made dynamic too and moved into fs/proc/proc_misc.c * fs/proc/net.c is gone. RIP. * proc_net_inode_operations is gone. It was an 100% equivalent of proc_file_inode_operations, which is default. * generic cleanup in fs/proc/* - things that should be static made static, etc.
TODO: * pseudo-revoke() for /proc/<pid>/* stuff. That will kill most of permission checks. It will take an addition of new field (cyclic list head) to task_struct and switching to dynamic inumbers for procfs. Then we will be able to invalidate those inodes on exit(), so that even if a new process would get the same PID old inodes would stay dissociated from all processes. * Extermination of scanning the whole struct file list upon removal of procfs entry. Mostly the same technics as above. * Cleanup of situation with symlinks.
Patched kernel compiles, but the patch is almost untested. You are warned. It is early alpha. With all usual implications. Help with testing is welcome, indeed, just don't forget to mount a scratch monkey ;-) Cheers, Al
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |