Messages in this thread | | | Subject | Re: smbfs: page write patch, dropping mounts, tunneling in 2.2.1 | From | Trond Myklebust <> | Date | 18 Feb 1999 00:25:45 +0100 |
| |
vherva@turing.netspan.fi writes:
> I'm not aware of what kind of symtoms the bug shoud have caused, but it seems > that it fixed at least one on my 2.2.1-ac5. Before, the mounts would drop when > NT peer would close the socket. After this, any access to the mount -- ls, cd,
The problem in stock 2.2.1 is that a page that is written to smbfs, will clear the page lock and serve the wait-queue twice (once in smb_writepage_sync, then again in generic_file_write). If some other process has grabbed the page lock in the meanwhile, then the page lock will vanish from underneath it, and chaos and anarchy ensue...
The patch should not be applied to 2.2.1-ac5 because I reverted to the old behaviour for generic_file_write in which the smb_writepage routine takes on the responsability both for setting and clearing the page lock. This old behaviour is wanted for the case where you have asynchronous writes that temporarily relinquish the page lock (i.e. NFS) since you don't want to copy new data over a dirty page before you have flushed out all the asynchronous write requests not owned by yourself. (Actually I'm a bit worried about that in the stock 2.2.1 code. It checks only after the data has been copied to the page).
All this to say, that my patch should not do anything for your problem with 2.2.1-ac5 other than to make it worse. Sorry 8-(...
Cheers, Trond
- 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/
| |