Messages in this thread | | | Date | Tue, 22 Jun 1999 11:34:21 -0500 | From | Mitchell Blank Jr <> | Subject | Re: parallel writes to the same file, 2.3.7 |
| |
Alexander Viro wrote: > Caution: side effect of that change was not pretty. truncate() used to be > serialized wrt write(). Now it isn't. We might contract some races here.
Sure sounds that way. Note that you don't really want truncate()/write() to be strictly serialized, you just want to serialize things that will change the length of the file (i.e. truncate and append) So does the locking on the inode save us here?
If not, there will be a need for a new lock protecting the length of the file. This might also make the simultaneous-writers case that people were talking about work out better.
-Mitch
- 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/
| |