Re: Linux-2.3.x: new page locking semantics must break NFS.

Trond Myklebust (trond.myklebust@fys.uio.no)
Fri, 15 Oct 1999 14:17:31 +0200 (CEST)


>>>>> " " == Stephen C Tweedie <sct@redhat.com> writes:

> Hmm? Look at the page_cache_read() code more closely:
> add_to_page_cache_unique() calls __add_to_page_cache, which
> sets up the new page to be locked.

> The locking is not quite so visible now, but it is still there.
> Indeed, the locking is now done at the lower level precisely so
> that we can be 100% sure that a page which has been newly added
> to the page cache and which does not yet have valid contents is
> guaranteed to be both locked and !uptodate, so that nobody will
> ever look at the contents before they have been initialised.

Yup. I found that out the hard way (lockups). It's been corrected
now. I still found the io-flag useful though for serializing writing
to pages and actually syncing those same pages to the server.

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/