From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from alogconduit1ah.ccr.net (ccr@alogconduit1am.ccr.net [208.130.159.13]) by kvack.org (8.8.7/8.8.7) with ESMTP id WAA14118 for ; Mon, 10 May 1999 22:42:26 -0400 Subject: Re: [PATCH] dirty pages in memory & co. References: <14135.13698.659905.454361@dukat.scot.redhat.com> From: ebiederm+eric@ccr.net (Eric W. Biederman) Date: 10 May 1999 19:30:00 -0500 In-Reply-To: "Stephen C. Tweedie"'s message of "Mon, 10 May 1999 20:37:38 +0100 (BST)" Message-ID: Sender: owner-linux-mm@kvack.org To: "Stephen C. Tweedie" Cc: linux-mm@kvack.org List-ID: >>>>> "ST" == Stephen C Tweedie writes: ST> Hi, ST> On 07 May 1999 09:56:00 -0500, ebiederm+eric@ccr.net (Eric W. Biederman) ST> said: >> It looks like I need 2 variations on generic_file_write at the >> moment. >> 1) for network filesystems that can get away without filling >> the page on a partial write. >> 2) for block based filesystems that must fill the page on a >> partial write because they can't write arbitrary chunks of >> data. ST> I'd be very worried by (1): sounds like a partial write followed by a ST> read of the full page could show up garbage in the page cache if you do ST> this. If NFS skips the page clearing for partial writes, how does it ST> avoid returning garbage later? Actually (1) is current behaviour. I really don't like it but I can see how it can potentially improve performance. Partial writes are handled by not setting PG_uptodate. Reads are handled by always flushing the per page dirty data before reading. I don't especially like it but it's what we have now. Eric -- To unsubscribe, send a message with 'unsubscribe linux-mm my@address' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://humbolt.geo.uu.nl/Linux-MM/