linux-kernel - Re: [PATCH v2 00/16] Allow readpage to return a locked page

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20201015114949.GY20115@casper.infradead.org>
Date:   Thu, 15 Oct 2020 12:49:49 +0100
From:   Matthew Wilcox <willy@...radead.org>
To:     Christoph Hellwig <hch@...radead.org>
Cc:     linux-fsdevel@...r.kernel.org, linux-mm@...ck.org,
        v9fs-developer@...ts.sourceforge.net, linux-kernel@...r.kernel.org,
        linux-afs@...ts.infradead.org, ceph-devel@...r.kernel.org,
        linux-cifs@...r.kernel.org, ecryptfs@...r.kernel.org,
        linux-um@...ts.infradead.org, linux-mtd@...ts.infradead.org,
        Richard Weinberger <richard@....at>, linux-xfs@...r.kernel.org
Subject: Re: [PATCH v2 00/16] Allow readpage to return a locked page

On Thu, Oct 15, 2020 at 10:02:42AM +0100, Christoph Hellwig wrote:
> On Fri, Oct 09, 2020 at 03:30:48PM +0100, Matthew Wilcox (Oracle) wrote:
> > Ideally all filesystems would return from ->readpage with the page
> > Uptodate and Locked, but it's a bit painful to convert all the
> > asynchronous readpage implementations to synchronous.  The first 14
> > filesystems converted are already synchronous.  The last two patches
> > convert iomap to synchronous readpage.
> 
> Is it really that bad?  It seems like a lot of the remainig file systems
> use the generic mpage/buffer/nobh helpers.
> 
> But I guess this series is a good first step.

I'm just testing a patch to mpage_readpage():

+++ b/fs/mpage.c
@@ -406,11 +406,17 @@ int mpage_readpage(struct page *page, get_block_t get_block)
                .nr_pages = 1,
                .get_block = get_block,
        };
+       int err;
 
        args.bio = do_mpage_readpage(&args);
-       if (args.bio)
-               mpage_bio_submit(REQ_OP_READ, 0, args.bio);
-       return 0;
+       if (!args.bio)
+               return 0;
+       bio_set_op_attrs(args.bio, REQ_OP_READ, 0);
+       guard_bio_eod(args.bio);
+       err = submit_bio_wait(args.bio);
+       if (!err)
+               err = AOP_UPDATED_PAGE;
+       return err;
 }
 EXPORT_SYMBOL(mpage_readpage);
 

but I'm not looking forward to block_read_full_page().