linux-kernel - Re: [PATCH v5 8/9] iov_iter, block: Make bio structs pin pages rather than ref'ing if appropriate

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: <Y8AWY991ilrO5Yco@infradead.org>
Date:   Thu, 12 Jan 2023 06:17:07 -0800
From:   Christoph Hellwig <hch@...radead.org>
To:     David Howells <dhowells@...hat.com>
Cc:     Christoph Hellwig <hch@...radead.org>,
        Al Viro <viro@...iv.linux.org.uk>,
        Jens Axboe <axboe@...nel.dk>, Jan Kara <jack@...e.cz>,
        Christoph Hellwig <hch@....de>,
        Matthew Wilcox <willy@...radead.org>,
        Logan Gunthorpe <logang@...tatee.com>,
        linux-block@...r.kernel.org, Jeff Layton <jlayton@...nel.org>,
        linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v5 8/9] iov_iter, block: Make bio structs pin pages
 rather than ref'ing if appropriate

On Thu, Jan 12, 2023 at 06:09:16AM -0800, Christoph Hellwig wrote:
> On Thu, Jan 12, 2023 at 10:28:41AM +0000, David Howells wrote:
> > Christoph Hellwig <hch@...radead.org> wrote:
> > 
> > > 	if (cleanup_mode & FOLL_GET) {
> > > 		WARN_ON_ONCE(bio_test_flag(bio, BIO_PAGE_PINNED));
> > > 		bio_set_flag(bio, BIO_PAGE_REFFED);
> > > 	}
> > > 	if (cleanup_mode & FOLL_PIN) {
> > > 		WARN_ON_ONCE(bio_test_flag(bio, BIO_PAGE_REFFED));
> > > 		bio_set_flag(bio, BIO_PAGE_PINNED);
> > > 	}
> > 
> > That won't necessarily work as you might get back cleanup_mode == 0, in which
> > case both flags are cleared - and neither warning will trip on the next
> > addition.
> 
> Well, it will work for the intended use case even with
> cleanup_mode == 0, we just won't get the debug check.  Or am I missing
> something fundamental?

In fact looking at the code we can debug check that case too by doing:

	if (cleanup_mode & FOLL_GET) {
 		WARN_ON_ONCE(bio_test_flag(bio, BIO_PAGE_PINNED));
 		bio_set_flag(bio, BIO_PAGE_REFFED);
 	} else if (cleanup_mode & FOLL_PIN) {
 		WARN_ON_ONCE(bio_test_flag(bio, BIO_PAGE_REFFED));
 		bio_set_flag(bio, BIO_PAGE_PINNED);
 	} else {
		WARN_ON_ONCE(bio_test_flag(bio, BIO_PAGE_PINNED) ||
			     bio_test_flag(bio, BIO_PAGE_REFFED));
	}

But given that all calls for the same iter type return the same
cleanup_mode by defintion I'm not even sure we need any of this
debug checking, and might as well just do:

	if (cleanup_mode & FOLL_GET)
 		bio_set_flag(bio, BIO_PAGE_REFFED);
 	else if (cleanup_mode & FOLL_PIN)
 		bio_set_flag(bio, BIO_PAGE_PINNED);