linux-kernel - Re: [dm-devel] [PATCH v3] block: trace completion of all bios.

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] [day] [month] [year] [list]

Message-ID: <87tw6e70dn.fsf@notabene.neil.brown.name>
Date:   Tue, 28 Mar 2017 10:42:12 +1100
From:   NeilBrown <neilb@...e.com>
To:     Christoph Hellwig <hch@...radead.org>
Cc:     Jens Axboe <axboe@...nel.dk>, linux-block@...r.kernel.org,
        linux-raid@...r.kernel.org,
        "Martin K . Petersen" <martin.petersen@...cle.com>,
        Mike Snitzer <snitzer@...hat.com>,
        Ming Lei <tom.leiming@...il.com>, linux-kernel@...r.kernel.org,
        Christoph Hellwig <hch@...radead.org>, dm-devel@...hat.com,
        Shaohua Li <shli@...nel.org>, Alasdair Kergon <agk@...hat.com>
Subject: Re: [dm-devel] [PATCH v3] block: trace completion of all bios.

On Mon, Mar 27 2017, Christoph Hellwig wrote:

> On Mon, Mar 27, 2017 at 08:49:57PM +1100, NeilBrown wrote:
>> On Mon, Mar 27 2017, Christoph Hellwig wrote:
>> 
>> > I don't really like the flag at all.  I'd much prefer a __bio_endio
>> > with a 'bool trace' flag.  Also please remove the manual tracing in
>> > dm.ċ.  Once that is done I suspect we can also remove the
>> > block_bio_complete export.
>> 
>> Can you say why you don't like it?
>
> It uses up a precious bit in the bio for something that should be state
> that can be determined in the caller at compile time.

I've already demonstrated that the bit is not "precious" at all.  I have
shown how I could easily give you 20 unused flag bits without increasing
the size of struct bio.
Yes, the state could be determined in the caller at compiler time.  That
would require developers to make the correct choice between two very
similar interfaces, where the consequences of an correct choice are not
immediately obvious.
I think that spending one bit (out of 20) to relieve developers of the
burden of choice (and to spare as all of the consequences of wrong
choice) is a price worth paying.

NeilBrown

Download attachment "signature.asc" of type "application/pgp-signature" (833 bytes)