From: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
To: Jens Axboe <axboe@suse.de>
Cc: Linus Torvalds <torvalds@transmeta.com>,
Linux Kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] 2.5 ide 48-bit usage
Date: Wed, 7 May 2003 21:58:12 +0200 (MET DST) [thread overview]
Message-ID: <Pine.SOL.4.30.0305072119530.27561-100000@mion.elka.pw.edu.pl> (raw)
In-Reply-To: <20030507175033.GR823@suse.de>
On Wed, 7 May 2003, Jens Axboe wrote:
> On Wed, May 07 2003, Linus Torvalds wrote:
> >
> > On Wed, 7 May 2003, Jens Axboe wrote:
> > > >
> > > > And testing. In particular, you might want to test whether a device
> > > > properly supports 48-bit addressing, either from the kernel or from user
> > > > programs.
> > >
> > > For that, a forced 48-bit hwif->addressing inherited by drives will
> > > suffice. And I agree, we should have that.
> >
> > No no no.
> >
> > You definitely do NOT want to set "hwif->addressing" to 1 before you've
> > tested whether it even _works_.
>
> Well duh, of course not. Whether a given request is executed in 48-bit
> or not is a check that _includes_ drive capabilities too of course.
Yeah, we test drive capabilities properly in idedisk_setup(),
but Linus is right speaking about _hwif_ capabilities.
Jens you your patch sets hwif->rqsize to 65535 in setup-pci.c for all
PCI hwifs which is simply wrong as not all of them supports LBA48.
You should check for hwif->addressing and if true set rqsize to 65536
(not 65535) and not in IDE PCI code but in ide_init_queue() in ide-probe.c.
I also think that max request size should be printed for all drives,
not only 48-bit capable.
> > Imagine something like "hdparm" - other things are already in progress,
> > the system is up, and IDE commands are potentially executing concurrently.
> > What something like that wants to do is to send one request out to check
> > whether 48-bit addressing works, but it absolutely does NOT want to set
> > some interface-global flag that affects other commands.
>
> Then it just puts a taskfile request on the request queue and lets it
> reach the drive, nicely syncronized with the other requests. There's no
> need to toggle any special bits for that.
Yes, but patch subtly breakes taskfile :-).
Taskfile ioctl uses do_rw_taskfile() or flagged_taskfile().
Patch replaces drive->adrressing checks by task->addressing,
but ide_taskfile_ioctl() doesn't know about it so task->addressing
will be always equal 0.
You should add checking for 48-bit commands and setting task->addressing
to 1 if neccessary to ide_taskfile_ioctl().
Also changes for pdc202xx_old.c are wrong, we should check for
task->addressing not rq_lba48(rq) as taskfile requests also use this
codepath.
Patch also misses updates for many uses of drive->addressing
(in ide.c, ide-io.c, icside.c, ide-tcq.c and even in ide-taskfile.c).
> > Only after it has verified that 48-bit addressing does work should it set
> > the global flag.
>
> Sounds fine.
Jens, I like the general idea of the patch, but it needs some more work.
Linus, please don't apply for now.
--
Bartlomiej
> --
> Jens Axboe
next prev parent reply other threads:[~2003-05-07 19:46 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-05-07 8:49 [PATCH] 2.5 ide 48-bit usage Jens Axboe
2003-05-07 16:28 ` Linus Torvalds
2003-05-07 16:46 ` Jens Axboe
2003-05-07 17:15 ` Linus Torvalds
2003-05-07 17:33 ` Jens Axboe
2003-05-07 17:42 ` Linus Torvalds
2003-05-07 17:50 ` Jens Axboe
2003-05-07 19:58 ` Bartlomiej Zolnierkiewicz [this message]
2003-05-07 20:19 ` Jens Axboe
2003-05-08 7:56 ` Jens Axboe
2003-05-08 11:01 ` Alan Cox
2003-05-08 12:01 ` Jens Axboe
2003-05-12 21:41 ` Mike Fedyk
2003-05-13 6:44 ` Jens Axboe
2003-05-08 11:34 ` Bartlomiej Zolnierkiewicz
2003-05-08 11:59 ` Jens Axboe
2003-05-08 12:20 ` Bartlomiej Zolnierkiewicz
2003-05-08 12:26 ` Jens Axboe
2003-05-08 12:36 ` Jens Axboe
2003-05-08 13:16 ` Bartlomiej Zolnierkiewicz
2003-05-08 13:23 ` Jens Axboe
2003-05-08 13:35 ` Bartlomiej Zolnierkiewicz
2003-05-08 13:37 ` Jens Axboe
2003-05-08 14:47 ` Jens Axboe
2003-05-08 14:51 ` Jens Axboe
2003-05-08 14:46 ` Alan Cox
2003-05-08 15:49 ` Bartlomiej Zolnierkiewicz
2003-05-08 16:16 ` Jens Axboe
2003-05-08 16:27 ` Linus Torvalds
2003-05-08 16:34 ` Jens Axboe
2003-05-08 16:59 ` Bartlomiej Zolnierkiewicz
2003-05-09 7:40 ` Jens Axboe
2003-05-08 22:06 ` Alan Cox
2003-05-09 7:06 ` Jens Axboe
2003-05-09 8:28 ` [PATCH][RFC] Sanitize hwif/drive addressing (was Re: [PATCH] 2.5 ide 48-bit usage) Jens Axboe
2003-05-09 11:07 ` Bartlomiej Zolnierkiewicz
2003-05-09 12:03 ` Jens Axboe
2003-05-07 21:45 ` [PATCH] 2.5 ide 48-bit usage Henning P. Schmiedehausen
2003-05-07 22:03 ` Alan Cox
2003-05-07 22:55 ` H. Peter Anvin
2003-05-07 18:29 ` Alan Cox
2003-05-07 19:30 ` Jens Axboe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=Pine.SOL.4.30.0305072119530.27561-100000@mion.elka.pw.edu.pl \
--to=b.zolnierkiewicz@elka.pw.edu.pl \
--cc=axboe@suse.de \
--cc=linux-kernel@vger.kernel.org \
--cc=torvalds@transmeta.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).