lkml.org 
[lkml]   [2008]   [Nov]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [git patches] libata fixes
Greg Freemyer wrote:
> On Sun, Nov 2, 2008 at 8:21 AM, Jeff Garzik <jeff@garzik.org> wrote:
>> Greg Freemyer wrote:
>>> On Fri, Oct 31, 2008 at 1:49 AM, Jeff Garzik <jeff@garzik.org> wrote:
>>>> Notes:
>>>>
>>>> 1) 1.5TB drive fix from Roland
>>>>
>>>> 2) Tejun's sata_via brings a non-working via configuration thanks to a
>>>> new facility also being used in recent (ICH9/10) ata_piix.
>>>>
>>>> Change is longer than one might like, but it accurately describes the
>>>> hardware now, the previous stuff wasn't working, and the newly added
>>>> support code shouldn't touch non-broken VIA controllers.
>>>>
>>>>
>>>>
>>>> Please pull from 'upstream-linus' branch of
>>>> master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev.git
>>>> upstream-linus
>>>>
>>>> to receive the following updates:
>>>>
>>>> drivers/ata/ata_piix.c | 1 -
>>>> drivers/ata/libata-core.c | 11 +++-
>>>> drivers/ata/sata_via.c | 155
>>>> +++++++++++++++++++++++++++++++++++++++++----
>>>> include/linux/libata.h | 1 +
>>>> 4 files changed, 152 insertions(+), 16 deletions(-)
>>>>
>>>> Jens Axboe (1):
>>>> libata: add whitelist for devices with known good pata-sata bridges
>>>>
>>>> Randy Dunlap (1):
>>>> ATA: remove excess kernel-doc notation
>>>>
>>>> Roland Dreier (1):
>>>> libata: Avoid overflow in ata_tf_to_lba48() when tf->hba_lbal > 127
>>>>
>>>> Tejun Heo (1):
>>>> sata_via: fix support for 5287
>>> <snip>
>>>
>>>> diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
>>>> index 2ff633c..82af701 100644
>>>> --- a/drivers/ata/libata-core.c
>>>> +++ b/drivers/ata/libata-core.c
>>>> @@ -1268,7 +1268,7 @@ u64 ata_tf_to_lba48(const struct ata_taskfile *tf)
>>>>
>>>> sectors |= ((u64)(tf->hob_lbah & 0xff)) << 40;
>>>> sectors |= ((u64)(tf->hob_lbam & 0xff)) << 32;
>>>> - sectors |= (tf->hob_lbal & 0xff) << 24;
>>>> + sectors |= ((u64)(tf->hob_lbal & 0xff)) << 24;
>>>> sectors |= (tf->lbah & 0xff) << 16;
>>>> sectors |= (tf->lbam & 0xff) << 8;
>>>> sectors |= (tf->lbal & 0xff);
>>> <snip>
>>>
>>> That looks really serious.
>>>
>>> What happens with all previous / stable kernels when connecting up a
>>> 1.5TB drive and the user tries to use the last portion of the drive?
>> tf-to-lba48 is a far less common operation, generally used for error
>> reporting (and in Host-Protected Area code as well, it appears).
>>
>> Jeff
>
> So are you saying it is not on any data read / write paths?
>
> Or that it is only on rarely used data read/write paths?
>
> Because if this is on a data read/write path at all, it looks like it
> could cause major data corruption with 1.5 TB drives (or larger).
..

It's on the initialization path for all lba48 devices which have
a configured HPA (Host Protected Area, mostly seen in brand-name
consumer-level systems).

The data it generates is then used on subsequent R/W commands.
I recommend the fix be backported for earlier kernels.

Cheers


\
 
 \ /
  Last update: 2008-11-03 00:45    [W:0.041 / U:0.112 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site