lkml.org 
[lkml]   [2005]   [Feb]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [patch ide-dev 7/9] convert disk flush functions to use REQ_DRIVE_TASKFILE
Bartlomiej Zolnierkiewicz wrote:
> On Sunday 27 February 2005 05:51, Tejun Heo wrote:
>
>> Hello, Bartlomiej,
>>
>>On Thu, Feb 24, 2005 at 03:45:39PM +0100, Bartlomiej Zolnierkiewicz wrote:
>>
>>>Original patch from Tejun Heo <htejun@gmail.com>,
>>>ported over recent IDE changes by me.
>>>
>>>* teach ide_tf_get_address() about CHS
>>
>> IMHO, as patch #4 moves LBA/CHS selection into taskfile, I think
>>using taskfile->device to determine whether LBA or CHS is used instead
>>of drive->select makes more sense.
>>
>>
>>>* use ide_tf_get_address() and remove ide_get_error_location()
>>
>> IIRC, error responses for WIN_FLUSH_CACHE is in CHS if the LBA bit in
>>the device register is zero; likewise, in LBA if the LBA bit is one.
>>I don't know if drives can change the LBA bit when posting error
>>result. The original code reads back the device register on error to
>>determine whether to interpret the error response in CHS or LBA.
>>(ATA/ATAPI-6 isn't clear on this issue. Is ATA/ATAPI-7 updated?)
>
>
> The thing is that LBA bit is marked as "na" for FLUSH CACHE
> in all ATA/ATAPI drafts that I've seen.
>

Yeah, and nothing is mentioned about the LBA bit in the normal output
description even though the lcyl, hcyl.. registers are described to be
in either form of CHS or LBA. Strange...

>
>> This change combined with patch #2/#5 can make error address
>>calculation wrong on LBA28 drives. I think setting the LBA bit in the
>>device register according to the drive's addressing mode in
>>ide_task_init_flush() and use taskfile->device in ide_tf_get_address()
>>should fix the problem.
>
>
> I will change the code to set LBA bit, it shouldn't hurt. :-)

IMHO, there can be cases where just setting the LBA bit isn't enough.
Theoretically, drives can report CHS and clear the LBA bit in the
device register even when the LBA bit was set when the command was
issued. Maybe the intention of the original code was to handle
something like this?

--
tejun

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:10    [W:0.128 / U:0.008 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site