Hello Andre,
Thank you very much for implementing the 48-bit ATA controller
support in your recent IDE kernel patches (ide.2.4.14.11062001.patch).
I am using a Maxtor 160GB hard disk with your patches on linux-2.4.15-pre5,
and it seems to be working well so far (two hours).
I do have a couple of minor notes about your patch. I could
generate some diffs, but they're simple and I'm not completely sure
about the right solution.
1. Your patch creates a circular dependency between the ide-mod.o
and ide-probe-mod.o modules, which is only noticible when IDE support
is compiled as a module. The problem is that ide.c has the
EXPORT_SYMBOL declarations for export_ide_init_queue and
export_probe_for_drive in ide-probe.c. At the moment, I have
moved the two EXPORT_SYMBOL declarations to the ide-probe.c, but I
believe the correct solution is just to remove the two routines
from your patch, since it appears that nothing uses them yet.
2. A while ago, I posted a change that modularizes partition
support (in reality, I never use the kernel-based partition code, but
that's another matter). Your declaration of ide_xlate_1024_hook to
fs/partitions/msdos.c creates a circular dependency in my kernel (but
not in Linus's), which I fixed by moving the declatation to
fs/partitions/check.c. I do not yet understand the purpose of
ide_xlate_1024 to understand whether it really is specific to the
MSDOS style of partition labeling.
Anyhow, I hope this information is helpful. Please let me know
if you want me to geneate a patch or test anything. So far, your patch
seems to be working very well. Thank you very much for developing it.
Adam J. Richter __ ______________ 4880 Stevens Creek Blvd, Suite 104
[email protected] \ / San Jose, California 95129-1034
+1 408 261-6630 | g g d r a s i l United States of America
fax +1 408 261-6631 "Free Software For The Rest Of Us."
On Sat, Nov 17, 2001 at 04:59:21AM -0800, Adam J. Richter wrote:
> I do not yet understand the purpose of ide_xlate_1024
> to understand whether it really is specific to the
> MSDOS style of partition labeling.
Yes, it is.
When DOS had a 528 MB limit various kludges were developed
to make larger disks available. BIOS-kludges went under the
name of "translation". Kludges that avoided the BIOS went
under the name "Disk Manager". Linux has some detailed knowledge
about these kludges, enough to enable Linux to successfully
share a disk with a DOS using translation and/or Disk Manager.
All of this stuff is totally obsolete today.
I think ide_xlate_1024 and family can be ripped out of 2.5
and nobody will notice, especially since it will be a long
time before 2.6.
Andries
Hi, Adam,
Thanks a lot that you have tested Maxtor 160G drives under Linux, Could you
tell me which former version linux kernel are you using? can we use the
newest 2.4.16 downloaded from http://www.kernel.org?
Do we need to check some special options in the configuration file for
version 2.4.16 to support Promise ata133 card and Maxtor 160G drive? Because
from the kernel configuration file for 2.4.16, only ata33/ata66/ata100 are
supported for Promise card.
The last question, do we need to manually change any files beside
downloading the former newest kernel?
Thanks a lot.
Yours sincerely,
Guolin Cheng
-----Original Message-----
From: Adam J. Richter [mailto:[email protected]]
Sent: Saturday, November 17, 2001 4:59 AM
To: [email protected]
Cc: [email protected]
Subject: Notes on ATA/133 patch (ide.2.4.14.11062001.patch)
Hello Andre,
Thank you very much for implementing the 48-bit ATA controller
support in your recent IDE kernel patches (ide.2.4.14.11062001.patch).
I am using a Maxtor 160GB hard disk with your patches on linux-2.4.15-pre5,
and it seems to be working well so far (two hours).
I do have a couple of minor notes about your patch. I could
generate some diffs, but they're simple and I'm not completely sure
about the right solution.
1. Your patch creates a circular dependency between the ide-mod.o
and ide-probe-mod.o modules, which is only noticible when IDE support
is compiled as a module. The problem is that ide.c has the
EXPORT_SYMBOL declarations for export_ide_init_queue and
export_probe_for_drive in ide-probe.c. At the moment, I have
moved the two EXPORT_SYMBOL declarations to the ide-probe.c, but I
believe the correct solution is just to remove the two routines
from your patch, since it appears that nothing uses them yet.
2. A while ago, I posted a change that modularizes partition
support (in reality, I never use the kernel-based partition code, but
that's another matter). Your declaration of ide_xlate_1024_hook to
fs/partitions/msdos.c creates a circular dependency in my kernel (but
not in Linus's), which I fixed by moving the declatation to
fs/partitions/check.c. I do not yet understand the purpose of
ide_xlate_1024 to understand whether it really is specific to the
MSDOS style of partition labeling.
Anyhow, I hope this information is helpful. Please let me know
if you want me to geneate a patch or test anything. So far, your patch
seems to be working very well. Thank you very much for developing it.
Adam J. Richter __ ______________ 4880 Stevens Creek Blvd, Suite
104
[email protected] \ / San Jose, California 95129-1034
+1 408 261-6630 | g g d r a s i l United States of America
fax +1 408 261-6631 "Free Software For The Rest Of Us."