* [PATCH] PIIX: config_drive_for_dma() cleanup
@ 2006-05-11 19:02 Sergei Shtylyov
2006-05-11 20:29 ` Alan Cox
0 siblings, 1 reply; 3+ messages in thread
From: Sergei Shtylyov @ 2006-05-11 19:02 UTC (permalink / raw)
To: Andrew Morton
Cc: Bartlomiej Zolnierkiewicz, Alan Cox, linux-ide, liknux-kernel
[-- Attachment #1: Type: text/plain, Size: 496 bytes --]
Hello.
Clean up config_drive_for_dma() from the useless code:
- there's no need for PIO fallback here since this is done by the caller,
piix_config_drive_xfer_rate();
- there's no need for MWDMA mode 0 check, since this mode is not supported by
the driver (hwif->mwdma_mask == 0x6), and hence can't be selected by
ide_dma_speed().
Also, while at it, simplify piix_config_drive_xfer_rate() a bit.
MBR, Sergei
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
[-- Attachment #2: piix-config_drive_for_dma-cleanup.patch --]
[-- Type: text/plain, Size: 3639 bytes --]
Index: linus/drivers/ide/pci/piix.c
===================================================================
--- linus.orig/drivers/ide/pci/piix.c
+++ linus/drivers/ide/pci/piix.c
@@ -4,10 +4,11 @@
* Copyright (C) 1998-1999 Andrzej Krzysztofowicz, Author and Maintainer
* Copyright (C) 1998-2000 Andre Hedrick <andre@linux-ide.org>
* Copyright (C) 2003 Red Hat Inc <alan@redhat.com>
+ * Copyright (C) 2006 MontaVista Software, Inc. <source@mvista.com>
*
* May be copied or modified under the terms of the GNU General Public License
*
- * PIO mode setting function for Intel chipsets.
+ * PIO mode setting function for Intel chipsets.
* For use instead of BIOS settings.
*
* 40-41
@@ -25,7 +26,7 @@
* sitre = word42 & 0x4000; secondary
*
* 44 8421|8421 hdd|hdb
- *
+ *
* 48 8421 hdd|hdc|hdb|hda udma enabled
*
* 0001 hda
@@ -73,8 +74,6 @@
* PIIX4 errata #15 - Must not read control registers
* during a PIO transfer
* 440MX errata #13 - As PIIX4 errata #15
- * ICH2 errata #21 - DMA mode 0 doesn't work right
- * ICH0/1 errata #55 - As ICH2 errata #21
* ICH2 spec c #9 - Extra operations needed to handle
* drive hotswap [NOT YET SUPPORTED]
* ICH2 spec c #20 - IDE PRD must not cross a 64K boundary
@@ -333,56 +332,24 @@ static int piix_tune_chipset (ide_drive_
}
/**
- * piix_faulty_dma0 - check for DMA0 errata
- * @hwif: IDE interface to check
- *
- * If an ICH/ICH0/ICH2 interface is is operating in multi-word
- * DMA mode with 600nS cycle time the IDE PIO prefetch buffer will
- * inadvertently provide an extra piece of secondary data to the primary
- * device resulting in data corruption.
- *
- * With such a device this test function returns true. This allows
- * our tuning code to follow Intel recommendations and use PIO on
- * such devices.
- */
-
-static int piix_faulty_dma0(ide_hwif_t *hwif)
-{
- switch(hwif->pci_dev->device)
- {
- case PCI_DEVICE_ID_INTEL_82801AA_1: /* ICH */
- case PCI_DEVICE_ID_INTEL_82801AB_1: /* ICH0 */
- case PCI_DEVICE_ID_INTEL_82801BA_8: /* ICH2 */
- case PCI_DEVICE_ID_INTEL_82801BA_9: /* ICH2 */
- return 1;
- }
- return 0;
-}
-
-/**
* piix_config_drive_for_dma - configure drive for DMA
* @drive: IDE drive to configure
*
* Set up a PIIX interface channel for the best available speed.
- * We prefer UDMA if it is available and then MWDMA. If DMA is
- * not available we switch to PIO and return 0.
+ * We prefer UDMA if it is available and then MWDMA. If DMA is
+ * not available we return 0.
*/
static int piix_config_drive_for_dma (ide_drive_t *drive)
{
u8 speed = ide_dma_speed(drive, piix_ratemask(drive));
-
- /* Some ICH devices cannot support DMA mode 0 */
- if(speed == XFER_MW_DMA_0 && piix_faulty_dma0(HWIF(drive)))
- speed = 0;
-
- /* If no DMA speed was available or the chipset has DMA bugs
- then disable DMA and use PIO */
-
- if (!speed || no_piix_dma) {
- u8 tspeed = ide_get_best_pio_mode(drive, 255, 5, NULL);
- speed = piix_dma_2_pio(XFER_PIO_0 + tspeed);
- }
+
+ /*
+ * If no DMA speed was available or the chipset has DMA bugs
+ * then disable DMA and use PIO
+ */
+ if (!speed || no_piix_dma)
+ return 0;
(void) piix_tune_chipset(drive, speed);
return ide_dma_enable(drive);
@@ -405,10 +372,8 @@ static int piix_config_drive_xfer_rate (
if ((id->capability & 1) && drive->autodma) {
- if (ide_use_dma(drive)) {
- if (piix_config_drive_for_dma(drive))
- return hwif->ide_dma_on(drive);
- }
+ if (ide_use_dma(drive) && piix_config_drive_for_dma(drive))
+ return hwif->ide_dma_on(drive);
goto fast_ata_pio;
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: [PATCH] PIIX: config_drive_for_dma() cleanup
2006-05-11 19:02 [PATCH] PIIX: config_drive_for_dma() cleanup Sergei Shtylyov
@ 2006-05-11 20:29 ` Alan Cox
2006-05-11 20:34 ` Sergei Shtylyov
0 siblings, 1 reply; 3+ messages in thread
From: Alan Cox @ 2006-05-11 20:29 UTC (permalink / raw)
To: Sergei Shtylyov
Cc: Andrew Morton, Bartlomiej Zolnierkiewicz, linux-ide,
liknux-kernel
On Iau, 2006-05-11 at 23:02 +0400, Sergei Shtylyov wrote:
> @@ -73,8 +74,6 @@
> * PIIX4 errata #15 - Must not read control registers
> * during a PIO transfer
> * 440MX errata #13 - As PIIX4 errata #15
> - * ICH2 errata #21 - DMA mode 0 doesn't work right
> - * ICH0/1 errata #55 - As ICH2 errata #21
NAK. The errata are documented carefully so that people know what the
problems are. They have not gone away.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] PIIX: config_drive_for_dma() cleanup
2006-05-11 20:29 ` Alan Cox
@ 2006-05-11 20:34 ` Sergei Shtylyov
0 siblings, 0 replies; 3+ messages in thread
From: Sergei Shtylyov @ 2006-05-11 20:34 UTC (permalink / raw)
To: Alan Cox; +Cc: Andrew Morton, Bartlomiej Zolnierkiewicz, linux-ide,
liknux-kernel
Hello.
Alan Cox wrote:
> On Iau, 2006-05-11 at 23:02 +0400, Sergei Shtylyov wrote:
>>@@ -73,8 +74,6 @@
>> * PIIX4 errata #15 - Must not read control registers
>> * during a PIO transfer
>> * 440MX errata #13 - As PIIX4 errata #15
>>- * ICH2 errata #21 - DMA mode 0 doesn't work right
>>- * ICH0/1 errata #55 - As ICH2 errata #21
> NAK. The errata are documented carefully so that people know what the
> problems are. They have not gone away.
OK. The patch was bad for another reason -- PIO fallback won't work since
piix_tune_drive() currently "forgets" to set the drive's own transfer mode.
The same goes to SLC90E66 patch. I'll resend...
MBR, Sergei
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2006-05-11 20:35 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-05-11 19:02 [PATCH] PIIX: config_drive_for_dma() cleanup Sergei Shtylyov
2006-05-11 20:29 ` Alan Cox
2006-05-11 20:34 ` Sergei Shtylyov
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).