Skip to contentSkip to conversation

hpt366: fix HPT370 DMA timeouts

PATCH Applied Sergei Shtylyov 2 participants 2 messages Apr 7, 2009 – Apr 16, 2009 libata

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 hpt366: fix HPT370 DMA timeouts

The big driver change in 2.4.19-rc1 introduced a regression for many HPT370[A] chips -- DMA stopped to work completely, only causing endless timeouts...

The culprit has been identified (at last!): it turned to be the code resetting the DMA state machine before each transfer. Stop doing it now as this counter- measure has clearly caused more harm than good.

This should fix the kernel.org bug #7703.

---
This patch is against the recent Linus' tree.
It should go into all stable kernels as it fixes post-2.4.18 (!) regression.

 drivers/ide/hpt366.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletion(-)
drivers/ide/hpt366.c +3 -1
Index: linux-2.6/drivers/ide/hpt366.c
===================================================================
@@ -114,6 +114,8 @@
  *   the register setting lists into the table indexed by the clock selected
  * - set the correct hwif->ultra_mask for each individual chip
  * - add Ultra and MW DMA mode filtering for the HPT37[24] based SATA cards
+ * - stop resetting HPT370's state machine before each DMA transfer as that has
+ *   caused more harm than good
  *	Sergei Shtylyov, <[email protected]> or <[email protected]>
  */
 
@@ -133,7 +135,7 @@
 #define DRV_NAME "hpt366"
 
 /* various tuning parameters */
-#define HPT_RESET_STATE_ENGINE
+#undef	HPT_RESET_STATE_ENGINE
 #undef	HPT_DELAY_INTERRUPT
 
 static const char *quirk_drives[] = {
#1 lore raw
Quoted 10 lines · On Tuesday 07 April 2009 18:09:32 Sergei Shtylyov wrote:

The big driver change in 2.4.19-rc1 introduced a regression for many HPT370[A] chips -- DMA stopped to work completely, only causing endless timeouts...

The culprit has been identified (at last!): it turned to be the code resetting the DMA state machine before each transfer. Stop doing it now as this counter- measure has clearly caused more harm than good.

This should fix the kernel.org bug #7703.

Wow!

applied