pdc202xx_old: ignore "FIFO empty" bit in test_irq() method

The driver takes into account not only the interrupt status bit but also "FIFO empty" bit in its test_irq() method. This actually is a superfluous check since for the DMA commands calling the dma_test_irq() method further in the interrupt handler makes sure FIFO is emptied.

---
The patch is atop of ide-2.6.git tree...

Russell, would you mind giving this patch a test?

 drivers/ide/pdc202xx_old.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Index: ide-2.6/drivers/ide/pdc202xx_old.c
===================================================================
@@ -93,13 +93,13 @@ static int pdc202xx_test_irq(ide_hwif_t 
 		 * bit 7: error, bit 6: interrupting,
 		 * bit 5: FIFO full, bit 4: FIFO empty
 		 */
-		return ((sc1d & 0x50) == 0x50) ? 1 : 0;
+		return (sc1d & 0x40) ? 1 : 0;
 	} else	{
 		/*
 		 * bit 3: error, bit 2: interrupting,
 		 * bit 1: FIFO full, bit 0: FIFO empty
 		 */
-		return ((sc1d & 0x05) == 0x05) ? 1 : 0;
+		return (sc1d & 0x04) ? 1 : 0;
 	}
 }
 

From: Sergei Shtylyov <[email protected]>
Date: Wed, 14 Apr 2010 19:52:59 +0400

Applied, but I don't want to push this out until there is positive testing from Russell.

Russell, please test Sergei's patch, thanks!

From: David Miller <[email protected]>
Date: Thu, 22 Apr 2010 19:12:36 -0700 (PDT)

I don't think Russell has seen any of these postings or requests for testing since as vger postmaster I'm now seeing his email address bouncing with 5 day delivery timeouts... sigh :-/

RMK is temporarily offline due to ISP problems. He's managed to sneak a few messages about it to the ARM list via some other route, but that's it.