[PATCH] sata_sil: update device hotplug handling, take #2

SIEN on some 3112 controllers doesn't mask SATA IRQ properly.  IRQ
stays asserted even after SIEN is masked and IRQ is acked.  Also, even
while frozen, any SATA PHY event including hardreset raises SATA IRQ.
Clearing SError seems to be the only way to deassert SATA IRQ.

This patch makes sil_host_intr() clear SError on SATA IRQs and ignore
SATA IRQs reported while frozen so that hardreset doesn't trigger
hotplug event (which ends up hardresetting again).

In such cases, the port still gets re-frozen to minimize the danger of
screaming interrupts.  This results in one nil EH repeat on
controllers with broken SIEN but other than that does no harm.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

drivers/scsi/sata_sil.c

@@ -344,7 +344,25 @@ static void sil_host_intr(struct ata_port *ap, u32 bmdma2)
 	u8 status;
 
 	if (unlikely(bmdma2 & SIL_DMA_SATA_IRQ)) {
-		ata_ehi_hotplugged(&ap->eh_info);
+		u32 serror;
+
+		/* SIEN doesn't mask SATA IRQs on some 3112s.  Those
+		 * controllers continue to assert IRQ as long as
+		 * SError bits are pending.  Clear SError immediately.
+		 */
+		serror = sil_scr_read(ap, SCR_ERROR);
+		sil_scr_write(ap, SCR_ERROR, serror);
+
+		/* Trigger hotplug and accumulate SError only if the
+		 * port isn't already frozen.  Otherwise, PHY events
+		 * during hardreset makes controllers with broken SIEN
+		 * repeat probing needlessly.
+		 */
+		if (!(ap->flags & ATA_FLAG_FROZEN)) {
+			ata_ehi_hotplugged(&ap->eh_info);
+			ap->eh_info.serror |= serror;
+		}
+
 		goto freeze;
 	}