[haiku-commits] r35083 - haiku/trunk/src/add-ons/kernel/generic/ata_adapter
- From: mmlr@xxxxxxxx
- To: haiku-commits@xxxxxxxxxxxxx
- Date: Fri, 15 Jan 2010 03:56:39 +0100 (CET)
Author: mmlr
Date: 2010-01-15 03:56:39 +0100 (Fri, 15 Jan 2010)
New Revision: 35083
Changeset: http://dev.haiku-os.org/changeset/35083/haiku
Modified:
haiku/trunk/src/add-ons/kernel/generic/ata_adapter/ata_adapter.c
Log:
Don't return early when we're not DMAing. This causes the bus master status
to be read/acknowledged (new) and the status to be read (like before in that
case) which helps discarding spurious interrupts on one controller I have that
would otherwise cause an interrupt storm as the bus master interrupt would never
be acknowledged. The ATA stack will discard the interrupt if it's not expecting
a transfer, so no real harm done.
Modified: haiku/trunk/src/add-ons/kernel/generic/ata_adapter/ata_adapter.c
===================================================================
--- haiku/trunk/src/add-ons/kernel/generic/ata_adapter/ata_adapter.c
2010-01-14 23:40:13 UTC (rev 35082)
+++ haiku/trunk/src/add-ons/kernel/generic/ata_adapter/ata_adapter.c
2010-01-15 02:56:39 UTC (rev 35083)
@@ -201,14 +201,6 @@
TRACE_INT("ata_adapter_inthand\n");
- if (!channel->dmaing) {
- // this could be a spurious interrupt, so read status
- // register unconditionally to acknowledge those
- TRACE_INT("ata_adapter_inthand: not DMA\n");
- pci->read_io_8(device, channel->command_block_base + 7);
- return B_UNHANDLED_INTERRUPT;
- }
-
// need to read bus master status first, because some controllers
// will clear the interrupt status bit once ATA status is read
statusBM = pci->read_io_8(device, channel->bus_master_base
Other related posts:
- » [haiku-commits] r35083 - haiku/trunk/src/add-ons/kernel/generic/ata_adapter - mmlr
