Re: Lost interrupts with IDE DMA on 2.5.x

[Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 2721 bytes --]


Attached patch should help, please try.

--
Bartlomiej

On Fri, 25 Apr 2003, Marek Habersack wrote:

> Hello,
>
>  I've recently added a second drive to my workstation and since then I'm
> getting the following error from time to time:
>
>   Apr 25 20:42:06 beowulf kernel: hda: dma_timer_expiry: dma status == 0x64
>   Apr 25 20:42:06 beowulf kernel: hda: lost interrupt
>   Apr 25 20:42:06 beowulf kernel: hda: dma_intr: bad DMA status (dma_stat=70)
>   Apr 25 20:42:06 beowulf kernel: hda: dma_intr: status=0x50 { DriveReady
>    SeekComplete }
>
> Both drives are new Maxtors (60 and 40GB) on the VIA KT266 chipset (the mobo
> is MSI K7T266 Pro2-A mobo):
>
> ----------VIA BusMastering IDE Configuration----------------
> Driver Version:                     3.36
> South Bridge:                       VIA vt8233a
> Revision:                           ISA 0x0 IDE 0x6
> Highest DMA rate:                   UDMA133
> BM-DMA base:                        0xfc00
> PCI clock:                          33.3MHz
> Master Read  Cycle IRDY:            0ws
> Master Write Cycle IRDY:            0ws
> BM IDE Status Register Read Retry:  yes
> Max DRDY Pulse Width:               No limit
> -----------------------Primary IDE-------Secondary IDE------
> Read DMA FIFO flush:          yes                 yes
> End Sector FIFO flush:         no                  no
> Prefetch Buffer:              yes                 yes
> Post Write Buffer:            yes                 yes
> Enabled:                      yes                 yes
> Simplex only:                  no                  no
> Cable Type:                   80w                 40w
> -------------------drive0----drive1----drive2----drive3-----
> Transfer Mode:       UDMA      UDMA       PIO       DMA
> Address Setup:      120ns     120ns     120ns     120ns
> Cmd Active:          90ns      90ns      90ns      90ns
> Cmd Recovery:        30ns      30ns      30ns      30ns
> Data Active:         90ns      90ns     330ns      90ns
> Data Recovery:       30ns      30ns     270ns      30ns
> Cycle Time:          15ns      15ns     600ns     120ns
> Transfer Rate:  133.3MB/s 133.3MB/s   3.3MB/s  16.6MB/s
>
>   Of course, when the above happens, all disk I/O freezes. The above happens
> only when there's simultaneous activity on both devices. It doesn't happen
> when the devices are on different IDE interfaces. The transfer is always
> retried and completed successfully, so it's not a bad hdd and I can only
> guess the problem is somewhere in the DMA/IRQ handling by the IDE driver. If
> there's not enough information to diagnose/solve the problem, I can do more
> tests (run with 2.4 for a while, run with the generic IDE drive etc.).
>
> TIA,
>
> marek

[-- Attachment #2: Type: TEXT/PLAIN, Size: 926 bytes --]

# Fix masked_irq arg handling for ide_do_request().
# Solves "hdx: lost interrupt" bug.
#
# Bartlomiej Zolnierkiewicz <bzolnier@elka.pw.edu.pl>

--- linux-2.5.68-bk6/drivers/ide/ide-io.c	Fri Apr 25 16:08:53 2003
+++ linux/drivers/ide/ide-io.c	Fri Apr 25 16:13:37 2003
@@ -850,14 +850,14 @@
 		 * happens anyway when any interrupt comes in, IDE or otherwise
 		 *  -- the kernel masks the IRQ while it is being handled.
 		 */
-		if (hwif->irq != masked_irq)
+		if (masked_irq != IDE_NO_IRQ && hwif->irq != masked_irq)
 			disable_irq_nosync(hwif->irq);
 		spin_unlock(&ide_lock);
 		local_irq_enable();
 			/* allow other IRQs while we start this request */
 		startstop = start_request(drive, rq);
 		spin_lock_irq(&ide_lock);
-		if (hwif->irq != masked_irq)
+		if (masked_irq != IDE_NO_IRQ && hwif->irq != masked_irq)
 			enable_irq(hwif->irq);
 		if (startstop == ide_released)
 			goto queue_next;