Re: 2.6.32-rc2: Intel(R) I/OAT DMA Engine init failed

[Dan Williams <dan.j.williams@intel.com>]

On Fri, 2009-10-02 at 02:46 -0700, Alexander Beregalov wrote:
> Hi
> The kerneil is 2.6.32-rc2-00196-g0efe5e3.
> Should I bisect?
> 
> ioatdma: Intel(R) QuickData Technology Driver 4.00
> ioatdma 0000:00:08.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
> ioatdma 0000:00:08.0: setting latency timer to 64
>   alloc irq_desc for 33 on node -1
>   alloc kstat_irqs on node -1
> ioatdma 0000:00:08.0: irq 33 for MSI/MSI-X
> ioatdma 0000:00:08.0: Self-test copy timed out, disabling
> ioatdma 0000:00:08.0: Intel(R) I/OAT DMA Engine init failed

I may ask for a bisect later, but for now let's try and see why it is
failing.  It is something platform specific as 2.6.32-rc3 loads fine on
my v1.2 platform.  Please reproduce this with the patch below.

Thanks,
Dan

diff --git a/drivers/dma/ioat/dma.c b/drivers/dma/ioat/dma.c
index c524d36..8f807b4 100644
--- a/drivers/dma/ioat/dma.c
+++ b/drivers/dma/ioat/dma.c
@@ -24,6 +24,7 @@
  * This driver supports an Intel I/OAT DMA engine, which does asynchronous
  * copy operations.
  */
+#define DEBUG
 
 #include <linux/init.h>
 #include <linux/module.h>
@@ -815,6 +816,7 @@ int __devinit ioat_dma_self_test(struct ioatdma_device *device)
 	struct completion cmp;
 	unsigned long tmo;
 	unsigned long flags;
+	enum dma_status stat;
 
 	src = kzalloc(sizeof(u8) * IOAT_TEST_SIZE, GFP_KERNEL);
 	if (!src)
@@ -864,10 +866,10 @@ int __devinit ioat_dma_self_test(struct ioatdma_device *device)
 
 	tmo = wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000));
 
-	if (tmo == 0 ||
-	    dma->device_is_tx_complete(dma_chan, cookie, NULL, NULL)
-					!= DMA_SUCCESS) {
-		dev_err(dev, "Self-test copy timed out, disabling\n");
+	stat = dma->device_is_tx_complete(dma_chan, cookie, NULL, NULL);
+	if (tmo == 0 || stat != DMA_SUCCESS) {
+		dev_err(dev, "Self-test copy timed out, disabling %ld:%d\n",
+			tmo, stat);
 		err = -ENODEV;
 		goto free_resources;
 	}