[PATCH 5/15] megaraid_sas: Fix probe_one to clear MSI-X flags in kdump

[PATCH 5/15] megaraid_sas: Fix probe_one to clear MSI-X flags in kdump

[adam radford]

[-- Attachment #1: Type: text/plain, Size: 1755 bytes --]

James/Linux-scsi,

The following patch for megaraid_sas fixes megasas_probe_one() to
clear MSI-X flags in kdump when the 'reset_devices' kernel parameter
is passed in.

Signed-off-by: Adam Radford <aradford@gmail.com>

diff -Naur linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas_base.c
linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas_base.c
--- linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas_base.c	2011-02-19
13:58:52.649363761 -0800
+++ linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas_base.c	2011-02-19
14:01:52.407362948 -0800
@@ -3901,9 +3901,26 @@
 static int __devinit
 megasas_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
 {
-	int rval;
+	int rval, pos;
 	struct Scsi_Host *host;
 	struct megasas_instance *instance;
+	u16 control = 0;
+
+	/* Reset MSI-X in the kdump kernel */
+	if (reset_devices) {
+		pos = pci_find_capability(pdev, PCI_CAP_ID_MSIX);
+		if (pos) {
+			pci_read_config_word(pdev, msi_control_reg(pos),
+					     &control);
+			if (control & PCI_MSIX_FLAGS_ENABLE) {
+				dev_info(&pdev->dev, "resetting MSI-X\n");
+				pci_write_config_word(pdev,
+						      msi_control_reg(pos),
+						      control &
+						      ~PCI_MSIX_FLAGS_ENABLE);
+			}
+		}
+	}

 	/*
 	 * Announce PCI information
diff -Naur linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas.h
linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas.h
--- linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas.h	2011-02-15
19:23:45.000000000 -0800
+++ linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas.h	2011-02-19
14:04:00.565333377 -0800
@@ -1477,4 +1477,7 @@
 	int max_index;
 };

+#define msi_control_reg(base) (base + PCI_MSI_FLAGS)
+#define PCI_MSIX_FLAGS_ENABLE (1 << 15)
+
 #endif				/*LSI_MEGARAID_SAS_H */

[-- Attachment #2: megaraid_sas.patch5 --]
[-- Type: application/octet-stream, Size: 1538 bytes --]

diff -Naur linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas_base.c linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas_base.c
--- linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas_base.c	2011-02-19 13:58:52.649363761 -0800
+++ linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas_base.c	2011-02-19 14:01:52.407362948 -0800
@@ -3901,9 +3901,26 @@
 static int __devinit
 megasas_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
 {
-	int rval;
+	int rval, pos;
 	struct Scsi_Host *host;
 	struct megasas_instance *instance;
+	u16 control = 0;
+
+	/* Reset MSI-X in the kdump kernel */
+	if (reset_devices) {
+		pos = pci_find_capability(pdev, PCI_CAP_ID_MSIX);
+		if (pos) {
+			pci_read_config_word(pdev, msi_control_reg(pos),
+					     &control);
+			if (control & PCI_MSIX_FLAGS_ENABLE) {
+				dev_info(&pdev->dev, "resetting MSI-X\n");
+				pci_write_config_word(pdev,
+						      msi_control_reg(pos),
+						      control &
+						      ~PCI_MSIX_FLAGS_ENABLE);
+			}
+		}
+	}
 
 	/*
 	 * Announce PCI information
diff -Naur linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas.h linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas.h
--- linux-2.6.38-rc5/drivers/scsi/megaraid/megaraid_sas.h	2011-02-15 19:23:45.000000000 -0800
+++ linux-2.6.38-rc5.new/drivers/scsi/megaraid/megaraid_sas.h	2011-02-19 14:04:00.565333377 -0800
@@ -1477,4 +1477,7 @@
 	int max_index;
 };
 
+#define msi_control_reg(base) (base + PCI_MSI_FLAGS)
+#define PCI_MSIX_FLAGS_ENABLE (1 << 15)
+
 #endif				/*LSI_MEGARAID_SAS_H */

Re: [PATCH 5/15] megaraid_sas: Fix probe_one to clear MSI-X flags in kdump

[Hannes Reinecke]

On 02/20/2011 03:22 AM, adam radford wrote:
> James/Linux-scsi,
> 
> The following patch for megaraid_sas fixes megasas_probe_one() to
> clear MSI-X flags in kdump when the 'reset_devices' kernel parameter
> is passed in.
> 
This is actually the one bit where I'm quite puzzled by.
Problem here is the pci_enable_msix() actually fails when MSI-X is
already enabled. Which is odd, given that MSI-X _is_ enabled
already, so taken at face value pci_enable_msix() shouldn't fail here.

I know that there are several stages to the MSI-X initialisation, so
this is the safe route here. But I also have seen _LOTS_ of failures
with kdump and raid HBAs, most of which point to the same problem here.

Can't we put in a 'force' parameter to pci_enable_msix() to avoid
having to code this one in each and every driver?

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		      zSeries & Storage
hare@suse.de			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Markus Rex, HRB 16746 (AG Nürnberg)
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html