AMD756 PCI IRQ Routing Patch 0.2.0

AMD756 PCI IRQ Routing Patch 0.2.0

[Jhon H. Caicedo O.]

[-- Attachment #1: Type: text/plain, Size: 309 bytes --]


Hi,

This is an updated version of the patch for AMD756 PCI IRQ Routing,
the changes are to use the read/write_config_nybble functions,
this makes the code shorter.

Thanks,

--
Jhon H. Caicedo O. <jhcaiced@osso.org.co>
Observatorio Sismológico del SurOccidente O.S.S.O
http://www.osso.org.co
Cali - Colombia

[-- Attachment #2: linux-2.4.5_amd756-pci-irq-routing-0.2.0.patch --]
[-- Type: text/plain, Size: 1641 bytes --]

--- linux-2.4.5/arch/i386/kernel/pci-irq.c	Wed May 16 12:25:39 2001
+++ linux/arch/i386/kernel/pci-irq.c	Fri Jun 22 07:46:41 2001
@@ -391,6 +391,38 @@
 	return 1;
 }
 
+/* Support for AMD756 PCI IRQ Routing
+ * Jhon H. Caicedo <jhcaiced@osso.org.co>
+ * Jun/21/2001 0.2.0 Release, fixed to use "nybble" functions... (jhcaiced)
+ * Jun/19/2001 Alpha Release 0.1.0 (jhcaiced)
+ * The AMD756 pirq rules are nibble-based
+ * offset 0x56 0-3 PIRQA  4-7  PIRQB
+ * offset 0x57 0-3 PIRQC  4-7  PIRQD
+ */
+static int pirq_amd756_get(struct pci_dev *router, struct pci_dev *dev, int pirq)
+{
+	u8 irq;
+	irq = 0;
+	if (pirq <= 4)
+	{
+		irq = read_config_nybble(router, 0x56, pirq - 1);
+	}
+	printk(KERN_INFO "AMD756: dev %04x:%04x, router pirq : %d get irq : %2d\n",
+		dev->vendor, dev->device, pirq, irq);
+	return irq;
+}
+
+static int pirq_amd756_set(struct pci_dev *router, struct pci_dev *dev, int pirq, int irq)
+{
+	printk(KERN_INFO "AMD756: dev %04x:%04x, router pirq : %d SET irq : %2d\n", 
+		dev->vendor, dev->device, pirq, irq);
+	if (pirq <= 4)
+	{
+		write_config_nybble(router, 0x56, pirq - 1, irq);
+	}
+	return 1;
+}
+
 #ifdef CONFIG_PCI_BIOS
 
 static int pirq_bios_set(struct pci_dev *router, struct pci_dev *dev, int pirq, int irq)
@@ -426,6 +458,8 @@
 	{ "VLSI 82C534", PCI_VENDOR_ID_VLSI, PCI_DEVICE_ID_VLSI_82C534, pirq_vlsi_get, pirq_vlsi_set },
 	{ "ServerWorks", PCI_VENDOR_ID_SERVERWORKS, PCI_DEVICE_ID_SERVERWORKS_OSB4,
 	  pirq_serverworks_get, pirq_serverworks_set },
+	{ "AMD756 VIPER", PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_VIPER_740B,
+		pirq_amd756_get, pirq_amd756_set },
 
 	{ "default", 0, 0, NULL, NULL }
 };

Re: AMD756 PCI IRQ Routing Patch 0.2.0

[Jeff Garzik]

"Jhon H. Caicedo O." wrote:
> This is an updated version of the patch for AMD756 PCI IRQ Routing,
> the changes are to use the read/write_config_nybble functions,
> this makes the code shorter.

Looks much better, thanks!

> +       printk(KERN_INFO "AMD756: dev %04x:%04x, router pirq : %d get irq : %2d\n",
> +               dev->vendor, dev->device, pirq, irq);
[...]
> +       printk(KERN_INFO "AMD756: dev %04x:%04x, router pirq : %d SET irq : %2d\n",
> +               dev->vendor, dev->device, pirq, irq);

None of the other PCI IRQ routines print out IRQ routing messages, so
these shouldn't either.  I assume this is debugging code?

Further, the printks are potentially misleading, because pirq_amd756_get
might not receive a valid irq, if 'pirq' is greater than 4.

	Jeff


-- 
Jeff Garzik      | Andre the Giant has a posse.
Building 1024    |
MandrakeSoft     |

Re: AMD756 PCI IRQ Routing Patch 0.2.0

[Jhon H. Caicedo O.]

Hi,

Jeff Garzik wrote:
> None of the other PCI IRQ routines print out IRQ routing messages, so
> these shouldn't either.  I assume this is debugging code?

Yes, the printks are only for debug and surely will be removed 
after some tests of the patch.
> 
> Further, the printks are potentially misleading, because pirq_amd756_get
> might not receive a valid irq, if 'pirq' is greater than 4.

With pirq > 4 pirq_amd756_get should return 0, i left the
printk just to see if this happens in some test.

In the original 2.4.5 kernel (without patch), I get irq=0 for pirq=4
and that was the cause of the error with a SMC Lucent CardBus Bridge.

Thanks,
--
Jhon H. Caicedo O. <jhcaiced@osso.org.co>
Observatorio Sismológico del SurOccidente O.S.S.O
http://www.osso.org.co
Cali - Colombia