[PATCH] SSB PCI core driver fixes

The patch below against 2.6.23-rc1-mm2 fixes various things on the SSB
PCI core driver:
 - Correctly write the configuration register value in 
   ssb_extpci_write_config() for len = 1 or len = 2.
 - Set the PCI latency timer to handle devices on the PCI bus.
 - Set the PCI arbiter control to internal.
 - Add some delay between the configuration of the PCI controller 
   and its registration.

Note: this is the latest SSB patch from my local tree. Next ones are 
only BCM947xx or CFE related.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

--- a/drivers/ssb/driver_pcicore.c
+++ b/drivers/ssb/driver_pcicore.c

@@ -99,6 +99,9 @@
 
 	/* Enable PCI bridge BAR1 prefetch and burst */
 	pci_write_config_dword(dev, SSB_BAR1_CONTROL, 3);
+
+	/* Make sure our latency is high enough to handle the devices behind us */
+	pci_write_config_byte(dev, PCI_LATENCY_TIMER, 0xa8);
 }
 DECLARE_PCI_FIXUP_EARLY(PCI_ANY_ID, PCI_ANY_ID, ssb_fixup_pcibridge);
 

@@ -230,7 +233,7 @@
 		val = *((const u32 *)buf);
 		break;
 	}
-	writel(*((const u32 *)buf), mmio);
+	writel(val, mmio);
 
 	err = 0;
 unmap:

@@ -311,6 +314,8 @@
 	udelay(150); /* Assertion time demanded by the PCI standard */
 	val |= SSB_PCICORE_CTL_RST; /* Deassert RST# */
 	pcicore_write32(pc, SSB_PCICORE_CTL, val);
+	val = SSB_PCICORE_ARBCTL_INTERN;
+	pcicore_write32(pc, SSB_PCICORE_ARBCTL, val);
 	udelay(1); /* Assertion time demanded by the PCI standard */
 
 	/*TODO cardbus mode */

@@ -340,6 +345,7 @@
 	 * The following needs change, if we want to port hostmode
 	 * to non-MIPS platform. */
 	set_io_port_base((unsigned long)ioremap_nocache(SSB_PCI_MEM, 0x04000000));
+	mdelay(300);
 	register_pci_controller(&ssb_pcicore_controller);
 }

-- 
  .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
 : :' :  Debian developer           | Electrical Engineer
 `. `'   aurel32@debian.org         | aurelien@aurel32.net
   `-    people.debian.org/~aurel32 | www.aurel32.net

On Thursday 09 August 2007, Aurelien Jarno wrote:

 - Add some delay between the configuration of the PCI controller 
   and its registration.

Why? It is _huge_ and people won't like it ;)
At least add a comment why this is needed.

On Thu, Aug 09, 2007 at 02:38:24AM +0200, Aurelien Jarno wrote:

Michael Buesch a écrit :

quoted

On Thursday 09 August 2007, Aurelien Jarno wrote:

quoted

 - Add some delay between the configuration of the PCI controller 
   and its registration.

Why? It is _huge_ and people won't like it ;)
At least add a comment why this is needed.

It is need, otherwise the PCI controller gets confused, which causes a
reset of the machine. Then CFE goes into a loop booting again and again
without being able to get up to the prompt.

I agree this is a huge value, so I will try to find the minimum value
and then add some margin.

I will send an updated patch.

A few experiments have shown that the minimum value is 3ms on my WGT634U
machine. I haved changed the value to 10ms in the new patch below, which
gives some margin, and added a comment.

Aurelien



The patch below against 2.6.23-rc1-mm2 fixes various things on the SSB
PCI core driver:
 - Correctly write the configuration register value in 
   ssb_extpci_write_config() for len = 1 or len = 2.
 - Set the PCI_LATENCY_TIMER to handle devices on the PCI bus.
 - Set the PCI arbiter control to internal.
 - Add some delay between the configuration of the PCI controller 
   and its registration.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

--- a/drivers/ssb/driver_pcicore.c
+++ b/drivers/ssb/driver_pcicore.c

@@ -99,6 +99,9 @@
 
 	/* Enable PCI bridge BAR1 prefetch and burst */
 	pci_write_config_dword(dev, SSB_BAR1_CONTROL, 3);
+
+	/* Make sure our latency is high enough to handle the devices behind us */
+	pci_write_config_byte(dev, PCI_LATENCY_TIMER, 0xa8);
 }
 DECLARE_PCI_FIXUP_EARLY(PCI_ANY_ID, PCI_ANY_ID, ssb_fixup_pcibridge);
 

@@ -230,7 +233,7 @@
 		val = *((const u32 *)buf);
 		break;
 	}
-	writel(*((const u32 *)buf), mmio);
+	writel(val, mmio);
 
 	err = 0;
 unmap:

@@ -311,6 +314,8 @@
 	udelay(150); /* Assertion time demanded by the PCI standard */
 	val |= SSB_PCICORE_CTL_RST; /* Deassert RST# */
 	pcicore_write32(pc, SSB_PCICORE_CTL, val);
+	val = SSB_PCICORE_ARBCTL_INTERN;
+	pcicore_write32(pc, SSB_PCICORE_ARBCTL, val);
 	udelay(1); /* Assertion time demanded by the PCI standard */
 
 	/*TODO cardbus mode */

@@ -340,6 +345,9 @@
 	 * The following needs change, if we want to port hostmode
 	 * to non-MIPS platform. */
 	set_io_port_base((unsigned long)ioremap_nocache(SSB_PCI_MEM, 0x04000000));
+	/* Give some time to the PCI controller to configure itself with the new
+	 * values. Not waiting at this point causes crashes of the machine. */
+	mdelay(10);
 	register_pci_controller(&ssb_pcicore_controller);
 }
 

-- 
  .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
 : :' :  Debian developer           | Electrical Engineer
 `. `'   aurel32@debian.org         | aurelien@aurel32.net
   `-    people.debian.org/~aurel32 | www.aurel32.net

Michael Buesch a écrit :

On Thursday 09 August 2007, Aurelien Jarno wrote:

quoted

 - Add some delay between the configuration of the PCI controller 
   and its registration.

Why? It is _huge_ and people won't like it ;)
At least add a comment why this is needed.

It is need, otherwise the PCI controller gets confused, which causes a
reset of the machine. Then CFE goes into a loop booting again and again
without being able to get up to the prompt.

I agree this is a huge value, so I will try to find the minimum value
and then add some margin.

I will send an updated patch.

quoted

@@ -340,6 +345,7 @@
 	 * The following needs change, if we want to port hostmode
 	 * to non-MIPS platform. */
 	set_io_port_base((unsigned long)ioremap_nocache(SSB_PCI_MEM, 0x04000000));
+	mdelay(300);
 	register_pci_controller(&ssb_pcicore_controller);
 }

-- 
  .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
 : :' :  Debian developer           | Electrical Engineer
 `. `'   aurel32@debian.org         | aurelien@aurel32.net
   `-    people.debian.org/~aurel32 | www.aurel32.net