More ATI North Bridges like RS780 can't do MSI like its predecessors

in linux. Disable MSIs on them.
Signed-off-by: Shane Huang <shane.huang@amd.com>
Since there is some word wrapping problem with my mail client MS outlook

if I copy the patch into the text, so I'll have to attach the patch as

an attachment. Please check it.
Thanks

Best Regards

Shane

This patch recover Tejun's commit

4be8f906435a6af241821ab5b94b2b12cb7d57d8

 because there is one MSI bug on RS690+SB600 board which will lead to

boot failure. This bug is NOT same as the one in SB700 SATA controller,

quirk_msi_intx_disable_bug does not work to SB600. Disablement the MSI

of RS690 is the workaround.
Signed-off-by: Shane Huang <shane.huang@amd.com>
Since there is some word wrapping problem with my mail client MS outlook

if I copy the patch into the text, so I'll also attach the patch as an

attachment. Please check it.
diff -ruN old/drivers/pci/quirks.c new/drivers/pci/quirks.c

--- old/drivers/pci/quirks.c	2008-01-07 05:45:38.000000000 +0800

+++ new/drivers/pci/quirks.c	2008-01-22 11:02:00.000000000 +0800

@@ -1623,6 +1623,7 @@

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_SERVERWORKS,

PCI_DEVICE_ID_SERVERWORKS_GCNB_LE, quirk_disable_all_msi);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RS400_200,

quirk_disable_all_msi);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RS480,

quirk_disable_all_msi);

+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_RS690,

quirk_disable_all_msi);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_VIA, PCI_DEVICE_ID_VIA_VT3351,

quirk_disable_all_msi);

=20

 /* Disable MSI on chipsets that are known to not support it */

diff -ruN old/include/linux/pci_ids.h new/include/linux/pci_ids.h

--- old/include/linux/pci_ids.h	2008-01-07 05:45:38.000000000 +0800

+++ new/include/linux/pci_ids.h	2008-01-22 11:01:55.000000000 +0800

@@ -360,6 +360,7 @@

 #define PCI_DEVICE_ID_ATI_RS400_166     0x5a32

 #define PCI_DEVICE_ID_ATI_RS400_200     0x5a33

 #define PCI_DEVICE_ID_ATI_RS480         0x5950

+#define PCI_DEVICE_ID_ATI_RS690         0x7910

 /* ATI IXP Chipset */

 #define PCI_DEVICE_ID_ATI_IXP200_IDE	0x4349

 #define PCI_DEVICE_ID_ATI_IXP200_SMBUS	0x4353
Thanks

Best Regards

Shane

This patch disable MSI for the _whole_ system, not only behind the

RS690. Is this on purpose? Is MSI really going to be broken on any

bus that's _not_ behind RS690. If not, you might want to use

quirk_disable_msi() instead (as we do for AMD8131).
Brice

--

quirk_disable_msi() can not fix the issue in my debug,

quirk_msi_intx_disable_bug() which can fix SB700 SATA MSI bug does not

work either.

quirk_disable_all_msi is the only workaround I found.
If there is any other guy who also has one SB600+RS690 board, and can

help

to  verify this RS690 MSI disablement patch with a new kernel version

such as

2.6.24-rc7,  that's great.
BTW:

RS690 MSI disablement should NOT affect SB700 MSI, because as I know,

there will not be the combination of RS690+SB700 on the market.
Thanks

Shane
--

SB700 SATA MSI bug will be fixed in SB700 revision A21 at hardware

level,

but the SB700 revision older than A21 will also be found in the market.

This patch modify the original quirk commit

bc38b411fe696fad32b261f492cb4afbf1835256 instead of withdrawing it.
Signed-off-by: Shane Huang <shane.huang@amd.com>
Since there is some word wrapping problem with my mail client MS

outlook, I also attach the patch as an attachment. Please check it.
diff -ruN old/drivers/pci/quirks.c new/drivers/pci/quirks.c

--- old/drivers/pci/quirks.c	2008-01-07 05:45:38.000000000 +0800

+++ new/drivers/pci/quirks.c	2008-01-22 11:31:09.000000000 +0800

@@ -1709,6 +1709,22 @@

 {

 	dev->dev_flags |=3D PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG;

 }

+static void __devinit quirk_msi_intx_disable_ati_bug(struct pci_dev

*dev)

+{

+	struct pci_dev *p;

+	u8 rev =3D 0;

+

+	/* SB700 MSI issue will be fixed at HW level from revision A21,

+	 * we need check PCI REVISION ID of SMBus controller to get

SB700 revision.

+	 */

+	p =3D pci_get_device(PCI_VENDOR_ID_ATI,

PCI_DEVICE_ID_ATI_SBX00_SMBUS, NULL);

+	if (p!=3DNULL) {

+		pci_read_config_byte(p, PCI_CLASS_REVISION, &rev);

+	}

+	if ((rev < 0x3B) && (rev >=3D 0x30)) {

+		dev->dev_flags |=3D PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG;

+	}

+}

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_BROADCOM,

 			PCI_DEVICE_ID_TIGON3_5780,

 			quirk_msi_intx_disable_bug);

@@ -1729,17 +1745,17 @@

 			quirk_msi_intx_disable_bug);

=20

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4390,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4391,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4392,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4393,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI...
[ message continues ]

I did some modification to this patch and send it again, Please check

it.

The quirk to 0x4395 has been removed because 0x4395 only belongs to

SB800.
Thanks
diff -ruN linux-2.6.24-rc7_org/drivers/pci/quirks.c

linux-2.6.24-rc7_new/drivers/pci/quirks.c

--- linux-2.6.24-rc7_org/drivers/pci/quirks.c	2008-01-23

14:44:53.000000000 +0800

+++ linux-2.6.24-rc7_new/drivers/pci/quirks.c	2008-01-25

10:55:21.000000000 +0800

@@ -1709,6 +1709,24 @@

 {

 	dev->dev_flags |=3D PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG;

 }

+static void __devinit quirk_msi_intx_disable_ati_bug(struct pci_dev

*dev)

+{

+	struct pci_dev *p;

+	u8 rev =3D 0;

+

+	/* SB700 MSI issue will be fixed at HW level from revision A21,

+	 * we need check PCI REVISION ID of SMBus controller to get

SB700

+	 * revision.

+	 */

+	p =3D pci_get_device(PCI_VENDOR_ID_ATI,

PCI_DEVICE_ID_ATI_SBX00_SMBUS,

+			   NULL);

+	if (p) {

+		pci_read_config_byte(p, PCI_CLASS_REVISION, &rev);

+	}

+	if ((rev < 0x3B) && (rev >=3D 0x30)) {

+		dev->dev_flags |=3D PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG;

+	}

+}

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_BROADCOM,

 			PCI_DEVICE_ID_TIGON3_5780,

 			quirk_msi_intx_disable_bug);

@@ -1729,17 +1747,15 @@

 			quirk_msi_intx_disable_bug);

=20

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4390,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4391,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4392,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4393,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4394,

-			quirk_msi_intx_disable_bug);

-DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4395,

-			quirk_msi_intx_disable_bug);

+			quirk_msi_intx_disable_ati_bug);

=20

 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4...
[ message continues ]

After S-O-B, you can put --- and between it and the patch body, you can

say things which you wanna mention but don't think should be included in
Hmm... So, if there's no SMBUS device the quirk applies.  Is this

intended?  If that can't happen, just do if (!p) return;
You tested this, right?
--

tejun

--

OK, I'll have to submit another update patch later.
Thanks

Shane
--

I recommend running the scripts/checkpatch.pl script on any proposed

patches like this before you send them.  It will find a lot of these

problems for you :)
thanks,
greg k-h

--

Can you please get a decent email client?  Thunderbird + toggle word
It would be nice if things stay under 80col limit although we don't
--

tejun

--

From: "Shane Huang" <Shane.Huang@amd.com>
Can we get some detail as to why these north bridges have to have MSI

disabled completely?  I can't believe that all 6 of these ATI

controllers which will now be listed in the quirk table cannot use MSI

at all.
I've discovered several cases where it was a buggy device or a driver

bug that caused someone to erroneously submit patches that disable MSI

completely for the bridge they were behind.
I don't want that to happen any more.  One way to prevent that is to

have a full detailed justification for the MSI disabling in the

changelog or in the comments for the MSI quirk.
Thank you.

-

Hi Miller:
Thank you for your response.
The reason why MSIs of these northbridges do not work is still under

further debug, we are NOT able to tell its hardware issue or software

issue at this  time. But enablement of them will lead to the OS

installation failure in many distributions like openSUSE, Ubuntu etc:

https://bugzilla.novell.com/show_bug.cgi?id=302016
So we have to disable them firstly before we find out the root cause,

maybe they are just workarounds.
If you guys know much more about this MSI problem, don't hesitate to

tell us, we can debug on some of the hardware platforms.
BTW, There already some disablements to some ATI NB MSIs in the kernel,

the reasons are similar.
Thanks
-

From: "Shane Huang" <Shane.Huang@amd.com>
This logic seems backwards, to me.  "shoot first, ask questions later"

To me this it not how to approach this problem.
Once you turn MSI off, there is next to no incentive to fix the

problem because users aren't running into it any longer.
The only two devices in that bug report which should be using MSI

would be the SATA controller and the broadcom ethernet NIC.  And by

the failed bootup logs provided by the user the problem is clearly

with the SATA controller.
One common problem we're finding is that some devices have a hardware

bug where setting INTX_DISABLE in the PCI COMMAND register masks MSI

interrupts too.
I mention this because the user in that report mentions that the

kernel upgrade causes the failure, and one thing we started doing not

too long ago was to set the INTX_DISABLE bit when MSI is enabled for a

device.
So maybe this SATA controller has this problem too.  It is easy to

test, simply comment out all of the pci_intx() function calls in

drivers/pci/msi.c and perform a test boot with MSI enabled.
I would rather you approach analysis of these kinds of MSI bugs in

this manner, instead of disabling MSI wholesale.  Because with the

latter approach it is nearly guarenteed that the real reason will only

be discovered with an extremely low priority.
Thank you.

-

And the same SATA controller could show up behind a different northbridge.

It would be unfortunate to hit the same device bug independantly on each

system and work around it by doing something that won't help the next 
Have we gotten around to having a device quirk for this? I bet it won't be

too long before we see a system where the SATA controller doesn't work

with INTX disabled and the ethernet controller doesn't work with it

enabled, since we've seen devices with each of these bugs.
	-Daniel

*This .sig left intentionally blank*

-

I agree with David here.  Please work to find the root cause of this

problem.  If it turns out that all of these chipsets have broken MSI and

can't handle it at all (oops, that's a major bug...) then we will accept

this patch.
But for now, let's not take a band-aid that prevents others from working

to solve the real issues here.
thanks,
greg k-h

-