Skip to contentSkip to conversation

PCI fixes for 2.6.29

GIT PULL Matthew Wilcox 4 participants 8 messages Mar 11, 2009 – Mar 12, 2009 PCI

lore
Share & tools
lore
#0 lore raw

Hi Linus,

These are the PCI fixes that have been piling up for the last week while Jesse's been on vacation. I don't think there's anything questionable here. All the patches have been posted to linux-pci.

The following changes since commit 16b71fdf97599f1b1b7f38418ee9922d9f117396:
Samuel CUELLA (1):
i810: fix kernel crash fix when struct fb_var_screeninfo is supplied

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/willy/pci.git fixes-20090310

Alex Chiang (2):
PCIe: AER: during disable, check subordinate before walking
PCIe: portdrv: call pci_disable_device during remove

Alexander Duyck (1):
PCI: Add PCI quirk to disable L0s ASPM state for 82575 and 82598

Michael Ellerman (1):
powerpc/pseries: The RPA PCI hotplug driver depends on EEH

Prakash Punnoor (2):
pci: don't disable too many HT MSI mapping
pci: Fix typo in message while disabling HT MSI mapping

 drivers/pci/hotplug/Kconfig        |    2 +-
 drivers/pci/pcie/aer/aerdrv_core.c |    3 +++
 drivers/pci/pcie/portdrv_pci.c     |    1 +
 drivers/pci/quirks.c               |   31 ++++++++++++++++++++++++++++++-
 4 files changed, 35 insertions(+), 2 deletions(-)
Signature · 4 lines
Matthew Wilcox				Intel Open Source Technology Centre
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."
#1 lore raw
Quoted 4 lines · Matthew Wilcox

These are the PCI fixes that have been piling up for the last week while Jesse's been on vacation. I don't think there's anything questionable here. All the patches have been posted to linux-pci.

Grrr. It doesn't even compile cleanly:

drivers/pci/quirks.c: In function ‘ht_check_msi_mapping’:
drivers/pci/quirks.c:2142: warning: ‘return’ with no value, in function returning non-void

and yes, it's a real bug.

I'm not happy.

I assume the proper fix is this trivial one-liner, but people should double-check.

Not pulled.

Linus

---
 drivers/pci/quirks.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/pci/quirks.c +1 -1
index a13f3d7..4ba55d0 100644
@@ -2139,7 +2139,7 @@ static int __devinit ht_check_msi_mapping(struct pci_dev *dev)
 
 	/* Enabling HT MSI mapping on this device breaks MCP51 */
 	if (dev->device == 0x270)
-		return;
+		return 0;
 
 	/* check if there is HT MSI cap or enabled on this device */
 	pos = pci_find_ht_capability(dev, HT_CAPTYPE_MSI_MAPPING);
#2 lore raw
Quoted 11 lines · Linus Torvalds
Quoted 3 lines · Matthew Wilcox

These are the PCI fixes that have been piling up for the last week while Jesse's been on vacation. I don't think there's anything questionable here. All the patches have been posted to linux-pci.

Grrr. It doesn't even compile cleanly:

drivers/pci/quirks.c: In function ?ht_check_msi_mapping?:
drivers/pci/quirks.c:2142: warning: ?return? with no value, in function returning non-void

and yes, it's a real bug.

Agreed, it's a real bug. I also think it should return 0, but I want Prakash to verify that, since I don't have a system of this nature.

I wish this warning were an error. According to the GCC docs, -Werror=return-type should do this, but it seems to have no effect.

Quoted 1 line

I'm not happy.

Me neither. I'm sorry I didn't catch the warning.

Signature · 4 lines
Matthew Wilcox				Intel Open Source Technology Centre
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."
#3 lore raw
Quoted 16 lines · Matthew Wilcox
Quoted 11 lines · Linus Torvalds
Quoted 3 lines · Matthew Wilcox

These are the PCI fixes that have been piling up for the last week while Jesse's been on vacation.  I don't think there's anything questionable here.  All the patches have been posted to linux-pci.

Grrr. It doesn't even compile cleanly:

drivers/pci/quirks.c: In function ?ht_check_msi_mapping?:
drivers/pci/quirks.c:2142: warning: ?return? with no value, in function returning non-void

and yes, it's a real bug.

Agreed, it's a real bug.  I also think it should return 0, but I want Prakash to verify that, since I don't have a system of this nature.

Quoted 7 lines

I wish this warning were an error.  According to the GCC docs, -Werror=return-type should do this, but it seems to have no effect.

Quoted 1 line

I'm not happy.

Me neither.  I'm sorry I didn't catch the warning.

please don't use that one. and use "pci: don't disable too many HT MSI mapping -v4" instead

http://lkml.org/lkml/2009/3/5/458

YH

#4 lore raw
Quoted 2 lines · Yinghai Lu

please don't use that one. and use "pci: don't disable too many HT MSI mapping -v4" instead

I already said I'm not comfortable applying a patch that intrusive at this stage of -rc7. I don't understand it nearly well enough, and given that v1 ended up *not* solving the problem, and neither did v2 or v3, I'm not convinced you understand it well enough either. I think v4 is appropriate for applying during the 2.6.30 merge window.

If you can find someone who really does understand it to review it and agree it's OK for a post-rc7 merge, I'll consider changing my mind.

Signature · 4 lines
Matthew Wilcox				Intel Open Source Technology Centre
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."
#5 lore raw
Quoted 4 lines · Matthew Wilcox
Quoted 2 lines · Yinghai Lu

please don't use that one. and use "pci: don't disable too many HT MSI mapping -v4" instead

Quoted 9 lines

I already said I'm not comfortable applying a patch that intrusive at this stage of -rc7. I don't understand it nearly well enough, and given that v1 ended up *not* solving the problem, and neither did v2 or v3, I'm not convinced you understand it well enough either. I think v4 is appropriate for applying during the 2.6.30 merge window.

If you can find someone who really does understand it to review it and agree it's OK for a post-rc7 merge, I'll consider changing my mind.

OK
v4 for 2.6.30. and later could push that back to 29 stabe later if needed.

YH

#6 lore raw
In reply to Linus Torvalds
Quoted 11 lines · On Mittwoch 11 März 2009 20:22:56 Linus Torvalds wrote:
Quoted 3 lines · Matthew Wilcox

These are the PCI fixes that have been piling up for the last week while Jesse's been on vacation. I don't think there's anything questionable here. All the patches have been posted to linux-pci.

Grrr. It doesn't even compile cleanly:

drivers/pci/quirks.c: In function ‘ht_check_msi_mapping’:
drivers/pci/quirks.c:2142: warning: ‘return’ with no value, in function returning non-void

Quoted 8 lines

and yes, it's a real bug.

I'm not happy.

I assume the proper fix is this trivial one-liner, but people should double-check.

Not pulled.

As I explained to Matthew Wilcox my small patch was incorrectly applied. It originally was in

static void __devinit __nv_msi_ht_cap_quirk(struct pci_dev *dev, int all)

I don't know who or how it moved into

static int __devinit ht_check_msi_mapping(struct pci_dev *dev)

where it breaks for obvious reasons.

Quoted 26 lines

Linus

---
 drivers/pci/quirks.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/pci/quirks.c +1 -1
index a13f3d7..4ba55d0 100644
@@ -2139,7 +2139,7 @@ static int __devinit ht_check_msi_mapping(struct
 pci_dev *dev)
 
 	/* Enabling HT MSI mapping on this device breaks MCP51 */
 	if (dev->device == 0x270)
-		return;
+		return 0;
 
 	/* check if there is HT MSI cap or enabled on this device */

pos = pci_find_ht_capability(dev, HT_CAPTYPE_MSI_MAPPING);
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#7 lore raw
In reply to Linus Torvalds
Quoted 11 lines · Linus Torvalds
Quoted 3 lines · Matthew Wilcox

These are the PCI fixes that have been piling up for the last week while Jesse's been on vacation. I don't think there's anything questionable here. All the patches have been posted to linux-pci.

Grrr. It doesn't even compile cleanly:

drivers/pci/quirks.c: In function ?ht_check_msi_mapping?:
drivers/pci/quirks.c:2142: warning: ?return? with no value, in function returning non-void

and yes, it's a real bug.

Sigh. I just went back through my logs of what I did ... and it's all my fault. The patch from Prakash didn't apply cleanly, and rather than kick it back to him and ask him to provide a patch that did, I tried to apply it by hand. And I got it wrong.

So I've adjusted it, and pushed out a new git tree:

The following changes since commit 16b71fdf97599f1b1b7f38418ee9922d9f117396:
Samuel CUELLA (1):
i810: fix kernel crash fix when struct fb_var_screeninfo is supplied

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/willy/pci.git fixes-20090312

Alex Chiang (2):
PCIe: AER: during disable, check subordinate before walking
PCIe: portdrv: call pci_disable_device during remove

Alexander Duyck (1):
PCI: Add PCI quirk to disable L0s ASPM state for 82575 and 82598

Michael Ellerman (1):
powerpc/pseries: The RPA PCI hotplug driver depends on EEH

Prakash Punnoor (2):
pci: don't disable too many HT MSI mapping
pci: Fix typo in message while disabling HT MSI mapping

 drivers/pci/hotplug/Kconfig        |    2 +-
 drivers/pci/pcie/aer/aerdrv_core.c |    3 +++
 drivers/pci/pcie/portdrv_pci.c     |    1 +
 drivers/pci/quirks.c               |   31 ++++++++++++++++++++++++++++++-
 4 files changed, 35 insertions(+), 2 deletions(-)
Signature · 4 lines
Matthew Wilcox				Intel Open Source Technology Centre
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."