Skip to contentSkip to conversation

IB/hfi: Only read capability registers if the capability exists

PATCH In review Dennis Dalessandro 2 participants 3 messages Dec 22, 2017 1 message not in the archive

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 IB/hfi: Only read capability registers if the capability exists

From: Michael J. Ruhl <[email protected]>

During driver init, various registers are saved to allow restoration after an FLR or gen3 bump. Some of these registers are not available in some circumstances (i.e. Virtual machines).

Delete unnecessary register read/write, and only access register if the capability exists.

Cc <[email protected]> # 4.14.x
Fixes a618b7e40af2 ("IB/hfi1: Move saving PCI values to a separate function")
---
 drivers/infiniband/hw/hfi1/hfi.h  |    1 -
 drivers/infiniband/hw/hfi1/pcie.c |   30 ++++++++++++------------------
 2 files changed, 12 insertions(+), 19 deletions(-)
drivers/infiniband/hw/hfi1/hfi.h +0 -1
index dc5b133..2fa651c 100644
@@ -1131,7 +1131,6 @@ struct hfi1_devdata {
 	u16 pcie_lnkctl;
 	u16 pcie_devctl2;
 	u32 pci_msix0;
-	u32 pci_lnkctl3;
 	u32 pci_tph2;
 
 	/*
drivers/infiniband/hw/hfi1/pcie.c +12 -18
index 09e50fd..8c7e7a6 100644
@@ -411,15 +411,12 @@ int restore_pci_variables(struct hfi1_devdata *dd)
 	if (ret)
 		goto error;
 
-	ret = pci_write_config_dword(dd->pcidev, PCIE_CFG_SPCIE1,
-				     dd->pci_lnkctl3);
-	if (ret)
-		goto error;
-
-	ret = pci_write_config_dword(dd->pcidev, PCIE_CFG_TPH2, dd->pci_tph2);
-	if (ret)
-		goto error;
-
+	if (pci_find_ext_capability(dd->pcidev, PCI_EXT_CAP_ID_TPH)) {
+		ret = pci_write_config_dword(dd->pcidev, PCIE_CFG_TPH2,
+					     dd->pci_tph2);
+		if (ret)
+			goto error;
+	}
 	return 0;
 
 error:
@@ -469,15 +466,12 @@ int save_pci_variables(struct hfi1_devdata *dd)
 	if (ret)
 		goto error;
 
-	ret = pci_read_config_dword(dd->pcidev, PCIE_CFG_SPCIE1,
-				    &dd->pci_lnkctl3);
-	if (ret)
-		goto error;
-
-	ret = pci_read_config_dword(dd->pcidev, PCIE_CFG_TPH2, &dd->pci_tph2);
-	if (ret)
-		goto error;
-
+	if (pci_find_ext_capability(dd->pcidev, PCI_EXT_CAP_ID_TPH)) {
+		ret = pci_read_config_dword(dd->pcidev, PCIE_CFG_TPH2,
+					    &dd->pci_tph2);
+		if (ret)
+			goto error;
+	}
 	return 0;
 
 error:

A message that is not in the archive (sent off-list or not yet received): 20171222164717.31842.12554.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org

#2 lore raw
Quoted 9 lines · Dennis Dalessandro

From: Michael J. Ruhl <[email protected]>

During driver init, various registers are saved to allow restoration after an FLR or gen3 bump. Some of these registers are not available in some circumstances (i.e. Virtual machines).

Delete unnecessary register read/write, and only access register if the capability exists.

Quoted 8 lines
Cc <[email protected]> # 4.14.x
Fixes a618b7e40af2 ("IB/hfi1: Move saving PCI values to a separate function")
 drivers/infiniband/hw/hfi1/hfi.h  |    1 -
 drivers/infiniband/hw/hfi1/pcie.c |   30 ++++++++++++------------------
 2 files changed, 12 insertions(+), 19 deletions(-)

Since we are getting late in the RC cycle can you help me clarify the user visible bug here? No need to resend

I'm guessing when hfi1 is used in a virtual machine with passthrough (SRIOV?) it fails in ?? way?

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw
Quoted 19 lines · Jason Gunthorpe
Quoted 9 lines · Dennis Dalessandro

From: Michael J. Ruhl <[email protected]>

During driver init, various registers are saved to allow restoration after an FLR or gen3 bump. Some of these registers are not available in some circumstances (i.e. Virtual machines).

Delete unnecessary register read/write, and only access register if the capability exists.

Quoted 8 lines
Cc <[email protected]> # 4.14.x
Fixes a618b7e40af2 ("IB/hfi1: Move saving PCI values to a separate function")
  drivers/infiniband/hw/hfi1/hfi.h  |    1 -
  drivers/infiniband/hw/hfi1/pcie.c |   30 ++++++++++++------------------
  2 files changed, 12 insertions(+), 19 deletions(-)
Quoted 5 lines

Since we are getting late in the RC cycle can you help me clarify the user visible bug here? No need to resend

I'm guessing when hfi1 is used in a virtual machine with passthrough (SRIOV?) it fails in ?? way?

Ugh. I actually amended the commit message to include a blurb on the impact. Sorry about that, not sure what happened. It should have had this line of text:

"
Without this patch the driver will not load on a virtual machine. "

So basically we just don't work on VMs. There is no panic or anything. I was on the fence whether it could go for this late of an -rc or not.

-Denny