Xen/PCIback: PCI reset using 'reset' SysFS attribute
Subsystems (2)
v3 cover Xen/PCIback: PCI reset using 'reset' SysFS attribute
This patch contains Xen pciback driver changes to support PCI reset (flr/slot/bus) based on SysFS 'reset' attribute. The following Xen libxl patch depends on these kernel patches.
- Xen/libxl: Perform PCI reset using 'reset' SysFS attribute
Govinda Tatti (2):
Drivers/PCI: Export pcie_has_flr() interface Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS
attribute
Documentation/ABI/testing/sysfs-driver-pciback | 15 +++ drivers/pci/pci.c | 3 +- drivers/xen/xen-pciback/pci_stub.c | 128 +++++++++++++++++++++++++ include/linux/pci.h | 1 + 4 files changed, 146 insertions(+), 1 deletion(-)
Signature · 1 line
2.9.5
PATCH v3 1/2 Drivers/PCI: Export pcie_has_flr() interface
This patch exports pcie_has_flr() and it is being used by Xen pciback driver to reset (flr/slot/bus) PCI devices based on 'reset' SysFS attribute.
---
v3: -New
drivers/pci/pci.c | 3 ++- include/linux/pci.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)
drivers/pci/pci.c +2 -1
@@ -3872,7 +3872,7 @@ static void pci_flr_wait(struct pci_dev *dev)* Returns true if the device advertises support for PCIe function level * resets. */ -static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;
@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);/** * pcie_flr - initiate a PCIe function level reset
include/linux/pci.h +1 -0
@@ -1089,6 +1089,7 @@ int pcie_get_mps(struct pci_dev *dev);int pcie_set_mps (struct pci_dev *dev,int mps);int pcie_get_minimum_link (struct pci_dev *dev,enum pci_bus_speed *speed,enum pcie_link_width *width); +bool pcie_has_flr (struct pci_dev *dev);void pcie_flr (struct pci_dev *dev);int __pci_reset_function (struct pci_dev *dev);int __pci_reset_function_locked (struct pci_dev *dev);
Signature · 7 lines
2.9.5 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 26 lines · Govinda Tatti
This patch exports
pcie_has_flr()and it is being used by Xen pciback driver to reset (flr/slot/bus) PCI devices based on 'reset' SysFS attribute.Signed-off-by Govinda Tatti <[email protected]>---
v3: -Newdrivers/pci/pci.c | 3 ++- include/linux/pci.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)drivers/pci/pci.c +1
-1@@ -3872,7 +3872,7 @@ static void pci_flr_wait(struct pci_dev *dev)* Returns true if the device advertises support for PCIe function level * resets. */ -static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 4 lines
drivers/pci/pci.c +1
-0@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);
I'd rather change pcie_flr() so you could *always* call it, and it would return 0, -ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
I don't like the "Can I do this? Ok, do this" style of interfaces. It's racy (not really applicable in this case) and seems clunky.
Quoted 17 lines
drivers/pci/pci.c +0
-0/** * pcie_flr - initiate a PCIe function level reset include/linux/pci.h +1
-0@@ -1089,6 +1089,7 @@ int pcie_get_mps(struct pci_dev *dev);int pcie_set_mps (struct pci_dev *dev,int mps);int pcie_get_minimum_link (struct pci_dev *dev,enum pci_bus_speed *speed,enum pcie_link_width *width); +bool pcie_has_flr (struct pci_dev *dev);void pcie_flr (struct pci_dev *dev);int __pci_reset_function (struct pci_dev *dev);int __pci_reset_function_locked (struct pci_dev *dev);Signature · 1 line
2.9.5
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Thanks Bjorn for your review comments. Please see below for my comments.
Quoted 26 lines · Bjorn Helgaas
Quoted 25 lines · Govinda Tatti
This patch exports
pcie_has_flr()and it is being used by Xen pciback driver to reset (flr/slot/bus) PCI devices based on 'reset' SysFS attribute.Signed-off-by Govinda Tatti <[email protected]>---
v3: -Newdrivers/pci/pci.c | 3 ++- include/linux/pci.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)drivers/pci/pci.c +1
-1@@ -3872,7 +3872,7 @@ static void pci_flr_wait(struct pci_dev *dev)* Returns true if the device advertises support for PCIe function level * resets. */ -static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;
Quoted 8 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 5 lines
drivers/pci/pci.c +1
-0@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
Sure, I will add pcie_has_flr() logic inside pcie_flr() and return appropriate values as suggested by you. Do we still want to retain pcie_has_flr() and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.
Please let me know.
Cheers
GOVINDA
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 37 lines · Govinda Tatti
Thanks Bjorn for your review comments. Please see below for my comments.
Quoted 25 lines · Bjorn Helgaas
Quoted 24 lines · Govinda Tatti
This patch exports
pcie_has_flr()and it is being used by Xen pciback driver to reset (flr/slot/bus) PCI devices based on 'reset' SysFS attribute.Signed-off-by Govinda Tatti <[email protected]>---
v3: -Newdrivers/pci/pci.c | 3 ++- include/linux/pci.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)drivers/pci/pci.c +1
-1@@ -3872,7 +3872,7 @@ static void pci_flr_wait(struct pci_dev *dev)* Returns true if the device advertises support for PCIe function level * resets. */ -static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;Quoted 8 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 5 lines
drivers/pci/pci.c +1
-0@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate
values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.
If you can restructure the code and remove pcie_has_flr() while retaining the existing behavior of its callers, that would be great.
Bjorn
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 10 lines
drivers/pci/pci.c +2
-1-static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
Quoted 6 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate
values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.
I checked the current usage of pcie_has_flr() and pcie_flr(). I have a couple of questions or need some clarification.
1. pcie_has_flr() usage inside pci_probe_reset_function().
This function is only calling pcie_has_flr() but not pcie_flr(). Rest of the code is trying to do specific type of reset except pcie_flr().
rc = pci_dev_specific_reset(dev, 1);
if (rc != -ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc = pci_af_flr(dev, 1);
if (rc != -ENOTTY)
return rc;
In other-words, I can remove usage of pcie_has_flr() in all other places in pci.c except in above function.
2. W.r.t pcie_flr(), I am planning to return error code. Currently, the following file/modules are calling this function. My plan is to add a check for return code and print a WANRING message if return code is NON-ZERO. I hope this is sufficient for this patch.
drivers/crypto/qat/qat_common/adf_aer.c drivers/infiniband/hw/hfi1/chip.c (2 places) drivers/net/ethernet/cavium/liquidio/lio_vf_main.c drivers/net/ethernet/intel/ixgbe/ixgbe_main.c (2 places) drivers/pci/quirks.c (2 places)
Cheers
GOVINDA
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
[+cc Christoph]
Quoted 38 lines · Govinda Tatti
Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 10 lines
drivers/pci/pci.c +2
-1-static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?Quoted 6 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate
values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.I checked the current usage of
pcie_has_flr()andpcie_flr(). I have a couple
of questions or need some clarification.1.
pcie_has_flr()usage insidepci_probe_reset_function().This function is only calling
pcie_has_flr()but notpcie_flr(). Rest of the code is trying to do specific type of reset exceptpcie_flr().rc =
pci_dev_specific_reset(dev, 1);
if (rc !=-ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc =pci_af_flr(dev, 1);
if (rc !=-ENOTTY)
return rc;
Quoted 2 lines
In other-words, I can remove usage of
pcie_has_flr()in all other places inpci.cexcept in above function.
I think we should keep the EXPORT_SYMBOL_GPL() part of a60a2b73ba69 ("PCI: Export pcie_flr()"), but revert the restructuring part.
Prior to a60a2b73ba69, we had
int pcie_flr(struct pci_dev *dev, int probe);
like all the other reset methods. AFAICT, the addition of pcie_has_flr() was to optimize the path slightly because when drivers call pcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and set PCI_DEV_FLAGS_NO_FLR_RESET accordingly.
Christoph, chime in if I'm missing something here.
Quoted 8 lines
2. W.r.t
pcie_flr(), I am planning to return error code. Currently, the following
file/modules are calling this function. My plan is to add a check for return
code and print a WANRING message if return code is NON-ZERO. I hope this is
sufficient for this patch.
Quoted 5 lines
drivers/crypto/qat/qat_common/adf_aer.cdrivers/infiniband/hw/hfi1/chip.c(2 places)drivers/net/ethernet/cavium/liquidio/lio_vf_main.cdrivers/net/ethernet/intel/ixgbe/ixgbe_main.c(2 places)drivers/pci/quirks.c(2 places)
Checking the return code is probably overkill, since pcie_flr() is void and returns no errors now. The only point of the return value is to tell whether the device supports FLR. If we call it with "probe == 0" there's no useful error to return.
Bjorn
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 12 lines · Bjorn Helgaas
Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.
Quoted 1 line
Christoph, chime in if I'm missing something here.
Didn't we just have that discussion in another thread a few days ago? I think that the pcie_has_flr was a mistake in retrospective but I think the bool probe API was an even bigger mistake. The only use of it is to hide the reset attribute in sysfs. I'd much rather always have it and have it return EOPNOTSUPP if no reset method is supported.
I can send a patch for that if it sounds fine to you.
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 13 lines · Christoph Hellwig
Quoted 12 lines · Bjorn Helgaas
Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.
Quoted 4 lines
Quoted 1 line
Christoph, chime in if I'm missing something here.
Didn't we just have that discussion in another thread a few days ago?
Probably, but I didn't have a clear picture of what you were suggesting.
Quoted 6 lines
I think that the
pcie_has_flrwas a mistake in retrospective but I think the bool probe API was an even bigger mistake. The only use of it is to hide the reset attribute in sysfs. I'd much rather always have it and have it return EOPNOTSUPP if no reset method is supported.I can send a patch for that if it sounds fine to you.
If you can get rid of the whole probe infrastructure, that sounds good to me.
Bjorn
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Christoph HellwigQuoted 13 lines · Christoph Hellwig
Quoted 12 lines · Bjorn Helgaas
Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.
Quoted 4 lines
Quoted 1 line
Christoph, chime in if I'm missing something here.
Didn't we just have that discussion in another thread a few days ago?
Probably, but I didn't have a clear picture of what you were suggesting.
Quoted 6 lines
I think that the
pcie_has_flrwas a mistake in retrospective but I think the bool probe API was an even bigger mistake. The only use of it is to hide the reset attribute in sysfs. I'd much rather always have it and have it return EOPNOTSUPP if no reset method is supported.I can send a patch for that if it sounds fine to you.
If you can get rid of the whole probe infrastructure, that sounds good to me.
Bjorn
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 12 lines · Bjorn Helgaas
Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.
Quoted 1 line
Christoph, chime in if I'm missing something here.
Didn't we just have that discussion in another thread a few days ago? I think that the pcie_has_flr was a mistake in retrospective but I think the bool probe API was an even bigger mistake. The only use of it is to hide the reset attribute in sysfs. I'd much rather always have it and have it return EOPNOTSUPP if no reset method is supported.
I can send a patch for that if it sounds fine to you.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasThanks Bjorn and Christophfor your response. Please see below for my comments.
Quoted 57 lines · Bjorn Helgaas
[+cc Christoph]
Quoted 37 lines · Govinda Tatti
Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 9 lines
drivers/pci/pci.c +2
-1-static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?Quoted 6 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate
values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.I checked the current usage of
pcie_has_flr()andpcie_flr(). I have a couple
of questions or need some clarification.1.
pcie_has_flr()usage insidepci_probe_reset_function().This function is only calling
pcie_has_flr()but notpcie_flr(). Rest of the code is trying to do specific type of reset exceptpcie_flr().rc =
pci_dev_specific_reset(dev, 1);
if (rc !=-ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc =pci_af_flr(dev, 1);
if (rc !=-ENOTTY)
return rc;Quoted 2 lines
In other-words, I can remove usage of
pcie_has_flr()in all other places inpci.cexcept in above function.I think we should keep the
EXPORT_SYMBOL_GPL()part ofa60a2b73ba69("PCI: Exportpcie_flr()"), but revert the restructuring part.Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.
Quoted 1 line
Christoph, chime in if I'm missing something here.
Not all code paths are aware of FLR capability and also, not using pcie_flr(). For example,
arch/powerpc/platforms/powernv/eeh-powernv.c drivers/crypto/cavium/nitrox/nitrox_main.c drivers/net/ethernet/cavium/liquidio/octeon_mailbox.c
So, we should consider one of these options.
- set PCI_DEV_FLAGS_NO_FLR_RESET if it is not supported.
- pcie_flr() should return if it is not supported
If we modify pcie_flr() to return error codes, then we need to modify all existing modules that are calling this function.
Please let me know your preference, so that I can move accordingly. Thanks.
Cheers
GOVINDA
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
[+cc Russell, Sinan, Herbert, Srikanth, Derek, Satanand, Felix, Raghu]
Quoted 62 lines · Govinda Tatti
Quoted 53 lines · Bjorn Helgaas
Quoted 41 lines · Govinda Tatti
Quoted 14 lines
Quoted 13 lines
Quoted 13 lines
Review static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 9 lines
(quoted diff) +2
-1-static bool pcie_has_flr(struct pci_dev *dev) +bool pcie_has_flr(struct pci_dev *dev) { u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev) pcie_capability_read_dword(dev, PCI_EXP_DEVCAP, &cap); return cap & PCI_EXP_DEVCAP_FLR; } +EXPORT_SYMBOL_GPL(pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?Quoted 8 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.I checked the current usage of
pcie_has_flr()andpcie_flr(). I have a couple of questions or need some clarification.1.
pcie_has_flr()usage insidepci_probe_reset_function().This function is only calling
pcie_has_flr()but notpcie_flr(). Rest of the code is trying to do specific type of reset except
pcie_flr().rc =
pci_dev_specific_reset(dev, 1);
if (rc !=-ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc =pci_af_flr(dev, 1);
if (rc !=-ENOTTY)
return rc;Quoted 2 lines
In other-words, I can remove usage of
pcie_has_flr()in all
other places inpci.cexcept in above function.I think we should keep the
EXPORT_SYMBOL_GPL()part ofa60a2b73ba69("PCI: Exportpcie_flr()"), but revert the restructuring part.Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);Quoted 7 lines
like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.
Quoted 4 lines
Not all code paths are aware of FLR capability and also, not using pcie_flr(). For example,
arch/powerpc/platforms/powernv/eeh-powernv.c
I assume you're referring to pnv_eeh_do_flr() (which contains code similar to pcie_flr()) and pnv_eeh_do_af_flr() (which has code similar to pci_af_flr()). I agree that those are problematic and would ideally be unified with the PCI core implementations.
Powerpc has quite a bit of this sort of special-case code for several reasons, some just historical and some more concrete, so I don't know how feasible this is.
Quoted 1 line
drivers/crypto/cavium/nitrox/nitrox_main.c
This has nitrox_reset_device(), which should definitely be replaced with a core interface.
Quoted 1 line
drivers/net/ethernet/cavium/liquidio/octeon_mailbox.c
And this has octeon_mbox_process_cmd() which also does a home-grown PCI_EXP_DEVCTL_BCR_FLR request and also should definitely use a core interface.
Quoted 7 lines
So, we should consider one of these options.
- set
PCI_DEV_FLAGS_NO_FLR_RESETif it is not supported.
-pcie_flr()should return if it is not supportedIf we modify
pcie_flr()to return error codes, then we need to modify all existing modules that are calling this function.
Yes, of course.
Quoted 1 line
Please let me know your preference, so that I can move accordingly. Thanks.
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
The other paths that use PCI_EXP_DEVCTL_BCR_FLR are definitely issues and should be fixed, but again should wait for the revised pcie_flr() interface. And if they're not actually required for your Xen issue, they sound like "nice to have" cleanups that will not gate your Xen fixes. I added this to my ever-growing list of cleanups to do.
Bjorn
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Thanks Bjorn for your response. Please see below for my comments.
Quoted 5 lines
Quoted 5 lines
So, we should consider one of these options.
- set
PCI_DEV_FLAGS_NO_FLR_RESETif it is not supported.
-pcie_flr()should return if it is not supported
Quoted 8 lines
Quoted 2 lines
If we modify
pcie_flr()to return error codes, then we need to modify all existing modules that are calling this function.Yes, of course.
Quoted 1 line
Please let me know your preference, so that I can move accordingly. Thanks.
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
OK.
Quoted 6 lines
The other paths that use
PCI_EXP_DEVCTL_BCR_FLRare definitely issues and should be fixed, but again should wait for the revisedpcie_flr()interface. And if they're not actually required for your Xen issue, they sound like "nice to have" cleanups that will not gate your Xen fixes. I added this to my ever-growing list of cleanups to do.
For now, I am planning to use existing pcie_flr() after checking FLR capability inside Xenpciback driver (like other existing pcie_flr() usage). We will switch to revised pcie_flr() once it is available.
Cheers
GOVINDA
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasThanks Bjorn for your response. Please see below for my comments.
Quoted 5 lines
Quoted 5 lines
So, we should consider one of these options.
- set
PCI_DEV_FLAGS_NO_FLR_RESETif it is not supported.
-pcie_flr()should return if it is not supported
Quoted 8 lines
Quoted 2 lines
If we modify
pcie_flr()to return error codes, then we need to modify all existing modules that are calling this function.Yes, of course.
Quoted 1 line
Please let me know your preference, so that I can move accordingly. Thanks.
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
OK.
Quoted 6 lines
The other paths that use
PCI_EXP_DEVCTL_BCR_FLRare definitely issues and should be fixed, but again should wait for the revisedpcie_flr()interface. And if they're not actually required for your Xen issue, they sound like "nice to have" cleanups that will not gate your Xen fixes. I added this to my ever-growing list of cleanups to do.
For now, I am planning to use existing pcie_flr() after checking FLR capability inside Xenpciback driver (like other existing pcie_flr() usage). We will switch to revised pcie_flr() once it is available.
Cheers
GOVINDA
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 70 lines · Bjorn Helgaas
[+cc Russell, Sinan, Herbert, Srikanth, Derek, Satanand, Felix, Raghu]
Quoted 62 lines · Govinda Tatti
Quoted 53 lines · Bjorn Helgaas
Quoted 41 lines · Govinda Tatti
Quoted 14 lines
Quoted 13 lines
Quoted 13 lines
Review static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 9 lines
(quoted diff) +2
-1-static bool pcie_has_flr(struct pci_dev *dev) +bool pcie_has_flr(struct pci_dev *dev) { u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev) pcie_capability_read_dword(dev, PCI_EXP_DEVCAP, &cap); return cap & PCI_EXP_DEVCAP_FLR; } +EXPORT_SYMBOL_GPL(pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?Quoted 8 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.I checked the current usage of
pcie_has_flr()andpcie_flr(). I have a couple of questions or need some clarification.1.
pcie_has_flr()usage insidepci_probe_reset_function().This function is only calling
pcie_has_flr()but notpcie_flr(). Rest of the code is trying to do specific type of reset except
pcie_flr().rc =
pci_dev_specific_reset(dev, 1);
if (rc !=-ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc =pci_af_flr(dev, 1);
if (rc !=-ENOTTY)
return rc;Quoted 2 lines
In other-words, I can remove usage of
pcie_has_flr()in all
other places inpci.cexcept in above function.I think we should keep the
EXPORT_SYMBOL_GPL()part ofa60a2b73ba69("PCI: Exportpcie_flr()"), but revert the restructuring part.Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);Quoted 7 lines
like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.Quoted 4 lines
Not all code paths are aware of FLR capability and also, not using pcie_flr(). For example,
arch/powerpc/platforms/powernv/eeh-powernv.c
Quoted 8 lines
I assume you're referring to
pnv_eeh_do_flr()(which contains code similar topcie_flr()) andpnv_eeh_do_af_flr()(which has code similar topci_af_flr()). I agree that those are problematic and would ideally be unified with the PCI core implementations.Powerpc has quite a bit of this sort of special-case code for several reasons, some just historical and some more concrete, so I don't know how feasible this is.
It would be lovely if pnv-eeh code used pci_af_flr() but since pnv_eeh_do_flr() uses different config space accessors (not sure why exactly, probably to avoid freezing the entire PHB), it is harder than just trivial change. I'll try and have a deeper look though.
Signature · 1 line
Alexey
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 70 lines · Bjorn Helgaas
[+cc Russell, Sinan, Herbert, Srikanth, Derek, Satanand, Felix, Raghu]
Quoted 62 lines · Govinda Tatti
Quoted 53 lines · Bjorn Helgaas
Quoted 41 lines · Govinda Tatti
Quoted 14 lines
Quoted 13 lines
Quoted 13 lines
Review static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 9 lines
(quoted diff) +2
-1-static bool pcie_has_flr(struct pci_dev *dev) +bool pcie_has_flr(struct pci_dev *dev) { u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev) pcie_capability_read_dword(dev, PCI_EXP_DEVCAP, &cap); return cap & PCI_EXP_DEVCAP_FLR; } +EXPORT_SYMBOL_GPL(pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?Quoted 8 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.I checked the current usage of
pcie_has_flr()andpcie_flr(). I have a couple of questions or need some clarification.1.
pcie_has_flr()usage insidepci_probe_reset_function().This function is only calling
pcie_has_flr()but notpcie_flr(). Rest of the code is trying to do specific type of reset except
pcie_flr().rc =
pci_dev_specific_reset(dev, 1);
if (rc !=-ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc =pci_af_flr(dev, 1);
if (rc !=-ENOTTY)
return rc;Quoted 2 lines
In other-words, I can remove usage of
pcie_has_flr()in all
other places inpci.cexcept in above function.I think we should keep the
EXPORT_SYMBOL_GPL()part ofa60a2b73ba69("PCI: Exportpcie_flr()"), but revert the restructuring part.Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);Quoted 7 lines
like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.Quoted 4 lines
Not all code paths are aware of FLR capability and also, not using pcie_flr(). For example,
arch/powerpc/platforms/powernv/eeh-powernv.c
Quoted 8 lines
I assume you're referring to
pnv_eeh_do_flr()(which contains code similar topcie_flr()) andpnv_eeh_do_af_flr()(which has code similar topci_af_flr()). I agree that those are problematic and would ideally be unified with the PCI core implementations.Powerpc has quite a bit of this sort of special-case code for several reasons, some just historical and some more concrete, so I don't know how feasible this is.
It would be lovely if pnv-eeh code used pci_af_flr() but since pnv_eeh_do_flr() uses different config space accessors (not sure why exactly, probably to avoid freezing the entire PHB), it is harder than just trivial change. I'll try and have a deeper look though.
Signature · 6 lines
Alexey _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 3 lines · Bjorn Helgaas
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
I'll have some time over the holidays. If you need it more urgent than that feel free to take over.
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 6 lines · Christoph Hellwig
Quoted 3 lines · Bjorn Helgaas
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
I'll have some time over the holidays. If you need it more urgent than that feel free to take over.
We will wait for your changes. Thanks Christoph.
Cheers
GOVINDA
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Christoph HellwigQuoted 6 lines · Christoph Hellwig
Quoted 3 lines · Bjorn Helgaas
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
I'll have some time over the holidays. If you need it more urgent than that feel free to take over.
We will wait for your changes. Thanks Christoph.
Cheers
GOVINDA
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Christoph HellwigQuoted 8 lines · Christoph Hellwig
Quoted 3 lines · Bjorn Helgaas
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
I'll have some time over the holidays. If you need it more urgent than that feel free to take over.
Replying to an old thread..
I noticed pcie_has_flr() has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when pcie_has_flr() wasn't exported at the time)
Thanks,
-- Pasi
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 6 lines · Pasi Kärkkäinen
I noticed
pcie_has_flr()has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when
pcie_has_flr()wasn't exported at the time)
Exporting pcie_has_flr() is a very simple change which could have been done by the XEN porting effort.
Maybe, the right question is what is so special about XEN reset?
What feature PCI core is missing to support XEN FLR reset that caused the effort to stall?
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Hi,
Quoted 8 lines · Sinan Kaya
Quoted 6 lines · Pasi Kärkkäinen
I noticed
pcie_has_flr()has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when
pcie_has_flr()wasn't exported at the time)
Quoted 8 lines
Exporting
pcie_has_flr()is a very simple change which could have been done by the XEN porting effort.Maybe, the right question is what is so special about XEN reset?
What feature PCI core is missing to support XEN FLR reset that caused the effort to stall?
Well one of the reasons probably was because Christoph was planning to deprecate the pcie_has_flr() functionality..
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01057.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01252.html
But now that pcie_has_flr() is exported and available I guess it's fine to continue using it also for xen-pciback :)
Thanks,
-- Pasi
Re: [Xen-devel] [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
Quoted 20 lines · Pasi Kärkkäinen
Hi,
Quoted 8 lines · Sinan Kaya
Quoted 6 lines · Pasi Kärkkäinen
I noticed
pcie_has_flr()has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when
pcie_has_flr()wasn't exported at the time)Quoted 8 lines
Exporting
pcie_has_flr()is a very simple change which could have been done by the XEN porting effort.Maybe, the right question is what is so special about XEN reset?
What feature PCI core is missing to support XEN FLR reset that caused the effort to stall?
Quoted 7 lines
Well one of the reasons probably was because Christoph was planning to deprecate the
pcie_has_flr()functionality..https://lists.xen.org/archives/html/xen-devel/2017-12/msg01057.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01252.htmlBut now that
pcie_has_flr()is exported and available I guess it's fine to continue using it also for xen-pciback :)
Yeah, I would go ahead with the implementation. Refactoring can be done independently.
Quoted 6 lines
Thanks,
-- Pasi
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Pasi KärkkäinenQuoted 20 lines · Pasi Kärkkäinen
Hi,
Quoted 8 lines · Sinan Kaya
Quoted 6 lines · Pasi Kärkkäinen
I noticed
pcie_has_flr()has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when
pcie_has_flr()wasn't exported at the time)Quoted 8 lines
Exporting
pcie_has_flr()is a very simple change which could have been done by the XEN porting effort.Maybe, the right question is what is so special about XEN reset?
What feature PCI core is missing to support XEN FLR reset that caused the effort to stall?
Quoted 7 lines
Well one of the reasons probably was because Christoph was planning to deprecate the
pcie_has_flr()functionality..https://lists.xen.org/archives/html/xen-devel/2017-12/msg01057.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01252.htmlBut now that
pcie_has_flr()is exported and available I guess it's fine to continue using it also for xen-pciback :)
Yeah, I would go ahead with the implementation. Refactoring can be done independently.
Quoted 6 lines
Thanks,
-- Pasi
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Sinan KayaHi,
Quoted 8 lines · Sinan Kaya
Quoted 6 lines · Pasi Kärkkäinen
I noticed
pcie_has_flr()has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when
pcie_has_flr()wasn't exported at the time)
Quoted 8 lines
Exporting
pcie_has_flr()is a very simple change which could have been done by the XEN porting effort.Maybe, the right question is what is so special about XEN reset?
What feature PCI core is missing to support XEN FLR reset that caused the effort to stall?
Well one of the reasons probably was because Christoph was planning to deprecate the pcie_has_flr() functionality..
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01057.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01252.html
But now that pcie_has_flr() is exported and available I guess it's fine to continue using it also for xen-pciback :)
Thanks,
-- Pasi
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Pasi KärkkäinenQuoted 6 lines · Pasi Kärkkäinen
I noticed
pcie_has_flr()has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when
pcie_has_flr()wasn't exported at the time)
Exporting pcie_has_flr() is a very simple change which could have been done by the XEN porting effort.
Maybe, the right question is what is so special about XEN reset?
What feature PCI core is missing to support XEN FLR reset that caused the effort to stall?
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Christoph HellwigQuoted 8 lines · Christoph Hellwig
Quoted 3 lines · Bjorn Helgaas
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
I'll have some time over the holidays. If you need it more urgent than that feel free to take over.
Replying to an old thread..
I noticed pcie_has_flr() has been recently exported in upstream Linux:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
Are there more changes / cleanups planned to these interfaces, as mentioned last year?
(context: xen-pciback reset/do_flr features upstreaming, which kind of stalled last year when pcie_has_flr() wasn't exported at the time)
Thanks,
-- Pasi
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 3 lines · Bjorn Helgaas
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
I'll have some time over the holidays. If you need it more urgent than that feel free to take over.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Govinda Tatti[+cc Russell, Sinan, Herbert, Srikanth, Derek, Satanand, Felix, Raghu]
Quoted 62 lines · Govinda Tatti
Quoted 53 lines · Bjorn Helgaas
Quoted 41 lines · Govinda Tatti
Quoted 14 lines
Quoted 13 lines
Quoted 13 lines
Review static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 9 lines
(quoted diff) +2
-1-static bool pcie_has_flr(struct pci_dev *dev) +bool pcie_has_flr(struct pci_dev *dev) { u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev) pcie_capability_read_dword(dev, PCI_EXP_DEVCAP, &cap); return cap & PCI_EXP_DEVCAP_FLR; } +EXPORT_SYMBOL_GPL(pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?Quoted 8 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.I checked the current usage of
pcie_has_flr()andpcie_flr(). I have a couple of questions or need some clarification.1.
pcie_has_flr()usage insidepci_probe_reset_function().This function is only calling
pcie_has_flr()but notpcie_flr(). Rest of the code is trying to do specific type of reset except
pcie_flr().rc =
pci_dev_specific_reset(dev, 1);
if (rc !=-ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc =pci_af_flr(dev, 1);
if (rc !=-ENOTTY)
return rc;Quoted 2 lines
In other-words, I can remove usage of
pcie_has_flr()in all
other places inpci.cexcept in above function.I think we should keep the
EXPORT_SYMBOL_GPL()part ofa60a2b73ba69("PCI: Exportpcie_flr()"), but revert the restructuring part.Prior to
a60a2b73ba69, we hadint pcie_flr(struct pci_dev *dev, int probe);Quoted 7 lines
like all the other reset methods. AFAICT, the addition of
pcie_has_flr()was to optimize the path slightly because when drivers callpcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and setPCI_DEV_FLAGS_NO_FLR_RESETaccordingly.
Quoted 4 lines
Not all code paths are aware of FLR capability and also, not using pcie_flr(). For example,
arch/powerpc/platforms/powernv/eeh-powernv.c
I assume you're referring to pnv_eeh_do_flr() (which contains code similar to pcie_flr()) and pnv_eeh_do_af_flr() (which has code similar to pci_af_flr()). I agree that those are problematic and would ideally be unified with the PCI core implementations.
Powerpc has quite a bit of this sort of special-case code for several reasons, some just historical and some more concrete, so I don't know how feasible this is.
Quoted 1 line
drivers/crypto/cavium/nitrox/nitrox_main.c
This has nitrox_reset_device(), which should definitely be replaced with a core interface.
Quoted 1 line
drivers/net/ethernet/cavium/liquidio/octeon_mailbox.c
And this has octeon_mbox_process_cmd() which also does a home-grown PCI_EXP_DEVCTL_BCR_FLR request and also should definitely use a core interface.
Quoted 7 lines
So, we should consider one of these options.
- set
PCI_DEV_FLAGS_NO_FLR_RESETif it is not supported.
-pcie_flr()should return if it is not supportedIf we modify
pcie_flr()to return error codes, then we need to modify all existing modules that are calling this function.
Yes, of course.
Quoted 1 line
Please let me know your preference, so that I can move accordingly. Thanks.
I think Christoph volunteered to do some restructuring, but I don't know his timeframe. If you can, I would probably wait for that because there's so much overlap here.
The other paths that use PCI_EXP_DEVCTL_BCR_FLR are definitely issues and should be fixed, but again should wait for the revised pcie_flr() interface. And if they're not actually required for your Xen issue, they sound like "nice to have" cleanups that will not gate your Xen fixes. I added this to my ever-growing list of cleanups to do.
Bjorn
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Govinda Tatti[+cc Christoph]
Quoted 38 lines · Govinda Tatti
Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 10 lines
drivers/pci/pci.c +2
-1-static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?Quoted 6 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate
values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.I checked the current usage of
pcie_has_flr()andpcie_flr(). I have a couple
of questions or need some clarification.1.
pcie_has_flr()usage insidepci_probe_reset_function().This function is only calling
pcie_has_flr()but notpcie_flr(). Rest of the code is trying to do specific type of reset exceptpcie_flr().rc =
pci_dev_specific_reset(dev, 1);
if (rc !=-ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc =pci_af_flr(dev, 1);
if (rc !=-ENOTTY)
return rc;
Quoted 2 lines
In other-words, I can remove usage of
pcie_has_flr()in all other places inpci.cexcept in above function.
I think we should keep the EXPORT_SYMBOL_GPL() part of a60a2b73ba69 ("PCI: Export pcie_flr()"), but revert the restructuring part.
Prior to a60a2b73ba69, we had
int pcie_flr(struct pci_dev *dev, int probe);
like all the other reset methods. AFAICT, the addition of pcie_has_flr() was to optimize the path slightly because when drivers call pcie_flr(), they should already know that their hardware supports FLR. But I don't think that optimization is worth the extra code complexity. If we do need to optimize it, we can check this in the core during enumeration and set PCI_DEV_FLAGS_NO_FLR_RESET accordingly.
Christoph, chime in if I'm missing something here.
Quoted 8 lines
2. W.r.t
pcie_flr(), I am planning to return error code. Currently, the following
file/modules are calling this function. My plan is to add a check for return
code and print a WANRING message if return code is NON-ZERO. I hope this is
sufficient for this patch.
Quoted 5 lines
drivers/crypto/qat/qat_common/adf_aer.cdrivers/infiniband/hw/hfi1/chip.c(2 places)drivers/net/ethernet/cavium/liquidio/lio_vf_main.cdrivers/net/ethernet/intel/ixgbe/ixgbe_main.c(2 places)drivers/pci/quirks.c(2 places)
Checking the return code is probably overkill, since pcie_flr() is void and returns no errors now. The only point of the return value is to tell whether the device supports FLR. If we call it with "probe == 0" there's no useful error to return.
Bjorn
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 13 lines
Quoted 13 lines
Quoted 13 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 10 lines
drivers/pci/pci.c +2
-1-static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
Quoted 6 lines
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate
values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.If you can restructure the code and remove
pcie_has_flr()while retaining the existing behavior of its callers, that would be great.
I checked the current usage of pcie_has_flr() and pcie_flr(). I have a couple of questions or need some clarification.
1. pcie_has_flr() usage inside pci_probe_reset_function().
This function is only calling pcie_has_flr() but not pcie_flr(). Rest of the code is trying to do specific type of reset except pcie_flr().
rc = pci_dev_specific_reset(dev, 1);
if (rc != -ENOTTY)
return rc;
if (pcie_has_flr(dev))
return 0;
rc = pci_af_flr(dev, 1);
if (rc != -ENOTTY)
return rc;
In other-words, I can remove usage of pcie_has_flr() in all other places in pci.c except in above function.
2. W.r.t pcie_flr(), I am planning to return error code. Currently, the following file/modules are calling this function. My plan is to add a check for return code and print a WANRING message if return code is NON-ZERO. I hope this is sufficient for this patch.
drivers/crypto/qat/qat_common/adf_aer.c drivers/infiniband/hw/hfi1/chip.c (2 places) drivers/net/ethernet/cavium/liquidio/lio_vf_main.c drivers/net/ethernet/intel/ixgbe/ixgbe_main.c (2 places) drivers/pci/quirks.c (2 places)
Cheers
GOVINDA
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Govinda TattiQuoted 37 lines · Govinda Tatti
Thanks Bjorn for your review comments. Please see below for my comments.
Quoted 25 lines · Bjorn Helgaas
Quoted 24 lines · Govinda Tatti
This patch exports
pcie_has_flr()and it is being used by Xen pciback driver to reset (flr/slot/bus) PCI devices based on 'reset' SysFS attribute.Signed-off-by Govinda Tatti <[email protected]>---
v3: -Newdrivers/pci/pci.c | 3 ++- include/linux/pci.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)drivers/pci/pci.c +1
-1@@ -3872,7 +3872,7 @@ static void pci_flr_wait(struct pci_dev *dev)* Returns true if the device advertises support for PCIe function level * resets. */ -static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;Quoted 8 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 5 lines
drivers/pci/pci.c +1
-0@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
Quoted 4 lines
Sure, I will add
pcie_has_flr()logic insidepcie_flr()and return appropriate
values as suggested by you. Do we still want to retainpcie_has_flr()and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.
If you can restructure the code and remove pcie_has_flr() while retaining the existing behavior of its callers, that would be great.
Bjorn
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasThanks Bjorn for your review comments. Please see below for my comments.
Quoted 26 lines · Bjorn Helgaas
Quoted 25 lines · Govinda Tatti
This patch exports
pcie_has_flr()and it is being used by Xen pciback driver to reset (flr/slot/bus) PCI devices based on 'reset' SysFS attribute.Signed-off-by Govinda Tatti <[email protected]>---
v3: -Newdrivers/pci/pci.c | 3 ++- include/linux/pci.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)drivers/pci/pci.c +1
-1@@ -3872,7 +3872,7 @@ static void pci_flr_wait(struct pci_dev *dev)* Returns true if the device advertises support for PCIe function level * resets. */ -static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;
Quoted 8 lines
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 5 lines
drivers/pci/pci.c +1
-0@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
Sure, I will add pcie_has_flr() logic inside pcie_flr() and return appropriate values as suggested by you. Do we still want to retain pcie_has_flr() and its usage inside pci.c?.Otherwise, I will remove it and do required cleanup.
Please let me know.
Cheers
GOVINDA
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 6 lines · Bjorn Helgaas
I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?I don't like the "Can I do this? Ok, do this" style of interfaces. It's racy (not really applicable in this case) and seems clunky.
I was tempted to change all that for the whole reset sequence but didn't dare to do that because someone probably put some thought into the current scheme.
I'd love to get rid of the "can I do this" interfaces entirely, and also change to EOPNOTSUPP for the not supported case.
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Bjorn HelgaasQuoted 6 lines · Bjorn Helgaas
I'd rather change
pcie_flr()so you could *always* call it, and it would return 0,-ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?I don't like the "Can I do this? Ok, do this" style of interfaces. It's racy (not really applicable in this case) and seems clunky.
I was tempted to change all that for the whole reset sequence but didn't dare to do that because someone probably put some thought into the current scheme.
I'd love to get rid of the "can I do this" interfaces entirely, and also change to EOPNOTSUPP for the not supported case.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 1/2] Drivers/PCI: Export pcie_has_flr() interface
In reply to Govinda TattiQuoted 26 lines · Govinda Tatti
This patch exports
pcie_has_flr()and it is being used by Xen pciback driver to reset (flr/slot/bus) PCI devices based on 'reset' SysFS attribute.Signed-off-by Govinda Tatti <[email protected]>---
v3: -Newdrivers/pci/pci.c | 3 ++- include/linux/pci.h | 1 + 2 files changed, 3 insertions(+), 1 deletion(-)drivers/pci/pci.c +1
-1@@ -3872,7 +3872,7 @@ static void pci_flr_wait(struct pci_dev *dev)* Returns true if the device advertises support for PCIe function level * resets. */ -static bool pcie_has_flr (struct pci_dev *dev) +bool pcie_has_flr (struct pci_dev *dev) {u32 cap;
drivers/pci/pci.c static bool pcie_has_flr(struct pci_dev *dev) · line 3885
Quoted 4 lines
drivers/pci/pci.c +1
-0@@ -3882,6 +3882,7 @@ static bool pcie_has_flr(struct pci_dev *dev)pcie_capability_read_dword (dev,PCI_EXP_DEVCAP , &cap);return cap &PCI_EXP_DEVCAP_FLR ; } +EXPORT_SYMBOL_GPL (pcie_has_flr);
I'd rather change pcie_flr() so you could *always* call it, and it would return 0, -ENOTTY, or whatever, based on whether FLR is supported. Is that feasible?
I don't like the "Can I do this? Ok, do this" style of interfaces. It's racy (not really applicable in this case) and seems clunky.
Quoted 17 lines
drivers/pci/pci.c +0
-0/** * pcie_flr - initiate a PCIe function level reset include/linux/pci.h +1
-0@@ -1089,6 +1089,7 @@ int pcie_get_mps(struct pci_dev *dev);int pcie_set_mps (struct pci_dev *dev,int mps);int pcie_get_minimum_link (struct pci_dev *dev,enum pci_bus_speed *speed,enum pcie_link_width *width); +bool pcie_has_flr (struct pci_dev *dev);void pcie_flr (struct pci_dev *dev);int __pci_reset_function (struct pci_dev *dev);int __pci_reset_function_locked (struct pci_dev *dev);Signature · 1 line
2.9.5
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v3 2/2 Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Govinda TattiThe life-cycle of a PCI device in Xen pciback is complex and is constrained by the generic PCI locking mechanism.
- It starts with the device being bound to us, for which we do a function reset (done via SysFS so the PCI lock is held).
- If the device is unbound from us, we also do a function reset (done via SysFS so the PCI lock is held).
- If the device is un-assigned from a guest - we do a function reset (no PCI lock is held).
All reset operations are done on the individual PCI function level (so bus:device:function).
The reset for an individual PCI function means device must support FLR (PCIe or AF), PM reset on D3hot->D0 device specific reset, or a secondary bus reset for a singleton device on a bus but FLR does not have widespread support or it is not reliable in some cases. So, we need to provide an alternate mechanism to users to perform a slot or bus level reset.
Currently, a slot or bus reset is not exposed in SysFS as there is no good way of exposing a bus topology there. This is due to the complexity - we MUST know that the different functions of a PCIe device are not in use by other drivers, or if they are in use (say one of them is assigned to a guest and the other is idle) - it is still OK to reset the slot (assuming both of them are owned by Xen pciback).
This patch does that providing an option to perform a flr/slot/bus reset when a PCI device is owned by Xen PCI backend. It will try to execute one of these reset method, starting with FLR if it is supported. Otherwise, it tries slot or bus reset method. For slot or bus reset method, it also checks to make sure that all of the devices under the bridge are owned by Xen PCI backend before applying those resets.
Due to the complexity with the PCI lock we cannot do the reset when a device is bound ('echo $BDF > bind') or when unbound ('echo $BDF > unbind') as the pci_[slot|bus]_reset also takes the same lock resulting in a dead-lock.
Putting the reset function in a work-queue or thread won't work either - as we have to do the reset function outside the 'unbind' context (it holds the PCI lock). But once you 'unbind' a device the device is no longer under the ownership of Xen pciback and the pci_set_drvdata has been reset, so we cannot use a thread for this.
Instead of doing all this complex dance, we depend on the tool-stack doing the right thing. As such, we implement 'reset' SysFS attribute which 'xl' uses when a device is detached or attached from/to a guest. It bypasses the need to worry about the PCI lock. BTW, previously defined "do_flr" attribute has been renamed to "reset" since "do_flr" name doesn't represent all PCI reset methods and plus, currently it is not being used. So, there is no impact in renaming this sysfs attribute.
To not inadvertently do a bus reset that would affect devices that are in use by other drivers (other than Xen pciback) prior to the reset, we check that all of the devices under the bridge are owned by Xen pciback. If they are not, we refrain from executing the bus (or slot) reset.
---
v1: - First posting
v2: - struct pcistub_args: Changed docunt field as unsigned int
- pcistub_reset_dev: initialization of "struct pcistub_args"
- pcistub_reset_dev: combined multiple if-statements
- pcistub_do_flr: removed goto statement
v3: - Resynced with linux kernel 4.14.4 and latest pci_stub.c changes.
- Renamed "do_flr" SysFS attribute to "reset". Plus, modified "reset" SysFS attribute code as per the latest changes in 4.14.4.
- struct pcistub_args: added "const" to "struct pci_dev *dev" field
- pcistub_device_search: Renamed found_dev to found
- pcistub_device_search: Modified comments and return statements
- pcistub_device_reset: introduced FLR reset code
- pcistub_device_reset: Modified all dev_dbg messages
Documentation/ABI/testing/sysfs-driver-pciback | 15 +++ drivers/xen/xen-pciback/pci_stub.c | 128 +++++++++++++++++++++++++ 2 files changed, 143 insertions(+)
Documentation/ABI/testing/sysfs-driver-pciback +15 -0
@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.F + will cause the pciback driver to perform a flr or slot or bus + reset if the device supports it. It will try to execute one + of these reset method, starting with FLR if it is supported. + Otherwise, it tries slot or bus reset methods. For slot or + bus reset method, it also checks to make sure that all of the + devices under the bridge are owned by Xen PCI backend before + performing those resets.
drivers/xen/xen-pciback/pci_stub.c +128 -0
152 lines not on this page · raw message
Signature · 1 line
2.9.5
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 5 lines
Quoted 1 line
Quoted 1 line · On 07.12.17 at 23:21, <[email protected]> wrote:
Due to the complexity with the PCI lock we cannot do the reset when a device is bound ('echo $BDF > bind') or when unbound ('echo $BDF > unbind') as the pci_[slot|bus]_reset also takes the same lock resulting in a dead-lock.
It took me a moment to figure that here you're referring to the process of (un)binding, not the state. To avoid that ambiguity in wording, how about "... we cannot do the reset while a device is being bound (...) or while it is being unbound ..."?
Quoted 4 lines
Documentation/ABI/testing/sysfs-driver-pciback +0
-0@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks
Documentation/ABI/testing/sysfs-driver-pciback Description: · line 21
Quoted 10 lines
Documentation/ABI/testing/sysfs-driver-pciback +8
-0will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.F
SSSS:BB:DD.F (or else the D-s are ambiguous, the more that "domain" in Xen code is ambiguous anyway - I continue to be mislead by struct pcistub_device_id's domain field)
Also I assume the SSSS part is optional (default zero), which probably can and should be expressed in some way.
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 318
Quoted 9 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0@@ -313,6 +313,102 @@ void pcistub_put_pci_dev(struct pci_dev *dev)up_write (&pcistub_sem); } +struct pcistub_args { +const struct pci_dev *dev; +unsigned int dcount;
The sole use of this field is for a debug message. Why not drop it and make "dev" the "data" argument without further indirection?
Quoted 4 lines
drivers/xen/xen-pciback/pci_stub.c +4
-0+static int pcistub_device_search (struct pci_dev *dev,void *data) +{ +struct pcistub_device *psdev; +struct pcistub_args *arg = data;
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 334
Quoted 10 lines
drivers/xen/xen-pciback/pci_stub.c +10
-0+bool found = false; +unsigned long flags; + +spin_lock_irqsave (&pcistub_devices_lock, flags); + +list_for_each_entry (psdev, &pcistub_devices, dev_list) { +if (psdev->dev == dev) { + found = true; + arg->dcount++; +break ;
Neither here nor in the caller I can see a check whether the device is currently assigned to a guest. Ownership by pciback alone imo is not sufficient to allow a reset to be performed.
Quoted 6 lines
drivers/xen/xen-pciback/pci_stub.c +6
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; +
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 10 lines
drivers/xen/xen-pciback/pci_stub.c +10
-0+if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) { +dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);
The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 371
Quoted 7 lines
drivers/xen/xen-pciback/pci_stub.c +7
-0+return 0 ; + } + +if (!pci_probe_reset_slot (dev->slot)) + slot = true; +else if ((!pci_probe_reset_bus (dev->bus)) && + (!pci_is_root_bus (dev->bus)))
Too many parentheses for my taste.
Quoted 15 lines
drivers/xen/xen-pciback/pci_stub.c +15
-0+static ssize_t reset_store (struct device_driver *drv,const char *buf, +size_t count) +{ +struct pcistub_device *psdev; +int domain, bus, slot, func; +int err; + + err =str_to_slot (buf, &domain, &bus, &slot, &func); +if (err) +return err; + + psdev =pcistub_device_find (domain, bus, slot, func); +if (psdev) { + err =pcistub_device_reset (psdev->dev); +pcistub_device_put (psdev);
drivers/xen/xen-pciback/pci_stub.c static ssize_t permissive_show(struct device_driver *drv, char *buf) · line 1553
Quoted 10 lines
drivers/xen/xen-pciback/pci_stub.c +10
-0+ }else { + err = -ENODEV ; + } + +if (!err) + err = count; + +return err; +} +static DRIVER_ATTR_WO (reset);
Would it be worth for reads of the file to return whether the device can be reset this way (i.e. the result of the checks you do before actually doing the reset)?
Jan
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 2 lines
Quoted 1 line
Quoted 1 line · On 12.12.17 at 15:48, <[email protected]> wrote:
Thanks Jan for your review comments. Please see below for my comments.
First of all - can you please do something about your reply style? HTML mail should be avoided. You'll see that the (plain text) reply as a result is rather hard to follow, too.
Quoted 20 lines
Documentation/ABI/testing/sysfs-driver-pciback +8
-0@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.FSSSS:BB:DD.F (or else the D-s are ambiguous, the more that "domain" in Xen code is ambiguous anyway - I continue to be mislead by struct pcistub_device_id's domain field) Thanks for catching this issue. I will fix it.
Quoted 3 lines
Also I assume the SSSS part is optional (default zero), which probably can and should be expressed in some way. SSSS can be 0 or non-zero, subject to system configuration.
The question isn't system configuration, but whether the field can be omitted on input, with zero being assumed in such a case. That's a common shorthand, considering that the vast majority of x86 (and maybe other) systems aren't using segments other than zero.
Jan
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 5 lines · Jan Beulich
Quoted 2 lines
Quoted 1 line
Quoted 1 line · On 12.12.17 at 15:48, <[email protected]> wrote:
Thanks Jan for your review comments. Please see below for my comments.
First of all - can you please do something about your reply style? HTML mail should be avoided. You'll see that the (plain text) reply as a result is rather hard to follow, too.
Sorry about it. I had an issue with my Thunderbird setting.
Quoted 21 lines
Quoted 21 lines
Documentation/ABI/testing/sysfs-driver-pciback +8
-0@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.FSSSS:BB:DD.F (or else the D-s are ambiguous, the more that "domain" in Xen code is ambiguous anyway - I continue to be mislead by struct pcistub_device_id's domain field) Thanks for catching this issue. I will fix it.
Quoted 7 lines
Quoted 3 lines
Also I assume the SSSS part is optional (default zero), which probably can and should be expressed in some way. SSSS can be 0 or non-zero, subject to system configuration.
The question isn't system configuration, but whether the field can be omitted on input, with zero being assumed in such a case. That's a common shorthand, considering that the vast majority of x86 (and maybe other) systems aren't using segments other than zero
Yes, it can be omitted if SSSS is zero.I will add this information to above documentation file.
Cheers
GOVINDA
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Jan BeulichQuoted 5 lines · Jan Beulich
Quoted 2 lines
Quoted 1 line
Quoted 1 line · On 12.12.17 at 15:48, <[email protected]> wrote:
Thanks Jan for your review comments. Please see below for my comments.
First of all - can you please do something about your reply style? HTML mail should be avoided. You'll see that the (plain text) reply as a result is rather hard to follow, too.
Sorry about it. I had an issue with my Thunderbird setting.
Quoted 21 lines
Quoted 21 lines
Documentation/ABI/testing/sysfs-driver-pciback +8
-0@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.FSSSS:BB:DD.F (or else the D-s are ambiguous, the more that "domain" in Xen code is ambiguous anyway - I continue to be mislead by struct pcistub_device_id's domain field) Thanks for catching this issue. I will fix it.
Quoted 7 lines
Quoted 3 lines
Also I assume the SSSS part is optional (default zero), which probably can and should be expressed in some way. SSSS can be 0 or non-zero, subject to system configuration.
The question isn't system configuration, but whether the field can be omitted on input, with zero being assumed in such a case. That's a common shorthand, considering that the vast majority of x86 (and maybe other) systems aren't using segments other than zero
Yes, it can be omitted if SSSS is zero.I will add this information to above documentation file.
Cheers
GOVINDA
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Govinda TattiQuoted 2 lines
Quoted 1 line
Quoted 1 line · On 12.12.17 at 15:48, <[email protected]> wrote:
Thanks Jan for your review comments. Please see below for my comments.
First of all - can you please do something about your reply style? HTML mail should be avoided. You'll see that the (plain text) reply as a result is rather hard to follow, too.
Quoted 20 lines
Documentation/ABI/testing/sysfs-driver-pciback +8
-0@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.FSSSS:BB:DD.F (or else the D-s are ambiguous, the more that "domain" in Xen code is ambiguous anyway - I continue to be mislead by struct pcistub_device_id's domain field) Thanks for catching this issue. I will fix it.
Quoted 3 lines
Also I assume the SSSS part is optional (default zero), which probably can and should be expressed in some way. SSSS can be 0 or non-zero, subject to system configuration.
The question isn't system configuration, but whether the field can be omitted on input, with zero being assumed in such a case. That's a common shorthand, considering that the vast majority of x86 (and maybe other) systems aren't using segments other than zero.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Govinda TattiJan,
One quick update on pcie_flr() specific implementation. Please see below.
Quoted 21 lines
Quoted 13 lines
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing pcie_flr() after checking FLR capability. We will switchto revised pcie_flr() once it is available.
I hope you are fine with this approach. Please let me know. Thanks.
Cheers
GOVINDA
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Govinda TattiJan,
One quick update on pcie_flr() specific implementation. Please see below.
Quoted 21 lines
Quoted 13 lines
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing pcie_flr() after checking FLR capability. We will switchto revised pcie_flr() once it is available.
I hope you are fine with this approach. Please let me know. Thanks.
Cheers
GOVINDA
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.
Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.
Jan
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.
This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
-boris
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Hi,
Quoted 43 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.
Quoted 7 lines
This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Meanwhile the pcie_has_flr() has been exported in upstream Linux kernel, so that's already available for use now.
"PCI: Export pcie_has_flr()":
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
Quoted 2 lines
-boris
Thanks,
-- Pasi
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Boris OstrovskyHi,
Quoted 43 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.
Quoted 7 lines
This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Meanwhile the pcie_has_flr() has been exported in upstream Linux kernel, so that's already available for use now.
"PCI: Export pcie_has_flr()":
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
Quoted 2 lines
-boris
Thanks,
-- Pasi
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 49 lines · Pasi Kärkkäinen
Hi,
Quoted 46 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
(quoted diff) +13
-0+static int pcistub_device_reset(struct pci_dev *dev) +{ + struct xen_pcibk_dev_data *dev_data; + bool slot = false, bus = false; + struct pcistub_args arg = {}; + + if (!dev) + return -EINVAL; + + dev_dbg(&dev->dev, "[%s]\n", __func__); + + /* First check and try FLR */ + if (pcie_has_flr(dev)) {Quoted 8 lines
Review
Quoted 3 lines
(quoted diff) +3
-0+ dev_dbg(&dev->dev, "resetting %s device using FLR\n", + pci_name(dev)); + pcie_flr(dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Quoted 5 lines
Quoted 3 lines
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Govinda is no longer at Oracle.
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
-boris
Quoted 9 lines
Meanwhile the
pcie_has_flr()has been exported in upstream Linux kernel, so that's already available for use now."PCI: Export
pcie_has_flr()":
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Pasi KärkkäinenQuoted 49 lines · Pasi Kärkkäinen
Hi,
Quoted 46 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
(quoted diff) +13
-0+static int pcistub_device_reset(struct pci_dev *dev) +{ + struct xen_pcibk_dev_data *dev_data; + bool slot = false, bus = false; + struct pcistub_args arg = {}; + + if (!dev) + return -EINVAL; + + dev_dbg(&dev->dev, "[%s]\n", __func__); + + /* First check and try FLR */ + if (pcie_has_flr(dev)) {Quoted 8 lines
Review
Quoted 3 lines
(quoted diff) +3
-0+ dev_dbg(&dev->dev, "resetting %s device using FLR\n", + pci_name(dev)); + pcie_flr(dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Quoted 5 lines
Quoted 3 lines
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Govinda is no longer at Oracle.
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
-boris
Quoted 9 lines
Meanwhile the
pcie_has_flr()has been exported in upstream Linux kernel, so that's already available for use now."PCI: Export
pcie_has_flr()":
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Hi,
Quoted 50 lines · Boris Ostrovsky
Quoted 49 lines · Pasi Kärkkäinen
Hi,
Quoted 46 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
(quoted diff) +13
-0+static int pcistub_device_reset(struct pci_dev *dev) +{ + struct xen_pcibk_dev_data *dev_data; + bool slot = false, bus = false; + struct pcistub_args arg = {}; + + if (!dev) + return -EINVAL; + + dev_dbg(&dev->dev, "[%s]\n", __func__); + + /* First check and try FLR */ + if (pcie_has_flr(dev)) {Quoted 8 lines
Review
Quoted 3 lines
(quoted diff) +3
-0+ dev_dbg(&dev->dev, "resetting %s device using FLR\n", + pci_name(dev)); + pcie_flr(dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Quoted 9 lines
Quoted 5 lines
Quoted 3 lines
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Govinda is no longer at Oracle.
Yep, thought so. Removed from CC list.
Quoted 4 lines
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html
"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
George asked for some clarifications:
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01044.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01116.html
Quoted 3 lines
-boris
Thanks,
-- Pasi
Quoted 11 lines
Quoted 10 lines
Meanwhile the
pcie_has_flr()has been exported in upstream Linux kernel, so that's already available for use now."PCI: Export
pcie_has_flr()":
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Boris OstrovskyHi,
Quoted 50 lines · Boris Ostrovsky
Quoted 49 lines · Pasi Kärkkäinen
Hi,
Quoted 46 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
(quoted diff) +13
-0+static int pcistub_device_reset(struct pci_dev *dev) +{ + struct xen_pcibk_dev_data *dev_data; + bool slot = false, bus = false; + struct pcistub_args arg = {}; + + if (!dev) + return -EINVAL; + + dev_dbg(&dev->dev, "[%s]\n", __func__); + + /* First check and try FLR */ + if (pcie_has_flr(dev)) {Quoted 8 lines
Review
Quoted 3 lines
(quoted diff) +3
-0+ dev_dbg(&dev->dev, "resetting %s device using FLR\n", + pci_name(dev)); + pcie_flr(dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Quoted 9 lines
Quoted 5 lines
Quoted 3 lines
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Govinda is no longer at Oracle.
Yep, thought so. Removed from CC list.
Quoted 4 lines
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html
"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
George asked for some clarifications:
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01044.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01116.html
Quoted 3 lines
-boris
Thanks,
-- Pasi
Quoted 11 lines
Quoted 10 lines
Meanwhile the
pcie_has_flr()has been exported in upstream Linux kernel, so that's already available for use now."PCI: Export
pcie_has_flr()":
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d2917f7747805a1f4188672f308d82a8ba01700
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 81 lines · Pasi Kärkkäinen
Hi,
Quoted 50 lines · Boris Ostrovsky
Quoted 49 lines · Pasi Kärkkäinen
Hi,
Quoted 46 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
(quoted diff) +13
-0+static int pcistub_device_reset(struct pci_dev *dev) +{ + struct xen_pcibk_dev_data *dev_data; + bool slot = false, bus = false; + struct pcistub_args arg = {}; + + if (!dev) + return -EINVAL; + + dev_dbg(&dev->dev, "[%s]\n", __func__); + + /* First check and try FLR */ + if (pcie_has_flr(dev)) {Quoted 8 lines
Review
Quoted 3 lines
(quoted diff) +3
-0+ dev_dbg(&dev->dev, "resetting %s device using FLR\n", + pci_name(dev)); + pcie_flr(dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Quoted 9 lines
Quoted 5 lines
Quoted 3 lines
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Govinda is no longer at Oracle.
Yep, thought so. Removed from CC list.
Quoted 4 lines
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html
Quoted 6 lines
"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.htmlGeorge asked for some clarifications:
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01044.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01116.html
Right, the description of the patch didn’t actually tell you what was going on. It should have said something like, “xl currently attempts to reset a device using X; but that’s never been implemented in Linux. Instead, use Y, which [is better for whatever reason]”.
-George
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Pasi KärkkäinenQuoted 81 lines · Pasi Kärkkäinen
Hi,
Quoted 50 lines · Boris Ostrovsky
Quoted 49 lines · Pasi Kärkkäinen
Hi,
Quoted 46 lines · Boris Ostrovsky
Quoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
(quoted diff) +13
-0+static int pcistub_device_reset(struct pci_dev *dev) +{ + struct xen_pcibk_dev_data *dev_data; + bool slot = false, bus = false; + struct pcistub_args arg = {}; + + if (!dev) + return -EINVAL; + + dev_dbg(&dev->dev, "[%s]\n", __func__); + + /* First check and try FLR */ + if (pcie_has_flr(dev)) {Quoted 8 lines
Review
Quoted 3 lines
(quoted diff) +3
-0+ dev_dbg(&dev->dev, "resetting %s device using FLR\n", + pci_name(dev)); + pcie_flr(dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Quoted 9 lines
Quoted 5 lines
Quoted 3 lines
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
It seems Govinda's email address is giving an error, so I assume someone else needs to pick up this pciback 'reset' feature. Is it likely someone else from Oracle can/will pick up and refresh this patch, with the review comments addressed?
Govinda is no longer at Oracle.
Yep, thought so. Removed from CC list.
Quoted 4 lines
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html
Quoted 6 lines
"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.htmlGeorge asked for some clarifications:
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01044.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01116.html
Right, the description of the patch didn’t actually tell you what was going on. It should have said something like, “xl currently attempts to reset a device using X; but that’s never been implemented in Linux. Instead, use Y, which [is better for whatever reason]”.
-George
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 10 lines · George Dunlap
Quoted 10 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
Quoted 8 lines
Quoted 8 lines
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Will this patch work for *BSD? Roger?
Quoted 5 lines
Quoted 4 lines
George asked for some clarifications:
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01044.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01116.htmlRight, the description of the patch didn’t actually tell you what was going on. It should have said something like, “xl currently attempts to reset a device using X; but that’s never been implemented in Linux. Instead, use Y, which [is better for whatever reason]”.
Yes, the description can be tightened a bit ;-)
-boris
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to George DunlapQuoted 10 lines · George Dunlap
Quoted 10 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
Quoted 8 lines
Quoted 8 lines
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Will this patch work for *BSD? Roger?
Quoted 5 lines
Quoted 4 lines
George asked for some clarifications:
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01044.html
https://lists.xen.org/archives/html/xen-devel/2017-12/msg01116.htmlRight, the description of the patch didn’t actually tell you what was going on. It should have said something like, “xl currently attempts to reset a device using X; but that’s never been implemented in Linux. Instead, use Y, which [is better for whatever reason]”.
Yes, the description can be tightened a bit ;-)
-boris
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 14 lines · Boris Ostrovsky
Quoted 13 lines · George Dunlap
Quoted 13 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
Quoted 8 lines
Quoted 5 lines
Quoted 5 lines
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.htmlWill this patch work for *BSD? Roger?
At least FreeBSD don't support pci-passthrough, so none of this works ATM. There's no sysfs on BSD, so much of what's in libxl_pci.c will have to be moved to libxl_linux.c when BSD support is added.
Thanks, Roger.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Boris OstrovskyQuoted 14 lines · Boris Ostrovsky
Quoted 13 lines · George Dunlap
Quoted 13 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
Quoted 8 lines
Quoted 5 lines
Quoted 5 lines
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.htmlWill this patch work for *BSD? Roger?
At least FreeBSD don't support pci-passthrough, so none of this works ATM. There's no sysfs on BSD, so much of what's in libxl_pci.c will have to be moved to libxl_linux.c when BSD support is added.
Thanks, Roger.
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 22 lines · Roger Pau Monné
Quoted 21 lines · Boris Ostrovsky
Quoted 18 lines · George Dunlap
Quoted 18 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Quoted 6 lines
Quoted 1 line
Will this patch work for *BSD? Roger?
At least FreeBSD don't support pci-passthrough, so none of this works ATM. There's no sysfs on BSD, so much of what's in
libxl_pci.cwill have to be moved tolibxl_linux.cwhen BSD support is added.
Ok. That sounds like it's OK for the initial pci 'reset' implementation in xl/libxl to be linux-only..
Thanks,
-- Pasi
Quoted 1 line
Thanks, Roger.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Roger Pau MonnéQuoted 22 lines · Roger Pau Monné
Quoted 21 lines · Boris Ostrovsky
Quoted 18 lines · George Dunlap
Quoted 18 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Quoted 6 lines
Quoted 1 line
Will this patch work for *BSD? Roger?
At least FreeBSD don't support pci-passthrough, so none of this works ATM. There's no sysfs on BSD, so much of what's in
libxl_pci.cwill have to be moved tolibxl_linux.cwhen BSD support is added.
Ok. That sounds like it's OK for the initial pci 'reset' implementation in xl/libxl to be linux-only..
Thanks,
-- Pasi
Quoted 1 line
Thanks, Roger.
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Quoted 21 lines · Pasi Kärkkäinen
Quoted 20 lines · Roger Pau Monné
Quoted 19 lines · Boris Ostrovsky
Quoted 17 lines · George Dunlap
Quoted 17 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Quoted 7 lines
Quoted 5 lines
Quoted 1 line
Will this patch work for *BSD? Roger?
At least FreeBSD don't support pci-passthrough, so none of this works ATM. There's no sysfs on BSD, so much of what's in
libxl_pci.cwill have to be moved tolibxl_linux.cwhen BSD support is added.Ok. That sounds like it's OK for the initial pci 'reset' implementation in xl/libxl to be linux-only..
Are these two patches still needed? ISTR they were written originally to deal with guest trying to use device that was previously assigned to another guest. But pcistub_put_pci_dev() calls __pci_reset_function_locked() which first tries FLR, and it looks like it was added relatively recently.
-boris
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Pasi KärkkäinenQuoted 21 lines · Pasi Kärkkäinen
Quoted 20 lines · Roger Pau Monné
Quoted 19 lines · Boris Ostrovsky
Quoted 17 lines · George Dunlap
Quoted 17 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Quoted 7 lines
Quoted 5 lines
Quoted 1 line
Will this patch work for *BSD? Roger?
At least FreeBSD don't support pci-passthrough, so none of this works ATM. There's no sysfs on BSD, so much of what's in
libxl_pci.cwill have to be moved tolibxl_linux.cwhen BSD support is added.Ok. That sounds like it's OK for the initial pci 'reset' implementation in xl/libxl to be linux-only..
Are these two patches still needed? ISTR they were written originally to deal with guest trying to use device that was previously assigned to another guest. But pcistub_put_pci_dev() calls __pci_reset_function_locked() which first tries FLR, and it looks like it was added relatively recently.
-boris
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Den 08. okt. 2018 16:32, skrev Boris Ostrovsky:
Quoted 8 lines
Are these two patches still needed? ISTR they were written originally to deal with guest trying to use device that was previously assigned to another guest. But
pcistub_put_pci_dev()calls__pci_reset_function_locked()which first tries FLR, and it looks like it was added relatively recently.
Sorry for the late reply, but I just now booted xen staging-4.11 (94fba9f438a2c36ad9bf3a481a6013ddc7cf8cd9), with gentoo-sources-4.19.0 as dom0. Shut down and started again a VM that has a secondary GPU passed through, and the whole machine hung. I haven't had time to look more closely into this, other than that my old "do_flr" patch no longer applies to gentoo-sources (i.e. the linux kernel sources) . "do_flr" worked for me on linux-4.18.? , with appropriate patch to the linux kernel.
So, something is definitely needed. No "reset" , or "do_flr" is present in linux-4.19.0, viz:
<code>
$ cd /usr/src/linux/drivers/xen/xen-pciback $ grep DRIVER * pci_stub.c:#define PCISTUB_DRIVER_NAME "pciback" pci_stub.c: !strcmp(dev->driver_override, PCISTUB_DRIVER_NAME)) || pci_stub.c: .name = PCISTUB_DRIVER_NAME, pci_stub.c:static DRIVER_ATTR_WO(new_slot); pci_stub.c:static DRIVER_ATTR_WO(remove_slot); pci_stub.c:static DRIVER_ATTR_RO(slots); pci_stub.c:static DRIVER_ATTR_RO(irq_handlers); pci_stub.c:static DRIVER_ATTR_WO(irq_handler_state); pci_stub.c:static DRIVER_ATTR_RW(quirks); pci_stub.c:static DRIVER_ATTR_RW(permissive); pci_stub.c: if (action != BUS_NOTIFY_UNBIND_DRIVER) $ </code>
I'd be happy to test patches. Seems I only got one corrupt file from my test this morning :-D .
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
Hi,
Quoted 38 lines · Håkon Alstadheim
Den 08. okt. 2018 16:32, skrev Boris Ostrovsky:
Quoted 8 lines
Are these two patches still needed? ISTR they were written originally to deal with guest trying to use device that was previously assigned to another guest. But
pcistub_put_pci_dev()calls__pci_reset_function_locked()which first tries FLR, and it looks like it was added relatively recently.Sorry for the late reply, but I just now booted xen staging-4.11 (
94fba9f438a2c36ad9bf3a481a6013ddc7cf8cd9), with gentoo-sources-4.19.0 as dom0. Shut down and started again a VM that has a secondary GPU passed through, and the whole machine hung. I haven't had time to look more closely into this, other than that my old "do_flr" patch no longer applies to gentoo-sources (i.e. the linux kernel sources) . "do_flr" worked for me on linux-4.18.? , with appropriate patch to the linux kernel.So, something is definitely needed. No "reset" , or "
do_flr" is present in linux-4.19.0, viz:
<code>$ cd /usr/src/linux/drivers/xen/xen-pciback $ grep DRIVER * pci_stub.c:#define PCISTUB_DRIVER_NAME "pciback" pci_stub.c: !strcmp(dev->driver_override, PCISTUB_DRIVER_NAME)) || pci_stub.c: .name = PCISTUB_DRIVER_NAME, pci_stub.c:static DRIVER_ATTR_WO(new_slot); pci_stub.c:static DRIVER_ATTR_WO(remove_slot); pci_stub.c:static DRIVER_ATTR_RO(slots); pci_stub.c:static DRIVER_ATTR_RO(irq_handlers); pci_stub.c:static DRIVER_ATTR_WO(irq_handler_state); pci_stub.c:static DRIVER_ATTR_RW(quirks); pci_stub.c:static DRIVER_ATTR_RW(permissive); pci_stub.c: if (action != BUS_NOTIFY_UNBIND_DRIVER) $ </code>
Quoted 3 lines
I'd be happy to test patches. Seems I only got one corrupt file from my test this morning :-D .
Håkon: Please do test the patches and report how it works for you! Here are the links:
Linux kernel xen-pciback 'reset' patches:
"[PATCH V3 0/2] Xen/PCIback: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00659.html
(Patch 1/2 is not needed anymore in upstream Linux kernel, as pcie_has_flr() is already exported meanwhile)
"[PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00661.html
Xen libxl 'reset' patches:
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute": https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html "[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute": https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Thanks,
-- Pasi
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH 1/1 For linux-4.19.x . Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Håkon AlstadheimDen 23.10.2018 20:40, skrev Håkon Alstadheim:
Quoted 9 lines
Den 08. okt. 2018 16:32, skrev Boris Ostrovsky:
Quoted 7 lines
Are these two patches still needed? ISTR they were written originally to deal with guest trying to use device that was previously assigned to another guest. But
pcistub_put_pci_dev()calls__pci_reset_function_locked()which first tries FLR, and it looks like it was added relatively recently.
Quoted 7 lines
Sorry for the late reply, but I just now booted xen staging-4.11 (
94fba9f438a2c36ad9bf3a481a6013ddc7cf8cd9), with gentoo-sources-4.19.0 as dom0. Shut down and started again a VM that has a secondary GPU passed through, and the whole machine hung. I haven't had time to look more closely into this, other than that my old "do_flr" patch no longer applies to gentoo-sources (i.e. the linux kernel sources) . "do_flr" worked for me on linux-4.18.? , with appropriate patch to the linux kernel.
Without some kind of fix, my whole server (dom0) goes down whenever a domu with pci passed through is re-started.
NOTE: I am not a programmer. I have no idea what I am doing.
The patch I have as a starting-point does not compile correctly when applied to kernel version 4.19.x. I get implicit declarations of pci_try_reset_slot() and pci_try_reset_bus(). Replacing those with pci_reset_bus(dev) gives me the attached patch which applies cleanly to gentoo-sources-4.19.2, compiles without warnings, and works to let me restart a domU with pci-passthrough (modulo changing do_flr to reset in xen libxl).
I hope a dev will adopt these and give them proper care :-) .
drivers/xen/xen-pciback/pci_stub.c +118 -0
@@ -244,6 +244,90 @@return found_dev; } +struct pcistub_args { +struct pci_dev *dev; +unsigned int dcount; +}; + +static int pcistub_search_dev (struct pci_dev *dev,void *data) +{ +struct pcistub_device *psdev; +struct pcistub_args *arg = data; +bool found_dev = false; +unsigned long flags; + +spin_lock_irqsave (&pcistub_devices_lock, flags); + +list_for_each_entry (psdev, &pcistub_devices, dev_list) { +if (psdev->dev == dev) { + found_dev = true; + arg->dcount++; +break ; + } + } + +spin_unlock_irqrestore (&pcistub_devices_lock, flags); + +/* Device not owned by pcistub, someone owns it. Abort the walk */ +if (!found_dev) + arg->dev = dev; + +return found_dev ?0 :1 ; +} + +static int pcistub_reset_dev (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +if (!pci_probe_reset_slot (dev->slot)) + slot = true; +else if ((!pci_probe_reset_bus (dev->bus)) && + (!pci_is_root_bus (dev->bus))) + bus = true; + +if (!bus && !slot) +return -EOPNOTSUPP ; + +/* +* Make sure all devices on this bus are owned by the +* PCI backend so that we can safely reset the whole bus. +*/ +pci_walk_bus (dev->bus, pcistub_search_dev, &arg); + +/* All devices under the bus should be part of pcistub! */ +if (arg.dev) { +dev_err (&dev->dev,"%s device on bus 0x%x is not owned by pcistub\n" , +pci_name (arg.dev), dev->bus->number); + +return -EBUSY ; + } + +dev_dbg (&dev->dev,"pcistub owns %d devices on bus 0x%x\n" , + arg.dcount, dev->bus->number); + + dev_data =pci_get_drvdata (dev); +if (!pci_load_saved_state (dev, dev_data->pci_saved_state)) +pci_restore_state (dev); + +/* This disables the device. */ +xen_pcibk_reset_device (dev); + +/* Cleanup up any emulated fields */ +xen_pcibk_config_reset_dev (dev); + +dev_dbg (&dev->dev,"resetting %s device using %s reset\n" , +pci_name (dev), slot ?"slot" :"bus" ); + +return pci_reset_bus (dev); +} +/* * Called when: * - XenBus state has been reconfigure (pci unplug). See xen_pcibk_remove_device
@@ -1430,6 +1514,33 @@ }static DRIVER_ATTR_RW (permissive); +static ssize_t reset_store (struct device_driver *drv,const char *buf, +size_t count) +{ +struct pcistub_device *psdev; +int domain, bus, slot, func; +int err; + + err =str_to_slot (buf, &domain, &bus, &slot, &func); +if (err) +return err; + + psdev =pcistub_device_find (domain, bus, slot, func); +if (psdev) { + err =pcistub_reset_dev (psdev->dev); +pcistub_device_put (psdev); + }else { + err = -ENODEV ; + } + +if (!err) + err = count; + +return err; +} + +static DRIVER_ATTR_WO (reset); +static void pcistub_exit (void ) {driver_remove_file (&xen_pcibk_pci_driver.driver, &driver_attr_new_slot);
@@ -1443,6 +1554,8 @@ &driver_attr_irq_handlers);driver_remove_file (&xen_pcibk_pci_driver.driver, &driver_attr_irq_handler_state); +driver_remove_file (&xen_pcibk_pci_driver.driver, + &driver_attr_reset);pci_unregister_driver (&xen_pcibk_pci_driver); }
@@ -1536,6 +1649,11 @@if (!err) err =driver_create_file (&xen_pcibk_pci_driver.driver, &driver_attr_irq_handler_state); + +if (!err) + err =driver_create_file (&xen_pcibk_pci_driver.driver, + &driver_attr_reset); +if (err)pcistub_exit ();
Documentation/ABI/testing/sysfs-driver-pciback +12 -0
@@ -11,3 +11,15 @@ #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Nov 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a slot or bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.F + will cause the pciback driver to perform a slot or bus reset + if the device supports it. It also checks to make sure that + all of the devices under the bridge are owned by Xen PCI + backend.
tools/libxl/libxl_pci.c +1 -1
@@ -1119,7 +1119,7 @@char *reset;int fd, rc; - reset =GCSPRINTF ("%s/do_flr" ,SYSFS_PCIBACK_DRIVER ); + reset =GCSPRINTF ("%s/reset" ,SYSFS_PCIBACK_DRIVER ); fd =open (reset,O_WRONLY );if (fd >=0 ) {char *buf =GCSPRINTF (PCI_BDF , domain, bus, dev, func);
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Boris OstrovskyHi,
Quoted 27 lines · Boris Ostrovsky
Quoted 26 lines · Pasi Kärkkäinen
Quoted 25 lines · Roger Pau Monné
Quoted 13 lines · Boris Ostrovsky
Quoted 12 lines · George Dunlap
Quoted 12 lines · Pasi Kärkkäinen
Hi,
Quoted 4 lines · Boris Ostrovsky
What about the toolstack changes? Have they been accepted? I vaguely recall there was a discussion about those changes but don't remember how it ended.
I don't think toolstack/libxl patch has been applied yet either.
Quoted 7 lines
Quoted 5 lines
Quoted 5 lines
"[PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html"[PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.htmlWill this patch work for *BSD? Roger?
At least FreeBSD don't support pci-passthrough, so none of this works ATM. There's no sysfs on BSD, so much of what's in
libxl_pci.cwill have to be moved tolibxl_linux.cwhen BSD support is added.
Quoted 9 lines
Quoted 2 lines
Ok. That sounds like it's OK for the initial pci 'reset' implementation in xl/libxl to be linux-only..
Are these two patches still needed? ISTR they were written originally to deal with guest trying to use device that was previously assigned to another guest. But
pcistub_put_pci_dev()calls__pci_reset_function_locked()which first tries FLR, and it looks like it was added relatively recently.
Replying to an old thread.. I only now realized I forgot to reply to this message earlier.
afaik these patches are still needed. Håkon (CC'd) wrote to me in private that he gets a (dom0) Linux kernel crash if he doesn't have these patches applied.
Here are the links to both the linux kernel and libxl patches:
"[Xen-devel] [PATCH V3 0/2] Xen/PCIback: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00659.html
[Note that PATCH V3 1/2 "Drivers/PCI: Export pcie_has_flr() interface" is already applied in upstream linux kernel, so it's not needed anymore]
"[Xen-devel] [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00661.html
"[Xen-devel] [PATCH V1 0/1] Xen/Tools: PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00664.html
"[Xen-devel] [PATCH V1 1/1] Xen/libxl: Perform PCI reset using 'reset' SysFS attribute":
https://lists.xen.org/archives/html/xen-devel/2017-12/msg00663.html
Quoted 2 lines
-boris
Thanks,
-- Pasi
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Jan BeulichQuoted 36 lines · Jan Beulich
Quoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Quoted 4 lines
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.
This is now 4.16 material so we can at least wait until closer to opening of the merge window when we may have the PCI updates. (And I just noticed that you responded to Christoph.)
Besides, we don't want to make kernel changes until the interface is settled (i.e the toolstack changes are accepted).
-boris
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Govinda TattiQuoted 24 lines
Quoted 22 lines
Quoted 14 lines · On 15.12.17 at 20:52, <[email protected]> wrote:
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
Quoted 2 lines
We will change
pcie_flr()to return error code. I will make this change in the next version of this patch.
Quoted 7 lines
I exchanged some emails with Bjorn/Christoph and it looks like Christoph as some planto restructure pcie flr specific functions but I don't know the exact time-frame. For now,I am planning to use existing
pcie_flr()after checking FLR capability. We will switchto revisedpcie_flr()once it is available.I hope you are fine with this approach. Please let me know. Thanks.
I've seen that other discussion. I don't think the change here should be done prior to the error reporting being put in place, for security reasons. But in the end it'll be Konrad as the maintainer to judge.
Or wait, looks like there's some confusion in ./MAINTAINERS:
Konrad is listed as maintainer for "XEN PCI SUBSYSTEM", but the list of files doesn't include pciback. So it would instead be Boris or Jürgen to give you a final word.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Jan BeulichThanks Jan for your review comments. Please see below for my comments.
Quoted 9 lines · Jan Beulich
Quoted 5 lines
Quoted 1 line
Quoted 1 line · On 07.12.17 at 23:21,<[email protected]> wrote:
Due to the complexity with the PCI lock we cannot do the reset when a device is bound ('echo $BDF > bind') or when unbound ('echo $BDF > unbind') as the pci_[slot|bus]_reset also takes the same lock resulting in a dead-lock.
It took me a moment to figure that here you're referring to the process of (un)binding, not the state. To avoid that ambiguity in wording, how about "... we cannot do the reset while a device is being bound (...) or while it is being unbound ..."?
Sure, I will fix it.
Quoted 9 lines
Quoted 9 lines
Documentation/ABI/testing/sysfs-driver-pciback +3
-0@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017
Quoted 8 lines
Documentation/ABI/testing/sysfs-driver-pciback Description: · line 21
Quoted 5 lines
Documentation/ABI/testing/sysfs-driver-pciback +5
-0+KernelVersion: 4.15 +Contact:[email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.FSSSS:BB:DD.F (or else the D-s are ambiguous, the more that "domain" in Xen code is ambiguous anyway - I continue to be mislead by struct pcistub_device_id's domain field)
Thanks for catching this issue. I will fix it.
Quoted 2 lines
Also I assume the SSSS part is optional (default zero), which probably can and should be expressed in some way.
SSSS can be 0 or non-zero, subject to system configuration.
Quoted 11 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 318
Quoted 9 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0@@ -313,6 +313,102 @@ void pcistub_put_pci_dev(struct pci_dev *dev)up_write (&pcistub_sem); } +struct pcistub_args { +const struct pci_dev *dev; +unsigned int dcount;The sole use of this field is for a debug message. Why not drop it and make "dev" the "data" argument without further indirection?
I prefer to keep this data structure since it will be helpful to debug any issues orfor future enhancements.
Quoted 9 lines
Quoted 9 lines
drivers/xen/xen-pciback/pci_stub.c +9
-0+static int pcistub_device_search (struct pci_dev *dev,void *data) +{ +struct pcistub_device *psdev; +struct pcistub_args *arg = data; +bool found = false; +unsigned long flags; + +spin_lock_irqsave (&pcistub_devices_lock, flags); +
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 334
Quoted 5 lines
drivers/xen/xen-pciback/pci_stub.c +5
-0+list_for_each_entry (psdev, &pcistub_devices, dev_list) { +if (psdev->dev == dev) { + found = true; + arg->dcount++; +break ;Neither here nor in the caller I can see a check whether the device is currently assigned to a guest. Ownership by pciback alone imo is not sufficient to allow a reset to be performed.
I can add the following check
if ((psdev->dev == dev) && (pci_is_dev_assigned(dev)))
Quoted 13 lines
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
We will change pcie_flr() to return error code. I will make this change in the next version of this patch.
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 371
Quoted 7 lines
drivers/xen/xen-pciback/pci_stub.c +7
-0+return 0 ; + } + +if (!pci_probe_reset_slot (dev->slot)) + slot = true; +else if ((!pci_probe_reset_bus (dev->bus)) && + (!pci_is_root_bus (dev->bus)))Too many parentheses for my taste.
I will fix it.
Quoted 20 lines
Quoted 20 lines
drivers/xen/xen-pciback/pci_stub.c +20
-0+static ssize_t reset_store (struct device_driver *drv,const char *buf, +size_t count) +{ +struct pcistub_device *psdev; +int domain, bus, slot, func; +int err; + + err =str_to_slot (buf, &domain, &bus, &slot, &func); +if (err) +return err; + + psdev =pcistub_device_find (domain, bus, slot, func); +if (psdev) { + err =pcistub_device_reset (psdev->dev); +pcistub_device_put (psdev); + }else { + err = -ENODEV ; + } + +if (!err)
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c static ssize_t permissive_show(struct device_driver *drv, char *buf) · line 1553
Quoted 5 lines
drivers/xen/xen-pciback/pci_stub.c +5
-0+ err = count; + +return err; +} +static DRIVER_ATTR_WO (reset);Would it be worth for reads of the file to return whether the device can be reset this way (i.e. the result of the checks you do before actually doing the reset)?
I don't think so. Plus, it makes this interface and its usage more complicated.
Cheers
GOVINDA
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH V3 2/2] Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Jan BeulichThanks Jan for your review comments. Please see below for my comments.
Quoted 9 lines · Jan Beulich
Quoted 5 lines
Quoted 1 line
Quoted 1 line · On 07.12.17 at 23:21,<[email protected]> wrote:
Due to the complexity with the PCI lock we cannot do the reset when a device is bound ('echo $BDF > bind') or when unbound ('echo $BDF > unbind') as the pci_[slot|bus]_reset also takes the same lock resulting in a dead-lock.
It took me a moment to figure that here you're referring to the process of (un)binding, not the state. To avoid that ambiguity in wording, how about "... we cannot do the reset while a device is being bound (...) or while it is being unbound ..."?
Sure, I will fix it.
Quoted 9 lines
Quoted 9 lines
Documentation/ABI/testing/sysfs-driver-pciback +3
-0@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017
Quoted 8 lines
Documentation/ABI/testing/sysfs-driver-pciback Description: · line 21
Quoted 5 lines
Documentation/ABI/testing/sysfs-driver-pciback +5
-0+KernelVersion: 4.15 +Contact:[email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.FSSSS:BB:DD.F (or else the D-s are ambiguous, the more that "domain" in Xen code is ambiguous anyway - I continue to be mislead by struct pcistub_device_id's domain field)
Thanks for catching this issue. I will fix it.
Quoted 2 lines
Also I assume the SSSS part is optional (default zero), which probably can and should be expressed in some way.
SSSS can be 0 or non-zero, subject to system configuration.
Quoted 11 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 318
Quoted 9 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0@@ -313,6 +313,102 @@ void pcistub_put_pci_dev(struct pci_dev *dev)up_write (&pcistub_sem); } +struct pcistub_args { +const struct pci_dev *dev; +unsigned int dcount;The sole use of this field is for a debug message. Why not drop it and make "dev" the "data" argument without further indirection?
I prefer to keep this data structure since it will be helpful to debug any issues orfor future enhancements.
Quoted 9 lines
Quoted 9 lines
drivers/xen/xen-pciback/pci_stub.c +9
-0+static int pcistub_device_search (struct pci_dev *dev,void *data) +{ +struct pcistub_device *psdev; +struct pcistub_args *arg = data; +bool found = false; +unsigned long flags; + +spin_lock_irqsave (&pcistub_devices_lock, flags); +
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 334
Quoted 5 lines
drivers/xen/xen-pciback/pci_stub.c +5
-0+list_for_each_entry (psdev, &pcistub_devices, dev_list) { +if (psdev->dev == dev) { + found = true; + arg->dcount++; +break ;Neither here nor in the caller I can see a check whether the device is currently assigned to a guest. Ownership by pciback alone imo is not sufficient to allow a reset to be performed.
I can add the following check
if ((psdev->dev == dev) && (pci_is_dev_assigned(dev)))
Quoted 13 lines
Quoted 13 lines
drivers/xen/xen-pciback/pci_stub.c +13
-0+static int pcistub_device_reset (struct pci_dev *dev) +{ +struct xen_pcibk_dev_data *dev_data; +bool slot = false, bus = false; +struct pcistub_args arg = {}; + +if (!dev) +return -EINVAL ; + +dev_dbg (&dev->dev,"[%s]\n" , __func__); + +/* First check and try FLR */ +if (pcie_has_flr (dev)) {
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 364
Quoted 3 lines
drivers/xen/xen-pciback/pci_stub.c +3
-0+dev_dbg (&dev->dev,"resetting %s device using FLR\n" , +pci_name (dev)); +pcie_flr (dev);The lack of error check here puzzled me, but I see the function indeed returns void right now. I think the prereq patch should change this along with exporting the function - you really don't want the device to be handed to a guest when the FLR timed out.
We will change pcie_flr() to return error code. I will make this change in the next version of this patch.
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c void pcistub_put_pci_dev(struct pci_dev *dev) · line 371
Quoted 7 lines
drivers/xen/xen-pciback/pci_stub.c +7
-0+return 0 ; + } + +if (!pci_probe_reset_slot (dev->slot)) + slot = true; +else if ((!pci_probe_reset_bus (dev->bus)) && + (!pci_is_root_bus (dev->bus)))Too many parentheses for my taste.
I will fix it.
Quoted 20 lines
Quoted 20 lines
drivers/xen/xen-pciback/pci_stub.c +20
-0+static ssize_t reset_store (struct device_driver *drv,const char *buf, +size_t count) +{ +struct pcistub_device *psdev; +int domain, bus, slot, func; +int err; + + err =str_to_slot (buf, &domain, &bus, &slot, &func); +if (err) +return err; + + psdev =pcistub_device_find (domain, bus, slot, func); +if (psdev) { + err =pcistub_device_reset (psdev->dev); +pcistub_device_put (psdev); + }else { + err = -ENODEV ; + } + +if (!err)
Quoted 8 lines
drivers/xen/xen-pciback/pci_stub.c static ssize_t permissive_show(struct device_driver *drv, char *buf) · line 1553
Quoted 5 lines
drivers/xen/xen-pciback/pci_stub.c +5
-0+ err = count; + +return err; +} +static DRIVER_ATTR_WO (reset);Would it be worth for reads of the file to return whether the device can be reset this way (i.e. the result of the checks you do before actually doing the reset)?
I don't think so. Plus, it makes this interface and its usage more complicated.
Cheers
GOVINDA
PATCH v3 2/2 Xen/PCIback: Implement PCI flr/slot/bus reset with 'reset' SysFS attribute
In reply to Govinda TattiThe life-cycle of a PCI device in Xen pciback is complex and is constrained by the generic PCI locking mechanism.
- It starts with the device being bound to us, for which we do a function reset (done via SysFS so the PCI lock is held).
- If the device is unbound from us, we also do a function reset (done via SysFS so the PCI lock is held).
- If the device is un-assigned from a guest - we do a function reset (no PCI lock is held).
All reset operations are done on the individual PCI function level (so bus:device:function).
The reset for an individual PCI function means device must support FLR (PCIe or AF), PM reset on D3hot->D0 device specific reset, or a secondary bus reset for a singleton device on a bus but FLR does not have widespread support or it is not reliable in some cases. So, we need to provide an alternate mechanism to users to perform a slot or bus level reset.
Currently, a slot or bus reset is not exposed in SysFS as there is no good way of exposing a bus topology there. This is due to the complexity - we MUST know that the different functions of a PCIe device are not in use by other drivers, or if they are in use (say one of them is assigned to a guest and the other is idle) - it is still OK to reset the slot (assuming both of them are owned by Xen pciback).
This patch does that providing an option to perform a flr/slot/bus reset when a PCI device is owned by Xen PCI backend. It will try to execute one of these reset method, starting with FLR if it is supported. Otherwise, it tries slot or bus reset method. For slot or bus reset method, it also checks to make sure that all of the devices under the bridge are owned by Xen PCI backend before applying those resets.
Due to the complexity with the PCI lock we cannot do the reset when a device is bound ('echo $BDF > bind') or when unbound ('echo $BDF > unbind') as the pci_[slot|bus]_reset also takes the same lock resulting in a dead-lock.
Putting the reset function in a work-queue or thread won't work either - as we have to do the reset function outside the 'unbind' context (it holds the PCI lock). But once you 'unbind' a device the device is no longer under the ownership of Xen pciback and the pci_set_drvdata has been reset, so we cannot use a thread for this.
Instead of doing all this complex dance, we depend on the tool-stack doing the right thing. As such, we implement 'reset' SysFS attribute which 'xl' uses when a device is detached or attached from/to a guest. It bypasses the need to worry about the PCI lock. BTW, previously defined "do_flr" attribute has been renamed to "reset" since "do_flr" name doesn't represent all PCI reset methods and plus, currently it is not being used. So, there is no impact in renaming this sysfs attribute.
To not inadvertently do a bus reset that would affect devices that are in use by other drivers (other than Xen pciback) prior to the reset, we check that all of the devices under the bridge are owned by Xen pciback. If they are not, we refrain from executing the bus (or slot) reset.
---
v1: - First posting
v2: - struct pcistub_args: Changed docunt field as unsigned int
- pcistub_reset_dev: initialization of "struct pcistub_args"
- pcistub_reset_dev: combined multiple if-statements
- pcistub_do_flr: removed goto statement
v3: - Resynced with linux kernel 4.14.4 and latest pci_stub.c changes.
- Renamed "do_flr" SysFS attribute to "reset". Plus, modified "reset" SysFS attribute code as per the latest changes in 4.14.4.
- struct pcistub_args: added "const" to "struct pci_dev *dev" field
- pcistub_device_search: Renamed found_dev to found
- pcistub_device_search: Modified comments and return statements
- pcistub_device_reset: introduced FLR reset code
- pcistub_device_reset: Modified all dev_dbg messages
Documentation/ABI/testing/sysfs-driver-pciback | 15 +++ drivers/xen/xen-pciback/pci_stub.c | 128 +++++++++++++++++++++++++ 2 files changed, 143 insertions(+)
Documentation/ABI/testing/sysfs-driver-pciback +15 -0
@@ -11,3 +11,18 @@ Description: #echo 00:19.0-E0:2:FF > /sys/bus/pci/drivers/pciback/quirks will allow the guest to read and write to the configuration register 0x0E. + +What: /sys/bus/pci/drivers/pciback/reset +Date: Dec 2017 +KernelVersion: 4.15 +Contact: [email protected] +Description: + An option to perform a flr/slot/bus reset when a PCI device + is owned by Xen PCI backend. Writing a string of DDDD:BB:DD.F + will cause the pciback driver to perform a flr or slot or bus + reset if the device supports it. It will try to execute one + of these reset method, starting with FLR if it is supported. + Otherwise, it tries slot or bus reset methods. For slot or + bus reset method, it also checks to make sure that all of the + devices under the bridge are owned by Xen PCI backend before + performing those resets.
drivers/xen/xen-pciback/pci_stub.c +128 -0
152 lines not on this page · raw message
Signature · 7 lines
2.9.5 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel