PCI: Consolidate TLP Log reading and printing
Subsystems (2)
v6 cover PCI: Consolidate TLP Log reading and printing
This series has the remaining patches of the AER & DPC TLP Log handling consolidation and now includes a few minor improvements to the earlier accepted TLP Logging code.
v6:
- Preserve "AER:"/"DPC:" prefix on the printed TLP line
- New patch to add "AER:" also on other lines of the AER error dump
v5:
- Fix build with AER=y and DPC=n
- Match kerneldoc and function parameter name
v4:
- Added patches:
- Remove EXPORT of pcie_read_tlp_log()
- Moved code to pcie/tlp.c and build only with AER enabled
- Match variables in prototype and function
- int -> unsigned int conversion
- eetlp_prefix_max into own patch
- struct pcie_tlp_log param consistently called "log" within tlp.c
- Moved function prototypes into drivers/pci/pci.h
- Describe AER/DPC differences more clearly in one commit message
v3:
- Small rewording in a commit message
v2:
- Don't add EXPORT()s
- Don't include igxbe changes
- Don't use pr_cont() as it's incompatible with pci_err() and according to Andy Shevchenko should not be used in the first place
Ilpo Järvinen (8):
PCI: Don't expose pcie_read_tlp_log() outside of PCI subsystem
PCI: Move TLP Log handling to own file
PCI: Make pcie_read_tlp_log() signature same
PCI: Use unsigned int i in pcie_read_tlp_log()
PCI: Store # of supported End-End TLP Prefixes
PCI: Add TLP Prefix reading into pcie_read_tlp_log()
PCI: Create helper to print TLP Header and Prefix Log PCI/AER: Add prefixes to printouts
drivers/pci/ats.c | 2 +- drivers/pci/pci.c | 28 --------- drivers/pci/pci.h | 9 +++ drivers/pci/pcie/Makefile | 2 +- drivers/pci/pcie/aer.c | 26 ++++---- drivers/pci/pcie/dpc.c | 14 ++--- drivers/pci/pcie/tlp.c | 109 ++++++++++++++++++++++++++++++++++ drivers/pci/probe.c | 14 +++-- include/linux/aer.h | 3 +- include/linux/pci.h | 2 +- include/uapi/linux/pci_regs.h | 2 + 11 files changed, 149 insertions(+), 62 deletions(-) create mode 100644 drivers/pci/pcie/tlp.c
Signature · 1 line
2.39.2
PATCH v6 1/8 PCI: Don't expose pcie_read_tlp_log() outside of PCI subsystem
pcie_read_tlp_log() was exposed by the commit 0a5a46a6a61b ("PCI/AER:
Generalize TLP Header Log reading") but this is now considered a mistake. No drivers outside of PCI subsystem should build their own diagnostic logging but should rely on PCI core doing it for them.
There's currently one driver (ixgbe) doing it independently which was the initial reason why the export was added but it was decided by the PCI maintainer that it's something that should be eliminated.
Remove the unwanted EXPORT of pcie_read_tlp_log() and remove it from include/linux/aer.h.
--- drivers/pci/pci.c | 1 - drivers/pci/pci.h | 4 ++++ include/linux/aer.h | 2 -- 3 files changed, 4 insertions(+), 3 deletions(-)
drivers/pci/pci.c +0 -1
@@ -1121,7 +1121,6 @@ int pcie_read_tlp_log(struct pci_dev *dev, int where,return 0 ; } -EXPORT_SYMBOL_GPL (pcie_read_tlp_log);/** * pci_restore_bars - restore a device's BAR values (e.g. after wake-up)
drivers/pci/pci.h +4 -0
@@ -4,6 +4,8 @@#include <linux/pci.h> +struct pcie_tlp_log ; +/* Number of possible devfns: 0.0 to 1f.7 inclusive */ #define MAX_NR_DEVFNS 256
@@ -482,6 +484,8 @@ int aer_get_device_error_info(struct pci_dev *dev, struct aer_err_info *info);void aer_print_error (struct pci_dev *dev,struct aer_err_info *info);#endif /* CONFIG_PCIEAER */ +int pcie_read_tlp_log (struct pci_dev *dev,int where,struct pcie_tlp_log *log); +#ifdef CONFIG_PCIEPORTBUS /* Cached RCEC Endpoint Association */ struct rcec_ea {
include/linux/aer.h +0 -2
@@ -37,8 +37,6 @@ struct aer_capability_regs {u16 uncor_err_source; }; -int pcie_read_tlp_log (struct pci_dev *dev,int where,struct pcie_tlp_log *log); -#if defined(CONFIG_PCIEAER )int pci_aer_clear_nonfatal_status (struct pci_dev *dev);int pcie_aer_is_native (struct pci_dev *dev);
Signature · 1 line
2.39.2
Re: [PATCH v6 1/8] PCI: Don't expose pcie_read_tlp_log() outside of PCI subsystem
Quoted 9 lines · On Fri, 13 Sep 2024 17:36:25 +0300 Ilpo Järvinen…
pcie_read_tlp_log()was exposed by the commit0a5a46a6a61b("PCI/AER:
Generalize TLP Header Log reading") but this is now considered a mistake. No drivers outside of PCI subsystem should build their own diagnostic logging but should rely on PCI core doing it for them.There's currently one driver (ixgbe) doing it independently which was the initial reason why the export was added but it was decided by the PCI maintainer that it's something that should be eliminated.
Quoted 5 lines
Remove the unwanted EXPORT of
pcie_read_tlp_log()and remove it frominclude/linux/aer.h.
FWIW LGTM
PATCH v6 2/8 PCI: Move TLP Log handling to own file
In reply to Ilpo JärvinenTLP Log is PCIe feature and is processed only by AER and DPC. Configwise, DPC depends AER being enabled. In lack of better place, the TLP Log handling code was initially placed into pci.c but it can be easily placed in a separate file.
Move TLP Log handling code to own file under pcie/ subdirectory and include it only when AER is enabled.
--- drivers/pci/pci.c | 27 --------------------------- drivers/pci/pci.h | 2 +- drivers/pci/pcie/Makefile | 2 +- drivers/pci/pcie/tlp.c | 39 +++++++++++++++++++++++++++++++++++++++ 4 files changed, 41 insertions(+), 29 deletions(-) create mode 100644 drivers/pci/pcie/tlp.c
drivers/pci/pci.c +0 -27
@@ -1095,33 +1095,6 @@ static void pci_enable_acs(struct pci_dev *dev)pci_write_config_word (dev, pos +PCI_ACS_CTRL , caps.ctrl); } -/** -* pcie_read_tlp_log - read TLP Header Log -* @dev: PCIe device -* @where: PCI Config offset of TLP Header Log -* @tlp_log: TLP Log structure to fill -* -* Fill @tlp_log from TLP Header Log registers, e.g., AER or DPC. -* -* Return: 0 on success and filled TLP Log structure, <0 on error. -*/ -int pcie_read_tlp_log (struct pci_dev *dev,int where, -struct pcie_tlp_log *tlp_log) -{ -int i, ret; - -memset (tlp_log,0 ,sizeof (*tlp_log)); - -for (i =0 ; i <4 ; i++) { - ret =pci_read_config_dword (dev, where + i *4 , - &tlp_log->dw[i]); -if (ret) -return pcibios_err_to_errno (ret); - } - -return 0 ; -} -/** * pci_restore_bars - restore a device's BAR values (e.g. after wake-up) * @dev: PCI device to have its BARs restored
drivers/pci/pci.h +1 -1
@@ -482,9 +482,9 @@ struct aer_err_info {int aer_get_device_error_info (struct pci_dev *dev,struct aer_err_info *info);void aer_print_error (struct pci_dev *dev,struct aer_err_info *info); -#endif /* CONFIG_PCIEAER */ int pcie_read_tlp_log (struct pci_dev *dev,int where,struct pcie_tlp_log *log); +#endif /* CONFIG_PCIEAER */ #ifdef CONFIG_PCIEPORTBUS /* Cached RCEC Endpoint Association */
drivers/pci/pcie/Makefile +1 -1
@@ -7,7 +7,7 @@ pcieportdrv-y := portdrv.o rcec.o obj-$(CONFIG_PCIEPORTBUS) += pcieportdrv.o obj-y += aspm.o -obj-$(CONFIG_PCIEAER) += aer.o err.o +obj-$(CONFIG_PCIEAER) += aer.o err.o tlp.o obj-$(CONFIG_PCIEAER_INJECT) += aer_inject.o obj-$(CONFIG_PCIE_PME) += pme.o obj-$(CONFIG_PCIE_DPC) += dpc.o
drivers/pci/pcie/tlp.c new file +39 -0
@@ -0,0 +1,39 @@ +// SPDX-License-Identifier: GPL-2.0 +/* +* PCIe TLP Log handling +* +* Copyright (C) 2024 Intel Corporation +*/ + +#include <linux/aer.h> +#include <linux/pci.h> +#include <linux/string.h> + +#include "../pci.h" + +/** +* pcie_read_tlp_log - read TLP Header Log +* @dev: PCIe device +* @where: PCI Config offset of TLP Header Log +* @tlp_log: TLP Log structure to fill +* +* Fill @tlp_log from TLP Header Log registers, e.g., AER or DPC. +* +* Return: 0 on success and filled TLP Log structure, <0 on error. +*/ +int pcie_read_tlp_log (struct pci_dev *dev,int where, +struct pcie_tlp_log *tlp_log) +{ +int i, ret; + +memset (tlp_log,0 ,sizeof (*tlp_log)); + +for (i =0 ; i <4 ; i++) { + ret =pci_read_config_dword (dev, where + i *4 , + &tlp_log->dw[i]); +if (ret) +return pcibios_err_to_errno (ret); + } + +return 0 ; +}
Signature · 1 line
2.39.2
Re: [PATCH v6 2/8] PCI: Move TLP Log handling to own file
Quoted 9 lines · On Fri, 13 Sep 2024 17:36:26 +0300 Ilpo Järvinen…
TLP Log is PCIe feature and is processed only by AER and DPC. Configwise, DPC depends AER being enabled. In lack of better place, the TLP Log handling code was initially placed into
pci.cbut it can be easily placed in a separate file.Move TLP Log handling code to own file under pcie/ subdirectory and include it only when AER is enabled.
Signed-off-by Ilpo Järvinen <[email protected]>
Can sort of see this might get used outside AER sometime in the future but for now this seems sensible.
Jonathan
PATCH v6 3/8 PCI: Make pcie_read_tlp_log() signature same
In reply to Ilpo Järvinenpcie_read_tlp_log()'s prototype and function signature diverged due to changes made while applying.
Make the parameters of pcie_read_tlp_log() named identically.
--- drivers/pci/pcie/tlp.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-)
drivers/pci/pcie/tlp.c +5 -6
@@ -15,22 +15,21 @@* pcie_read_tlp_log - read TLP Header Log * @dev: PCIe device * @where: PCI Config offset of TLP Header Log -* @tlp_log: TLP Log structure to fill +* @log: TLP Log structure to fill * -* Fill @tlp_log from TLP Header Log registers, e.g., AER or DPC. +* Fill @log from TLP Header Log registers, e.g., AER or DPC. * * Return: 0 on success and filled TLP Log structure, <0 on error. */ int pcie_read_tlp_log (struct pci_dev *dev,int where, -struct pcie_tlp_log *tlp_log) +struct pcie_tlp_log *log) {int i, ret; -memset (tlp_log,0 ,sizeof (*tlp_log)); +memset (log,0 ,sizeof (*log));for (i =0 ; i <4 ; i++) { - ret =pci_read_config_dword (dev, where + i *4 , - &tlp_log->dw[i]); + ret =pci_read_config_dword (dev, where + i *4 , &log->dw[i]);if (ret)return pcibios_err_to_errno (ret); }
Signature · 1 line
2.39.2
Re: [PATCH v6 3/8] PCI: Make pcie_read_tlp_log() signature same
Quoted 6 lines · On Fri, 13 Sep 2024 17:36:27 +0300 Ilpo Järvinen…
pcie_read_tlp_log()'s prototype and function signature diverged due to changes made while applying.
Make the parameters of
pcie_read_tlp_log()named identically.Signed-off-by Ilpo Järvinen <[email protected]>
Good to clean this up.
PATCH v6 4/8 PCI: Use unsigned int i in pcie_read_tlp_log()
In reply to Ilpo JärvinenLoop variable i counting from 0 upwards does not need to be signed so make it unsigned int.
--- drivers/pci/pcie/tlp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/pci/pcie/tlp.c +2 -1
@@ -24,7 +24,8 @@int pcie_read_tlp_log (struct pci_dev *dev,int where,struct pcie_tlp_log *log) { -int i, ret; +unsigned int i; +int ret;memset (log,0 ,sizeof (*log));
Signature · 1 line
2.39.2
Re: [PATCH v6 4/8] PCI: Use unsigned int i in pcie_read_tlp_log()
Quoted 4 lines · On Fri, 13 Sep 2024 17:36:28 +0300 Ilpo Järvinen…
Loop variable i counting from 0 upwards does not need to be signed so make it unsigned int.
Signed-off-by Ilpo Järvinen <[email protected]>
To me this is unnecessary noise, but it's harmless so no problem if you really like it.
Jonathan
Quoted 18 lines
--- drivers/pci/pcie/tlp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)drivers/pci/pcie/tlp.c +2
-1@@ -24,7 +24,8 @@int pcie_read_tlp_log (struct pci_dev *dev,int where,struct pcie_tlp_log *log) { -int i, ret; +unsigned int i; +int ret;memset (log,0 ,sizeof (*log));
PATCH v6 5/8 PCI: Store # of supported End-End TLP Prefixes
In reply to Ilpo Järvineneetlp_prefix_path in the struct pci_dev tells if End-End TLP Prefixes are supported by the path or not, the value is only calculated if CONFIG_PCI_PASID is set.
The Max End-End TLP Prefixes field in the Device Capabilities Register 2 also tells how many (1-4) End-End TLP Prefixes are supported (PCIe r6 sec 7.5.3.15). The number of supported End-End Prefixes is useful for reading correct number of DWORDs from TLP Prefix Log register in AER capability (PCIe r6 sec 7.8.4.12).
Replace eetlp_prefix_path with eetlp_prefix_max and determine the number of supported End-End Prefixes regardless of CONFIG_PCI_PASID so that an upcoming commit generalizing TLP Prefix Log register reading does not have to read extra DWORDs for End-End Prefixes that never will be there.
--- drivers/pci/ats.c | 2 +- drivers/pci/probe.c | 14 +++++++++----- include/linux/pci.h | 2 +- include/uapi/linux/pci_regs.h | 1 + 4 files changed, 12 insertions(+), 7 deletions(-)
drivers/pci/ats.c +1 -1
@@ -377,7 +377,7 @@ int pci_enable_pasid(struct pci_dev *pdev, int features)if (WARN_ON (pdev->pasid_enabled))return -EBUSY ; -if (!pdev->eetlp_prefix_path && !pdev->pasid_no_tlp) +if (!pdev->eetlp_prefix_max && !pdev->pasid_no_tlp)return -EINVAL ;if (!pasid)
drivers/pci/probe.c +9 -5
@@ -2228,8 +2228,8 @@ static void pci_configure_relaxed_ordering(struct pci_dev *dev)static void pci_configure_eetlp_prefix (struct pci_dev *dev) { -#ifdef CONFIG_PCI_PASID struct pci_dev *bridge; +unsigned int eetlp_max;int pcie_type;u32 cap;
@@ -2241,15 +2241,19 @@ static void pci_configure_eetlp_prefix(struct pci_dev *dev)return ; pcie_type =pci_pcie_type (dev); + + eetlp_max =FIELD_GET (PCI_EXP_DEVCAP2_EE_PREFIX_MAX , cap); +/* 00b means 4 */ + eetlp_max = eetlp_max ?:4 ; +if (pcie_type ==PCI_EXP_TYPE_ROOT_PORT || pcie_type ==PCI_EXP_TYPE_RC_END ) - dev->eetlp_prefix_path =1 ; + dev->eetlp_prefix_max = eetlp_max;else { bridge =pci_upstream_bridge (dev); -if (bridge && bridge->eetlp_prefix_path) - dev->eetlp_prefix_path =1 ; +if (bridge && bridge->eetlp_prefix_max) + dev->eetlp_prefix_max = eetlp_max; } -#endif }static void pci_configure_serr (struct pci_dev *dev)
include/linux/pci.h +1 -1
@@ -398,7 +398,7 @@ struct pci_dev {supported from root to here */ #endif unsigned int pasid_no_tlp:1; /* PASID works without TLP Prefix */ - unsigned int eetlp_prefix_path:1; /* End-to-End TLP Prefix */ + unsigned int eetlp_prefix_max:3; /* Max # of End-End TLP Prefixes, 0=not supported */pci_channel_state_t error_state;/* Current connectivity state */ struct device dev;/* Generic device interface */
include/uapi/linux/pci_regs.h +1 -0
@@ -661,6 +661,7 @@#define PCI_EXP_DEVCAP2_OBFF_MSG 0x00040000/* New message signaling */ #define PCI_EXP_DEVCAP2_OBFF_WAKE 0x00080000/* Re-use WAKE# for OBFF */ #define PCI_EXP_DEVCAP2_EE_PREFIX 0x00200000/* End-End TLP Prefix */ +#define PCI_EXP_DEVCAP2_EE_PREFIX_MAX 0x00c00000/* Max End-End TLP Prefixes */ #define PCI_EXP_DEVCTL2 0x28/* Device Control 2 */ #define PCI_EXP_DEVCTL2_COMP_TIMEOUT 0x000f/* Completion Timeout Value */ #define PCI_EXP_DEVCTL2_COMP_TMOUT_DIS 0x0010/* Completion Timeout Disable */
Signature · 1 line
2.39.2
Re: [PATCH v6 5/8] PCI: Store # of supported End-End TLP Prefixes
Quoted 59 lines · On Fri, 13 Sep 2024 17:36:29 +0300 Ilpo Järvinen…
eetlp_prefix_pathin the structpci_devtells if End-End TLP Prefixes are supported by the path or not, the value is only calculated ifCONFIG_PCI_PASIDis set.The Max End-End TLP Prefixes field in the Device Capabilities Register 2 also tells how many (1-4) End-End TLP Prefixes are supported (PCIe r6 sec 7.5.3.15). The number of supported End-End Prefixes is useful for reading correct number of DWORDs from TLP Prefix Log register in AER capability (PCIe r6 sec 7.8.4.12).
Replace
eetlp_prefix_pathwitheetlp_prefix_maxand determine the number of supported End-End Prefixes regardless ofCONFIG_PCI_PASIDso that an upcoming commit generalizing TLP Prefix Log register reading does not have to read extra DWORDs for End-End Prefixes that never will be there.Signed-off-by Ilpo Järvinen <[email protected]>--- drivers/pci/ats.c | 2 +- drivers/pci/probe.c | 14 +++++++++----- include/linux/pci.h | 2 +- include/uapi/linux/pci_regs.h | 1 + 4 files changed, 12 insertions(+), 7 deletions(-)drivers/pci/ats.c +1
-1@@ -377,7 +377,7 @@ int pci_enable_pasid(struct pci_dev *pdev, int features)if (WARN_ON (pdev->pasid_enabled))return -EBUSY ; -if (!pdev->eetlp_prefix_path && !pdev->pasid_no_tlp) +if (!pdev->eetlp_prefix_max && !pdev->pasid_no_tlp)return -EINVAL ;if (!pasid)drivers/pci/probe.c +5
-1@@ -2228,8 +2228,8 @@ static void pci_configure_relaxed_ordering(struct pci_dev *dev)static void pci_configure_eetlp_prefix (struct pci_dev *dev) { -#ifdef CONFIG_PCI_PASID struct pci_dev *bridge; +unsigned int eetlp_max;int pcie_type;u32 cap;@@ -2241,15 +2241,19 @@ static void pci_configure_eetlp_prefix(struct pci_dev *dev)return ; pcie_type =pci_pcie_type (dev); + + eetlp_max =FIELD_GET (PCI_EXP_DEVCAP2_EE_PREFIX_MAX , cap); +/* 00b means 4 */ + eetlp_max = eetlp_max ?:4 ;
drivers/pci/probe.c static void pci_configure_eetlp_prefix(struct pci_dev *dev) · line 2254
Quoted 10 lines
drivers/pci/probe.c +3
-3+if (pcie_type ==PCI_EXP_TYPE_ROOT_PORT || pcie_type ==PCI_EXP_TYPE_RC_END ) - dev->eetlp_prefix_path =1 ; + dev->eetlp_prefix_max = eetlp_max;else { bridge =pci_upstream_bridge (dev); -if (bridge && bridge->eetlp_prefix_path) - dev->eetlp_prefix_path =1 ; +if (bridge && bridge->eetlp_prefix_max)
What happens if they disagree? That is the bridge supports 2 and the device 3?
Quoted 8 lines
drivers/pci/probe.c +1
-1+ dev->eetlp_prefix_max = eetlp_max; } -#endif }
Re: [PATCH v6 5/8] PCI: Store # of supported End-End TLP Prefixes
Quoted 67 lines · Jonathan Cameron
Quoted 64 lines · On Fri, 13 Sep 2024 17:36:29 +0300 Ilpo Järvinen…
eetlp_prefix_pathin the structpci_devtells if End-End TLP Prefixes are supported by the path or not, the value is only calculated ifCONFIG_PCI_PASIDis set.The Max End-End TLP Prefixes field in the Device Capabilities Register 2 also tells how many (1-4) End-End TLP Prefixes are supported (PCIe r6 sec 7.5.3.15). The number of supported End-End Prefixes is useful for reading correct number of DWORDs from TLP Prefix Log register in AER capability (PCIe r6 sec 7.8.4.12).
Replace
eetlp_prefix_pathwitheetlp_prefix_maxand determine the number of supported End-End Prefixes regardless ofCONFIG_PCI_PASIDso that an upcoming commit generalizing TLP Prefix Log register reading does not have to read extra DWORDs for End-End Prefixes that never will be there.Signed-off-by Ilpo Järvinen <[email protected]>--- drivers/pci/ats.c | 2 +- drivers/pci/probe.c | 14 +++++++++----- include/linux/pci.h | 2 +- include/uapi/linux/pci_regs.h | 1 + 4 files changed, 12 insertions(+), 7 deletions(-)drivers/pci/ats.c +1
-1@@ -377,7 +377,7 @@ int pci_enable_pasid(struct pci_dev *pdev, int features)if (WARN_ON (pdev->pasid_enabled))return -EBUSY ; -if (!pdev->eetlp_prefix_path && !pdev->pasid_no_tlp) +if (!pdev->eetlp_prefix_max && !pdev->pasid_no_tlp)return -EINVAL ;if (!pasid)drivers/pci/probe.c +7
-2@@ -2228,8 +2228,8 @@ static void pci_configure_relaxed_ordering(struct pci_dev *dev)static void pci_configure_eetlp_prefix (struct pci_dev *dev) { -#ifdef CONFIG_PCI_PASID struct pci_dev *bridge; +unsigned int eetlp_max;int pcie_type;u32 cap;@@ -2241,15 +2241,19 @@ static void pci_configure_eetlp_prefix(struct pci_dev *dev)return ; pcie_type =pci_pcie_type (dev); + + eetlp_max =FIELD_GET (PCI_EXP_DEVCAP2_EE_PREFIX_MAX , cap); +/* 00b means 4 */ + eetlp_max = eetlp_max ?:4 ; +if (pcie_type ==PCI_EXP_TYPE_ROOT_PORT || pcie_type ==PCI_EXP_TYPE_RC_END ) - dev->eetlp_prefix_path =1 ; + dev->eetlp_prefix_max = eetlp_max;
Quoted 8 lines
drivers/pci/probe.c static void pci_configure_eetlp_prefix(struct pci_dev *dev) · line 2254
Quoted 5 lines
drivers/pci/probe.c +1
-2else { bridge =pci_upstream_bridge (dev); -if (bridge && bridge->eetlp_prefix_path) - dev->eetlp_prefix_path =1 ; +if (bridge && bridge->eetlp_prefix_max)What happens if they disagree? That is the bridge supports 2 and the device 3?
That's a good question.
The current code obviously only checks if Prefixes are supported or not so the max value doesn't matter for the existing code.
I went to read spec and my reading from TLP logging point of view is that the device's own maximum matters even if it might never get >2 Prefixes (r6.1 2.2.10.4 & 6.2.4).
AFAIK, things happen on low level and there's no way to program this value. So it's not like the kernel is telling that hw must only use x Prefixes at most if that's what you were worried about.
But there are more things to consider, e.g., I noticed End-End TLP Prefix Blocking in DevCtl2 which might impact the existing usage too and is not checked by the kernel currently.
My interest here lies on cleaning this up so I'm not sure if functional changes such as considering End-End TLP Prefix Blocking really matter for some case or not. I can obviously easily add the code for that too if it's required for this series to be acceptable but I don't have a test case for it. My main goal with this TLP logging consolidation series is to get to a point that extending it to support Flit mode is easier.
There's also TLP Prefix Log Present which I think the reader should consider, which matters to another patch in this series and I'm going to alter the length based on that flag.
Signature · 1 line
i.
PATCH v6 6/8 PCI: Add TLP Prefix reading into pcie_read_tlp_log()
In reply to Ilpo Järvinenpcie_read_tlp_log() handles only 4 Header Log DWORDs but TLP Prefix Log (PCIe r6.1 secs 7.8.4.12 & 7.9.14.13) may also be present.
Generalize pcie_read_tlp_log() and struct pcie_tlp_log to handle also TLP Prefix Log. The relevant registers are formatted identically in AER and DPC Capability, but has these variations:
a) The offsets of TLP Prefix Log registers vary.
b) DPC RP PIO TLP Prefix Log register can be < 4 DWORDs.
Therefore callers must pass the offset of the TLP Prefix Log register and the entire length to pcie_read_tlp_log() to be able to read the correct number of TLP Prefix DWORDs from the correct offset.
--- drivers/pci/pci.h | 5 +++- drivers/pci/pcie/aer.c | 4 ++- drivers/pci/pcie/dpc.c | 13 +++++----- drivers/pci/pcie/tlp.c | 49 +++++++++++++++++++++++++++++++---- include/linux/aer.h | 1 + include/uapi/linux/pci_regs.h | 1 + 6 files changed, 59 insertions(+), 14 deletions(-)
drivers/pci/pci.h +4 -1
@@ -483,7 +483,9 @@ struct aer_err_info {int aer_get_device_error_info (struct pci_dev *dev,struct aer_err_info *info);void aer_print_error (struct pci_dev *dev,struct aer_err_info *info); -int pcie_read_tlp_log (struct pci_dev *dev,int where,struct pcie_tlp_log *log); +int pcie_read_tlp_log (struct pci_dev *dev,int where,int where2, +unsigned int tlp_len,struct pcie_tlp_log *log); +unsigned int aer_tlp_log_len (struct pci_dev *dev);#endif /* CONFIG_PCIEAER */ #ifdef CONFIG_PCIEPORTBUS
@@ -502,6 +504,7 @@ void pci_dpc_init(struct pci_dev *pdev);void dpc_process_error (struct pci_dev *pdev);pci_ers_result_t dpc_reset_link (struct pci_dev *pdev);bool pci_dpc_recovered (struct pci_dev *pdev); +unsigned int dpc_tlp_log_len (struct pci_dev *dev);#else static inline void pci_save_dpc_state (struct pci_dev *dev) { }static inline void pci_restore_dpc_state (struct pci_dev *dev) { }
drivers/pci/pcie/aer.c +3 -1
@@ -1245,7 +1245,9 @@ int aer_get_device_error_info(struct pci_dev *dev, struct aer_err_info *info)if (info->status &AER_LOG_TLP_MASKS ) { info->tlp_header_valid =1 ; -pcie_read_tlp_log (dev, aer +PCI_ERR_HEADER_LOG , &info->tlp); +pcie_read_tlp_log (dev, aer +PCI_ERR_HEADER_LOG , + aer +PCI_ERR_PREFIX_LOG , +aer_tlp_log_len (dev), &info->tlp); } }
drivers/pci/pcie/dpc.c +6 -7
@@ -190,7 +190,7 @@ pci_ers_result_t dpc_reset_link(struct pci_dev *pdev)static void dpc_process_rp_pio_error (struct pci_dev *pdev) {u16 cap = pdev->dpc_cap, dpc_status, first_error; -u32 status, mask, sev, syserr, exc, log, prefix; +u32 status, mask, sev, syserr, exc, log;struct pcie_tlp_log tlp_log;int i;
@@ -217,20 +217,19 @@ static void dpc_process_rp_pio_error(struct pci_dev *pdev)if (pdev->dpc_rp_log_size <4 ) goto clear_status; -pcie_read_tlp_log (pdev, cap +PCI_EXP_DPC_RP_PIO_HEADER_LOG , &tlp_log); +pcie_read_tlp_log (pdev, cap +PCI_EXP_DPC_RP_PIO_HEADER_LOG , + cap +PCI_EXP_DPC_RP_PIO_TLPPREFIX_LOG , +dpc_tlp_log_len (pdev), &tlp_log);pci_err (pdev,"TLP Header: %#010x %#010x %#010x %#010x\n" , tlp_log.dw[0 ], tlp_log.dw[1 ], tlp_log.dw[2 ], tlp_log.dw[3 ]); +for (i =0 ; i < pdev->dpc_rp_log_size -5 ; i++) +pci_err (pdev,"TLP Prefix Header: dw%d, %#010x\n" , i, tlp_log.prefix[i]);if (pdev->dpc_rp_log_size <5 ) goto clear_status;pci_read_config_dword (pdev, cap +PCI_EXP_DPC_RP_PIO_IMPSPEC_LOG , &log);pci_err (pdev,"RP PIO ImpSpec Log %#010x\n" , log); -for (i =0 ; i < pdev->dpc_rp_log_size -5 ; i++) { -pci_read_config_dword (pdev, - cap +PCI_EXP_DPC_RP_PIO_TLPPREFIX_LOG + i *4 , &prefix); -pci_err (pdev,"TLP Prefix Header: dw%d, %#010x\n" , i, prefix); - } clear_status:pci_write_config_dword (pdev, cap +PCI_EXP_DPC_RP_PIO_STATUS , status); }
drivers/pci/pcie/tlp.c +44 -5
70 lines not on this page · raw message
include/linux/aer.h +1 -0
@@ -20,6 +20,7 @@ struct pci_dev;struct pcie_tlp_log {u32 dw[4 ]; +u32 prefix[4 ]; };struct aer_capability_regs {
include/uapi/linux/pci_regs.h +1 -0
@@ -803,6 +803,7 @@#define PCI_ERR_ROOT_FATAL_RCV 0x00000040/* Fatal Received */ #define PCI_ERR_ROOT_AER_IRQ 0xf8000000/* Advanced Error Interrupt Message Number */ #define PCI_ERR_ROOT_ERR_SRC 0x34/* Error Source Identification */ +#define PCI_ERR_PREFIX_LOG 0x38/* TLP Prefix LOG Register (up to 16 bytes) */ /* Virtual Channel */ #define PCI_VC_PORT_CAP1 0x04
Signature · 1 line
2.39.2
Re: [PATCH v6 6/8] PCI: Add TLP Prefix reading into pcie_read_tlp_log()
Quoted 7 lines · On Fri, 13 Sep 2024 17:36:30 +0300 Ilpo Järvinen…
pcie_read_tlp_log()handles only 4 Header Log DWORDs but TLP Prefix Log (PCIe r6.1 secs 7.8.4.12 & 7.9.14.13) may also be present.Generalize
pcie_read_tlp_log()and structpcie_tlp_logto handle also TLP Prefix Log. The relevant registers are formatted identically in AER and DPC Capability, but has these variations:
Quoted 8 lines
a) The offsets of TLP Prefix Log registers vary.
b) DPC RP PIO TLP Prefix Log register can be < 4 DWORDs.Therefore callers must pass the offset of the TLP Prefix Log register and the entire length to
pcie_read_tlp_log()to be able to read the correct number of TLP Prefix DWORDs from the correct offset.Signed-off-by Ilpo Järvinen <[email protected]>
Trivial comments below
Would have been nice if they'd just made the formats have the same sized holes etc!
drivers/pci/pcie/tlp.c line 13
Quoted 8 lines
drivers/pci/pcie/tlp.c +0
-0@@ -11,26 +11,65 @@/** * pcie_read_tlp_log - read TLP Header Log
Maybe update this to read TLP Header and Prefix Logs
drivers/pci/pcie/tlp.c line 46
Quoted 3 lines
drivers/pci/pcie/tlp.c +1
-0* @dev: PCIe device * @where: PCI Config offset of TLP Header Log +* @where2: PCI Config offset of TLP Prefix Log
Is it worth giving it a more specific name than where2? Possibly renaming where as well!
Quoted 12 lines
drivers/pci/pcie/tlp.c +3
-2+* @tlp_len: TLP Log length (Header Log + TLP Prefix Log in DWORDs) * @log: TLP Log structure to fill * * Fill @log from TLP Header Log registers, e.g., AER or DPC. * * Return: 0 on success and filled TLP Log structure, <0 on error. */ -int pcie_read_tlp_log (struct pci_dev *dev,int where, -struct pcie_tlp_log *log) +int pcie_read_tlp_log (struct pci_dev *dev,int where,int where2, +unsigned int tlp_len,struct pcie_tlp_log *log) {
Re: [PATCH v6 6/8] PCI: Add TLP Prefix reading into pcie_read_tlp_log()
Quoted 17 lines · Jonathan Cameron
Quoted 14 lines · On Fri, 13 Sep 2024 17:36:30 +0300 Ilpo Järvinen…
pcie_read_tlp_log()handles only 4 Header Log DWORDs but TLP Prefix Log (PCIe r6.1 secs 7.8.4.12 & 7.9.14.13) may also be present.Generalize
pcie_read_tlp_log()and structpcie_tlp_logto handle also TLP Prefix Log. The relevant registers are formatted identically in AER and DPC Capability, but has these variations:a) The offsets of TLP Prefix Log registers vary.
b) DPC RP PIO TLP Prefix Log register can be < 4 DWORDs.Therefore callers must pass the offset of the TLP Prefix Log register and the entire length to
pcie_read_tlp_log()to be able to read the correct number of TLP Prefix DWORDs from the correct offset.
Quoted 7 lines
Quoted 1 line
Signed-off-by Ilpo Järvinen <[email protected]>Trivial comments below
Reviewed-by Jonathan Cameron <[email protected]>Would have been nice if they'd just made the formats have the same sized holes etc!
That's not even the worst problem.
They managed to copy-paste most of the stuff into DPC (copy-paste is really obvious because the text still refers to AER in a DPC section :-)) but forgot to add a few capability fields into the DPC capability, most importantly, the bit that indicates whether TLP was logged in Flit mode or not And now we get to keep the pieces how to interpret the Log Registers (relates to the follow up series). :-(
Quoted 8 lines
drivers/pci/pcie/tlp.c line 13
Quoted 8 lines
drivers/pci/pcie/tlp.c +0
-0@@ -11,26 +11,65 @@/** * pcie_read_tlp_log - read TLP Header Log
Quoted 7 lines
Maybe update this to read TLP Header and Prefix Logs
drivers/pci/pcie/tlp.c line 46
Quoted 3 lines
drivers/pci/pcie/tlp.c +1
-0* @dev: PCIe device * @where: PCI Config offset of TLP Header Log +* @where2: PCI Config offset of TLP Prefix Log Is it worth giving it a more specific name than where2? Possibly renaming where as well!
Sure, why not.
Signature · 1 line
i.
Quoted 12 lines
Quoted 12 lines
drivers/pci/pcie/tlp.c +3
-2+* @tlp_len: TLP Log length (Header Log + TLP Prefix Log in DWORDs) * @log: TLP Log structure to fill * * Fill @log from TLP Header Log registers, e.g., AER or DPC. * * Return: 0 on success and filled TLP Log structure, <0 on error. */ -int pcie_read_tlp_log (struct pci_dev *dev,int where, -struct pcie_tlp_log *log) +int pcie_read_tlp_log (struct pci_dev *dev,int where,int where2, +unsigned int tlp_len,struct pcie_tlp_log *log) {
Re: [PATCH v6 6/8] PCI: Add TLP Prefix reading into pcie_read_tlp_log()
Quoted 45 lines · Ilpo Järvinen
Quoted 17 lines · Jonathan Cameron
Quoted 14 lines · On Fri, 13 Sep 2024 17:36:30 +0300 Ilpo Järvinen…
pcie_read_tlp_log()handles only 4 Header Log DWORDs but TLP Prefix Log (PCIe r6.1 secs 7.8.4.12 & 7.9.14.13) may also be present.Generalize
pcie_read_tlp_log()and structpcie_tlp_logto handle also TLP Prefix Log. The relevant registers are formatted identically in AER and DPC Capability, but has these variations:a) The offsets of TLP Prefix Log registers vary.
b) DPC RP PIO TLP Prefix Log register can be < 4 DWORDs.Therefore callers must pass the offset of the TLP Prefix Log register and the entire length to
pcie_read_tlp_log()to be able to read the correct number of TLP Prefix DWORDs from the correct offset.Quoted 7 lines
Quoted 1 line
Signed-off-by Ilpo Järvinen <[email protected]>Trivial comments below
Reviewed-by Jonathan Cameron <[email protected]>Would have been nice if they'd just made the formats have the same sized holes etc!
That's not even the worst problem.
They managed to copy-paste most of the stuff into DPC (copy-paste is really obvious because the text still refers to AER in a DPC section :-)) but forgot to add a few capability fields into the DPC capability, most importantly, the bit that indicates whether TLP was logged in Flit mode or not And now we get to keep the pieces how to interpret the Log Registers (relates to the follow up series). :-(
Quoted 9 lines
drivers/pci/pcie/tlp.c line 13
Quoted 8 lines
drivers/pci/pcie/tlp.c +0
-0@@ -11,26 +11,65 @@/** * pcie_read_tlp_log - read TLP Header Log Maybe update this to read TLP Header and Prefix Logs
Quoted 8 lines
Quoted 6 lines
drivers/pci/pcie/tlp.c line 46
Quoted 3 lines
drivers/pci/pcie/tlp.c +1
-0* @dev: PCIe device * @where: PCI Config offset of TLP Header Log +* @where2: PCI Config offset of TLP Prefix Log Is it worth giving it a more specific name than where2? Possibly renaming where as well!
Sure, why not.
Hi again,
After doing this rename, I rebased the Flit mode series on top of it and realized there's one small problem with naming where2. It will be overloaded between TLP Prefix Log (Non-Flit mode) and extended TLP Header Log (DW5+) (Flit mode) so I'm not sure if there's really going to be a good name for it.
Signature · 1 line
i.
Re: [PATCH v6 6/8] PCI: Add TLP Prefix reading into pcie_read_tlp_log()
Quoted 56 lines · On Thu, 12 Dec 2024 20:48:38 +0200 (EET) Ilpo Järvinen…
Quoted 45 lines · Ilpo Järvinen
Quoted 17 lines · Jonathan Cameron
Quoted 14 lines · On Fri, 13 Sep 2024 17:36:30 +0300 Ilpo Järvinen…
pcie_read_tlp_log()handles only 4 Header Log DWORDs but TLP Prefix Log (PCIe r6.1 secs 7.8.4.12 & 7.9.14.13) may also be present.Generalize
pcie_read_tlp_log()and structpcie_tlp_logto handle also TLP Prefix Log. The relevant registers are formatted identically in AER and DPC Capability, but has these variations:a) The offsets of TLP Prefix Log registers vary.
b) DPC RP PIO TLP Prefix Log register can be < 4 DWORDs.Therefore callers must pass the offset of the TLP Prefix Log register and the entire length to
pcie_read_tlp_log()to be able to read the correct number of TLP Prefix DWORDs from the correct offset.Quoted 7 lines
Quoted 1 line
Signed-off-by Ilpo Järvinen <[email protected]>Trivial comments below
Reviewed-by Jonathan Cameron <[email protected]>Would have been nice if they'd just made the formats have the same sized holes etc!
That's not even the worst problem.
They managed to copy-paste most of the stuff into DPC (copy-paste is really obvious because the text still refers to AER in a DPC section :-)) but forgot to add a few capability fields into the DPC capability, most importantly, the bit that indicates whether TLP was logged in Flit mode or not And now we get to keep the pieces how to interpret the Log Registers (relates to the follow up series). :-(
Quoted 9 lines
drivers/pci/pcie/tlp.c line 13
Quoted 8 lines
drivers/pci/pcie/tlp.c +0
-0@@ -11,26 +11,65 @@/** * pcie_read_tlp_log - read TLP Header Log Maybe update this to read TLP Header and Prefix Logs
Quoted 8 lines
Quoted 6 lines
drivers/pci/pcie/tlp.c line 46
Quoted 3 lines
drivers/pci/pcie/tlp.c +1
-0* @dev: PCIe device * @where: PCI Config offset of TLP Header Log +* @where2: PCI Config offset of TLP Prefix Log Is it worth giving it a more specific name than where2? Possibly renaming where as well!
Sure, why not.
Quoted 8 lines
Hi again,
After doing this rename, I rebased the Flit mode series on top of it and realized there's one small problem with naming where2. It will be overloaded between TLP Prefix Log (Non-Flit mode) and extended TLP Header Log (DW5+) (Flit mode) so I'm not sure if there's really going to be a good name for it.
Ah. I should read that series ;)
Absolutely fine - sometime generic naming is the best we can do!
Jonathan
PATCH v6 7/8 PCI: Create helper to print TLP Header and Prefix Log
In reply to Ilpo JärvinenAdd pcie_print_tlp_log() helper to print TLP Header and Prefix Log. Print End-End Prefixes only if they are non-zero.
Consolidate the few places which currently print TLP using custom formatting.
The first attempt used pr_cont() instead of building a string first but it turns out pr_cont() is not compatible with pci_err() and prints on a separate line. When I asked about this, Andy Shevchenko suggested pr_cont() should not be used in the first place (to eventually get rid of it) so pr_cont() is now replaced with building the string first.
--- drivers/pci/pci.h | 2 ++ drivers/pci/pcie/aer.c | 10 ++-------- drivers/pci/pcie/dpc.c | 5 +---- drivers/pci/pcie/tlp.c | 31 +++++++++++++++++++++++++++++++ 4 files changed, 36 insertions(+), 12 deletions(-)
drivers/pci/pci.h +2 -0
@@ -486,6 +486,8 @@ void aer_print_error(struct pci_dev *dev, struct aer_err_info *info);int pcie_read_tlp_log (struct pci_dev *dev,int where,int where2,unsigned int tlp_len,struct pcie_tlp_log *log);unsigned int aer_tlp_log_len (struct pci_dev *dev); +void pcie_print_tlp_log (const struct pci_dev *dev, +const struct pcie_tlp_log *log,const char *pfx);#endif /* CONFIG_PCIEAER */ #ifdef CONFIG_PCIEPORTBUS
drivers/pci/pcie/aer.c +2 -8
@@ -664,12 +664,6 @@ static void pci_rootport_aer_stats_incr(struct pci_dev *pdev, } } -static void __print_tlp_header (struct pci_dev *dev,struct pcie_tlp_log *t) -{ -pci_err (dev," TLP Header: %08x %08x %08x %08x\n" , - t->dw[0 ], t->dw[1 ], t->dw[2 ], t->dw[3 ]); -} -static void __aer_print_error (struct pci_dev *dev,struct aer_err_info *info) {
@@ -724,7 +718,7 @@ void aer_print_error(struct pci_dev *dev, struct aer_err_info *info)__aer_print_error (dev, info);if (info->tlp_header_valid) -__print_tlp_header (dev, &info->tlp); +pcie_print_tlp_log (dev, &info->tlp,dev_fmt (" " )); out:if (info->id && info->error_dev_num >1 && info->id == id)
@@ -796,7 +790,7 @@ void pci_print_aer(struct pci_dev *dev, int aer_severity, aer->uncor_severity);if (tlp_header_valid) -__print_tlp_header (dev, &aer->header_log); +pcie_print_tlp_log (dev, &aer->header_log,dev_fmt (" " ));trace_aer_event (dev_name (&dev->dev), (status & ~mask), aer_severity, tlp_header_valid, &aer->header_log);
drivers/pci/pcie/dpc.c +1 -4
@@ -220,10 +220,7 @@ static void dpc_process_rp_pio_error(struct pci_dev *pdev)pcie_read_tlp_log (pdev, cap +PCI_EXP_DPC_RP_PIO_HEADER_LOG , cap +PCI_EXP_DPC_RP_PIO_TLPPREFIX_LOG ,dpc_tlp_log_len (pdev), &tlp_log); -pci_err (pdev,"TLP Header: %#010x %#010x %#010x %#010x\n" , - tlp_log.dw[0 ], tlp_log.dw[1 ], tlp_log.dw[2 ], tlp_log.dw[3 ]); -for (i =0 ; i < pdev->dpc_rp_log_size -5 ; i++) -pci_err (pdev,"TLP Prefix Header: dw%d, %#010x\n" , i, tlp_log.prefix[i]); +pcie_print_tlp_log (pdev, &tlp_log,dev_fmt ("" ));if (pdev->dpc_rp_log_size <5 ) goto clear_status;
drivers/pci/pcie/tlp.c +31 -0
@@ -6,6 +6,7 @@*/ #include <linux/aer.h> +#include <linux/array_size.h> #include <linux/pci.h> #include <linux/string.h>
@@ -76,3 +77,33 @@ int pcie_read_tlp_log(struct pci_dev *dev, int where, int where2,return 0 ; } + +/** +* pcie_print_tlp_log - Print TLP Header / Prefix Log contents +* @dev: PCIe device +* @log: TLP Log structure +* @pfx: String prefix (for print out indentation) +* +* Prints TLP Header and Prefix Log information held by @log. +*/ +void pcie_print_tlp_log (const struct pci_dev *dev, +const struct pcie_tlp_log *log,const char *pfx) +{ +char buf[(10 +1 ) * (4 +ARRAY_SIZE (log->prefix)) +14 +1 ]; +unsigned int i; +int len; + + len =scnprintf (buf,sizeof (buf),"%#010x %#010x %#010x %#010x" , + log->dw[0 ], log->dw[1 ], log->dw[2 ], log->dw[3 ]); + +if (log->prefix[0 ]) + len +=scnprintf (buf + len,sizeof (buf) - len," E-E Prefixes:" ); +for (i =0 ; i <ARRAY_SIZE (log->prefix); i++) { +if (!log->prefix[i]) +break ; + len +=scnprintf (buf + len,sizeof (buf) - len, +" %#010x" , log->prefix[i]); + } + +pci_err (dev,"%sTLP Header: %s\n" , pfx, buf); +}
Signature · 1 line
2.39.2
Re: [PATCH v6 7/8] PCI: Create helper to print TLP Header and Prefix Log
Quoted 6 lines · On Fri, 13 Sep 2024 17:36:31 +0300 Ilpo Järvinen…
Add
pcie_print_tlp_log()helper to print TLP Header and Prefix Log. Print End-End Prefixes only if they are non-zero.Consolidate the few places which currently print TLP using custom formatting.
Quoted 7 lines
The first attempt used
pr_cont()instead of building a string first but it turns outpr_cont()is not compatible withpci_err()and prints on a separate line. When I asked about this, Andy Shevchenko suggestedpr_cont()should not be used in the first place (to eventually get rid of it) sopr_cont()is now replaced with building the string first.Signed-off-by Ilpo Järvinen <[email protected]>
A couple of trivial things inline but looks good to me either way.
Quoted 13 lines
drivers/pci/pcie/tlp.c +1
-0@@ -6,6 +6,7 @@*/ #include <linux/aer.h> +#include <linux/array_size.h> #include <linux/pci.h> #include <linux/string.h>
drivers/pci/pcie/tlp.c int pcie_read_tlp_log(struct pci_dev *dev, int where, int where2, · line 85
Quoted 9 lines
drivers/pci/pcie/tlp.c +6
-0@@ -76,3 +77,33 @@ int pcie_read_tlp_log(struct pci_dev *dev, int where, int where2,return 0 ; } + +/** +* pcie_print_tlp_log - Print TLP Header / Prefix Log contents +* @dev: PCIe device +* @log: TLP Log structure +* @pfx: String prefix (for print out indentation)
Code doesn't care if it is indentation or ponies. So does it make sense to say anything beyond String prefix?
drivers/pci/pcie/tlp.c int pcie_read_tlp_log(struct pci_dev *dev, int where, int where2, · line 92
Quoted 7 lines
drivers/pci/pcie/tlp.c +7
-0+* +* Prints TLP Header and Prefix Log information held by @log. +*/ +void pcie_print_tlp_log (const struct pci_dev *dev, +const struct pcie_tlp_log *log,const char *pfx) +{ +char buf[(10 +1 ) * (4 +ARRAY_SIZE (log->prefix)) +14 +1 ];
Can we associate the 14 with the prefixes string by having that as a const char * and using strlen() It was a tiny bit irritating to count the characters whilst reviewing this ;)
Quoted 17 lines
drivers/pci/pcie/tlp.c +17
-0+unsigned int i; +int len; + + len =scnprintf (buf,sizeof (buf),"%#010x %#010x %#010x %#010x" , + log->dw[0 ], log->dw[1 ], log->dw[2 ], log->dw[3 ]); + +if (log->prefix[0 ]) + len +=scnprintf (buf + len,sizeof (buf) - len," E-E Prefixes:" ); +for (i =0 ; i <ARRAY_SIZE (log->prefix); i++) { +if (!log->prefix[i]) +break ; + len +=scnprintf (buf + len,sizeof (buf) - len, +" %#010x" , log->prefix[i]); + } + +pci_err (dev,"%sTLP Header: %s\n" , pfx, buf); +}
PATCH v6 8/8 PCI/AER: Add prefixes to printouts
In reply to Ilpo JärvinenOnly part of the AER diagnostic printouts use "AER:" prefix because they use low-level pci_printk() directly to allow selecting level.
Add "AER:" prefix to lines that are printed with pci_printk().
--- drivers/pci/pcie/aer.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/pci/pcie/aer.c +6 -6
@@ -685,8 +685,8 @@ static void __aer_print_error(struct pci_dev *dev,if (!errmsg) errmsg ="Unknown Error Bit" ; -pci_printk (level, dev," [%2d] %-22s%s\n" , i, errmsg, - info->first_error == i ?" (First)" :"" ); +pci_printk (level, dev,"%s [%2d] %-22s%s\n" ,dev_fmt ("" ), i, + errmsg, info->first_error == i ?" (First)" :"" ); }pci_dev_aer_stats_incr (dev, info); }
@@ -708,12 +708,12 @@ void aer_print_error(struct pci_dev *dev, struct aer_err_info *info) level = (info->severity ==AER_CORRECTABLE ) ?KERN_WARNING :KERN_ERR ; -pci_printk (level, dev,"PCIe Bus Error: severity=%s, type=%s, (%s)\n" , - aer_error_severity_string[info->severity], +pci_printk (level, dev,"%sPCIe Bus Error: severity=%s, type=%s, (%s)\n" , +dev_fmt ("" ), aer_error_severity_string[info->severity], aer_error_layer[layer], aer_agent_string[agent]); -pci_printk (level, dev," device [%04x:%04x] error status/mask=%08x/%08x\n" , - dev->vendor, dev->device, info->status, info->mask); +pci_printk (level, dev,"%s device [%04x:%04x] error status/mask=%08x/%08x\n" , +dev_fmt ("" ), dev->vendor, dev->device, info->status, info->mask);__aer_print_error (dev, info);
Signature · 1 line
2.39.2
Re: [PATCH v6 8/8] PCI/AER: Add prefixes to printouts
Quoted 6 lines · On Fri, 13 Sep 2024 17:36:32 +0300 Ilpo Järvinen…
Only part of the AER diagnostic printouts use "AER:" prefix because they use low-level
pci_printk()directly to allow selecting level.Add "AER:" prefix to lines that are printed with
pci_printk().Signed-off-by Ilpo Järvinen <[email protected]>
Makes sense.
Quoted 7 lines · Ilpo Järvinen
This series has the remaining patches of the AER & DPC TLP Log handling consolidation and now includes a few minor improvements to the earlier accepted TLP Logging code.
v6:
- Preserve "AER:"/"DPC:" prefix on the printed TLP line
- New patch to add "AER:" also on other lines of the AER error dump
Hi Bjorn,
A small reminder this series exists.
Signature · 1 line
i.
Quoted 49 lines
v5:
- Fix build with AER=y and DPC=n
- Match kerneldoc and function parameter namev4:
- Added patches:
- Remove EXPORT ofpcie_read_tlp_log()
- Moved code topcie/tlp.cand build only with AER enabled
- Match variables in prototype and function
- int -> unsigned int conversion
-eetlp_prefix_maxinto own patch
- structpcie_tlp_logparam consistently called "log" withintlp.c
- Moved function prototypes intodrivers/pci/pci.h
- Describe AER/DPC differences more clearly in one commit messagev3:
- Small rewording in a commit messagev2:
- Don't add EXPORT()s
- Don't include igxbe changes
- Don't usepr_cont()as it's incompatible withpci_err()and according to Andy Shevchenko should not be used in the first placeIlpo Järvinen (8):
PCI: Don't exposepcie_read_tlp_log()outside of PCI subsystem
PCI: Move TLP Log handling to own file
PCI: Makepcie_read_tlp_log()signature same
PCI: Use unsigned int i inpcie_read_tlp_log()
PCI: Store # of supported End-End TLP Prefixes
PCI: Add TLP Prefix reading intopcie_read_tlp_log()
PCI: Create helper to print TLP Header and Prefix Log PCI/AER: Add prefixes to printoutsdrivers/pci/ats.c | 2 +- drivers/pci/pci.c | 28 --------- drivers/pci/pci.h | 9 +++ drivers/pci/pcie/Makefile | 2 +- drivers/pci/pcie/aer.c | 26 ++++---- drivers/pci/pcie/dpc.c | 14 ++--- drivers/pci/pcie/tlp.c | 109 ++++++++++++++++++++++++++++++++++ drivers/pci/probe.c | 14 +++-- include/linux/aer.h | 3 +- include/linux/pci.h | 2 +- include/uapi/linux/pci_regs.h | 2 + 11 files changed, 149 insertions(+), 62 deletions(-) create mode 100644 drivers/pci/pcie/tlp.c
Quoted 6 lines · Ilpo Järvinen
Quoted 4 lines · Ilpo Järvinen
This series has the remaining patches of the AER & DPC TLP Log handling consolidation and now includes a few minor improvements to the earlier accepted TLP Logging code.
Quoted 7 lines
Quoted 3 lines
v6:
- Preserve "AER:"/"DPC:" prefix on the printed TLP line
- New patch to add "AER:" also on other lines of the AER error dumpHi Bjorn,
A small reminder this series exists.
Another reminder about this series. :-)
Signature · 1 line
i.