| Message ID | 20121204170436.31397.40857.stgit@grignak.americas.hpqcorp.net (mailing list archive) |
|---|---|
| State | New, archived |
| Delegated to: | Bjorn Helgaas |
| Headers | show |
Em Tue, 04 Dec 2012 10:04:36 -0700 Lance Ortiz <lance.ortiz@hp.com> escreveu: > This patch will provide a more reliable and easy way for user-space > applications to have access to AER logs rather than reading them from the > message buffer. It also provides a way to notify user-space when an AER > event occurs. > > The aer driver is updated to generate a trace event of function 'aer_event' > when a PCIe error is reported over the AER interface. The trace event was > added to both the interrupt based aer path and the firmware first path. > > v1-v2 fix compile errors in ifdefs. > v2-v3 Update to new location of trace header. Update print to remove > warning. > v3-v4 Reworked logic when getting ready to call cper_print_aer > Signed-off-by: Lance Ortiz <lance.ortiz@hp.com> > --- > > 0 files changed, 0 insertions(+), 0 deletions(-) > > diff --git a/drivers/acpi/apei/cper.c b/drivers/acpi/apei/cper.c > index e6defd8..4a3e945 100644 > --- a/drivers/acpi/apei/cper.c > +++ b/drivers/acpi/apei/cper.c > @@ -29,6 +29,7 @@ > #include <linux/time.h> > #include <linux/cper.h> > #include <linux/acpi.h> > +#include <linux/pci.h> > #include <linux/aer.h> > > /* > @@ -249,6 +250,10 @@ static const char *cper_pcie_port_type_strs[] = { > static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie, > const struct acpi_hest_generic_data *gdata) > { > +#ifdef CONFIG_ACPI_APEI_PCIEAER > + struct pci_dev *dev; > +#endif > + > if (pcie->validation_bits & CPER_PCIE_VALID_PORT_TYPE) > printk("%s""port_type: %d, %s\n", pfx, pcie->port_type, > pcie->port_type < ARRAY_SIZE(cper_pcie_port_type_strs) ? > @@ -281,10 +286,18 @@ static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie, > "%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n", > pfx, pcie->bridge.secondary_status, pcie->bridge.control); > #ifdef CONFIG_ACPI_APEI_PCIEAER > - if (pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) { > - struct aer_capability_regs *aer_regs = (void *)pcie->aer_info; > - cper_print_aer(pfx, gdata->error_severity, aer_regs); > + dev = pci_get_domain_bus_and_slot(pcie->device_id.segment, > + pcie->device_id.bus, pcie->device_id.function); > + if (!dev) { > + pr_info("PCI AER Cannot get PCI device %04x:%02x:%02x.%d\n", > + pcie->device_id.segment, pcie->device_id.bus, > + pcie->device_id.slot, pcie->device_id.function); Hmm... please correct if I'm wrong, but an error happened at PCI, and also a kernel bug that prevented it to get the proper PCI device... IMHO, the message here should be stronger, and likely printed via pr_err(). > + return; > } > + if (pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) > + cper_print_aer(dev, gdata->error_severity, > + (struct aer_capability_regs *) pcie->aer_info); > + pci_dev_put(dev); > #endif > } > > diff --git a/drivers/pci/pcie/aer/aerdrv_errprint.c b/drivers/pci/pcie/aer/aerdrv_errprint.c > index 3ea5173..34d96e4 100644 > --- a/drivers/pci/pcie/aer/aerdrv_errprint.c > +++ b/drivers/pci/pcie/aer/aerdrv_errprint.c > @@ -23,6 +23,9 @@ > > #include "aerdrv.h" > > +#define CREATE_TRACE_POINTS > +#include <trace/events/ras.h> > + > #define AER_AGENT_RECEIVER 0 > #define AER_AGENT_REQUESTER 1 > #define AER_AGENT_COMPLETER 2 > @@ -194,6 +197,8 @@ void aer_print_error(struct pci_dev *dev, struct aer_err_info *info) > if (info->id && info->error_dev_num > 1 && info->id == id) > printk("%s"" Error of this Agent(%04x) is reported first\n", > prefix, id); > + trace_aer_event(dev_name(&dev->dev), (info->status & ~info->mask), > + info->severity); > } > > void aer_print_port_info(struct pci_dev *dev, struct aer_err_info *info) > @@ -217,12 +222,13 @@ int cper_severity_to_aer(int cper_severity) > } > EXPORT_SYMBOL_GPL(cper_severity_to_aer); > > -void cper_print_aer(const char *prefix, int cper_severity, > +void cper_print_aer(struct pci_dev *dev, int cper_severity, > struct aer_capability_regs *aer) > { > int aer_severity, layer, agent, status_strs_size, tlp_header_valid = 0; > u32 status, mask; > const char **status_strs; > + char *prefix = NULL; > > aer_severity = cper_severity_to_aer(cper_severity); > if (aer_severity == AER_CORRECTABLE) { > @@ -259,5 +265,7 @@ void cper_print_aer(const char *prefix, int cper_severity, > *(tlp + 8), *(tlp + 15), *(tlp + 14), > *(tlp + 13), *(tlp + 12)); > } > + trace_aer_event(dev_name(&dev->dev), (status & ~mask), > + aer_severity); > } > #endif > diff --git a/include/linux/aer.h b/include/linux/aer.h > index 544abdb..7b86dc6 100644 > --- a/include/linux/aer.h > +++ b/include/linux/aer.h > @@ -49,7 +49,7 @@ static inline int pci_cleanup_aer_uncorrect_error_status(struct pci_dev *dev) > } > #endif > > -extern void cper_print_aer(const char *prefix, int cper_severity, > +extern void cper_print_aer(struct pci_dev *dev, int cper_severity, > struct aer_capability_regs *aer); > extern int cper_severity_to_aer(int cper_severity); > extern void aer_recover_queue(int domain, unsigned int bus, unsigned int devfn, > After addressing the above: Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
> > + if (!dev) { > > + pr_info("PCI AER Cannot get PCI device > %04x:%02x:%02x.%d\n", > > + pcie->device_id.segment, pcie->device_id.bus, > > + pcie->device_id.slot, pcie->device_id.function); > > Hmm... please correct if I'm wrong, but an error happened at PCI, and > also a > kernel bug that prevented it to get the proper PCI device... > > IMHO, the message here should be stronger, and likely printed via > pr_err(). > Mauro, I modeled this message after other places in the kernel where this function failed. So I figured it would be safe to be consistent there. I agree though that it should be pr_err(). I can make that change. -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Em Tue, 4 Dec 2012 20:14:10 +0000 "Ortiz, Lance E" <lance.oritz@hp.com> escreveu: > > > + if (!dev) { > > > + pr_info("PCI AER Cannot get PCI device > > %04x:%02x:%02x.%d\n", > > > + pcie->device_id.segment, pcie->device_id.bus, > > > + pcie->device_id.slot, pcie->device_id.function); > > > > Hmm... please correct if I'm wrong, but an error happened at PCI, and > > also a > > kernel bug that prevented it to get the proper PCI device... > > > > IMHO, the message here should be stronger, and likely printed via > > pr_err(). > > > > Mauro, > > I modeled this message after other places in the kernel where this function failed. So I figured it would be safe to be consistent there. I agree though that it should be pr_err(). I can make that change. I understand. On most cases, this may not be a critical issue. However, in this particular case, if PCI AER got an error, but the device is not found when trying to handle it, it can be an indication that the PCI device has a more serious issue. So, I'm in favor of changing it, and likely be more verbose at the error message, saying that the device was not found while trying to report an error condition that happened there. It could make sense to even send a trace for the daemon to be aware of the error, on some pci device that vanished likely due to the error. Regards, Mauro -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Dec 04, 2012 at 06:31:10PM -0200, Mauro Carvalho Chehab wrote: > However, in this particular case, if PCI AER got an error, but the > device is not found when trying to handle it, it can be an indication > that the PCI device has a more serious issue. So, I'm in favor > of changing it, and likely be more verbose at the error message, > saying that the device was not found while trying to report an error > condition that happened there. It could make sense to even send a > trace for the daemon to be aware of the error, on some pci device that > vanished likely due to the error. Let's leave it at pr_err now and if we actually start seeing errors like that and decide that they need more verbose reporting, then to change it to whatever's best. IOW, do the empirical approach. Thanks.
diff --git a/drivers/acpi/apei/cper.c b/drivers/acpi/apei/cper.c index e6defd8..4a3e945 100644 --- a/drivers/acpi/apei/cper.c +++ b/drivers/acpi/apei/cper.c @@ -29,6 +29,7 @@ #include <linux/time.h> #include <linux/cper.h> #include <linux/acpi.h> +#include <linux/pci.h> #include <linux/aer.h> /* @@ -249,6 +250,10 @@ static const char *cper_pcie_port_type_strs[] = { static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie, const struct acpi_hest_generic_data *gdata) { +#ifdef CONFIG_ACPI_APEI_PCIEAER + struct pci_dev *dev; +#endif + if (pcie->validation_bits & CPER_PCIE_VALID_PORT_TYPE) printk("%s""port_type: %d, %s\n", pfx, pcie->port_type, pcie->port_type < ARRAY_SIZE(cper_pcie_port_type_strs) ? @@ -281,10 +286,18 @@ static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie, "%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n", pfx, pcie->bridge.secondary_status, pcie->bridge.control); #ifdef CONFIG_ACPI_APEI_PCIEAER - if (pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) { - struct aer_capability_regs *aer_regs = (void *)pcie->aer_info; - cper_print_aer(pfx, gdata->error_severity, aer_regs); + dev = pci_get_domain_bus_and_slot(pcie->device_id.segment, + pcie->device_id.bus, pcie->device_id.function); + if (!dev) { + pr_info("PCI AER Cannot get PCI device %04x:%02x:%02x.%d\n", + pcie->device_id.segment, pcie->device_id.bus, + pcie->device_id.slot, pcie->device_id.function); + return; } + if (pcie->validation_bits & CPER_PCIE_VALID_AER_INFO) + cper_print_aer(dev, gdata->error_severity, + (struct aer_capability_regs *) pcie->aer_info); + pci_dev_put(dev); #endif } diff --git a/drivers/pci/pcie/aer/aerdrv_errprint.c b/drivers/pci/pcie/aer/aerdrv_errprint.c index 3ea5173..34d96e4 100644 --- a/drivers/pci/pcie/aer/aerdrv_errprint.c +++ b/drivers/pci/pcie/aer/aerdrv_errprint.c @@ -23,6 +23,9 @@ #include "aerdrv.h" +#define CREATE_TRACE_POINTS +#include <trace/events/ras.h> + #define AER_AGENT_RECEIVER 0 #define AER_AGENT_REQUESTER 1 #define AER_AGENT_COMPLETER 2 @@ -194,6 +197,8 @@ void aer_print_error(struct pci_dev *dev, struct aer_err_info *info) if (info->id && info->error_dev_num > 1 && info->id == id) printk("%s"" Error of this Agent(%04x) is reported first\n", prefix, id); + trace_aer_event(dev_name(&dev->dev), (info->status & ~info->mask), + info->severity); } void aer_print_port_info(struct pci_dev *dev, struct aer_err_info *info) @@ -217,12 +222,13 @@ int cper_severity_to_aer(int cper_severity) } EXPORT_SYMBOL_GPL(cper_severity_to_aer); -void cper_print_aer(const char *prefix, int cper_severity, +void cper_print_aer(struct pci_dev *dev, int cper_severity, struct aer_capability_regs *aer) { int aer_severity, layer, agent, status_strs_size, tlp_header_valid = 0; u32 status, mask; const char **status_strs; + char *prefix = NULL; aer_severity = cper_severity_to_aer(cper_severity); if (aer_severity == AER_CORRECTABLE) { @@ -259,5 +265,7 @@ void cper_print_aer(const char *prefix, int cper_severity, *(tlp + 8), *(tlp + 15), *(tlp + 14), *(tlp + 13), *(tlp + 12)); } + trace_aer_event(dev_name(&dev->dev), (status & ~mask), + aer_severity); } #endif diff --git a/include/linux/aer.h b/include/linux/aer.h index 544abdb..7b86dc6 100644 --- a/include/linux/aer.h +++ b/include/linux/aer.h @@ -49,7 +49,7 @@ static inline int pci_cleanup_aer_uncorrect_error_status(struct pci_dev *dev) } #endif -extern void cper_print_aer(const char *prefix, int cper_severity, +extern void cper_print_aer(struct pci_dev *dev, int cper_severity, struct aer_capability_regs *aer); extern int cper_severity_to_aer(int cper_severity); extern void aer_recover_queue(int domain, unsigned int bus, unsigned int devfn,
This patch will provide a more reliable and easy way for user-space applications to have access to AER logs rather than reading them from the message buffer. It also provides a way to notify user-space when an AER event occurs. The aer driver is updated to generate a trace event of function 'aer_event' when a PCIe error is reported over the AER interface. The trace event was added to both the interrupt based aer path and the firmware first path. v1-v2 fix compile errors in ifdefs. v2-v3 Update to new location of trace header. Update print to remove warning. v3-v4 Reworked logic when getting ready to call cper_print_aer Signed-off-by: Lance Ortiz <lance.ortiz@hp.com> --- 0 files changed, 0 insertions(+), 0 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html