| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAErSpo4_h+yrjGTEpJ-Q=nwfcd3xi0qa0z6NQad1EQDZom8kUQ@mail.gmail.com>
Date: Fri, 24 May 2013 13:36:14 -0600
From: Bjorn Helgaas <bhelgaas@...gle.com>
To: Lance Ortiz <lance.ortiz@...com>
Cc: "lance_ortiz@...mail.com" <lance_ortiz@...mail.com>,
Jiang Liu <jiang.liu@...wei.com>,
Tony Luck <tony.luck@...el.com>,
Borislav Petkov <bp@...en8.de>,
Steven Rostedt <rostedt@...dmis.org>,
Mauro Carvalho Chehab <mchehab@...hat.com>,
"linux-acpi@...r.kernel.org" <linux-acpi@...r.kernel.org>,
"linux-pci@...r.kernel.org" <linux-pci@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] aerdrv: Move cper_print_aer() call out of interrupt context
On Thu, May 9, 2013 at 1:51 PM, Lance Ortiz <lance.ortiz@...com> wrote:
> The following warning was seen on 3.9 when a corrected PCIe error was being
> handled by the AER subsystem.
>
> WARNING: at .../drivers/pci/search.c:214 pci_get_dev_by_id+0x8a/0x90()
>
> This occurred because a call to pci_get_domain_bus_and_slot() was added to
> cper_print_pcie() to setup for the call to cper_print_aer(). The warning
> showed up because cper_print_pcie() is called in an interrupt context and
> pci_get* functions are not supposed to be called in that context.
>
> The solution is to move the cper_print_aer() call out of the interrupt
> context and into aer_recover_work_func() to avoid any warnings when calling
> pci_get* functions.
>
> v2 - Re-worded header text. Removed prefix arg from cper_print_aer().
> Added TODO message in cper_print_aer().
>
> Signed-off-by: Lance Ortiz <lance.ortiz@...com>
> Acked-by: Borislav Petkov <bp@...e.de>
> ---
>
> drivers/acpi/apei/cper.c | 18 ------------------
> drivers/acpi/apei/ghes.c | 3 ++-
> drivers/pci/pcie/aer/aerdrv_core.c | 6 +++++-
> drivers/pci/pcie/aer/aerdrv_errprint.c | 10 ++++++++--
> include/linux/aer.h | 4 ++--
> 5 files changed, 17 insertions(+), 24 deletions(-)
>
> diff --git a/drivers/acpi/apei/cper.c b/drivers/acpi/apei/cper.c
> index 1e5d8a4..8713229 100644
> --- a/drivers/acpi/apei/cper.c
> +++ b/drivers/acpi/apei/cper.c
> @@ -250,10 +250,6 @@ 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) ?
> @@ -285,20 +281,6 @@ static void cper_print_pcie(const char *pfx, const struct cper_sec_pcie *pcie,
> printk(
> "%s""bridge: secondary_status: 0x%04x, control: 0x%04x\n",
> pfx, pcie->bridge.secondary_status, pcie->bridge.control);
> -#ifdef CONFIG_ACPI_APEI_PCIEAER
> - dev = pci_get_domain_bus_and_slot(pcie->device_id.segment,
> - pcie->device_id.bus, pcie->device_id.function);
> - if (!dev) {
> - pr_err("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(pfx, dev, gdata->error_severity,
> - (struct aer_capability_regs *) pcie->aer_info);
> - pci_dev_put(dev);
> -#endif
> }
>
> static const char *apei_estatus_section_flag_strs[] = {
> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
> index d668a8a..f2084b5 100644
> --- a/drivers/acpi/apei/ghes.c
> +++ b/drivers/acpi/apei/ghes.c
> @@ -454,7 +454,8 @@ static void ghes_do_proc(struct ghes *ghes,
> aer_severity = cper_severity_to_aer(sev);
> aer_recover_queue(pcie_err->device_id.segment,
> pcie_err->device_id.bus,
> - devfn, aer_severity);
> + devfn, aer_severity,
> + pcie_err->aer_info);
> }
>
> }
> diff --git a/drivers/pci/pcie/aer/aerdrv_core.c b/drivers/pci/pcie/aer/aerdrv_core.c
> index 564d97f..ceba124 100644
> --- a/drivers/pci/pcie/aer/aerdrv_core.c
> +++ b/drivers/pci/pcie/aer/aerdrv_core.c
> @@ -582,6 +582,7 @@ struct aer_recover_entry
> u8 devfn;
> u16 domain;
> int severity;
> + u8 *regs;
> };
>
> static DEFINE_KFIFO(aer_recover_ring, struct aer_recover_entry,
> @@ -595,7 +596,7 @@ static DEFINE_SPINLOCK(aer_recover_ring_lock);
> static DECLARE_WORK(aer_recover_work, aer_recover_work_func);
>
> void aer_recover_queue(int domain, unsigned int bus, unsigned int devfn,
> - int severity)
> + int severity, u8 *aer_regs)
> {
> unsigned long flags;
> struct aer_recover_entry entry = {
> @@ -603,6 +604,7 @@ void aer_recover_queue(int domain, unsigned int bus, unsigned int devfn,
> .devfn = devfn,
> .domain = domain,
> .severity = severity,
> + .regs = aer_regs,
> };
>
> spin_lock_irqsave(&aer_recover_ring_lock, flags);
> @@ -629,6 +631,8 @@ static void aer_recover_work_func(struct work_struct *work)
> PCI_SLOT(entry.devfn), PCI_FUNC(entry.devfn));
> continue;
> }
> + cper_print_aer(pdev, entry.severity,
> + (struct aer_capability_regs *)entry.regs);
I like this change a lot because pci_get_domain_bus_and_slot() is sort
of a dubious interface to begin with, and we were using it twice in
this path -- once in cper_print_pcie() and then again (for the exact
same device) here in aer_recover_work_func() -- and this patch gets
rid of the first one.
Normally we shouldn't have to look up a pci_dev from a bus
number/device number. This is an exception because there's no
connection between the life of a pci_dev in the kernel and the
information from firmware, so there's not really a better way to do
it.
> do_recovery(pdev, entry.severity);
> pci_dev_put(pdev);
> }
> diff --git a/drivers/pci/pcie/aer/aerdrv_errprint.c b/drivers/pci/pcie/aer/aerdrv_errprint.c
> index 5ab1425..f277dc3 100644
> --- a/drivers/pci/pcie/aer/aerdrv_errprint.c
> +++ b/drivers/pci/pcie/aer/aerdrv_errprint.c
> @@ -220,13 +220,19 @@ int cper_severity_to_aer(int cper_severity)
> }
> EXPORT_SYMBOL_GPL(cper_severity_to_aer);
>
> -void cper_print_aer(const char *prefix, struct pci_dev *dev, 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;
>
> + /*
> + * TODO: This function needs to be re-written so that it's output
> + * matches the output of aer_print_error(). Right now, the output
> + * is formatted very differently.
> + */
> +
> aer_severity = cper_severity_to_aer(cper_severity);
> if (aer_severity == AER_CORRECTABLE) {
> status = aer->cor_status;
> @@ -244,7 +250,7 @@ void cper_print_aer(const char *prefix, struct pci_dev *dev, int cper_severity,
> agent = AER_GET_AGENT(aer_severity, status);
> dev_err(&dev->dev, "aer_status: 0x%08x, aer_mask: 0x%08x\n",
> status, mask);
> - cper_print_bits(prefix, status, status_strs, status_strs_size);
> + cper_print_bits("", status, status_strs, status_strs_size);
> dev_err(&dev->dev, "aer_layer=%s, aer_agent=%s\n",
> aer_error_layer[layer], aer_agent_string[agent]);
> if (aer_severity != AER_CORRECTABLE)
> diff --git a/include/linux/aer.h b/include/linux/aer.h
> index ec10e1b..d8cf12b 100644
> --- a/include/linux/aer.h
> +++ b/include/linux/aer.h
> @@ -49,10 +49,10 @@ static inline int pci_cleanup_aer_uncorrect_error_status(struct pci_dev *dev)
> }
> #endif
>
> -extern void cper_print_aer(const char *prefix, struct pci_dev *dev,
> +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,
> - int severity);
> + int severity, u8 *aer_regs);
Can the aer_regs type be made "struct aer_capability_regs *"? It'd be
nice if we could avoid the casting when passing the pointer around.
We might still need a cast in ghes_do_proc() when calling
aer_recover_queue(), but I *think* everyplace else would be OK with
keeping the exact type.
Bjorn
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists