]> git.karo-electronics.de Git - karo-tx-linux.git/commitdiff
powerpc/eeh: Lock module while handling EEH event
authorGavin Shan <shangw@linux.vnet.ibm.com>
Mon, 17 Sep 2012 04:34:27 +0000 (04:34 +0000)
committerBen Hutchings <ben@decadent.org.uk>
Wed, 17 Oct 2012 02:48:32 +0000 (03:48 +0100)
commit feadf7c0a1a7c08c74bebb4a13b755f8c40e3bbc upstream.

The EEH core is talking with the PCI device driver to determine the
action (purely reset, or PCI device removal). During the period, the
driver might be unloaded and in turn causes kernel crash as follows:

EEH: Detected PCI bus error on PHB#4-PE#10000
EEH: This PCI device has failed 3 times in the last hour
lpfc 0004:01:00.0: 0:2710 PCI channel disable preparing for reset
Unable to handle kernel paging request for data at address 0x00000490
Faulting instruction address: 0xd00000000e682c90
cpu 0x1: Vector: 300 (Data Access) at [c000000fc75ffa20]
    pc: d00000000e682c90: .lpfc_io_error_detected+0x30/0x240 [lpfc]
    lr: d00000000e682c8c: .lpfc_io_error_detected+0x2c/0x240 [lpfc]
    sp: c000000fc75ffca0
   msr: 8000000000009032
   dar: 490
 dsisr: 40000000
  current = 0xc000000fc79b88b0
  paca    = 0xc00000000edb0380  softe: 0  irq_happened: 0x00
    pid   = 3386, comm = eehd
enter ? for help
[c000000fc75ffca0c000000fc75ffd30 (unreliable)
[c000000fc75ffd30c00000000004fd3c .eeh_report_error+0x7c/0xf0
[c000000fc75ffdc0c00000000004ee00 .eeh_pe_dev_traverse+0xa0/0x180
[c000000fc75ffe70c00000000004ffd8 .eeh_handle_event+0x68/0x300
[c000000fc75fff00c0000000000503a0 .eeh_event_handler+0x130/0x1a0
[c000000fc75fff90c000000000020138 .kernel_thread+0x54/0x70
1:mon>

The patch increases the reference of the corresponding driver modules
while EEH core does the negotiation with PCI device driver so that the
corresponding driver modules can't be unloaded during the period and
we're safe to refer the callbacks.

Reported-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[bwh: Backported to 3.2:
 - Adjust context
 - Reporting functions return int (success = 0), not void * (success = NULL)
 - Assume that the 'dev' arguments are non-null]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
arch/powerpc/platforms/pseries/eeh_driver.c

index 1b6cb10589e01cd8ad44d91edb39e82794a84e11..a0a4e8abacebe29cfe3cc1f6313ecfd5181a74aa 100644 (file)
@@ -25,6 +25,7 @@
 #include <linux/delay.h>
 #include <linux/interrupt.h>
 #include <linux/irq.h>
+#include <linux/module.h>
 #include <linux/pci.h>
 #include <asm/eeh.h>
 #include <asm/eeh_event.h>
@@ -41,6 +42,41 @@ static inline const char * pcid_name (struct pci_dev *pdev)
        return "";
 }
 
+/**
+ * eeh_pcid_get - Get the PCI device driver
+ * @pdev: PCI device
+ *
+ * The function is used to retrieve the PCI device driver for
+ * the indicated PCI device. Besides, we will increase the reference
+ * of the PCI device driver to prevent that being unloaded on
+ * the fly. Otherwise, kernel crash would be seen.
+ */
+static inline struct pci_driver *eeh_pcid_get(struct pci_dev *pdev)
+{
+       if (!pdev || !pdev->driver)
+               return NULL;
+
+       if (!try_module_get(pdev->driver->driver.owner))
+               return NULL;
+
+       return pdev->driver;
+}
+
+/**
+ * eeh_pcid_put - Dereference on the PCI device driver
+ * @pdev: PCI device
+ *
+ * The function is called to do dereference on the PCI device
+ * driver of the indicated PCI device.
+ */
+static inline void eeh_pcid_put(struct pci_dev *pdev)
+{
+       if (!pdev || !pdev->driver)
+               return;
+
+       module_put(pdev->driver->driver.owner);
+}
+
 #if 0
 static void print_device_node_tree(struct pci_dn *pdn, int dent)
 {
@@ -109,18 +145,20 @@ static void eeh_enable_irq(struct pci_dev *dev)
 static int eeh_report_error(struct pci_dev *dev, void *userdata)
 {
        enum pci_ers_result rc, *res = userdata;
-       struct pci_driver *driver = dev->driver;
+       struct pci_driver *driver;
 
        dev->error_state = pci_channel_io_frozen;
 
-       if (!driver)
-               return 0;
+       driver = eeh_pcid_get(dev);
+       if (!driver) return 0;
 
        eeh_disable_irq(dev);
 
        if (!driver->err_handler ||
-           !driver->err_handler->error_detected)
+           !driver->err_handler->error_detected) {
+               eeh_pcid_put(dev);
                return 0;
+       }
 
        rc = driver->err_handler->error_detected (dev, pci_channel_io_frozen);
 
@@ -128,6 +166,7 @@ static int eeh_report_error(struct pci_dev *dev, void *userdata)
        if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
        if (*res == PCI_ERS_RESULT_NONE) *res = rc;
 
+       eeh_pcid_put(dev);
        return 0;
 }
 
@@ -142,12 +181,15 @@ static int eeh_report_error(struct pci_dev *dev, void *userdata)
 static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
 {
        enum pci_ers_result rc, *res = userdata;
-       struct pci_driver *driver = dev->driver;
+       struct pci_driver *driver;
 
-       if (!driver ||
-           !driver->err_handler ||
-           !driver->err_handler->mmio_enabled)
+       driver = eeh_pcid_get(dev);
+       if (!driver) return 0;
+       if (!driver->err_handler ||
+           !driver->err_handler->mmio_enabled) {
+               eeh_pcid_put(dev);
                return 0;
+       }
 
        rc = driver->err_handler->mmio_enabled (dev);
 
@@ -155,6 +197,7 @@ static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
        if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
        if (*res == PCI_ERS_RESULT_NONE) *res = rc;
 
+       eeh_pcid_put(dev);
        return 0;
 }
 
@@ -165,18 +208,20 @@ static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
 static int eeh_report_reset(struct pci_dev *dev, void *userdata)
 {
        enum pci_ers_result rc, *res = userdata;
-       struct pci_driver *driver = dev->driver;
-
-       if (!driver)
-               return 0;
+       struct pci_driver *driver;
 
        dev->error_state = pci_channel_io_normal;
 
+       driver = eeh_pcid_get(dev);
+       if (!driver) return 0;
+
        eeh_enable_irq(dev);
 
        if (!driver->err_handler ||
-           !driver->err_handler->slot_reset)
+           !driver->err_handler->slot_reset) {
+               eeh_pcid_put(dev);
                return 0;
+       }
 
        rc = driver->err_handler->slot_reset(dev);
        if ((*res == PCI_ERS_RESULT_NONE) ||
@@ -184,6 +229,7 @@ static int eeh_report_reset(struct pci_dev *dev, void *userdata)
        if (*res == PCI_ERS_RESULT_DISCONNECT &&
             rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
 
+       eeh_pcid_put(dev);
        return 0;
 }
 
@@ -193,21 +239,24 @@ static int eeh_report_reset(struct pci_dev *dev, void *userdata)
 
 static int eeh_report_resume(struct pci_dev *dev, void *userdata)
 {
-       struct pci_driver *driver = dev->driver;
+       struct pci_driver *driver;
 
        dev->error_state = pci_channel_io_normal;
 
-       if (!driver)
-               return 0;
+       driver = eeh_pcid_get(dev);
+       if (!driver) return 0;
 
        eeh_enable_irq(dev);
 
        if (!driver->err_handler ||
-           !driver->err_handler->resume)
+           !driver->err_handler->resume) {
+               eeh_pcid_put(dev);
                return 0;
+       }
 
        driver->err_handler->resume(dev);
 
+       eeh_pcid_put(dev);
        return 0;
 }
 
@@ -220,21 +269,24 @@ static int eeh_report_resume(struct pci_dev *dev, void *userdata)
 
 static int eeh_report_failure(struct pci_dev *dev, void *userdata)
 {
-       struct pci_driver *driver = dev->driver;
+       struct pci_driver *driver;
 
        dev->error_state = pci_channel_io_perm_failure;
 
-       if (!driver)
-               return 0;
+       driver = eeh_pcid_get(dev);
+       if (!driver) return 0;
 
        eeh_disable_irq(dev);
 
        if (!driver->err_handler ||
-           !driver->err_handler->error_detected)
+           !driver->err_handler->error_detected) {
+               eeh_pcid_put(dev);
                return 0;
+       }
 
        driver->err_handler->error_detected(dev, pci_channel_io_perm_failure);
 
+       eeh_pcid_put(dev);
        return 0;
 }