powerpc/eeh: Fix a bug when pci structure is null
authorBreno Leitao <leitao@linux.vnet.ibm.com>
Wed, 3 Feb 2010 05:56:41 +0000 (05:56 +0000)
committerBenjamin Herrenschmidt <benh@kernel.crashing.org>
Wed, 17 Feb 2010 03:02:47 +0000 (14:02 +1100)
During a EEH recover, the pci_dev structure can be null, mainly if an
eeh event is detected during cpi config operation. In this case, the
pci_dev will not be known (and will be null) the kernel will crash
with the following message:

Unable to handle kernel paging request for data at address 0x000000a0
Faulting instruction address: 0xc00000000006b8b4
Oops: Kernel access of bad area, sig: 11 [#1]

NIP [c00000000006b8b4] .eeh_event_handler+0x10c/0x1a0
LR [c00000000006b8a8] .eeh_event_handler+0x100/0x1a0
Call Trace:
[c0000003a80dff00] [c00000000006b8a8] .eeh_event_handler+0x100/0x1a0
[c0000003a80dff90] [c000000000031f1c] .kernel_thread+0x54/0x70

The bug occurs because pci_name() tries to access a null pointer.
This patch just guarantee that pci_name() is not called on Null pointers.

Signed-off-by: Breno Leitao <leitao@linux.vnet.ibm.com>
Signed-off-by: Linas Vepstas <linasvepstas@gmail.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
arch/powerpc/include/asm/ppc-pci.h
arch/powerpc/platforms/pseries/eeh.c
arch/powerpc/platforms/pseries/eeh_driver.c
arch/powerpc/platforms/pseries/eeh_event.c

index 2828f9d0f66ddb1d66be47662c999282fc25fdaf..42fdff0e4b329a266b6398893fda81bd80de0b8d 100644 (file)
@@ -137,6 +137,11 @@ struct device_node * find_device_pe(struct device_node *dn);
 void eeh_sysfs_add_device(struct pci_dev *pdev);
 void eeh_sysfs_remove_device(struct pci_dev *pdev);
 
+static inline const char *eeh_pci_name(struct pci_dev *pdev) 
+{ 
+       return pdev ? pci_name(pdev) : "<null>";
+} 
+
 #endif /* CONFIG_EEH */
 
 #else /* CONFIG_PCI */
index ccd8dd03b8c987e701e476fe43a8f84348901d13..3304f32fc7b897c5ce2183bd02f426fdaff522ab 100644 (file)
@@ -491,7 +491,7 @@ int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev)
            pdn->eeh_mode & EEH_MODE_NOCHECK) {
                ignored_check++;
                pr_debug("EEH: Ignored check (%x) for %s %s\n",
-                        pdn->eeh_mode, pci_name (dev), dn->full_name);
+                        pdn->eeh_mode, eeh_pci_name(dev), dn->full_name);
                return 0;
        }
 
@@ -515,7 +515,7 @@ int eeh_dn_check_failure(struct device_node *dn, struct pci_dev *dev)
                        printk (KERN_ERR "EEH: %d reads ignored for recovering device at "
                                "location=%s driver=%s pci addr=%s\n",
                                pdn->eeh_check_count, location,
-                               dev->driver->name, pci_name(dev));
+                               dev->driver->name, eeh_pci_name(dev));
                        printk (KERN_ERR "EEH: Might be infinite loop in %s driver\n",
                                dev->driver->name);
                        dump_stack();
index 0c252c3a5f9fbc9966cdeb134f0110c62dc8526e..b8d70f5d9aa984de1a44b91f762493820c58b4c8 100644 (file)
@@ -337,7 +337,7 @@ struct pci_dn * handle_eeh_events (struct eeh_event *event)
                location = location ? location : "unknown";
                printk(KERN_ERR "EEH: Error: Cannot find partition endpoint "
                                "for location=%s pci addr=%s\n",
-                       location, pci_name(event->dev));
+                       location, eeh_pci_name(event->dev));
                return NULL;
        }
 
@@ -368,7 +368,7 @@ struct pci_dn * handle_eeh_events (struct eeh_event *event)
                pci_str = pci_name (frozen_pdn->pcidev);
                drv_str = pcid_name (frozen_pdn->pcidev);
        } else {
-               pci_str = pci_name (event->dev);
+               pci_str = eeh_pci_name(event->dev);
                drv_str = pcid_name (event->dev);
        }
        
index ddb80f5d850b77231783b7f05c1d7adadd6f9aed..ec5df8f519c7417327923bcadafdbda76874ec04 100644 (file)
@@ -80,7 +80,7 @@ static int eeh_event_handler(void * dummy)
        eeh_mark_slot(event->dn, EEH_MODE_RECOVERING);
 
        printk(KERN_INFO "EEH: Detected PCI bus error on device %s\n",
-              pci_name(event->dev));
+              eeh_pci_name(event->dev));
 
        pdn = handle_eeh_events(event);