powerpc: IOMMU fault injection
Add the ability to inject IOMMU faults. We enable this per device
via a fail_iommu sysfs property, similar to fault injection on other
subsystems.
An example:
...
0003:01:00.1 Ethernet controller: Emulex Corporation OneConnect 10Gb NIC (be3) (rev 02)
To inject one error to this device:
echo 1 > /sys/bus/pci/devices/0003:01:00.1/fail_iommu
echo 1 > /sys/kernel/debug/fail_iommu/probability
echo 1 > /sys/kernel/debug/fail_iommu/times
As feared, the first failure injected on the be3 results in an
unrecoverable error, taking down both functions of the card
permanently:
be2net 0003:01:00.1: Unrecoverable error in the card
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
diff --git a/arch/powerpc/Kconfig.debug b/arch/powerpc/Kconfig.debug
index e5f2689..5416e28 100644
--- a/arch/powerpc/Kconfig.debug
+++ b/arch/powerpc/Kconfig.debug
@@ -331,4 +331,13 @@
If you are unsure, say Y.
+config FAIL_IOMMU
+ bool "Fault-injection capability for IOMMU"
+ depends on FAULT_INJECTION
+ help
+ Provide fault-injection capability for IOMMU. Each device can
+ be selectively enabled via the fail_iommu property.
+
+ If you are unsure, say N.
+
endmenu
diff --git a/arch/powerpc/include/asm/device.h b/arch/powerpc/include/asm/device.h
index 63d5ca4..77e97dd 100644
--- a/arch/powerpc/include/asm/device.h
+++ b/arch/powerpc/include/asm/device.h
@@ -34,6 +34,9 @@
#ifdef CONFIG_EEH
struct eeh_dev *edev;
#endif
+#ifdef CONFIG_FAIL_IOMMU
+ int fail_iommu;
+#endif
};
struct pdev_archdata {
diff --git a/arch/powerpc/kernel/iommu.c b/arch/powerpc/kernel/iommu.c
index 7bc94da..fbefe72 100644
--- a/arch/powerpc/kernel/iommu.c
+++ b/arch/powerpc/kernel/iommu.c
@@ -34,6 +34,8 @@
#include <linux/iommu-helper.h>
#include <linux/crash_dump.h>
#include <linux/hash.h>
+#include <linux/fault-inject.h>
+#include <linux/pci.h>
#include <asm/io.h>
#include <asm/prom.h>
#include <asm/iommu.h>
@@ -41,6 +43,7 @@
#include <asm/machdep.h>
#include <asm/kdump.h>
#include <asm/fadump.h>
+#include <asm/vio.h>
#define DBG(...)
@@ -79,6 +82,94 @@
}
subsys_initcall(setup_iommu_pool_hash);
+#ifdef CONFIG_FAIL_IOMMU
+
+static DECLARE_FAULT_ATTR(fail_iommu);
+
+static int __init setup_fail_iommu(char *str)
+{
+ return setup_fault_attr(&fail_iommu, str);
+}
+__setup("fail_iommu=", setup_fail_iommu);
+
+static bool should_fail_iommu(struct device *dev)
+{
+ return dev->archdata.fail_iommu && should_fail(&fail_iommu, 1);
+}
+
+static int __init fail_iommu_debugfs(void)
+{
+ struct dentry *dir = fault_create_debugfs_attr("fail_iommu",
+ NULL, &fail_iommu);
+
+ return IS_ERR(dir) ? PTR_ERR(dir) : 0;
+}
+late_initcall(fail_iommu_debugfs);
+
+static ssize_t fail_iommu_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%d\n", dev->archdata.fail_iommu);
+}
+
+static ssize_t fail_iommu_store(struct device *dev,
+ struct device_attribute *attr, const char *buf,
+ size_t count)
+{
+ int i;
+
+ if (count > 0 && sscanf(buf, "%d", &i) > 0)
+ dev->archdata.fail_iommu = (i == 0) ? 0 : 1;
+
+ return count;
+}
+
+static DEVICE_ATTR(fail_iommu, S_IRUGO|S_IWUSR, fail_iommu_show,
+ fail_iommu_store);
+
+static int fail_iommu_bus_notify(struct notifier_block *nb,
+ unsigned long action, void *data)
+{
+ struct device *dev = data;
+
+ if (action == BUS_NOTIFY_ADD_DEVICE) {
+ if (device_create_file(dev, &dev_attr_fail_iommu))
+ pr_warn("Unable to create IOMMU fault injection sysfs "
+ "entries\n");
+ } else if (action == BUS_NOTIFY_DEL_DEVICE) {
+ device_remove_file(dev, &dev_attr_fail_iommu);
+ }
+
+ return 0;
+}
+
+static struct notifier_block fail_iommu_bus_notifier = {
+ .notifier_call = fail_iommu_bus_notify
+};
+
+static int __init fail_iommu_setup(void)
+{
+#ifdef CONFIG_PCI
+ bus_register_notifier(&pci_bus_type, &fail_iommu_bus_notifier);
+#endif
+#ifdef CONFIG_IBMVIO
+ bus_register_notifier(&vio_bus_type, &fail_iommu_bus_notifier);
+#endif
+
+ return 0;
+}
+/*
+ * Must execute after PCI and VIO subsystem have initialised but before
+ * devices are probed.
+ */
+arch_initcall(fail_iommu_setup);
+#else
+static inline bool should_fail_iommu(struct device *dev)
+{
+ return false;
+}
+#endif
+
static unsigned long iommu_range_alloc(struct device *dev,
struct iommu_table *tbl,
unsigned long npages,
@@ -107,6 +198,9 @@
return DMA_ERROR_CODE;
}
+ if (should_fail_iommu(dev))
+ return DMA_ERROR_CODE;
+
/*
* We don't need to disable preemption here because any CPU can
* safely use any IOMMU pool.
