[PATCH V4] iommu/amd: Add logic to decode AMD IOMMU event flag

From: suravee.suthikulpanit
Date: Tue Apr 09 2013 - 15:04:19 EST


From: Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>

Add logic to decode AMD IOMMU event flag based on information from AMD IOMMU specification.
This should simplify debugging IOMMU errors. Also, dump DTE information in additional cases.

This is an example:
AMD-Vi: Event logged [INVALID_DEVICE_REQUEST device=51:00.0 address=0x0000000000000000]
AMD-Vi: Flags details: Gst NX Usr Int P W N-Perm Rsrv Trnslt
AMD-Vi: Type of error: (0x7) Invalid transaction to the system management address range
AMD-Vi: DTE[0..3]: 600000bfa74e8403 0000000000000014 2000005fa718cc11 0000000000000000

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>
---
V4:
* Change print out format to reduce noise
* Remove "(Note:...)"
V3:
* Move comments to end of line
* Shorten the print out to be within one line
V2:
* Fix printing format to reduce noise
* Use string table instead of switch/case
* Use pr_cont instead of printk

drivers/iommu/amd_iommu.c | 198 ++++++++++++++++++++++++++++++++++-----------
1 file changed, 151 insertions(+), 47 deletions(-)

diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
index b287ca3..1f8e5fd 100644
--- a/drivers/iommu/amd_iommu.c
+++ b/drivers/iommu/amd_iommu.c
@@ -601,13 +601,101 @@ static void amd_iommu_stats_init(void)
*
****************************************************************************/

+struct _event_log_flags {
+ u32 gn:1, /* 16 */
+ nx:1, /* 17 */
+ us:1, /* 18 */
+ i:1, /* 19 */
+ pr:1, /* 20 */
+ rw:1, /* 21 */
+ pe:1, /* 22 */
+ rz:1, /* 23 */
+ tr:1, /* 24 */
+ type:3, /* [27:25] */
+ _reserved_:20; /* Reserved */
+};
+
+static const char * const _type_field_encodings[] = {
+ "Reserved", /* 00 */
+ "Master Abort", /* 01 */
+ "Target Abort", /* 10 */
+ "Data Error", /* 11 */
+};
+
+static const char * const _invalid_transaction_desc[] = {
+ "Read request or non-posted write in the interrupt "
+ "addres range", /* 000 */
+ "Pretranslated transaction received from an "
+ "I/O device that has I=0 or V=0 in DTE", /* 001 */
+ "Port I/O space transaction received from an "
+ "I/O device that has IoCtl=00b in DTE", /* 010 */
+ "Posted write to invalid address range", /* 011 */
+ "Invalid read request or non-posted write", /* 100 */
+ "Posted write to the interrupt/EOI range from an "
+ "I/O device that has IntCtl=00b in DTE", /* 101 */
+ "Posted write to a reserved interrupt address range", /* 110 */
+ "Invalid transaction to the system management "
+ "address range", /* 111 */
+};
+
+static const char * const _invalid_translation_desc[] = {
+ "Translation request received from an I/O device "
+ "that has I=0, or has V=0, or has V=1 and "
+ "TV=0 in DTE", /* 000 */
+ "Translation request in invalid address range", /* 001 */
+ "Invalid translation request", /* 010 */
+ "Reserved", /* 011 */
+ "Reserved", /* 100 */
+ "Reserved", /* 101 */
+ "Reserved", /* 110 */
+ "Reserved", /* 111 */
+};
+
+static void dump_flags(int flags, int ev_type)
+{
+ struct _event_log_flags *p = (struct _event_log_flags *) &flags;
+ u32 err_type = p->type;
+
+ pr_err("AMD-Vi: Flags details: %s %s %s %s %s %s %s %s %s\n",
+ (p->gn ? "Gst" : "Nst"),
+ (p->nx ? "NX" : "EX"),
+ (p->us ? "Usr" : "Sup"),
+ (p->i ? "Int" : "Mem"),
+ (p->pr ? "P" : "NP"),
+ (p->rw ? "W" : "R"),
+ (p->pe ? "N-Perm" : "Perm"),
+ (p->rz ? "Rsrv" : "Ill"),
+ (p->tr ? "Trnslt" : "Trnsact"));
+
+ pr_err("AMD-Vi: Type of error: (0x%x) ", err_type);
+ if ((ev_type == EVENT_TYPE_DEV_TAB_ERR) ||
+ (ev_type == EVENT_TYPE_PAGE_TAB_ERR) ||
+ (ev_type == EVENT_TYPE_CMD_HARD_ERR)) {
+ if (err_type < ARRAY_SIZE(_type_field_encodings)) {
+ pr_cont("%s\n",
+ _type_field_encodings[err_type]);
+ }
+ } else if (ev_type == EVENT_TYPE_INV_DEV_REQ) {
+ if (p->tr == 0) {
+ if (err_type < ARRAY_SIZE(_invalid_translation_desc))
+ pr_cont("%s\n",
+ _invalid_translation_desc[err_type]);
+ } else {
+ if (err_type < ARRAY_SIZE(_invalid_transaction_desc))
+ pr_cont("%s\n",
+ _invalid_transaction_desc[err_type]);
+ }
+ }
+}
+
static void dump_dte_entry(u16 devid)
{
int i;

+ pr_err("AMD-Vi: DTE[0..3]:");
for (i = 0; i < 4; ++i)
- pr_err("AMD-Vi: DTE[%d]: %016llx\n", i,
- amd_iommu_dev_table[devid].data[i]);
+ pr_cont(" %016llx", amd_iommu_dev_table[devid].data[i]);
+ pr_cont("\n");
}

static void dump_command(unsigned long phys_addr)
@@ -619,82 +707,97 @@ static void dump_command(unsigned long phys_addr)
pr_err("AMD-Vi: CMD[%d]: %08x\n", i, cmd->data[i]);
}

-static void iommu_print_event(struct amd_iommu *iommu, void *__evt)
+void amd_iommu_print_event(int type, int devid, int domid,
+ int flags, u64 address)
{
- int type, devid, domid, flags;
- volatile u32 *event = __evt;
- int count = 0;
- u64 address;
-
-retry:
- type = (event[1] >> EVENT_TYPE_SHIFT) & EVENT_TYPE_MASK;
- devid = (event[0] >> EVENT_DEVID_SHIFT) & EVENT_DEVID_MASK;
- domid = (event[1] >> EVENT_DOMID_SHIFT) & EVENT_DOMID_MASK;
- flags = (event[1] >> EVENT_FLAGS_SHIFT) & EVENT_FLAGS_MASK;
- address = (u64)(((u64)event[3]) << 32) | event[2];
-
- if (type == 0) {
- /* Did we hit the erratum? */
- if (++count == LOOP_TIMEOUT) {
- pr_err("AMD-Vi: No event written to event log\n");
- return;
- }
- udelay(1);
- goto retry;
- }
-
- printk(KERN_ERR "AMD-Vi: Event logged [");
+ pr_err("AMD-Vi: Event logged [");

switch (type) {
case EVENT_TYPE_ILL_DEV:
- printk("ILLEGAL_DEV_TABLE_ENTRY device=%02x:%02x.%x "
- "address=0x%016llx flags=0x%04x]\n",
+ pr_cont("ILLEGAL_DEV_TABLE_ENTRY device=%02x:%02x.%x "
+ "address=0x%016llx]\n",
PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
- address, flags);
+ address);
+ dump_flags(flags, type);
dump_dte_entry(devid);
break;
case EVENT_TYPE_IO_FAULT:
- printk("IO_PAGE_FAULT device=%02x:%02x.%x "
- "domain=0x%04x address=0x%016llx flags=0x%04x]\n",
+ pr_cont("IO_PAGE_FAULT device=%02x:%02x.%x "
+ "domain=0x%04x address=0x%016llx]\n",
PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
- domid, address, flags);
+ domid, address);
+ dump_flags(flags, type);
+ dump_dte_entry(devid);
break;
case EVENT_TYPE_DEV_TAB_ERR:
- printk("DEV_TAB_HARDWARE_ERROR device=%02x:%02x.%x "
- "address=0x%016llx flags=0x%04x]\n",
+ pr_cont("DEV_TAB_HARDWARE_ERROR device=%02x:%02x.%x "
+ "address=0x%016llx]\n",
PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
- address, flags);
+ address);
+ dump_flags(flags, type);
break;
case EVENT_TYPE_PAGE_TAB_ERR:
- printk("PAGE_TAB_HARDWARE_ERROR device=%02x:%02x.%x "
- "domain=0x%04x address=0x%016llx flags=0x%04x]\n",
+ pr_cont("PAGE_TAB_HARDWARE_ERROR device=%02x:%02x.%x "
+ "domain=0x%04x address=0x%016llx]\n",
PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
- domid, address, flags);
+ domid, address);
+ dump_flags(flags, type);
break;
case EVENT_TYPE_ILL_CMD:
- printk("ILLEGAL_COMMAND_ERROR address=0x%016llx]\n", address);
+ pr_cont("ILLEGAL_COMMAND_ERROR address=0x%016llx]\n",
+ address);
dump_command(address);
break;
case EVENT_TYPE_CMD_HARD_ERR:
- printk("COMMAND_HARDWARE_ERROR address=0x%016llx "
- "flags=0x%04x]\n", address, flags);
+ pr_cont("COMMAND_HARDWARE_ERROR address=0x%016llx]\n",
+ address);
+ dump_flags(flags, type);
break;
case EVENT_TYPE_IOTLB_INV_TO:
- printk("IOTLB_INV_TIMEOUT device=%02x:%02x.%x "
+ pr_cont("IOTLB_INV_TIMEOUT device=%02x:%02x.%x "
"address=0x%016llx]\n",
PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
address);
break;
case EVENT_TYPE_INV_DEV_REQ:
- printk("INVALID_DEVICE_REQUEST device=%02x:%02x.%x "
- "address=0x%016llx flags=0x%04x]\n",
+ pr_cont("INVALID_DEVICE_REQUEST device=%02x:%02x.%x "
+ "address=0x%016llx]\n",
PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
- address, flags);
+ address);
+ dump_flags(flags, type);
+ dump_dte_entry(devid);
break;
default:
- printk(KERN_ERR "UNKNOWN type=0x%02x]\n", type);
+ pr_cont("UNKNOWN type=0x%02x]\n", type);
+ }
+}
+
+static void iommu_handle_event(struct amd_iommu *iommu, void *__evt)
+{
+ int type, devid, domid, flags;
+ u32 *event = __evt;
+ int count = 0;
+ u64 address;
+
+retry:
+ type = (event[1] >> EVENT_TYPE_SHIFT) & EVENT_TYPE_MASK;
+ devid = (event[0] >> EVENT_DEVID_SHIFT) & EVENT_DEVID_MASK;
+ domid = (event[1] >> EVENT_DOMID_SHIFT) & EVENT_DOMID_MASK;
+ flags = (event[1] >> EVENT_FLAGS_SHIFT) & EVENT_FLAGS_MASK;
+ address = (u64)(((u64)event[3]) << 32) | event[2];
+
+ if (type == 0) {
+ /* Did we hit the erratum? */
+ if (++count == LOOP_TIMEOUT) {
+ pr_err("AMD-Vi: No event written to event log\n");
+ return;
+ }
+ udelay(1);
+ goto retry;
}

+ amd_iommu_print_event(type, devid, domid, flags, address);
+
memset(__evt, 0, 4 * sizeof(u32));
}

@@ -709,7 +812,7 @@ static void iommu_poll_events(struct amd_iommu *iommu)
tail = readl(iommu->mmio_base + MMIO_EVT_TAIL_OFFSET);

while (head != tail) {
- iommu_print_event(iommu, iommu->evt_buf + head);
+ iommu_handle_event(iommu, iommu->evt_buf + head);
head = (head + EVENT_ENTRY_SIZE) % iommu->evt_buf_size;
}

@@ -3268,6 +3371,7 @@ static int __init alloc_passthrough_domain(void)

return 0;
}
+
static int amd_iommu_domain_init(struct iommu_domain *dom)
{
struct protection_domain *domain;
--
1.7.10.4


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/