aboutsummaryrefslogtreecommitdiff
path: root/drivers/infiniband/hw/mthca/mthca_catas.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/hw/mthca/mthca_catas.c')
-rw-r--r--drivers/infiniband/hw/mthca/mthca_catas.c104
1 files changed, 74 insertions, 30 deletions
diff --git a/drivers/infiniband/hw/mthca/mthca_catas.c b/drivers/infiniband/hw/mthca/mthca_catas.c
index c3bec7490f5..712d2a30fbe 100644
--- a/drivers/infiniband/hw/mthca/mthca_catas.c
+++ b/drivers/infiniband/hw/mthca/mthca_catas.c
@@ -28,12 +28,12 @@
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
- *
- * $Id$
*/
#include <linux/jiffies.h>
+#include <linux/module.h>
#include <linux/timer.h>
+#include <linux/workqueue.h>
#include "mthca_dev.h"
@@ -48,15 +48,53 @@ enum {
static DEFINE_SPINLOCK(catas_lock);
+static LIST_HEAD(catas_list);
+static struct workqueue_struct *catas_wq;
+static struct work_struct catas_work;
+
+static int catas_reset_disable;
+module_param_named(catas_reset_disable, catas_reset_disable, int, 0644);
+MODULE_PARM_DESC(catas_reset_disable, "disable reset on catastrophic event if nonzero");
+
+static void catas_reset(struct work_struct *work)
+{
+ struct mthca_dev *dev, *tmpdev;
+ LIST_HEAD(tlist);
+ int ret;
+
+ mutex_lock(&mthca_device_mutex);
+
+ spin_lock_irq(&catas_lock);
+ list_splice_init(&catas_list, &tlist);
+ spin_unlock_irq(&catas_lock);
+
+ list_for_each_entry_safe(dev, tmpdev, &tlist, catas_err.list) {
+ struct pci_dev *pdev = dev->pdev;
+ ret = __mthca_restart_one(dev->pdev);
+ /* 'dev' now is not valid */
+ if (ret)
+ printk(KERN_ERR "mthca %s: Reset failed (%d)\n",
+ pci_name(pdev), ret);
+ else {
+ struct mthca_dev *d = pci_get_drvdata(pdev);
+ mthca_dbg(d, "Reset succeeded\n");
+ }
+ }
+
+ mutex_unlock(&mthca_device_mutex);
+}
+
static void handle_catas(struct mthca_dev *dev)
{
struct ib_event event;
+ unsigned long flags;
const char *type;
int i;
event.device = &dev->ib_dev;
event.event = IB_EVENT_DEVICE_FATAL;
event.element.port_num = 0;
+ dev->active = false;
ib_dispatch_event(&event);
@@ -82,12 +120,19 @@ static void handle_catas(struct mthca_dev *dev)
for (i = 0; i < dev->catas_err.size; ++i)
mthca_err(dev, " buf[%02x]: %08x\n",
i, swab32(readl(dev->catas_err.map + i)));
+
+ if (catas_reset_disable)
+ return;
+
+ spin_lock_irqsave(&catas_lock, flags);
+ list_add(&dev->catas_err.list, &catas_list);
+ queue_work(catas_wq, &catas_work);
+ spin_unlock_irqrestore(&catas_lock, flags);
}
static void poll_catas(unsigned long dev_ptr)
{
struct mthca_dev *dev = (struct mthca_dev *) dev_ptr;
- unsigned long flags;
int i;
for (i = 0; i < dev->catas_err.size; ++i)
@@ -96,61 +141,60 @@ static void poll_catas(unsigned long dev_ptr)
return;
}
- spin_lock_irqsave(&catas_lock, flags);
- if (!dev->catas_err.stop)
- mod_timer(&dev->catas_err.timer,
- jiffies + MTHCA_CATAS_POLL_INTERVAL);
- spin_unlock_irqrestore(&catas_lock, flags);
-
- return;
+ mod_timer(&dev->catas_err.timer,
+ round_jiffies(jiffies + MTHCA_CATAS_POLL_INTERVAL));
}
void mthca_start_catas_poll(struct mthca_dev *dev)
{
- unsigned long addr;
+ phys_addr_t addr;
init_timer(&dev->catas_err.timer);
- dev->catas_err.stop = 0;
dev->catas_err.map = NULL;
addr = pci_resource_start(dev->pdev, 0) +
((pci_resource_len(dev->pdev, 0) - 1) &
dev->catas_err.addr);
- if (!request_mem_region(addr, dev->catas_err.size * 4,
- DRV_NAME)) {
- mthca_warn(dev, "couldn't request catastrophic error region "
- "at 0x%lx/0x%x\n", addr, dev->catas_err.size * 4);
- return;
- }
-
dev->catas_err.map = ioremap(addr, dev->catas_err.size * 4);
if (!dev->catas_err.map) {
mthca_warn(dev, "couldn't map catastrophic error region "
- "at 0x%lx/0x%x\n", addr, dev->catas_err.size * 4);
- release_mem_region(addr, dev->catas_err.size * 4);
+ "at 0x%llx/0x%x\n", (unsigned long long) addr,
+ dev->catas_err.size * 4);
return;
}
dev->catas_err.timer.data = (unsigned long) dev;
dev->catas_err.timer.function = poll_catas;
dev->catas_err.timer.expires = jiffies + MTHCA_CATAS_POLL_INTERVAL;
+ INIT_LIST_HEAD(&dev->catas_err.list);
add_timer(&dev->catas_err.timer);
}
void mthca_stop_catas_poll(struct mthca_dev *dev)
{
+ del_timer_sync(&dev->catas_err.timer);
+
+ if (dev->catas_err.map)
+ iounmap(dev->catas_err.map);
+
spin_lock_irq(&catas_lock);
- dev->catas_err.stop = 1;
+ list_del(&dev->catas_err.list);
spin_unlock_irq(&catas_lock);
+}
- del_timer_sync(&dev->catas_err.timer);
+int __init mthca_catas_init(void)
+{
+ INIT_WORK(&catas_work, catas_reset);
- if (dev->catas_err.map) {
- iounmap(dev->catas_err.map);
- release_mem_region(pci_resource_start(dev->pdev, 0) +
- ((pci_resource_len(dev->pdev, 0) - 1) &
- dev->catas_err.addr),
- dev->catas_err.size * 4);
- }
+ catas_wq = create_singlethread_workqueue("mthca_catas");
+ if (!catas_wq)
+ return -ENOMEM;
+
+ return 0;
+}
+
+void mthca_catas_cleanup(void)
+{
+ destroy_workqueue(catas_wq);
}