firewire: Don't time out command orbs, leave that to the scsi stack.
The mod_timer based timing out of orb was a little to agressive and
would time out legit, but long-lived scsi cmds. Besides, the scsi
stack keeps track of this already. Since we're only timing out
management orbs, go back to wait_for_completion_timeout.
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
diff --git a/drivers/firewire/fw-sbp2.c b/drivers/firewire/fw-sbp2.c
index 07e410b..2e5479b 100644
--- a/drivers/firewire/fw-sbp2.c
+++ b/drivers/firewire/fw-sbp2.c
@@ -68,9 +68,6 @@
int address_high;
int generation;
- /* Timer for flushing ORBs. */
- struct timer_list orb_timer;
-
int retries;
struct delayed_work work;
struct Scsi_Host *scsi_host;
@@ -342,9 +339,6 @@
list_add_tail(&orb->link, &sd->orb_list);
spin_unlock_irqrestore(&device->card->lock, flags);
- mod_timer(&sd->orb_timer,
- jiffies + DIV_ROUND_UP(SBP2_ORB_TIMEOUT * HZ, 1000));
-
fw_send_request(device->card, &orb->t, TCODE_WRITE_BLOCK_REQUEST,
node_id, generation,
device->node->max_speed, offset,
@@ -352,13 +346,14 @@
complete_transaction, orb);
}
-static void sbp2_cancel_orbs(struct fw_unit *unit)
+static int sbp2_cancel_orbs(struct fw_unit *unit)
{
struct fw_device *device = fw_device(unit->device.parent);
struct sbp2_device *sd = unit->device.driver_data;
struct sbp2_orb *orb, *next;
struct list_head list;
unsigned long flags;
+ int retval = -ENOENT;
INIT_LIST_HEAD(&list);
spin_lock_irqsave(&device->card->lock, flags);
@@ -366,19 +361,15 @@
spin_unlock_irqrestore(&device->card->lock, flags);
list_for_each_entry_safe(orb, next, &list, link) {
+ retval = 0;
if (fw_cancel_transaction(device->card, &orb->t) == 0)
continue;
orb->rcode = RCODE_CANCELLED;
orb->callback(orb, NULL);
}
-}
-static void orb_timer_callback(unsigned long data)
-{
- struct sbp2_device *sd = (struct sbp2_device *)data;
-
- sbp2_cancel_orbs(sd->unit);
+ return retval;
}
static void
@@ -447,20 +438,22 @@
init_completion(&orb->done);
orb->base.callback = complete_management_orb;
+
sbp2_send_orb(&orb->base, unit,
node_id, generation, sd->management_agent_address);
- wait_for_completion(&orb->done);
+ wait_for_completion_timeout(&orb->done,
+ msecs_to_jiffies(SBP2_ORB_TIMEOUT));
retval = -EIO;
- if (orb->base.rcode != RCODE_COMPLETE) {
- fw_error("management write failed, rcode 0x%02x\n",
+ if (sbp2_cancel_orbs(unit) == 0) {
+ fw_error("orb reply timed out, rcode=0x%02x\n",
orb->base.rcode);
goto out;
}
- if (orb->base.rcode == RCODE_CANCELLED) {
- fw_error("orb reply timed out, rcode=0x%02x\n",
+ if (orb->base.rcode != RCODE_COMPLETE) {
+ fw_error("management write failed, rcode 0x%02x\n",
orb->base.rcode);
goto out;
}
@@ -602,7 +595,6 @@
unit->device.driver_data = sd;
sd->unit = unit;
INIT_LIST_HEAD(&sd->orb_list);
- setup_timer(&sd->orb_timer, orb_timer_callback, (unsigned long)sd);
sd->address_handler.length = 0x100;
sd->address_handler.address_callback = sbp2_status_write;
@@ -675,7 +667,6 @@
SBP2_LOGOUT_REQUEST, sd->login_id, NULL);
remove_scsi_devices(unit);
- del_timer_sync(&sd->orb_timer);
fw_core_remove_address_handler(&sd->address_handler);
kfree(sd);