From: Holger Schurig <hs4233@mail.mn-solutions.de>
To: libertas-dev@lists.infradead.org, Dan Williams <dcbw@redhat.com>,
linux-wireless@vger.kernel.org,
"John W. Linville" <linville@tuxdriver.com>
Subject: [PATCH] libertas: reduce command retry time
Date: Wed, 30 Apr 2008 16:44:29 +0200 [thread overview]
Message-ID: <200804301644.29927.hs4233@mail.mn-solutions.de> (raw)
In the normal case, an unsuccessful command would be tried for
10*5 seconds, or 10*10 seconds in the worst case. This patch
reduces this to 3*1 seconds, or 3*5 seconds in the worst case.
Also changed the needless del_timer_sync(&priv->command_timer);
to del_timer().
Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
---
The story behind this patch is the following: let the driver
load, and let the hardware have a failure, e.g. because of a
bad power supply to the device. Now try "pccardctl eject".
It looks like this is now hanging, nothing happens. Only
when all the retries are done does the "pccardctrl eject"
do what it is supposed to do, that might have been in 50 or 100
seconds.
As I didn't have seen a case where it makes sense to retry a
faulty command 10 times I changed this number.
Also, I've never seen a command (except scan/assoc) to take more
than a few milliseconds. So giving them 5 seconds seemed excessive.
Index: wireless-testing/drivers/net/wireless/libertas/cmd.c
===================================================================
--- wireless-testing.orig/drivers/net/wireless/libertas/cmd.c 2008-04-30 15:10:14.000000000 +0200
+++ wireless-testing/drivers/net/wireless/libertas/cmd.c 2008-04-30 15:18:08.000000000 +0200
@@ -1144,7 +1144,7 @@ static void lbs_submit_command(struct lb
struct cmd_header *cmd;
uint16_t cmdsize;
uint16_t command;
- int timeo = 5 * HZ;
+ int timeo = HZ;
int ret;
lbs_deb_enter(LBS_DEB_HOST);
@@ -1162,7 +1162,7 @@ static void lbs_submit_command(struct lb
/* These commands take longer */
if (command == CMD_802_11_SCAN || command == CMD_802_11_ASSOCIATE ||
command == CMD_802_11_AUTHENTICATE)
- timeo = 10 * HZ;
+ timeo = 5 * HZ;
lbs_deb_cmd("DNLD_CMD: command 0x%04x, seq %d, size %d\n",
command, le16_to_cpu(cmd->seqnum), cmdsize);
@@ -1174,7 +1174,7 @@ static void lbs_submit_command(struct lb
lbs_pr_info("DNLD_CMD: hw_host_to_card failed: %d\n", ret);
/* Let the timer kick in and retry, and potentially reset
the whole thing if the condition persists */
- timeo = HZ;
+ timeo = HZ/2;
}
/* Setup the timer after transmit command */
Index: wireless-testing/drivers/net/wireless/libertas/main.c
===================================================================
--- wireless-testing.orig/drivers/net/wireless/libertas/main.c 2008-04-30 15:10:13.000000000 +0200
+++ wireless-testing/drivers/net/wireless/libertas/main.c 2008-04-30 15:16:27.000000000 +0200
@@ -749,15 +749,18 @@ static int lbs_thread(void *data)
if (priv->cmd_timed_out && priv->cur_cmd) {
struct cmd_ctrl_node *cmdnode = priv->cur_cmd;
- if (++priv->nr_retries > 10) {
- lbs_pr_info("Excessive timeouts submitting command %x\n",
- le16_to_cpu(cmdnode->cmdbuf->command));
+ if (++priv->nr_retries > 3) {
+ lbs_pr_info("excessive timeouts submitting "
+ "command 0x%04x\n",
+ le16_to_cpu(cmdnode->cmdbuf->command));
lbs_complete_command(priv, cmdnode, -ETIMEDOUT);
priv->nr_retries = 0;
} else {
priv->cur_cmd = NULL;
- lbs_pr_info("requeueing command %x due to timeout (#%d)\n",
- le16_to_cpu(cmdnode->cmdbuf->command), priv->nr_retries);
+ lbs_pr_info("requeueing command 0x%04x due "
+ "to timeout (#%d)\n",
+ le16_to_cpu(cmdnode->cmdbuf->command),
+ priv->nr_retries);
/* Stick it back at the _top_ of the pending queue
for immediate resubmission */
@@ -958,12 +961,11 @@ static void command_timer_fn(unsigned lo
lbs_deb_enter(LBS_DEB_CMD);
spin_lock_irqsave(&priv->driver_lock, flags);
- if (!priv->cur_cmd) {
- lbs_pr_info("Command timer expired; no pending command\n");
+ if (!priv->cur_cmd)
goto out;
- }
- lbs_pr_info("Command %x timed out\n", le16_to_cpu(priv->cur_cmd->cmdbuf->command));
+ lbs_pr_info("command 0x%04x timed out\n",
+ le16_to_cpu(priv->cur_cmd->cmdbuf->command));
priv->cmd_timed_out = 1;
wake_up_interruptible(&priv->waitq);
@@ -1278,7 +1280,7 @@ void lbs_stop_card(struct lbs_private *p
device_remove_file(&dev->dev, &dev_attr_lbs_mesh);
/* Flush pending command nodes */
- del_timer_sync(&priv->command_timer);
+ del_timer(&priv->command_timer);
spin_lock_irqsave(&priv->driver_lock, flags);
list_for_each_entry(cmdnode, &priv->cmdpendingq, list) {
cmdnode->result = -ENOENT;
next reply other threads:[~2008-04-30 14:47 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-04-30 14:44 Holger Schurig [this message]
2008-04-30 20:34 ` [PATCH] libertas: reduce command retry time Dan Williams
2008-04-30 20:43 ` David Woodhouse
2008-05-01 21:30 ` Holger Schurig
2008-05-01 21:57 ` David Woodhouse
2008-05-02 6:07 ` Holger Schurig
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=200804301644.29927.hs4233@mail.mn-solutions.de \
--to=hs4233@mail.mn-solutions.de \
--cc=dcbw@redhat.com \
--cc=libertas-dev@lists.infradead.org \
--cc=linux-wireless@vger.kernel.org \
--cc=linville@tuxdriver.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.