diff options
| author | msmith <msmith@FreeBSD.org> | 2000-03-23 18:33:19 +0000 |
|---|---|---|
| committer | msmith <msmith@FreeBSD.org> | 2000-03-23 18:33:19 +0000 |
| commit | 2054c7bab47b0600e4308a6dd0d6d9f7605067ee (patch) | |
| tree | 748794e9e4675cd48d8499df036d1faa3f4c1be6 /sys/dev/amr/amr.c | |
| parent | f3ae48fd505aa13a32ee063d82a80ed52e7a298f (diff) | |
| download | FreeBSD-src-2054c7bab47b0600e4308a6dd0d6d9f7605067ee.zip FreeBSD-src-2054c7bab47b0600e4308a6dd0d6d9f7605067ee.tar.gz | |
Fix detection of the card-ready-for-command state. Use an ultimate
timeout closer to the 1-second value that AMI use in their Linux
driver, and mark the mailbox structures as volatile so that gcc
doesn't over-optimise access to them.
This should fix the "controller wedged" bug.
Submitted by: Brian Dean <brdean@unx.sas.com>
Diffstat (limited to 'sys/dev/amr/amr.c')
| -rw-r--r-- | sys/dev/amr/amr.c | 9 |
1 files changed, 7 insertions, 2 deletions
diff --git a/sys/dev/amr/amr.c b/sys/dev/amr/amr.c index 6d7524d..3145e08 100644 --- a/sys/dev/amr/amr.c +++ b/sys/dev/amr/amr.c @@ -1126,7 +1126,7 @@ amr_start(struct amr_command *ac) /* spin waiting for the mailbox */ debug("wait for mailbox"); - for (i = 100000, done = 0, worked = 0; (i > 0) && !done; i--) { + for (i = 10000, done = 0, worked = 0; (i > 0) && !done; i--) { s = splbio(); /* is the mailbox free? */ @@ -1142,7 +1142,12 @@ amr_start(struct amr_command *ac) /* not free, try to clean up while we wait */ } else { debug("busy flag %x\n", sc->amr_mailbox->mb_busy); - worked = amr_done(sc); + /* try to kill some time being useful */ + if (amr_done(sc)) { + worked = 1; + } else { + DELAY(100); + } } splx(s); } |
