Tim Howe wrote:
Tim Howe <[EMAIL PROTECTED]> writes:
ata0-master: stat=0xd0 err=0xd0 lsb=0xd0 msb=0xd0
This turned out to be the key.
Version 1.51 of ata-lowlevel.c added a check for stat0/1, err, lsb, and
msb being identical. If they are, it aborts the probe. The attached
patch creates an option ATA_AGGRESSIVE_PROBE which disables this for the
old aggressive behavior (which may wait up to the full 31 seconds).
I also took the liberty of reworking the still-busy check from 3
equality tests to 2 bitmask tests. It seems simpler to my eye with
identical results, but if I missed something or the other style was
preferred please let me know.
The patch is against 5-STABLE because that's what I have.
Looks good except for one issue I talk about below. By the way, I don't
think this will ever appear back into the RELENG_5 tree unless it gets a
PR or at least someone who can check it in. Is there anyone willing to
ask the right people? You might look at the initials for whoever changed
the code recently and see if you can figure out who that was.
Meanwhile, I will try and test it within the week.
See below for my question about the patch.
Billy
--- sys/conf/options.stock Tue Sep 27 00:28:50 2005
+++ sys/conf/options Tue Sep 27 00:55:53 2005
@@ -301,6 +301,7 @@
# Options used in the 'ata' ATA/ATAPI driver
ATA_STATIC_ID opt_ata.h
ATA_NOPCI opt_ata.h
+ATA_AGGRESSIVE_PROBE opt_ata.h
DEV_ATADISK opt_ata.h
DEV_ATAPICD opt_ata.h
DEV_ATAPIST opt_ata.h
--- sys/conf/NOTES.stock Tue Sep 27 00:30:28 2005
+++ sys/conf/NOTES Tue Sep 27 00:56:39 2005
@@ -1501,8 +1501,11 @@
#
# ATA_STATIC_ID: controller numbering is static ie depends on location
# else the device numbers are dynamically allocated.
+# ATA_AGGRESSIVE_PROBE: Does not give up on probing when all err, lsb,
and
+ msb are identical.
Shouldn't there be a comment for the above line?
Like this:
+# msb are identical.
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
