RE: 2.6.13.2 aacraid regression

["Salyzyn, Mark" <mark_salyzyn@adaptec.com>]

Juan was kind enough to linger on site, and work on a production
machine, to try the parameter to make the system stable. He discovered
that reducing the maximum transfer size issued to the adapter to 128KB
stabilized his system. This is related to an earlier change for the
2.6.13 tree resulting from Martin Drab's testing where the transfer size
was reduced from 4G to 256KB; we needed to go still further in scaling
back the request size.

Here is the patch that tames this regression.

Applies to the 2.6.13.2 tree.

Signed-off-by: Mark Salyzyn <aacraid@adaptec.com>

Index: linux-2.6.13.2/drivers/scsi/aacraid/aacraid.h
===================================================================
--- linux-2.6.13.2/drivers/scsi/aacraid/aacraid.h       2005-10-05
12:45:16 -0400
+++ linux-2.6.13.2-aacraid-fix/drivers/scsi/aacraid/aacraid.h
2005-09-16 21:02:12 -0400
@@ -15,7 +15,7 @@
 #define AAC_MAX_LUN            (8)

 #define AAC_MAX_HOSTPHYSMEMPAGES (0xfffff)
-#define AAC_MAX_32BIT_SGBCOUNT ((unsigned short)512)
+#define AAC_MAX_32BIT_SGBCOUNT ((unsigned short)256)

 /*
  * These macros convert from physical channels to virtual channels

Sincerely -- Mark Salyzyn
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/