Re: [PATCH] libata-core Use more robust parsing for multi_count

[Mark Lord <liml@rtr.ca>]

Mark Lord wrote:
> Alan Cox wrote:
>>> +            switch (dev->id[59] & 0xff) {
>>
>> Umm surely...
>>
>>     if (n & (n - 1))
> ..
> 
> ????
..

Ah.. backwards.  Thanks, Alan!  Revised patch follows:

-----

Make libata a little more robust in parsing the multi_count
field from a drive's identify data.  This prevents libata from
attempting to use dubious multi_count values ad infinitum.

Signed-off-by: Mark Lord <mlord@pobox.com> 

--- upstream/drivers/ata/libata-core.c	2009-03-18 10:21:07.000000000 -0400
+++ new/drivers/ata/libata-core.c	2009-03-18 10:22:05.000000000 -0400
@@ -2426,9 +2426,12 @@
 
 		dev->n_sectors = ata_id_n_sectors(id);
 
-		if (dev->id[59] & 0x100)
-			dev->multi_count = dev->id[59] & 0xff;
-
+		dev->multi_count = 0;
+		if (dev->id[59] & 0x100) {
+			int mc = dev->id[59] & 0xff;
+			if ((mc & (mc - 1)) == 0) /* even power of two? */
+				dev->multi_count = mc;
+		}
 		if (ata_id_has_lba(id)) {
 			const char *lba_desc;
 			char ncq_desc[20];