Error: Wacky! Unable to decode failure status

[Michael Moedt]

--- Thayne Harbaugh <tharbaugh at lnxi.com> wrote:
> On Mon, 2003-05-05 at 13:42, Michael Moedt wrote:
> > > > I'm getting the error message "Wacky! Unable
> > > > to decode failure status".  See below:
> > > 
..
> > I am using 1.68. (last modified 04/21, by thayne. 
> > =])
> 
> Good.
> 
> > As far as we can tell, it looks like the erase
> > operation completes correctly, and we start
> > reading the flash again, getting the erased value
> > of 0xffff.
> > (It is in 16-bit wide mode)
> 
> Correct - all prior "MTD do_erase_oneblock(): Check"
> messages indicate that dq6 is toggling.
> 
> > The 'ones' variable that it compares the read
> > values to is only an 8-bit 0xff.  I traced this
> > back a little,  (the CMD macro in
include/linux/mtd/cfi.h
> > and the function right above it) and it keeps it
> > as 8-bit even if it knows that it's x16.
> 
> The status should only be a single byte because the
> erase or write (depending on the operation) is being
> sent to a single chip (not using interlaced chips).
> The final test value should be two bytes because you
> are using the x16 mode.
> 
> > Should it be masking off only the lowest byte?
> 
> Nope - not for the final data verification.
> 
> Try the attached patch.
> 
> > __________________________________
> > Do you Yahoo!?
> > The New Yahoo! Search - Faster. Easier. Bingo.
> > http://search.yahoo.com
> -- 
> Thayne Harbaugh
> Linux Networx
> > diff -u -r1.68 cfi_cmdset_0002.c

...

Thanks!  I guess that was the problem.  However, I
think that it should compare CFIDEV_BUSWIDTH_ 1,2,4
and 8 instead of 1,2,3,4.  Here's the corrected patch.

diff -u -r1.68 cfi_cmdset_0002.c
--- drivers/mtd/chips/cfi_cmdset_0002.c	21 Apr 2003
23:26:25 -0000	1.68
+++ drivers/mtd/chips/cfi_cmdset_0002.c	5 May 2003
20:21:53 -0000
@@ -925,7 +931,16 @@
 	DEBUG( MTD_DEBUG_LEVEL3, "MTD %s(): Check 0x%.8x
0x%.8x\n",
 	       __func__, oldstatus, status );
 
-	ones = CMD( (__u8)~0 );
+	switch(CFIDEV_BUSWIDTH) {
+	case(CFIDEV_BUSWIDTH_1): ones =  (__u8)~0; break;
+	case(CFIDEV_BUSWIDTH_2): ones = (__u16)~0; break;
+	case(CFIDEV_BUSWIDTH_4): ones = (__u32)~0; break;
+	case(CFIDEV_BUSWIDTH_8): ones = (__u64)~0; break;
+	default:
+	        printk(KERN_WARNING "Unsupported
buswidth\n");
+		goto erase_failed;
+	}
+	ones = CMD( ones );
 	
 	if ( oldstatus == ones && status == ones ) {
 		/* success - do nothing */
@@ -1095,8 +1111,17 @@
 	DEBUG( MTD_DEBUG_LEVEL3, "MTD %s(): Check 0x%.8x
0x%.8x\n",
 	       __func__, oldstatus, status );
 
-	ones = CMD( (__u8)~0 );
-
+	switch(CFIDEV_BUSWIDTH) {
+	case(CFIDEV_BUSWIDTH_1): ones =  (__u8)~0; break;
+	case(CFIDEV_BUSWIDTH_2): ones = (__u16)~0; break;
+	case(CFIDEV_BUSWIDTH_4): ones = (__u32)~0; break;
+	case(CFIDEV_BUSWIDTH_8): ones = (__u64)~0; break;
+	default:
+	        printk(KERN_WARNING "Unsupported
buswidth\n");
+		goto erase_failed;
+	}
+	ones = CMD( ones );
+	
 	if ( oldstatus == ones && status == ones ) {
 		/* success - do nothing */
 		goto erase_done;


Thanks again,
Mike

__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com