[PATCH] [MTD] CHIPS: 0xFF intolerance for M29W128G

[PATCH] [MTD] CHIPS: 0xFF intolerance for M29W128G

[massimo cirillo]

From: Massimo Cirillo <maxcir@gmail.com>

The M29W128G Numonyx flash devices are intolerant to any 0xFF command:
in the Cfi_util.c the function cfi_qry_mode_off() (that resets the device
after the autoselect mode) must have a 0xF0 command after the 0xFF command.
Intel-like devices are not influenced by a 0xF0 command.
This fix solves also the cause of the fixup_M29W128G_write_buffer() fix,
that can be commented out now.
The following patch applies to 2.6.30 kernel.

Signed-off-by: Massimo Cirillo <maxcir@gmail.com>
---
diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c
b/drivers/mtd/chips/cfi_cmdset_0002.c
old mode 100644
new mode 100755
index 61ea833..5a65e92
--- a/drivers/mtd/chips/cfi_cmdset_0002.c
+++ b/drivers/mtd/chips/cfi_cmdset_0002.c
@@ -308,7 +308,7 @@ static struct cfi_fixup cfi_fixup_table[] = {
 	{ CFI_MFR_AMD, 0x1301, fixup_s29gl064n_sectors, NULL, },
 	{ CFI_MFR_AMD, 0x1a00, fixup_s29gl032n_sectors, NULL, },
 	{ CFI_MFR_AMD, 0x1a01, fixup_s29gl032n_sectors, NULL, },
-	{ CFI_MFR_ST,  0x227E, fixup_M29W128G_write_buffer, NULL, },
+	/*{ CFI_MFR_ST,  0x227E, fixup_M29W128G_write_buffer, NULL, },*/
 #if !FORCE_WORD_WRITE
 	{ CFI_MFR_ANY, CFI_ID_ANY, fixup_use_write_buffers, NULL, },
 #endif
diff --git a/drivers/mtd/chips/cfi_util.c b/drivers/mtd/chips/cfi_util.c
old mode 100644
new mode 100755
index 34d40e2..9c897f9
--- a/drivers/mtd/chips/cfi_util.c
+++ b/drivers/mtd/chips/cfi_util.c
@@ -61,8 +61,8 @@ int __xipram cfi_qry_mode_on(uint32_t base,
 		return 1;
 	/* QRY not found probably we deal with some odd CFI chips */
 	/* Some revisions of some old Intel chips? */
-	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 	cfi_send_gen_cmd(0xFF, 0, base, map, cfi, cfi->device_type, NULL);
+	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 	cfi_send_gen_cmd(0x98, 0x55, base, map, cfi, cfi->device_type, NULL);
 	if (cfi_qry_present(map, base, cfi))
 		return 1;
@@ -79,8 +79,11 @@ EXPORT_SYMBOL_GPL(cfi_qry_mode_on);
 void __xipram cfi_qry_mode_off(uint32_t base, struct map_info *map,
 			       struct cfi_private *cfi)
 {
-	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 	cfi_send_gen_cmd(0xFF, 0, base, map, cfi, cfi->device_type, NULL);
+    /* Numonyx M29W128G requires two 0xF0 to exit succesfully from
+    autoselect mode. No consequences on Intel-like devices. */
+	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
+	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 }
 EXPORT_SYMBOL_GPL(cfi_qry_mode_off);

--

Re: [PATCH] [MTD] CHIPS: 0xFF intolerance for M29W128G

[Jamie Lokier]

massimo cirillo wrote:
> From: Massimo Cirillo <maxcir@gmail.com>
> 
> The M29W128G Numonyx flash devices are intolerant to any 0xFF command:
> in the Cfi_util.c the function cfi_qry_mode_off() (that resets the device
> after the autoselect mode) must have a 0xF0 command after the 0xFF command.
> Intel-like devices are not influenced by a 0xF0 command.
> This fix solves also the cause of the fixup_M29W128G_write_buffer() fix,
> that can be commented out now.
> The following patch applies to 2.6.30 kernel.

This change was discussed 1 year ago:

   http://lists.infradead.org/pipermail/linux-mtd/2008-August/022497.html

The conclusion was that 0xf0 after 0xff might not be safe for some
chips ("early revisions of L30"), but nobody could confirm which
chips, so Alexey Korolov suggested staying safe and using fixup
functions.

I'm inclined to think a per-manufacturer (ignoring chip-id) reset
function would be better than the attempt to poke several different
commands at a chip all mixed together in a careful order, knowing that
some commands break some chips but later commands fix them again.

-- Jamie

Re: [PATCH] [MTD] CHIPS: 0xFF intolerance for M29W128G

[massimo cirillo]

2009/8/26 Jamie Lokier <jamie@shareable.org>:
> massimo cirillo wrote:
>> From: Massimo Cirillo <maxcir@gmail.com>
>>
>> The M29W128G Numonyx flash devices are intolerant to any 0xFF command:
>> in the Cfi_util.c the function cfi_qry_mode_off() (that resets the device
>> after the autoselect mode) must have a 0xF0 command after the 0xFF command.
>> Intel-like devices are not influenced by a 0xF0 command.
>> This fix solves also the cause of the fixup_M29W128G_write_buffer() fix,
>> that can be commented out now.
>> The following patch applies to 2.6.30 kernel.
>
> This change was discussed 1 year ago:
>
>   http://lists.infradead.org/pipermail/linux-mtd/2008-August/022497.html
>
> The conclusion was that 0xf0 after 0xff might not be safe for some
> chips ("early revisions of L30"), but nobody could confirm which
> chips, so Alexey Korolov suggested staying safe and using fixup
> functions.
>
> I'm inclined to think a per-manufacturer (ignoring chip-id) reset
> function would be better than the attempt to poke several different
> commands at a chip all mixed together in a careful order, knowing that
> some commands break some chips but later commands fix them again.
>
> -- Jamie
>

Jamie,
I've rewritten the patch: now only the function cfi_qry_mode_off()
has been patched, and only if the device is a M29W128G (16bit or 8bit)
a final 0xF0 is given. The fixup_M29W128G_write_buffer() keeps on being
removed, because the buffer write failure derived from the unstable
state due to the missing 0xF0 command.
Please give comments about this new version.

Signed-off-by: Massimo Cirillo <maxcir@gmail.com>
---
diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c
b/drivers/mtd/chips/cfi_cmdset_0002.c
old mode 100644
new mode 100755
index 61ea833..86cc313
--- a/drivers/mtd/chips/cfi_cmdset_0002.c
+++ b/drivers/mtd/chips/cfi_cmdset_0002.c
@@ -308,7 +308,7 @@ static struct cfi_fixup cfi_fixup_table[] = {
 	{ CFI_MFR_AMD, 0x1301, fixup_s29gl064n_sectors, NULL, },
 	{ CFI_MFR_AMD, 0x1a00, fixup_s29gl032n_sectors, NULL, },
 	{ CFI_MFR_AMD, 0x1a01, fixup_s29gl032n_sectors, NULL, },
-	{ CFI_MFR_ST,  0x227E, fixup_M29W128G_write_buffer, NULL, },
+	/* { CFI_MFR_ST,  0x227E, fixup_M29W128G_write_buffer, NULL, }, */
 #if !FORCE_WORD_WRITE
 	{ CFI_MFR_ANY, CFI_ID_ANY, fixup_use_write_buffers, NULL, },
 #endif
diff --git a/drivers/mtd/chips/cfi_util.c b/drivers/mtd/chips/cfi_util.c
old mode 100644
new mode 100755
index 34d40e2..43511ab
--- a/drivers/mtd/chips/cfi_util.c
+++ b/drivers/mtd/chips/cfi_util.c
@@ -81,6 +81,9 @@ void __xipram cfi_qry_mode_off(uint32_t base, struct
map_info *map,
 {
 	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 	cfi_send_gen_cmd(0xFF, 0, base, map, cfi, cfi->device_type, NULL);
+	/* send a 0xF0 if the device is M29W128G */
+	if ((cfi->mfr == CFI_MFR_ST) && (cfi->id == 0x227E || cfi->id == 0x7E))
+		cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 }
 EXPORT_SYMBOL_GPL(cfi_qry_mode_off);

--
Best Regards,
Massimo Cirillo <maxcir@gmail.com>

Re:[PATCH] [MTD] CHIPS: 0xFF intolerance for M29W128G

[Alexey Korolev]

Hi Massimo,

> Jamie,
> I've rewritten the patch: now only the function cfi_qry_mode_off()
> has been patched, and only if the device is a M29W128G (16bit or 8bit)
> a final 0xF0 is given. The fixup_M29W128G_write_buffer() keeps on being
> removed, because the buffer write failure derived from the unstable
> state due to the missing 0xF0 command.
> Please give comments about this new version.

> @@ -308,7 +308,7 @@ static struct cfi_fixup cfi_fixup_table[] = {
>  	{ CFI_MFR_AMD, 0x1301, fixup_s29gl064n_sectors, NULL, },
>  	{ CFI_MFR_AMD, 0x1a00, fixup_s29gl032n_sectors, NULL, },
>  	{ CFI_MFR_AMD, 0x1a01, fixup_s29gl032n_sectors, NULL, },
> -	{ CFI_MFR_ST,  0x227E, fixup_M29W128G_write_buffer, NULL, },
> +	/* { CFI_MFR_ST,  0x227E, fixup_M29W128G_write_buffer, NULL, },
It does not need to be commented. Just remove this line :).

> @@ -81,6 +81,9 @@ void __xipram cfi_qry_mode_off(uint32_t base, struct
> map_info *map,
>  {
>  	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
>  	cfi_send_gen_cmd(0xFF, 0, base, map, cfi, cfi->device_type, NULL);
> +	/* send a 0xF0 if the device is M29W128G */
> +	if ((cfi->mfr == CFI_MFR_ST) && (cfi->id == 0x227E || cfi->id == 0x7E))
> +		cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
May be it would be better to add comment like this:
 "M29W128G flashes require an additional reset command when exit qry mode".

New version is much better. This implementation does not break
compatibility with other devices. Please add acked-by when update.

Thanks,
Alexey

p/s Mail thread is interrupted because I weren't subscribed. Sorry for that.

Re: [PATCH] [MTD] CHIPS: 0xFF intolerance for M29W128G

[massimo cirillo]

From: Massimo Cirillo <maxcir@gmail.com>

The M29W128G Numonyx flash devices are intolerant to any 0xFF command:
in the Cfi_util.c the function cfi_qry_mode_off() (that resets the device
after the autoselect mode) must have a 0xF0 command after the 0xFF command.
This fix solves also the cause of the fixup_M29W128G_write_buffer() fix,
that can be removed now.
The following patch applies to 2.6.30 kernel.

Signed-off-by: Massimo Cirillo <maxcir@gmail.com>
Acked-by: alexey Korolev <akorolev@infradead.org>
---
diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c
b/drivers/mtd/chips/cfi_cmdset_0002.c
old mode 100644
new mode 100755
index 61ea833..94bb61e
--- a/drivers/mtd/chips/cfi_cmdset_0002.c
+++ b/drivers/mtd/chips/cfi_cmdset_0002.c
@@ -282,16 +282,6 @@ static void fixup_s29gl032n_sectors(struct mtd_info *mtd,
 	}
 }

-static void fixup_M29W128G_write_buffer(struct mtd_info *mtd, void *param)
-{
-	struct map_info *map = mtd->priv;
-	struct cfi_private *cfi = map->fldrv_priv;
-	if (cfi->cfiq->BufWriteTimeoutTyp) {
-		pr_warning("Don't use write buffer on ST flash M29W128G\n");
-		cfi->cfiq->BufWriteTimeoutTyp = 0;
-	}
-}
-
 static struct cfi_fixup cfi_fixup_table[] = {
 	{ CFI_MFR_ATMEL, CFI_ID_ANY, fixup_convert_atmel_pri, NULL },
 #ifdef AMD_BOOTLOC_BUG
@@ -308,7 +298,6 @@ static struct cfi_fixup cfi_fixup_table[] = {
 	{ CFI_MFR_AMD, 0x1301, fixup_s29gl064n_sectors, NULL, },
 	{ CFI_MFR_AMD, 0x1a00, fixup_s29gl032n_sectors, NULL, },
 	{ CFI_MFR_AMD, 0x1a01, fixup_s29gl032n_sectors, NULL, },
-	{ CFI_MFR_ST,  0x227E, fixup_M29W128G_write_buffer, NULL, },
 #if !FORCE_WORD_WRITE
 	{ CFI_MFR_ANY, CFI_ID_ANY, fixup_use_write_buffers, NULL, },
 #endif
diff --git a/drivers/mtd/chips/cfi_util.c b/drivers/mtd/chips/cfi_util.c
old mode 100644
new mode 100755
index 34d40e2..8b87652
--- a/drivers/mtd/chips/cfi_util.c
+++ b/drivers/mtd/chips/cfi_util.c
@@ -81,6 +81,10 @@ void __xipram cfi_qry_mode_off(uint32_t base,
 {
 	cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 	cfi_send_gen_cmd(0xFF, 0, base, map, cfi, cfi->device_type, NULL);
+	/* M29W128G flashes require an additional reset command
+	   when exit qry mode */
+	if ((cfi->mfr == CFI_MFR_ST) && (cfi->id == 0x227E || cfi->id == 0x7E))
+		cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);
 }
 EXPORT_SYMBOL_GPL(cfi_qry_mode_off);