[PATCH|RFC] beagle: make nand ecc command based

[PATCH|RFC] beagle: make nand ecc command based

[Michael Grzeschik]

Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
---
This will only work once and bring the nand chip into a undefined state
after a second call. Any ideas for doing this save?

 arch/arm/boards/omap/board-beagle.c |   28 +++++++++++++++++++++++++++-
 1 files changed, 27 insertions(+), 1 deletions(-)

diff --git a/arch/arm/boards/omap/board-beagle.c b/arch/arm/boards/omap/board-beagle.c
index c0b6b43..99766af 100644
--- a/arch/arm/boards/omap/board-beagle.c
+++ b/arch/arm/boards/omap/board-beagle.c
@@ -53,6 +53,7 @@
  */
 
 #include <common.h>
+#include <command.h>
 #include <console.h>
 #include <init.h>
 #include <driver.h>
@@ -324,7 +325,6 @@ static int beagle_devices_init(void)
 	/* WP is made high and WAIT1 active Low */
 	gpmc_generic_init(0x10);
 #endif
-	gpmc_generic_nand_devices_init(0, 16, 1);
 
 	armlinux_add_dram(&sdram_dev);
 	armlinux_set_bootparams((void *)0x80000100);
@@ -333,3 +333,29 @@ failed:
 	return ret;
 }
 device_initcall(beagle_devices_init);
+
+static int do_ecc(struct command *cmdtp, int argc, char *argv[])
+{
+	if (*argv[0] == 's') {
+		gpmc_generic_nand_devices_init(0, 16, 0);
+	} else {
+		gpmc_generic_nand_devices_init(0, 16, 1);
+	}
+	return 0;
+}
+
+static const __maybe_unused char cmd_ecc_help[] =
+"Usage: (s|h)whwcc <device>\n"
+"set software or hardware ecc for nand controller \n";
+
+BAREBOX_CMD_START(hwecc)
+        .cmd            = do_ecc,
+        .usage          = "enable HWECC support in NAND",
+        BAREBOX_CMD_HELP(cmd_ecc_help)
+BAREBOX_CMD_END
+
+BAREBOX_CMD_START(swecc)
+        .cmd            = do_ecc,
+        .usage          = "enable SWECC support in NAND",
+        BAREBOX_CMD_HELP(cmd_ecc_help)
+BAREBOX_CMD_END
-- 
1.7.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH|RFC] beagle: make nand ecc command based

[Nishanth Menon]

On 08/17/2010 03:55 AM, Michael Grzeschik wrote:
> Signed-off-by: Michael Grzeschik<m.grzeschik@pengutronix.de>
> ---
> This will only work once and bring the nand chip into a undefined state
> after a second call. Any ideas for doing this save?
looking at the gpmc logic, it does a reset in gpmc_cs_config by 
disabling and re-enabling it -> so my guess is:
a) in the selection of ecc logic
b) reset of statemachines in mtd layers
c) nand chip not being reset from it's previous state (resetting the 
controller does not mean nand chip is reset) (if i recollect sometime 
back mtd used to do a 0xff and reset)..


personally, IMHO using s/w ecc has not much benefit other than being 
"legacy enabled"
>
>   arch/arm/boards/omap/board-beagle.c |   28 +++++++++++++++++++++++++++-
>   1 files changed, 27 insertions(+), 1 deletions(-)
>
> diff --git a/arch/arm/boards/omap/board-beagle.c b/arch/arm/boards/omap/board-beagle.c
> index c0b6b43..99766af 100644
> --- a/arch/arm/boards/omap/board-beagle.c
> +++ b/arch/arm/boards/omap/board-beagle.c
> @@ -53,6 +53,7 @@
>    */
>
>   #include<common.h>
> +#include<command.h>
>   #include<console.h>
>   #include<init.h>
>   #include<driver.h>
> @@ -324,7 +325,6 @@ static int beagle_devices_init(void)
>   	/* WP is made high and WAIT1 active Low */
>   	gpmc_generic_init(0x10);
>   #endif
> -	gpmc_generic_nand_devices_init(0, 16, 1);
>
>   	armlinux_add_dram(&sdram_dev);
>   	armlinux_set_bootparams((void *)0x80000100);
> @@ -333,3 +333,29 @@ failed:
>   	return ret;
>   }
>   device_initcall(beagle_devices_init);
> +
> +static int do_ecc(struct command *cmdtp, int argc, char *argv[])
> +{
> +	if (*argv[0] == 's') {
> +		gpmc_generic_nand_devices_init(0, 16, 0);
> +	} else {
> +		gpmc_generic_nand_devices_init(0, 16, 1);
> +	}
> +	return 0;
> +}
> +
> +static const __maybe_unused char cmd_ecc_help[] =
> +"Usage: (s|h)whwcc<device>\n"
> +"set software or hardware ecc for nand controller \n";
> +
> +BAREBOX_CMD_START(hwecc)
> +        .cmd            = do_ecc,
> +        .usage          = "enable HWECC support in NAND",
> +        BAREBOX_CMD_HELP(cmd_ecc_help)
> +BAREBOX_CMD_END
> +
> +BAREBOX_CMD_START(swecc)
> +        .cmd            = do_ecc,
> +        .usage          = "enable SWECC support in NAND",
> +        BAREBOX_CMD_HELP(cmd_ecc_help)
> +BAREBOX_CMD_END


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH|RFC] beagle: make nand ecc command based

[Sascha Hauer]

On Tue, Aug 17, 2010 at 05:40:09AM -0500, Nishanth Menon wrote:
> On 08/17/2010 03:55 AM, Michael Grzeschik wrote:
>> Signed-off-by: Michael Grzeschik<m.grzeschik@pengutronix.de>
>> ---
>> This will only work once and bring the nand chip into a undefined state
>> after a second call. Any ideas for doing this save?
> looking at the gpmc logic, it does a reset in gpmc_cs_config by  
> disabling and re-enabling it -> so my guess is:
> a) in the selection of ecc logic
> b) reset of statemachines in mtd layers
> c) nand chip not being reset from it's previous state (resetting the  
> controller does not mean nand chip is reset) (if i recollect sometime  
> back mtd used to do a 0xff and reset)..
>
>
> personally, IMHO using s/w ecc has not much benefit other than being  
> "legacy enabled"

It really seems odd to me that the omap internal ROM code expects hw
ecc while xloader and kernel expect hw ecc. This way we always need
two different ecc algorithms in place which is really inconvenient and
hard for users to get it right.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH|RFC] beagle: make nand ecc command based

[Nishanth Menon]

On 08/18/2010 01:42 AM, Sascha Hauer wrote:
> On Tue, Aug 17, 2010 at 05:40:09AM -0500, Nishanth Menon wrote:
>> On 08/17/2010 03:55 AM, Michael Grzeschik wrote:
>>> Signed-off-by: Michael Grzeschik<m.grzeschik@pengutronix.de>
>>> ---
>>> This will only work once and bring the nand chip into a undefined state
>>> after a second call. Any ideas for doing this save?
>> looking at the gpmc logic, it does a reset in gpmc_cs_config by
>> disabling and re-enabling it ->  so my guess is:
>> a) in the selection of ecc logic
>> b) reset of statemachines in mtd layers
>> c) nand chip not being reset from it's previous state (resetting the
>> controller does not mean nand chip is reset) (if i recollect sometime
>> back mtd used to do a 0xff and reset)..
>>
>>
>> personally, IMHO using s/w ecc has not much benefit other than being
>> "legacy enabled"
>
> It really seems odd to me that the omap internal ROM code expects hw
> ecc while xloader and kernel expect hw ecc. This way we always need
> two different ecc algorithms in place which is really inconvenient and
> hard for users to get it right.


I agree, but there has been a lot of history behind how this came to 
happen. there is still discussion in linux-omap in getting kernel's mtd 
driver to sanely handle the ecc strategy selectable by the board.

Regards,
Nishanth Menon


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox