Dear Dirk Behme,
Hi Marek,
On 30.05.2012 07:49, Marek Vasut wrote:
Dear Wolfgang Denk,
Dear Otavio Salvador,
In message1336866018-614-5-git-send-email-otavio@ossystems.com.br you
wrote:
Signed-off-by: Otavio Salvadorotavio@ossystems.com.br
tools/mxsboot.c | 110 +++++++++++++++++++++++++++---------------------------- 1 file changed, 55 insertions(+), 55 deletions(-)
diff --git a/tools/mxsboot.c b/tools/mxsboot.c index 6c05aa4..9661a48 100644 --- a/tools/mxsboot.c +++ b/tools/mxsboot.c @@ -1,5 +1,5 @@
/*
- Freescale i.MX28 image generator
- Freescale i.MXS image generator
I'm unhappy about the name "i.MXS".
I'm unhappy about the image generator ;-)
Ok, now Otavio will probably hate me, but I believe it'd be much more helpful to convert mxsboot into part of mkimage (though this is unrelated to this patch). And maybe even better, study the generation of boot headers and compare it with mx53/mx6q ones, as mx53 ones are supported by imximage and mx28 carries direct predecessor of mx53 bootrom and then try to implement support into imximage part of mkimage.
I wasn't aware of mxsboot before this thread, so I start looking at it. I'm no expert of mx28 nor how the boot ROMs handle the NAND. Therefore some comments and questions:
Looking at
http://git.denx.de/?p=u-boot.git;a=blob;f=tools/mxsboot.c;h=6c05aa479d4360c 0c14cd0f989c250ecf119126d;hb=HEAD#l453
mxsboot seem to support NAND and SD. Looking at the options given there, yes it sounds like a good idea to move that to imximage.
100%
One question regarding the NAND handling, though:
Is mxsboot assumed to replace the Freescale tool 'kobs-ng'
I believe so. It generates the necessary NAND headers for NAND boot. You can then write the resulting image directly at the begining of the NAND.
http://imxcommunity.org/forum/topics/i-mx6-nand-boot?commentId=4103961%3ACo mment%3A69314&xg_source=activity
?
It was my understanding that for the NAND the bad sector table is necessary to be able to write a bootable image to the NAND? And this can be done only on the target?
You mean DBBT? From what I remember, the bad block handling is done with the BCH module and the first block is always OK anyway. Or it was somehow like that, but I might be wrong. The point is, mxsboot doesn't use DBBT.
While mxsboot runs on the host? So it's not sufficient to give mxsboot the page/OOB/erase size options because the bad sector table is individual for each board/NAND chip?
I don't see why the bad sector table is needed at all as the BCH handles the ECC.
Best regards
Dirk
Best regards, Marek Vasut