Tom,
Thanks for the comments. I'll wait a couple of days for others then resubmit with fixes for these problems and any others that come up.
John
On Wed, Sep 30, 2009 at 1:22 PM, Tom Tom.Rix@windriver.com wrote:
John Rigby wrote:
New commands nand read.raw and write.raw read/write main and oob area.
Implement previously stubbed nand biterr command.
Document the above and also the previously undocumented read.oob and write.oob.
Signed-off-by: John Rigby jrigby@control4.com
common/cmd_nand.c | 115 +++++++++++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 112 insertions(+), 3 deletions(-)
diff --git a/common/cmd_nand.c b/common/cmd_nand.c index 158a55f..a488038 100644 --- a/common/cmd_nand.c +++ b/common/cmd_nand.c @@ -90,6 +90,95 @@ static int nand_dump(nand_info_t *nand, ulong off, int only_oob) return 0; } +#define NAND_RW_RAW_READ 0 +#define NAND_RW_RAW_WRITE 1
+static int nand_rdwr_raw(int rdwr, nand_info_t *nand, ulong off, u_char *buf,
size_t size)+{
struct mtd_oob_ops ops = {.len = nand->writesize,.ooblen = nand->oobsize,.mode = MTD_OOB_RAW,};int i;int nrblocks = size / nand->writesize;loff_t addr = (loff_t)(off & ~(nand->writesize - 1));Silently dropping bytes. Would it be better to require the size to be a multiple of the block size ? Or at least warn of the dropped bytes.
while (nrblocks--) {ops.datbuf = buf;<snip>
}@@ -494,13 +600,16 @@ U_BOOT_CMD(nand, CONFIG_SYS_MAXARGS, 1, do_nand, "nand write - addr off|partition size\n" " read/write 'size' bytes starting at offset 'off'\n" " to/from memory address 'addr', skipping bad blocks.\n"
" .oob - reads/writes oob only.\n"" .raw - reads/writes both main and oob with no error\n"" detection or correction\n"Writing the oob area directly is UNSAFE. Having done to myself, it was a pain to undo. You should put at least document that.
Tom