[U-Boot] what to do when the uboot can not mark the phisical bad block of nand flash?

[Marek Vasut]

> OK，guys,I can discribe this question in the other way.
> When I download the program image to the nand flash in the uboot.
> At firist, I need to erase the nand flash by running the command of "nand
> erase" or "nand scrub" in the cmdline. But sometimes this step will be
> failed because one block of this flash is bad, and this command will
> return "Failed erase".

Scrub wont, but it's stupid to use. Nand erase should mark the block bad into 
BBT. Anyway, what version of uboot do you use?
> 
> Guys what would you do when you got this problem? Do I need to mrak this
> block as bad block and add it to the bad block table with  software? how
> to do this ? Thank you !
> 
> At 2011-11-25 16:51:17,"Marek Vasut" <marek.vasut at gmail.com> wrote:
> >> Hi, guys                I got a problem. I move a part of source code of
> >> linux/driver/mtd/nand to the u-boot. But I found that the u-boot can not
> >> mark the phsical bad block . when I run nand_erase() or nand_scrub(), it
> >> return error num as below: linux/driver/mtd/nand/nand_base.c
> >> 
> >>     2589            chip->erase_cmd(mtd, page & chip->pagemask);    
> >>     2590
> >>     
> >>               status = chip->waitfunc(mtd, chip); 2592                
> >>               /*
> >>     
> >>     2593                 * See if operation failed and additional status
> >> 
> >> checks are 2594                 * available
> >> 
> >>     2595                 */
> >>     2596                 if ((status & NAND_STATUS_FAIL) &&
> >> 
> >> (chip->errstat)) 2597                         status =
> >> chip->errstat(mtd, chip, FL_ERASING, 2598
> >> status, page); 2600                 /* See if block erase succeeded */
> >> 
> >>     2601                 if (status & NAND_STATUS_FAIL) {
> >>     2602                         DEBUG(MTD_DEBUG_LEVEL0, "%s: Failed
> >>     erase,
> >> 
> >> " 2603                                         "page 0x%08x\n",
> >> __func__, page); 2604                         instr->state =
> >> MTD_ERASE_FAILED;
> >> 
> >>     2605                         instr->fail_addr =
> >>     2606                                 ((loff_t)page <<
> >> 
> >> chip->page_shift); 2607                         goto erase_exit;
> >> 
> >>     2608                 }
> >>     
> >>         you can find that if block erase is failed(status==-1), it only
> >> 
> >> mark the instr(we will not use the instr later), and go to the
> >> erase_exit. So there is no code to mark the oob of the phisical bad
> >> block, neither add it to the bad block table. so ,I add the code of
> >> writing oob of the phisical bad block when it is erased failed, such as
> >> below:
> >> chip->erase_cmd(mtd, page & chip->pagemask);2589
> >> 
> >> - Ignored:
> >>     2590                 status = chip->waitfunc(mtd, chip);
> >>     2592                 /*
> >>     2593                 * See if operation failed and additional status
> >> 
> >> checks are 2594                 * available
> >> 
> >>     2595                 */
> >>     2596                 if ((status & NAND_STATUS_FAIL) &&
> >> 
> >> (chip->errstat)) 2597                         status =
> >> chip->errstat(mtd, chip, FL_ERASING, 2598
> >> status, page); 2600                 /* See if block erase succeeded */
> >> 
> >>     2601                 if (status & NAND_STATUS_FAIL) {
> >>     2602                         DEBUG(MTD_DEBUG_LEVEL0, "%s: Failed
> >>     erase,
> >> 
> >> " 2603                                         "page 0x%08x\n",
> >> __func__, page); 2604                         instr->state =
> >> MTD_ERASE_FAILED;
> >> 
> >>     2605                         instr->fail_addr =
> >>     2606                                 ((loff_t)page <<
> >> 
> >> chip->page_shift); nand_write_oob(addr);
> >> 
> >>     2607                         goto erase_exit;
> >>     2608                 }
> >>     
> >>        Guys, what do you think of it , and what would you to if you  got
> >> 
> >> this problem.
> >
> >Can you please fix your mailer and send this stuff again? It's really hard
> >to decode what you're asking please.
> >
> >M