| Message ID | 1326507110-10893-1-git-send-email-computersforpeace@gmail.com |
|---|---|
| State | New, archived |
| Headers | show |
On Fri, 2012-01-13 at 18:11 -0800, Brian Norris wrote: > Many NAND flash systems (especially those with MLC NAND) cannot be > reliably written twice in a row. For instance, when marking a bad block, > the block may already have data written to it, and so we should attempt > to erase the block before writing a bad block marker to its OOB region. > > We can ignore erase failures, since the block may be bad such that it > cannot be erased properly; we still attempt to write zeros to its spare > area. > > Signed-off-by: Brian Norris <computersforpeace@gmail.com> All four look good, pushed to l2-mtd.git, thanks. What are your plans WRT marking blocks as bad both in the OOB and the table?
On Mon, Jan 16, 2012 at 4:07 AM, Artem Bityutskiy <dedekind1@gmail.com> wrote: > On Fri, 2012-01-13 at 18:11 -0800, Brian Norris wrote: >> Many NAND flash systems (especially those with MLC NAND) cannot be >> reliably written twice in a row. For instance, when marking a bad block, >> the block may already have data written to it, and so we should attempt >> to erase the block before writing a bad block marker to its OOB region. > > All four look good, pushed to l2-mtd.git, thanks. Sorry, but actually this "erase block before marking bad" patch is incorrect. I didn't test this particular revision, and so I forgot that the erase fails because the block is already marked as bad: nand_erase_nand: attempt to erase a bad block at page 0x... I'll send a fixup patch on top of l2-mtd-2.6. > What are your plans WRT marking blocks as bad both in the OOB and the > table? I don't have a 100% definite plan, but I've been trying to track the comments/suggestions in the other thread. I'll reply to that thread so we have the right reply-chain and CC list for discussion. Brian
diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c index 8a393f9..f0e768f 100644 --- a/drivers/mtd/nand/nand_base.c +++ b/drivers/mtd/nand/nand_base.c @@ -407,6 +407,14 @@ static int nand_default_block_markbad(struct mtd_info *mtd, loff_t ofs) ret = nand_update_bbt(mtd, ofs); else { struct mtd_oob_ops ops; + struct erase_info einfo; + + /* Attempt erase before marking OOB */ + memset(&einfo, 0, sizeof(einfo)); + einfo.mtd = mtd; + einfo.addr = ofs; + einfo.len = 1 << chip->phys_erase_shift; + nand_erase_nand(mtd, &einfo, 0); nand_get_device(chip, mtd, FL_WRITING);
Many NAND flash systems (especially those with MLC NAND) cannot be reliably written twice in a row. For instance, when marking a bad block, the block may already have data written to it, and so we should attempt to erase the block before writing a bad block marker to its OOB region. We can ignore erase failures, since the block may be bad such that it cannot be erased properly; we still attempt to write zeros to its spare area. Signed-off-by: Brian Norris <computersforpeace@gmail.com> --- drivers/mtd/nand/nand_base.c | 8 ++++++++ 1 files changed, 8 insertions(+), 0 deletions(-)