Message ID | 1389642280-11714-1-git-send-email-computersforpeace@gmail.com |
---|---|
State | Accepted |
Commit | b72f3dfb8ccf7e39b9434f548c1c98aa45c11426 |
Headers | show |
On Mon, Jan 13, 2014 at 11:44:36AM -0800, Brian Norris wrote: > ECC failures can be tracked at the page level, not the do_read_ops level > (i.e., a potentially multi-page transaction). > > This helps prepare for READ RETRY support. > > Signed-off-by: Brian Norris <computersforpeace@gmail.com> > --- > v3 -> v4: adjust whitespace > > (Earlier versions: no changes) > > drivers/mtd/nand/nand_base.c | 13 ++++++++----- > 1 file changed, 8 insertions(+), 5 deletions(-) > > diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c > index 404e83d7e3da..b5c3768b32ae 100644 > --- a/drivers/mtd/nand/nand_base.c > +++ b/drivers/mtd/nand/nand_base.c > @@ -1422,7 +1422,6 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, > { > int chipnr, page, realpage, col, bytes, aligned, oob_required; > struct nand_chip *chip = mtd->priv; > - struct mtd_ecc_stats stats; > int ret = 0; > uint32_t readlen = ops->len; > uint32_t oobreadlen = ops->ooblen; > @@ -1431,8 +1430,7 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, > > uint8_t *bufpoi, *oob, *buf; > unsigned int max_bitflips = 0; > - > - stats = mtd->ecc_stats; > + bool ecc_fail = false; > > chipnr = (int)(from >> chip->chip_shift); > chip->select_chip(mtd, chipnr); > @@ -1447,6 +1445,8 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, > oob_required = oob ? 1 : 0; > > while (1) { > + unsigned int ecc_failures = mtd->ecc_stats.failed; > + > bytes = min(mtd->writesize - col, readlen); > aligned = (bytes == mtd->writesize); > > @@ -1483,7 +1483,7 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, > /* Transfer not aligned data */ > if (!aligned) { > if (!NAND_HAS_SUBPAGE_READ(chip) && !oob && > - !(mtd->ecc_stats.failed - stats.failed) && > + !(mtd->ecc_stats.failed - ecc_failures) && > (ops->mode != MTD_OPS_RAW)) { > chip->pagebuf = realpage; > chip->pagebuf_bitflips = ret; > @@ -1513,6 +1513,9 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, > else > nand_wait_ready(mtd); > } > + > + if (mtd->ecc_stats.failed - ecc_failures) > + ecc_fail = true; > } else { > memcpy(buf, chip->buffers->databuf + col, bytes); > buf += bytes; > @@ -1547,7 +1550,7 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, > if (ret < 0) > return ret; > > - if (mtd->ecc_stats.failed - stats.failed) > + if (ecc_fail) > return -EBADMSG; > > return max_bitflips; > -- > 1.8.3.2 > > > thanks Please add my acks to all the patch set: Acked-by: Huang Shijie <b32955@freescale.com>
On Tue, Jan 14, 2014 at 10:34:40AM +0800, Huang Shijie wrote: > On Mon, Jan 13, 2014 at 11:44:36AM -0800, Brian Norris wrote: > > ECC failures can be tracked at the page level, not the do_read_ops level > > (i.e., a potentially multi-page transaction). > > > > This helps prepare for READ RETRY support. > > > > Signed-off-by: Brian Norris <computersforpeace@gmail.com> > > --- [...] > Please add my acks to all the patch set: > Acked-by: Huang Shijie <b32955@freescale.com> Thanks! Done, and pushed to l2-mtd.git. Brian
>From: Brian Norris [mailto:computersforpeace@gmail.com] >>On Tue, Jan 14, 2014 at 10:34:40AM +0800, Huang Shijie wrote: >> On Mon, Jan 13, 2014 at 11:44:36AM -0800, Brian Norris wrote: >> > ECC failures can be tracked at the page level, not the do_read_ops level >> > (i.e., a potentially multi-page transaction). >> > >> > This helps prepare for READ RETRY support. >> > >> > Signed-off-by: Brian Norris <computersforpeace@gmail.com> >> > --- > >[...] > >> Please add my acks to all the patch set: >> Acked-by: Huang Shijie <b32955@freescale.com> > >Thanks! Done, and pushed to l2-mtd.git. > Thanks Brian and Shijie for getting this READ_RETRY feature in kernel. Sorry, I din't had any MLC NAND locally at my place, so couldn't help in Testing these patches, but I know these would be helpful later. As we observed quite frequent failures due to read-disturb bit-flips on MLC NANDs, and I hope these patches would reduce such failures. with regards, pekon
diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c index 404e83d7e3da..b5c3768b32ae 100644 --- a/drivers/mtd/nand/nand_base.c +++ b/drivers/mtd/nand/nand_base.c @@ -1422,7 +1422,6 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, { int chipnr, page, realpage, col, bytes, aligned, oob_required; struct nand_chip *chip = mtd->priv; - struct mtd_ecc_stats stats; int ret = 0; uint32_t readlen = ops->len; uint32_t oobreadlen = ops->ooblen; @@ -1431,8 +1430,7 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, uint8_t *bufpoi, *oob, *buf; unsigned int max_bitflips = 0; - - stats = mtd->ecc_stats; + bool ecc_fail = false; chipnr = (int)(from >> chip->chip_shift); chip->select_chip(mtd, chipnr); @@ -1447,6 +1445,8 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, oob_required = oob ? 1 : 0; while (1) { + unsigned int ecc_failures = mtd->ecc_stats.failed; + bytes = min(mtd->writesize - col, readlen); aligned = (bytes == mtd->writesize); @@ -1483,7 +1483,7 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, /* Transfer not aligned data */ if (!aligned) { if (!NAND_HAS_SUBPAGE_READ(chip) && !oob && - !(mtd->ecc_stats.failed - stats.failed) && + !(mtd->ecc_stats.failed - ecc_failures) && (ops->mode != MTD_OPS_RAW)) { chip->pagebuf = realpage; chip->pagebuf_bitflips = ret; @@ -1513,6 +1513,9 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, else nand_wait_ready(mtd); } + + if (mtd->ecc_stats.failed - ecc_failures) + ecc_fail = true; } else { memcpy(buf, chip->buffers->databuf + col, bytes); buf += bytes; @@ -1547,7 +1550,7 @@ static int nand_do_read_ops(struct mtd_info *mtd, loff_t from, if (ret < 0) return ret; - if (mtd->ecc_stats.failed - stats.failed) + if (ecc_fail) return -EBADMSG; return max_bitflips;
ECC failures can be tracked at the page level, not the do_read_ops level (i.e., a potentially multi-page transaction). This helps prepare for READ RETRY support. Signed-off-by: Brian Norris <computersforpeace@gmail.com> --- v3 -> v4: adjust whitespace (Earlier versions: no changes) drivers/mtd/nand/nand_base.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)