On Wed, 2018-06-20 at 11:14 +0200, Boris Brezillon wrote:

CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

On Wed, 6 Jun 2018 12:13:29 +0200 Joakim Tjernlund joakim.tjernlund@infinera.com wrote:

...

cfi_ppb_unlock() walks all flash chips when unlocking sectors. testing lock status on each chip which causes relocking of already locked sectors. Test against offset to aviod this aliasing.

                                     ^ avoid

As I said before, I think the current code is doing worse than just relocking already locked sectors. As soon as you cross a chip boundary, addr is set back to 0, and the (addr < offs || adr >= (ofs + len)) might be true while it shouldn't be (absolute offset still in the unlock range), which means you'll lock sectors that the caller expect to be unlocked.

I don't see how, the code asks for its current lock status and will reapply those that are locked again.

...

Fixes: 1648eaaa1575 ("mtd: cfi_cmdset_0002: Support Persistent Protection Bits (PPB) locking") Cc: stable@vger.kernel.org Signed-off-by: Joakim Tjernlund joakim.tjernlund@infinera.com

---

v2 - Spilt into several patches

drivers/mtd/chips/cfi_cmdset_0002.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c b/drivers/mtd/chips/cfi_cmdset_0002.c index cb85cccc48c1..b6273ce83de7 100644 --- a/drivers/mtd/chips/cfi_cmdset_0002.c +++ b/drivers/mtd/chips/cfi_cmdset_0002.c @@ -2670,7 +2670,7 @@ static int __maybe_unused cfi_ppb_unlock(struct mtd_info *mtd, loff_t ofs, * sectors shall be unlocked, so lets keep their locking * status at "unlocked" (locked=0) for the final re-locking. */

•         if ((adr < ofs) || (adr >= (ofs + len))) {
  

•         if ((offset < ofs) || (offset >= (ofs + len))) {
                  sect[sectors].chip = &cfi->chips[chipnum];
                  sect[sectors].adr = adr;
                  sect[sectors].locked = do_ppb_xxlock(
  

On Wed, 2018-06-20 at 11:25 +0200, Boris Brezillon wrote:

On Wed, 6 Jun 2018 12:13:30 +0200 Joakim Tjernlund joakim.tjernlund@infinera.com wrote:

...

cfi_ppb_unlock() walks all flash chips when unlocking sectors, avoid walking chips unaffected by the unlock operation.

Fixes: 1648eaaa1575 ("mtd: cfi_cmdset_0002: Support Persistent Protection Bits (PPB) locking") Cc: stable@vger.kernel.org

That's clearly not a fix, just an optimization. You should drop the Fixes and Cc-stable tags.

It sure IS! The code never intended to do this and it is just bad luck that nothing bad happened and I sure don't want to walk all 4 chips we have, stealing CPU and keeping the flash busy just because I am using stable.

Given I have moved on now and we disagree, I will not reword and resubmit any time soon. Feel free to do needed edits though.

Jocke

...

Signed-off-by: Joakim Tjernlund joakim.tjernlund@infinera.com

v2 - Spilt into several patches

drivers/mtd/chips/cfi_cmdset_0002.c | 2 ++ 1 file changed, 2 insertions(+)

diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c b/drivers/mtd/chips/cfi_cmdset_0002.c index b6273ce83de7..62cd4ee280b3 100644 --- a/drivers/mtd/chips/cfi_cmdset_0002.c +++ b/drivers/mtd/chips/cfi_cmdset_0002.c @@ -2686,6 +2686,8 @@ static int __maybe_unused cfi_ppb_unlock(struct mtd_info *mtd, loff_t ofs, i++;

          if (adr >> cfi->chipshift) {

•                 if (offset >= (ofs + len))
  

•                         break;
                  adr = 0;
                  chipnum++;
  
  

On Wed, 2018-06-20 at 11:03 +0200, Boris Brezillon wrote:

CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

On Wed, 6 Jun 2018 12:13:27 +0200 Joakim Tjernlund joakim.tjernlund@infinera.com wrote:

...

do_ppb_xxlock() fails to add chip->start when quering for lock status(and chip_ready test),

^ querying?

...

which caused false status reports.

The 3 above lines are wrapped at less than 50 chars, is this normal?

I actually hit return every now and then when I type, sometimes the lines might become a bit short, but yes, this is normal for me.

...

Fix by adding adr += chip->start and adjust call sites accordingly.

Fixes: 1648eaaa1575 ("mtd: cfi_cmdset_0002: Support Persistent Protection Bits (PPB) locking") Cc: stable@vger.kernel.org Signed-off-by: Joakim Tjernlund joakim.tjernlund@infinera.com

---

v2 - Spilt into several patches

drivers/mtd/chips/cfi_cmdset_0002.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c b/drivers/mtd/chips/cfi_cmdset_0002.c index 53a976a8e614..8648b1adccd5 100644 --- a/drivers/mtd/chips/cfi_cmdset_0002.c +++ b/drivers/mtd/chips/cfi_cmdset_0002.c @@ -2554,8 +2554,9 @@ static int __maybe_unused do_ppb_xxlock(struct map_info *map, unsigned long timeo; int ret;

• adr += chip->start;
  mutex_lock(&chip->mutex);
  

• ret = get_chip(map, chip, adr + chip->start, FL_LOCKING);
  

• ret = get_chip(map, chip, adr, FL_LOCKING);
  if (ret) {
          mutex_unlock(&chip->mutex);
          return ret;
  

@@ -2573,8 +2574,8 @@ static int __maybe_unused do_ppb_xxlock(struct map_info *map,

  if (thunk == DO_XXLOCK_ONEBLOCK_LOCK) {
          chip->state = FL_LOCKING;

•         map_write(map, CMD(0xA0), chip->start + adr);
  

•         map_write(map, CMD(0x00), chip->start + adr);
  

•         map_write(map, CMD(0xA0), adr);
  

•         map_write(map, CMD(0x00), adr);
  } else if (thunk == DO_XXLOCK_ONEBLOCK_UNLOCK) {
          /*
           * Unlocking of one specific sector is not supported, so we
  

@@ -2612,7 +2613,7 @@ static int __maybe_unused do_ppb_xxlock(struct map_info *map, map_write(map, CMD(0x00), chip->start);

  chip->state = FL_READY;

• put_chip(map, chip, adr + chip->start);
  

• put_chip(map, chip, adr);
  mutex_unlock(&chip->mutex);
  
  return ret;