Mark,
Am Donnerstag, 12. Juli 2018, 07:22:13 CEST schrieb Heiko Schocher:
Hello Mark,
added Richard Weinberger to cc...
Am 12.07.2018 um 02:28 schrieb Mark Spieth:
Hi
In the process of investigating a boot failure on one of our devices, the
UBI: fixable bit-flip detected at PEB
message was seen with the following behaviour during kernel load in u-boot.
Read [2285568] bytes UBI: fixable bit-flip detected at PEB 415 UBI: schedule PEB 415 for scrubbing UBI: fixable bit-flip detected at PEB 415 UBI: fixable bit-flip detected at PEB 419 UBI: schedule PEB 419 for scrubbing UBI: fixable bit-flip detected at PEB 419 UBI: fixable bit-flip detected at PEB 420 UBI: schedule PEB 420 for scrubbing UBI: fixable bit-flip detected at PEB 420 UBI: fixable bit-flip detected at PEB 419 UBI: fixable bit-flip detected at PEB 420 UBI: fixable bit-flip detected at PEB 419 UBI: fixable bit-flip detected at PEB 420 UBI: fixable bit-flip detected at PEB 419 UBI: fixable bit-flip detected at PEB 420 UBI: fixable bit-flip detected at PEB 419 UBI: fixable bit-flip detected at PEB 420 UBI: fixable bit-flip detected at PEB 419 UBI: fixable bit-flip detected at PEB 420 UBI: fixable bit-flip detected at PEB 419
This repeats until reset.
Do you see the same symptom also on Linux? We need to be very sure that it is actually a UBI problem.
This fix is not a root cause fix though. Investigating further led to the following root cause solution. The following is AFAICT.
When the scrubber chooses a PEB to move the from the free balanced tree. This tree is sorted by EC (erase count) and then by PEB number.
The find_wl_entry call uses a max parameter of WL_FREE_MAX_DIFF which is 8192 in this config. So the find_wl_entry function will find a PEB that is better in error count that the current PEB EC. This
error count? You mean erase count?
can easily cause it to find the PEB that was just moved from if it is the lowest numbered PEB in the free tree. Waiting for EC to go above 8192 would take a long time and cause premature aging of the flash PEBs in question.
The easy solution is to change the max parameter to this call to 0 so it finds a PEB with a smaller EC than the one being replaced. This means it wont use the previously discarded PEB as its first choice.
For scrubbing this might be a good idea, but not for regular wear-leveling.
See comment in UBI: /* * When a physical eraseblock is moved, the WL sub-system has to pick the target * physical eraseblock to move to. The simplest way would be just to pick the * one with the highest erase counter. But in certain workloads this could lead * to an unlimited wear of one or few physical eraseblock. Indeed, imagine a * situation when the picked physical eraseblock is constantly erased after the * data is written to it. So, we have a constant which limits the highest erase * counter of the free physical eraseblock to pick. Namely, the WL sub-system * does not pick eraseblocks with erase counter greater than the lowest erase * counter plus %WL_FREE_MAX_DIFF. */ #define WL_FREE_MAX_DIFF (2*UBI_WL_THRESHOLD)
So we could change the logic such that for regular wear-leveling we keep using WL_FREE_MAX_DIFF, but for scrubbing (which is 1:1 wear-leveling but the source PEB is showing bit-flips) we use a lower value. IMHO WL_FREE_MAX_DIFF/2 would be a good choice. I'm not sure whether 0 is too extreme and might cause other distortions.
Mark, can you please file a patch and send it to linux-mtd mailing list? Such a change needs to go through Linux and then to u-boot. But first we need to think about and discuss it in detail.
I am not sure if it is so easy ...
This fix was implemented and fixable bit-flip errors no longer hang/freeze the boot process! UBI erase and reformat was used between re-tests to get consistent results.
Adding the above 75% correctable bitflip threshold is also a good thing as less movement will ensue when the FLASH is new, but as the flash ages, the root cause will once again be invoked causing un-recoverable boot failures.
Note this fault is also in the latest kernel drivers for UBI and may also exist in other wear leveling implementations. The kernel driver issue may be at fault for android devices locking up/freezing sporadically during FLASH read when scrubbing due to a relatively full flash and correctable errors causing ping pong PEB moves.
The question is, is my root cause solution sound or have I missed something?
I have to think about, before I write nonsene, but may Richard has here a deeper insight.
Please see my comments. :)
Thanks, //richard