F2fs reports a lot of issues in 4.4.70

For Developers
1

hi,

with the latest 17.01 commits which have upgraded the kernel to 4.4.69 and 4.4.70, the f2fs which is used for the squashfs builds as the overlay, I am getting a large number of errors such as:

[   46.773725] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:718 update_sit_entry+0x1c0/0x2b0()
[   47.059153] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   47.346193] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   47.637566] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   47.930638] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   48.219677] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:718 update_sit_entry+0x1c0/0x2b0()
[   48.475175] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   48.728958] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   48.998406] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   49.252144] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   49.505672] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   49.759231] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   50.013216] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   50.267375] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   50.520926] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   50.774467] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   51.033686] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   51.288065] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   51.542162] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   51.796311] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   52.050536] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   52.305259] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   52.559418] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   52.813531] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   53.067641] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   53.321276] WARNING: CPU: 0 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   53.576488] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   53.830074] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   54.083668] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   54.337270] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   54.590811] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   54.844448] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[   55.098011] WARNING: CPU: 1 PID: 435 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[  116.278549] WARNING: CPU: 0 PID: 6 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()
[  116.582175] WARNING: CPU: 0 PID: 6 at fs/f2fs/segment.c:713 update_sit_entry+0x14c/0x2b0()

These are the 2 tests in segment.c lines 713 and 718:

if (f2fs_test_and_set_bit(offset, se->cur_valid_map))
f2fs_bug_on(sbi, 1);

This is in UBNT ERLite using the Octeon build of Lede (which uses a USB flash drive as the device for the f2fs file system).

I have tried to wipe out the overlay with first boot or sysupgrade with -n but I am getting these issues even after this.

On older builds, I do not believe there have been this many issues reported, but there have always been some immediately on the first boot after sysupgrade -n. But I cannot say this 100% sure, but on some long running devices, I do not see these in the dmesg, while on some new devices under test, there are quite a few cases (but other things).

Has anyone seen anything similar on any other device?

NB: I am using an Octeon kernel with hires timers enabled (see another post on that), I have not tried to run without the hires timers, as I am using a packet scheduler which depends on the hires timers.

Thanks.

Petr

2

Have you tried current git ??

3

I have tried current Git on the lede-17.01 branch (kernel 4.4.70). That is what seemed to report the errors.

I have not tried the current Lede master branch yet, as I will need to rewrite my kernel packet scheduler to the newer kernel API first.

4

i would try current git without this driver.
v.4.9.x kernel has more fs2fs updates as v4.4.x

Did you try your USB flash drive on other systems i.e. with your PC ?

5

Trunk/Master is fine
https://bugs.lede-project.org/index.php?do=details&task_id=749&order=id&sort=desc