Critical error from Tree-checker

[Archange <archange@archlinux.org>]

Hi there,

Since today, my system started randomly becoming read-only. At that 
point I can still run dmesg in an open terminal, so I’ve seen it was 
related to a btrfs error, but did not try anything since I could not 
open a web browser anymore. But I’ve seen the error to be “BTRFS 
critical” and related to a “corrupt leaf”.

I’ve tried to run `btrfs scrub` on the device after rebooting, and in 
fact it aborted almost right away triggering the same error in dmesg 
(but not turning the system read-only, so I can copy paste it here):

[  365.268769] BTRFS info (device dm-0): scrub: started on devid 1
[  385.788000] page: refcount:3 mapcount:0 mapping:00000000d0054cae 
index:0x9678888 pfn:0x11ce15
[  385.788015] memcg:ffff9fc94db8f000
[  385.788021] aops:btree_aops [btrfs] ino:1
[  385.788235] flags: 
0x2ffffa000004020(lru|private|node=0|zone=2|lastcpupid=0x1ffff)
[  385.788248] raw: 02ffffa000004020 ffffea9a8574ff88 ffffea9a847385c8 
ffff9fc95b8365b0
[  385.788255] raw: 0000000009678888 ffff9fc9ae554000 00000003ffffffff 
ffff9fc94db8f000
[  385.788259] page dumped because: eb page dump
[  385.788264] BTRFS critical (device dm-0): corrupt leaf: 
block=646267305984 slot=92 extent bytenr=1182031872 len=106496 invalid 
data ref objectid value 257
[  385.788283] BTRFS error (device dm-0): read time tree block 
corruption detected on logical 646267305984 mirror 1
[  385.796803] BTRFS info (device dm-0): scrub: not finished on devid 1 
with status: -5

According to https://btrfs.readthedocs.io/en/latest/Tree-checker.html 
this is not really expected, and the last paragraph says to report 
troubles here. So here I am, in the search for advice about this error 
(web searches returned nothing with this specific error except the 
commit/ml messages that added the code for it) and how to fix my random 
lockups.

Regards.

P.S.: I’m not subscribed to the list, so please keep me in copy when 
answering.