Improve checkpoint_gc jank and logging
This switches over to setting gc_urgent to 2 for a lower gc setting
which will respect existing IO for lower performance impact. We skip
running if we're not using f2fs userdata. Added some more descriptive
comments and logs to make debugging any future issue easier.
2 > gc_urgent triggers GC_URGENT_LOW, which does not make use of
gc_urgent_sleep_time, so we drop that. We leave gc running at
GC_URGENT_LOW in case of a long delay between ota postinstall and
reboot. If something else disables it then, we'll rely on normal
background gc.
Test: Cause disk fragmentation, run checkpoint_gc.sh
Bug: 173524015
Change-Id: I948070fc169c0a085e4a75cb7adb8138656e99d5
1 file changed