UBIFS: Does init_constants_early ignore LEBs being erased? Complains about amount.

[Artem Bityutskiy]

On Fri, 2014-10-17 at 08:29 +0200, Rafał Miłecki wrote:
> UBI: available PEBs: 0, total reserved PEBs: 986, PEBs reserved for
> bad PEB handling: 20
> UBI: background thread "ubi_bgt0d" started, PID 257
> UBI: ubiblock0_0 created from ubi0:0(rootfs)
> UBIFS error (pid 1): init_constants_early: too few LEBs (15), min. is 17

UBIFS wants more LEBs in the volume.

> *** 5 seconds later ***
> UBIFS: default file-system created
> UBIFS: background thread "ubifs_bgt0_1" started, PID 302
> UBIFS: mounted UBI device 0, volume 1, name "rootfs_data"
> UBIFS: LEB size: 126976 bytes (124 KiB), min./max. I/O unit sizes:
> 2048 bytes/2048 bytes
> UBIFS: FS size: 118849536 bytes (113 MiB, 936 LEBs), journal size
> 5967872 bytes (5 MiB, 47 LEBs)

This is a different volume mounted, it seems, it has 47 LEBs.

> UBIFS: reserved for root: 4952683 bytes (4836 KiB)
> UBIFS: media format: w4/r0 (latest is w4/r0), UUID
> A3D6B177-D8CC-43AA-A15F-5A0480428338, small LPT model
> 
> My guess is there is nothing wrong happening above.

No, UBIFS refused to mount one of the UBI volumes. The
'init_constants_early()' function is called from the mount path.

>  I guess UBIFS
> aborted because there weren't enough LEBs and later (after UBI
> prepared empty LEBs) UBIFS was triggered again.
> 
> However this "UBIFS error" looks like a serious error, a bit scary ;)
> Would it be possible to detect that there are some LEBs being prepared
> and if so, change the above message to something like
> "UBIFS: too few LEBs (15), min. is 17, waiting for UBI to prepare more"

I do not think there is anything being prepared. The messages are for
different volumes. Do you have more than 1?

Tanya's patch should improve the messages - they will print the volume
ID they belong to. Now you see messages, but they may be for different
volumes.