As a follow up to this, we've noticed that we're seeing some kind of asynchronous issue when doing mass deletions. This is exposed by the Bonnie++ filesystem tester:

root@gcms:~# bonnie\+\+ -d /data -s 500 -u root -r 250

Using uid:0, gid:0.

Writing with putc()...done

Writing intelligently...done

Rewriting...done

Reading with getc()...done

Reading intelligently...done

start 'em...done...done...done...

Create files in sequential order...done.

Stat files in sequential order...done.

Delete files in sequential order...Bonnie: drastic I/O error (rmdir): Directory not empty

Cleaning up test directory after error.

root@gcms:~# ls /data/

Bonnie.19399/ lost+found/

root@gcms:~# rm -rf /data/Bonnie.19399/

rm: can't remove '/data/Bonnie.19399': Directory not empty

root@gcms:~# rm -rf /data/Bonnie.19399/

rm: can't remove '/data/Bonnie.19399': Directory not empty

root@gcms:~# rm -rf /data/Bonnie.19399/

root@gcms:~#

I don't think this is a YAFFS issue so much as it is an interaction with the VFS issue.

Any comments, or ideas on how best to debug this?

Thanks,

Andre

On Tue, 25 Oct 2016 at 14:20 Andre Renaud <andre@ignavus.net> wrote:

Hi,
We're porting one of our systems to Linux 4.8, and we've noticed that Yaffs doesn't seem to want to compile properly on it - there have been some changes around xattr, get_link, and a few other bits & pieces.

Charles - can you have a look at https://github.com/AndreRenaud/yaffs, and provide any feedback on whether this looks appropriate? We've only just done preliminary testing at this stage.

Thanks,
Andre