[PATCH 2/2] ubi: Dont call ubi_self_check_all_ff() in __wl_get_peb()

[Richard Weinberger]

Am Tue, 04 Dec 2012 10:15:01 +0200
schrieb Artem Bityutskiy <dedekind1 at gmail.com>:

> On Tue, 2012-12-04 at 09:56 +0200, Artem Bityutskiy wrote:
> > On Mon, 2012-12-03 at 20:57 +0100, Richard Weinberger wrote:
> > > As ubi_self_check_all_ff() might sleep we are not allowed
> > > to call it from atomic context.
> > > For now we call it only from ubi_wl_get_peb().
> > > There are some code paths where it would also make sense,
> > > but these paths are currently atomic and only enabled
> > > when fastmap is used.
> > > 
> > > Signed-off-by: Richard Weinberger <richard at nod.at>
> > 
> > Can you confirm that you enabled the 3 debugging options I sent and
> > tested both FM and non-FM configurations with those options enabled?
> 
> Let me elaborate why I am asking.
> 
> 1. If you did the tests and with this fix the debugging stuff works
> fine, then it is OK to take this patch now and let you improve the FM
> debugging coverage a bit later.

Of course I did the tests.
Namely chk_gen, chk_io, tst_emulate_bit_flips and tst_disable_bgt.
(Also in various combinations)

Please note, there is one issue regarding tst_disable_bgt and fastmap.
Fastmap needs the UBI background thread. Therefore ubi-tests fail if
fastmap is enabled and tst_disable_bgt=1.
For 3.8 I can try to rework my code such that the test passes...
But IMHO the tst_disable_bgt test is useless.

I'll rework the fastmap code paths such that ubi_self_check_all_ff()
can be used everywhere. This is planned for 3.8.
As we are at 3.7-rc8 now I get very nervous if I have to touch
much UBI/fastmap code.

Again, with CONFIG_MTD_UBI_FASTMAP=n I'm no longer able to reproduce
any UBI regression.
If there is still a regression, please report it to me I'll fix it
immediately.

Thanks,
//richard