UBIFS orphans and ro-mounts

UBIFS orphans and ro-mounts

[Richard Weinberger]

Artem,

I wonder why UBIFS processes orphan inodes also when mounting read-only.
Depending on the workload before a power cut this can take a few seconds
and since we mount read-only the updated TNC will not written back to flash,
so upon next mount the time is again wasted.

This hurts boot performance a lot on systems where u-boot loads the kernel
from UBIFS.
u-boot always mounts UBIFS read-only.
So even when Linux mounts UBIFS rw and a powercut happens orphan processing
will happen twice.

I'd suggest calling ubifs_mount_orphans() only when we mount rw or remount to
rw.
What do you think?

Thanks,
//richard
-- 
sigma star gmbh - Eduard-Bodem-Gasse 6 - 6020 Innsbruck - Austria
ATU66964118 - FN 374287y

Re: UBIFS orphans and ro-mounts

[Artem Bityutskiy]

On Wed, 2016-07-13 at 14:29 +0200, Richard Weinberger wrote:
> Artem,
> 
> I wonder why UBIFS processes orphan inodes also when mounting read-
> only.

Well, people expect UBIFS to report correct free/used space even if it
is R/O, right? :-)

> Depending on the workload before a power cut this can take a few
> seconds
> and since we mount read-only the updated TNC will not written back to
> flash,
> so upon next mount the time is again wasted.

I see the problem, but I do not agree with word "wasted", because this
is like doing it for nothing, while UBIFS is doing this for a reason.
So I'd rather used word "spent". :-)

> This hurts boot performance a lot on systems where u-boot loads the
> kernel
> from UBIFS.
> u-boot always mounts UBIFS read-only.
> So even when Linux mounts UBIFS rw and a powercut happens orphan
> processing
> will happen twice.

Sure, because in the R/O case the results of that processing cannot be
"saved" on the media, so they have to be repeated.

> I'd suggest calling ubifs_mount_orphans() only when we mount rw or
> remount to
> rw.
> What do you think?

I understand the problem. I do not think this is the right solution
because it'll introduce a regression (incorrect used/free space). But a
special mount option and/or superblock flag may be considered, if this
helps substantially.

Re: UBIFS orphans and ro-mounts

[Richard Weinberger]

Artem,

Am 13.07.2016 um 14:48 schrieb Artem Bityutskiy:
> On Wed, 2016-07-13 at 14:29 +0200, Richard Weinberger wrote:
>> Artem,
>>
>> I wonder why UBIFS processes orphan inodes also when mounting read-
>> only.
> 
> Well, people expect UBIFS to report correct free/used space even if it
> is R/O, right? :-)

Well, I'd say people will understand the on ro-mode free is less than
expected since UBIFS had no chance to garbage collect...

But I agree that changing the current behavior will confuse users. :-(

>> Depending on the workload before a power cut this can take a few
>> seconds
>> and since we mount read-only the updated TNC will not written back to
>> flash,
>> so upon next mount the time is again wasted.
> 
> I see the problem, but I do not agree with word "wasted", because this
> is like doing it for nothing, while UBIFS is doing this for a reason.
> So I'd rather used word "spent". :-)

;-)

>> This hurts boot performance a lot on systems where u-boot loads the
>> kernel
>> from UBIFS.
>> u-boot always mounts UBIFS read-only.
>> So even when Linux mounts UBIFS rw and a powercut happens orphan
>> processing
>> will happen twice.
> 
> Sure, because in the R/O case the results of that processing cannot be
> "saved" on the media, so they have to be repeated.
> 
>> I'd suggest calling ubifs_mount_orphans() only when we mount rw or
>> remount to
>> rw.
>> What do you think?
> 
> I understand the problem. I do not think this is the right solution
> because it'll introduce a regression (incorrect used/free space). But a
> special mount option and/or superblock flag may be considered, if this
> helps substantially.

A new mount option is an option, true.
That said, the real solution is that u-boot (and other bootloaders)
should not mount UBIFS and use static volumes.
But currently the opposite is happening on most systems I get my hands on. :(

Thanks,
//richard
-- 
sigma star gmbh - Eduard-Bodem-Gasse 6 - 6020 Innsbruck - Austria
ATU66964118 - FN 374287y

Re: UBIFS orphans and ro-mounts

[Artem Bityutskiy]

On Wed, 2016-07-13 at 14:55 +0200, Richard Weinberger wrote:
> A new mount option is an option, true.
> That said, the real solution is that u-boot (and other bootloaders)
> should not mount UBIFS and use static volumes.
> But currently the opposite is happening on most systems I get my
> hands on. :(

May be orphans could be handled in background?