linux-mtd.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
* ubihealthd status
@ 2020-11-02 19:20 Maurice Smulders
  2020-11-02 22:06 ` Richard Weinberger
  0 siblings, 1 reply; 2+ messages in thread
From: Maurice Smulders @ 2020-11-02 19:20 UTC (permalink / raw)
  To: linux-mtd

I am trying to determine how to be able to figure out if a PEB had an
uncorrectable bitflip and expose it to user space.
It would be good to determine if a specific volume is suspect
determined from the actual LEB, and have some exposure in for example
SYSFS. Would the actual volume be turned read-only?
I'm open to suggestions, either I build something into ubihealthd to
notify user programs to tell something bad has happened, or sysfs (but
that cannot notify)

The problem I have is that one of the apps i need to support this for
is commercial - and I couldn't change ubihealthd to a libary because
it is GPLv2, the other app is now Apache License based I think, it
used to be under NOSA.  Also UBIHEALTHD doesn't necessarily have
access to which LEB/Volume is potentially affected.


-Maurice

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: ubihealthd status
  2020-11-02 19:20 ubihealthd status Maurice Smulders
@ 2020-11-02 22:06 ` Richard Weinberger
  0 siblings, 0 replies; 2+ messages in thread
From: Richard Weinberger @ 2020-11-02 22:06 UTC (permalink / raw)
  To: Maurice Smulders; +Cc: linux-mtd

Maurice,

On Mon, Nov 2, 2020 at 8:29 PM Maurice Smulders
<maurice.smulders@genevatech.net> wrote:
>
> I am trying to determine how to be able to figure out if a PEB had an
> uncorrectable bitflip and expose it to user space.
> It would be good to determine if a specific volume is suspect
> determined from the actual LEB, and have some exposure in for example
> SYSFS. Would the actual volume be turned read-only?

It is a little more complicated.
There are many situations where a block is allowed to have uncorrectable
bit flips. e.g. the last data written by UBIFS after a power-cut. It
can recover from that.

On the other hand if all of a sudden committed data is no longer readable due to
ECC errors you are in deep trouble. UBI can fail to attach then, UBIFS
can refuse to
mount or fail much later, etc...

What situation are you trying to deal with?

> I'm open to suggestions, either I build something into ubihealthd to
> notify user programs to tell something bad has happened, or sysfs (but
> that cannot notify)
>
> The problem I have is that one of the apps i need to support this for
> is commercial - and I couldn't change ubihealthd to a libary because
> it is GPLv2, the other app is now Apache License based I think, it
> used to be under NOSA.  Also UBIHEALTHD doesn't necessarily have
> access to which LEB/Volume is potentially affected.

ubihealthd is a rather simple program. See source and use the kernel interface
directly in your application.

-- 
Thanks,
//richard

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2020-11-02 22:07 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-11-02 19:20 ubihealthd status Maurice Smulders
2020-11-02 22:06 ` Richard Weinberger

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).