* (no subject)
@ 2016-02-26 1:19 Fredrick Prashanth John Berchmans
2016-02-26 7:37 ` Richard Weinberger
0 siblings, 1 reply; 2+ messages in thread
From: Fredrick Prashanth John Berchmans @ 2016-02-26 1:19 UTC (permalink / raw
To: dwmw2; +Cc: linux-mtd, Suresh Siddha
Hi,
We are using UBIFS on our NOR flash.
We are observing that a lot of times the filesystem goes to read-only
unable to recover.
Most of the time its due to
a) ubifs_scan_a_node failing due to bad crc or unclean free space.
b) ubifs_leb_write failing to erase due to erase timeout
[ The above would have happened due to unclean power cuts. In our
environment this happens often ]
I checked the code in jffs2. Looking at jffs2 code it looks like jffs2
tolerates the above two
failures and moves on without mounting read-only.
Is my understanding right ?
Could we change the ubifs_scan_a_node to skip corrupted bytes and move
to next node,
instead of returning error ?
Thanks,
Fredrick
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re:
2016-02-26 1:19 Fredrick Prashanth John Berchmans
@ 2016-02-26 7:37 ` Richard Weinberger
0 siblings, 0 replies; 2+ messages in thread
From: Richard Weinberger @ 2016-02-26 7:37 UTC (permalink / raw
To: Fredrick Prashanth John Berchmans
Cc: David Woodhouse, linux-mtd@lists.infradead.org, Suresh Siddha
On Fri, Feb 26, 2016 at 2:19 AM, Fredrick Prashanth John Berchmans
<fredrickprashanth@gmail.com> wrote:
> We are using UBIFS on our NOR flash.
> We are observing that a lot of times the filesystem goes to read-only
> unable to recover.
> Most of the time its due to
> a) ubifs_scan_a_node failing due to bad crc or unclean free space.
> b) ubifs_leb_write failing to erase due to erase timeout
>
> [ The above would have happened due to unclean power cuts. In our
> environment this happens often ]
>
> I checked the code in jffs2. Looking at jffs2 code it looks like jffs2
> tolerates the above two
> failures and moves on without mounting read-only.
> Is my understanding right ?
>
> Could we change the ubifs_scan_a_node to skip corrupted bytes and move
> to next node,
> instead of returning error ?
Not without a detailed analysis what exactly is going on.
It sounds more like ad-hoc hack. :-)
--
Thanks,
//richard
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2016-02-26 7:38 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-02-26 1:19 Fredrick Prashanth John Berchmans
2016-02-26 7:37 ` Richard Weinberger
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.