[BUG] JFFS2 usage of write_begin and write_end functions causes kernel panic

[Alexander Belyakov]

On Sun, Apr 13, 2008 at 2:50 PM, Jörn Engel <joern at logfs.org> wrote:
> On Fri, 11 April 2008 21:05:33 +0300, Artem Bityutskiy wrote:
>  >
>  > So why did you start this project? How about just making all your work
>  > to go into mainline instead?
>
>  At least one of the patches should not go to mainline.  It is one of
>  those strange "If I change the code here, all symptoms disappear"
>  patches that doesn't cure the cause.  Someone should spend some time to
>  find the root couse, but noone has.

Joern, I do know the patch you are talking about ;) It was about jffs2
gc path bug.

There are two options: to constantly hit the bug because noone is
interested (have no time) in fixing it "properly" or to use cure
(tested and stressed) until someone would make better fix in the
mainline.

First time (more than half of year ago) I've submitted the patch (with
detailed bug analysis) it was simply ignored. The second time Joern
claimed it as cure not a fix. btw Joern was the only person who
replied - thanks.

So bug does exist, no fix in mainline, and no fruitful discussion in
mailing list. IMO answer to Artem's question becomes quite obvious.

-a