On 22/11/17 05:13, Michael Lyle wrote:
Reviewed-by: Michael Lyle <mlyle@xxxxxxxx>
<snip>
I have a little dilemma now related to this patch and I wonder if I
could ask you guys for your opinion?
I have been hit and am being hit by the bug this patch fixes every few
days at the moment, on writeback bcache resources. When it hits it
causes a failure in the layer above, and this leads to other problems
that can and has led to loss of data (not in bcache, just that the read
failure causes a chain of events in upper layer, separate issue, won't
clutter this mail with that). So, I am very keen to resolve this
urgently in my environment.
I am running stable 4.9, and I have cherry picked several recent bcache
patches, including "bcache: only permit to recovery read error
when cache device is clean", which I understand causes this issue.
I have a dilemma now whether I should apply the new patch in this
thread, which should fix the issue, or revert "only permit to recovery
read error" patch.
The dilemma is because I don't like the idea of reverting the "only
permit to recovery read error" patch since it appears to fix an
important problem. Or am I overestimating the problem (maybe that's why
the original patch has not been backported to stable kernels, maybe the
original issue quite rare)?
But I am also wary of applying the new patch as Michael only just added
to his tree so hasn't had much baking time.
I wondered what is the opinion of you guys which of these two I should
choose?
(BTW, please don't say "just don't cherry pick onto 4.9". I am aware of
the risks of cherry picking, as discussed recently on this list. I am
happy with the level of risk and I am being very careful in what I pick.
In this case I have not been "burnt" by cherry picking since I would
have hit the same issue if I would have just upgraded to 4.14 which
contains the original "only permit to recovery read error" patch.)
( N.B. I understand that any opinion/advice I receive is just that. I
take complete responsibility for what I decide :-) )
Thanks!
Eddie
--
To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
