Hi,
> On 31. Oct 2024, at 08:48, Yu Kuai <yukuai1@xxxxxxxxxxxxxxx> wrote:
>
>> I will try, but I’m not sure. I don’t have a deep enough understanding to resolve some of the conflicts. In my previous mail I wasn’t sure which change would be the right one:
>> I guess if 6.12 doesn’t have this line at all:
>> - atomic_set(&sh->count, 1);
>> … then setting it to 0 is fine?
>> + atomic_set(&sh→count, 0);
>
> My patch doesn't touch this field at all, why make such change? This is
> not OK.
Yeah, patch didn’t think that’s OK either, that’s why I came back instead of trying to run that. ;)
Here’s the part of the patch I extracted from the earlier emails:
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 58f71c3e1368..b2a75a904209 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -2369,6 +2369,7 @@ static struct stripe_head *alloc_stripe(struct kmem_cache *sc, gfp_t gfp,
atomic_set(&sh->count, 1);
sh->raid_conf = conf;
sh->log_start = MaxSector;
+ atomic_set(&sh->bitmap_counts, 0);
… aaand I just noticed that patch got confused and tried to apply your change 3 lines early, so I ended up with a conflict - correctly! :)
>> But again, I have no idea what’s actually going on there … ;)
>> If you want I can try to wade through and give you a list of questions where the patch doesn’t obviously apply and you can let me know …
>
> Perhaps can you try v6.12-rc5 directly? If not, I'll give a patch based
> on v6.11 later.
So. I’d like to avoid running 6.12rc5 and if it isn’t too much trouble I’d appreciate a 6.11 patch, but now that I understood what’s wrong I can try to create it myself in the next days.
Cheers,
Christian
--
Christian Theune · ct@xxxxxxxxxxxxxxx · +49 345 219401 0
Flying Circus Internet Operations GmbH · https://flyingcircus.io
Leipziger Str. 70/71 · 06108 Halle (Saale) · Deutschland
HR Stendal HRB 21169 · Geschäftsführer: Christian Theune, Christian Zagrodnick
