Re: [PATCH v2 3/3] md/raid10: skip clean regions in bulk during recovery

[Yunye Zhao <yunye.zhao@linux.alibaba.com>]

Hi Kuai,

On 2026/8/7, Yu Kuai wrote:
>> If I read you right, you would prefer this fixed on the bitmap side
>> rather than in raid10.c.
>
> I still don't get why

Sorry, I misread your earlier reply.  I had read it as "the whole thing
belongs in md-bitmap.c", i.e. that bitmap_start_sync() should hand back a
device-space span, which is why I said md-bitmap.c would need the raid10
geometry.  That is not what you were saying.

I think the split is this: the first half belongs in md, in array address
space - bitmap_start_sync() should report a much larger clean span,
ideally up to the full array size - and the second half, the array ->
member conversion and the minimum over the members, is what 3/3 already
does.

> Check the code, currently raid10 do not map the max_sync, which is RESYNC_PAGES, back
> to rdev sector, and use it directly. I wonder is this safe for raid10 with chunksize less
> than RESYNC_PAGES?

Good point, v2 does not address that.  Let me think it through properly
and handle it in the next version.

Thanks,
Yunye