Re: [PATCH v3 2/3] mm: vmalloc: Switch to find_unlink_vmap_area() in vm_unmap_ram()

From: Uladzislau Rezki
Date: Thu Dec 29 2022 - 07:48:20 EST


On Wed, Dec 28, 2022 at 03:47:07PM -0800, Andrew Morton wrote:
> On Fri, 23 Dec 2022 17:41:48 +0100 Uladzislau Rezki <urezki@xxxxxxxxx> wrote:
>
> > > Don't we also need to remove the manual unlink that was done
> > > here previously? Actually it seems like that manual unlink is missing
> > > after patch 1, creating a bisection hazard. So either add it there,
> > > or just fold this patch into the previous one.
> > >
> > Right. In terms of bisection it is not so good. I think folding is the
> > best.
> >
> > Andrew, could you please fold this patch into the:
>
> which patch ;)
>
Currently the next-20221226 contains three patches:

<snip>
[1]
commit c83b70c3cc1ecf99897ca0ea6e44aa2125a61ccb
Author: Uladzislau Rezki (Sony) <urezki@xxxxxxxxx>
Date: Wed Dec 21 18:44:54 2022 +0100

mm: vmalloc: replace BUG_ON() by WARN_ON_ONCE()

[2]
commit 8a85ea97b35924ee39d51e00ecb3f6d07f748a36
Author: Uladzislau Rezki (Sony) <urezki@xxxxxxxxx>
Date: Wed Dec 21 18:44:53 2022 +0100

mm: vmalloc: switch to find_unlink_vmap_area() in vm_unmap_ram()

[3]
commit a7c84c673c71cdfad20fe25e5d2051ed229859f7
Author: Uladzislau Rezki (Sony) <urezki@xxxxxxxxx>
Date: Wed Dec 21 18:44:52 2022 +0100

mm: vmalloc: avoid calling __find_vmap_area() twise in __vunmap()
<snip>

It would be good if you could fold [2] into [3] making it as one
patch. The problem is that, if we leave it as it is, the bisection
mechanism would consider [3] as a buggy patch, because it is not
fully accomplished and depends on [2].

Is that OK for you, i mean to squash on your own? Or i just should
resend one more time?

Thank you in advance!

--
Uladzislau Rezki