Re: [report] use-after-free in __irq_resolve_mapping()

From: Marc Zyngier
Date: Thu Jun 10 2021 - 08:37:38 EST


On Wed, 09 Jun 2021 12:59:26 +0100,
John Garry <john.garry@xxxxxxxxxx> wrote:
>
> JFYI, on -next 20210608, I find this on my arm64 D06 and D05:

[...]

Funny what happens when you blindly replace radix_tree_delete() with
radix_tree_insert(NULL) and fail to realise that no, they aren't
equivalent... I'll pick a brown paper bag from the stack.

I'll run some more tests on some of the bigger iron before rebuilding
irq/irqchip-next, but it certainly looks less broken now.

Thanks again,

M.

--
Without deviation from the norm, progress is not possible.