[<prev] [next>] [day] [month] [year] [list]
Message-ID: <2025012746-rebel-clanking-3f0e@gregkh>
Date: Mon, 27 Jan 2025 13:51:54 +0100
From: Greg KH <gregkh@...uxfoundation.org>
To: Zicheng Qu <quzicheng@...wei.com>
Cc: linux-cve-announce@...r.kernel.org, tanghui20@...wei.com,
zhangqiao22@...wei.com, judy.chenhui@...wei.com
Subject: Re: CVE-2021-47341: KVM: mmio: Fix use-after-free Read in
kvm_vm_ioctl_unregister_coalesced_mmio
On Mon, Jan 27, 2025 at 09:00:28AM +0000, Zicheng Qu wrote:
> Hi,
>
> I am submitting a request to cancel the CVE-2021-47341
> (https://lore.kernel.org/all/2024052137-CVE-2021-47341-f4e9@gregkh/).
> After reviewing the relevant code, I have identified that the described
> use-after-free (UAF) issue does not exist. Below is a detailed analysis:
>
> Issue Description:
> The CVE claims that `kvm_vm_ioctl_unregister_coalesced_mmio()` can lead to
> a UAF when `kvm_io_bus_unregister_dev()` returns `-ENOMEM`, supposedly
> freeing `struct kvm_coalesced_mmio_dev *dev`. A second call to
> `kvm_iodevice_destructor()` would then access freed memory.
The CVE also claims that you can trigger a BUG: callback, due to a KASAN
use-after-free warning which will reboot or crash your machine, right?
So because of that, this was issued a CVE. If that use-after-free is
not correct, why is KASAN triggered by it? Is that a bug in KASAN in
producing a false-positive? And what about systems that run with KASAN
enabled, will the BUG trigger not hit them as well?
thanks,
greg k-h
Powered by blists - more mailing lists