Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:7 as permitted sender) client-ip=2620:137:e000::3:7;
Date:   Mon, 25 Sep 2023 10:40:48 +0100
Message-ID: <87o7hqmvz3.wl-maz@kernel.org>
From:   Marc Zyngier <maz@kernel.org>
To:     Dinghao Liu <dinghao.liu@zju.edu.cn>
Cc:     Toan Le <toan@os.amperecomputing.com>,
        Lorenzo Pieralisi <lpieralisi@kernel.org>,
        Krzysztof =?UTF-8?B?V2lsY3p5xYRza2k=?= <kw@linux.com>,
        Rob Herring <robh@kernel.org>,
        Bjorn Helgaas <bhelgaas@google.com>, Duc Dang <dhdang@apm.com>,
        Tanmay Inamdar <tinamdar@apm.com>, linux-pci@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] PCI: xgene-msi: Fix a potential UAF in xgene_msi_probe
In-Reply-To: <20230925062133.14170-1-dinghao.liu@zju.edu.cn>
References: <20230925062133.14170-1-dinghao.liu@zju.edu.cn>
User-Agent: Wanderlust/2.15.9 (Almost Unreal) SEMI-EPG/1.14.7 (Harue)
 FLIM-LB/1.14.9 (=?UTF-8?B?R29qxY0=?=) APEL-LB/10.8 EasyPG/1.0.0 Emacs/28.2
 (x86_64-pc-linux-gnu) MULE/6.0 (HANACHIRUSATO)
MIME-Version: 1.0 (generated by SEMI-EPG 1.14.7 - "Harue")
Content-Type: text/plain; charset=US-ASCII
Precedence: bulk

On Mon, 25 Sep 2023 07:21:32 +0100,
Dinghao Liu <dinghao.liu@zju.edu.cn> wrote:
> 
> xgene_allocate_domains() will call irq_domain_remove() to free
> msi->inner_domain on failure. However, its caller, xgene_msi_probe(),
> will also call irq_domain_remove() through xgene_msi_remove() on the
> same failure, which may lead to a use-after-free. Set the freed pointer
> to NULL to fix this issue.
> 
> Fixes: dcd19de36775 ("PCI: xgene: Add APM X-Gene v1 PCIe MSI/MSIX termination driver")
> Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
> ---
>  drivers/pci/controller/pci-xgene-msi.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/pci/controller/pci-xgene-msi.c b/drivers/pci/controller/pci-xgene-msi.c
> index 3ce38dfd0d29..c0192c5ff0f3 100644
> --- a/drivers/pci/controller/pci-xgene-msi.c
> +++ b/drivers/pci/controller/pci-xgene-msi.c
> @@ -253,6 +253,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>  
>  	if (!msi->msi_domain) {
>  		irq_domain_remove(msi->inner_domain);
> +		msi->inner_domain = NULL;
>  		return -ENOMEM;
>  	}

Why can't we just drop the irq_domain_remove() call here instead, and
simply rely on xgene_msi_remove() to do the right thing? Something
like the untested patch below.

Thanks,

	M.

diff --git a/drivers/pci/controller/pci-xgene-msi.c b/drivers/pci/controller/pci-xgene-msi.c
index 0234e528b9a5..f98c9eb7bebf 100644
--- a/drivers/pci/controller/pci-xgene-msi.c
+++ b/drivers/pci/controller/pci-xgene-msi.c
@@ -251,10 +251,8 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
 						    &xgene_msi_domain_info,
 						    msi->inner_domain);
 
-	if (!msi->msi_domain) {
-		irq_domain_remove(msi->inner_domain);
+	if (!msi->msi_domain)
 		return -ENOMEM;
-	}
 
 	return 0;
 }

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