Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 9 Jun 2022 16:20:03 +0100
From:   Will Deacon <will@kernel.org>
To:     Chen Jun <chenjun102@huawei.com>
Cc:     linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        catalin.marinas@arm.com, xuqiang36@huawei.com
Subject: Re: [PATCH] arm64/smp: check !ipi_desc[i] in arch_show_interrupts
Message-ID: <20220609152003.GC3289@willie-the-truck>
References: <20220527082236.78037-1-chenjun102@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220527082236.78037-1-chenjun102@huawei.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Fri, May 27, 2022 at 08:22:36AM +0000, Chen Jun wrote:
> There is a potential dereferencing null pointer issue in
> arch_show_interrupts.
> 
> Problem 1:
> int arch_show_interrupts(struct seq_file *p, int prec)
>         for (i = 0; i < NR_IPI; i++) {
>                 seq_printf(p, "%10u ", irq_desc_kstat_cpu(ipi_desc[i],
> 			cpu));
> 
> Only ipi_desc[0..nr_ipi - 1] are initialized in set_smp_ipi_range.
> and ipi_desc[nr_ipi..NR_IPI] are NULL.
> irq_desc_kstat_cpu will dereference NULL pointer.
> For now, the problem can not be triggered, because NR_IPI is always
> equal to nr_ipi.
> 
> Problem 2:
> If request_percpu_irq failed in set_smp_ipi_range, ipi_desc[i]
> would be NULL.
> irq_desc_kstat_cpu will dereference NULL pointer.
> 
> check !ipi_desc[i] (as arm does) to avoid the problem.
> 
> Signed-off-by: Chen Jun <chenjun102@huawei.com>
> ---
>  arch/arm64/kernel/smp.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
> index 62ed361a4376..3d54f464428b 100644
> --- a/arch/arm64/kernel/smp.c
> +++ b/arch/arm64/kernel/smp.c
> @@ -781,6 +781,9 @@ int arch_show_interrupts(struct seq_file *p, int prec)
>  	unsigned int cpu, i;
>  
>  	for (i = 0; i < NR_IPI; i++) {
> +		if (!ipi_desc[i])
> +			continue;

Why not just use nr_ipi instead of NR_IPI?

Will