linux-kernel - Re: [PATCH] KVM: arm64: Fix the return value of smp_call_function

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Mon, 18 Jan 2021 11:18:20 +0000
From:   Marc Zyngier <maz@...nel.org>
To:     Yejune Deng <yejune.deng@...il.com>
Cc:     james.morse@....com, julien.thierry.kdev@...il.com,
        suzuki.poulose@....com, catalin.marinas@....com, will@...nel.org,
        linux-arm-kernel@...ts.infradead.org, kvmarm@...ts.cs.columbia.edu,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] KVM: arm64: Fix the return value of
 smp_call_function_single()

On 2021-01-18 09:31, Yejune Deng wrote:
> In smp_call_function_single(), the 3rd parameter isn't the return value
> and it's always positive. But it may return a negative value. So the
> 'ret' is should be the return value of the smp_call_function_single().
> 
> In check_kvm_target_cpu(), 'phys_target' is more readable than 'ret'.
> 
> Signed-off-by: Yejune Deng <yejune.deng@...il.com>
> ---
>  arch/arm64/kvm/arm.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
> index 04c44853b103..5fa5c04106de 100644
> --- a/arch/arm64/kvm/arm.c
> +++ b/arch/arm64/kvm/arm.c
> @@ -1815,9 +1815,9 @@ static int init_hyp_mode(void)
>  	return err;
>  }
> 
> -static void check_kvm_target_cpu(void *ret)
> +static void check_kvm_target_cpu(void *phys_target)
>  {
> -	*(int *)ret = kvm_target_cpu();
> +	*(int *)phys_target = kvm_target_cpu();
>  }
> 
>  struct kvm_vcpu *kvm_mpidr_to_vcpu(struct kvm *kvm, unsigned long 
> mpidr)
> @@ -1879,7 +1879,7 @@ void kvm_arch_irq_bypass_start(struct
> irq_bypass_consumer *cons)
>  int kvm_arch_init(void *opaque)
>  {
>  	int err;
> -	int ret, cpu;
> +	int ret, cpu, phys_target;
>  	bool in_hyp_mode;
> 
>  	if (!is_hyp_mode_available()) {
> @@ -1900,7 +1900,7 @@ int kvm_arch_init(void *opaque)
>  			 "Only trusted guests should be used on this system.\n");
> 
>  	for_each_online_cpu(cpu) {
> -		smp_call_function_single(cpu, check_kvm_target_cpu, &ret, 1);
> +		ret = smp_call_function_single(cpu, check_kvm_target_cpu, 
> &phys_target, 1);
>  		if (ret < 0) {
>  			kvm_err("Error, CPU %d not supported!\n", cpu);
>  			return -ENODEV;

That's not the purpose of this code. We check the target CPU
for so that we can decide to *fail* the KVM init if there is
a CPU we do not support (we definitely used to do that with
32bit), and the error message clearly states this.

So if you want to handle a cross-call failure, please do that.
But don't change the existing semantics of this code.

         M.
-- 
Jazz is not dead. It just smells funny...