Received-SPF: pass (google.com: domain of linux-kernel+bounces-570-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Fri, 15 Dec 2023 13:21:44 +0530
From: Naveen N Rao <naveen@kernel.org>
To: Nicholas Miehlbradt <nicholas@linux.ibm.com>
Cc: glider@google.com, elver@google.com, dvyukov@google.com, 
	akpm@linux-foundation.org, mpe@ellerman.id.au, npiggin@gmail.com, 
	christophe.leroy@csgroup.eu, linux-mm@kvack.org, kasan-dev@googlegroups.com, 
	iii@linux.ibm.com, linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 07/13] powerpc/kprobes: Unpoison instruction in kprobe
 struct
Message-ID: <xn274hbvxsfwii6lwis72ntnphiixvcob6hkopn5fygutht3qe@j4sau5ejaxwj>
References: <20231214055539.9420-1-nicholas@linux.ibm.com>
 <20231214055539.9420-8-nicholas@linux.ibm.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20231214055539.9420-8-nicholas@linux.ibm.com>

On Thu, Dec 14, 2023 at 05:55:33AM +0000, Nicholas Miehlbradt wrote:
> KMSAN does not unpoison the ainsn field of a kprobe struct correctly.
> Manually unpoison it to prevent false positives.
> 
> Signed-off-by: Nicholas Miehlbradt <nicholas@linux.ibm.com>
> ---
>  arch/powerpc/kernel/kprobes.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/arch/powerpc/kernel/kprobes.c b/arch/powerpc/kernel/kprobes.c
> index b20ee72e873a..1cbec54f2b6a 100644
> --- a/arch/powerpc/kernel/kprobes.c
> +++ b/arch/powerpc/kernel/kprobes.c
> @@ -27,6 +27,7 @@
>  #include <asm/sections.h>
>  #include <asm/inst.h>
>  #include <linux/uaccess.h>
> +#include <linux/kmsan-checks.h>
>  
>  DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
>  DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
> @@ -179,6 +180,7 @@ int arch_prepare_kprobe(struct kprobe *p)
>  
>  	if (!ret) {
>  		patch_instruction(p->ainsn.insn, insn);
> +		kmsan_unpoison_memory(p->ainsn.insn, sizeof(kprobe_opcode_t));

kprobe_opcode_t is u32, but we could be probing a prefixed instruction.  
You can pass the instruction length through ppc_inst_len(insn).


- Naveen