Date:   Wed, 15 Feb 2023 08:57:14 +0900
From:   Masami Hiramatsu (Google) <mhiramat@kernel.org>
To:     Nadav Amit <nadav.amit@gmail.com>
Cc:     x86@kernel.org, Borislav Petkov <bp@alien8.de>,
        Peter Zijlstra <peterz@infradead.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH -tip] kprobes/x86: Remove unneeded casting from
 immediate value
Message-Id: <20230215085714.8d46b98ce787a6e0c962ebf5@kernel.org>
In-Reply-To: <b79052ee-07d2-a6b9-03cb-b33f61985fe2@gmail.com>
References: <167595699689.1303232.13460109151246012198.stgit@mhiramat.roam.corp.google.com>
        <b79052ee-07d2-a6b9-03cb-b33f61985fe2@gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Fri, 10 Feb 2023 02:01:54 +0200
Nadav Amit <nadav.amit@gmail.com> wrote:

> 
> On 2/9/23 5:36 PM, Masami Hiramatsu (Google) wrote:
> > From: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> >
> > Remove unneeded casting from immediate value assignments for relative
> > jump offset. Since the immediate values in the 'insn' data structure are
> > assigned from immediate bytes as a signed value to insn.immediate.value
> > by insn_field_set(). Thus, if we need to access that value as a signed
> > value (in this kprobe's case), we don't need to cast it.
> > This is a kind of clean up (should not change behavior) follows Nadav's
> > bugfix.
> >
> > Link: https://lore.kernel.org/all/20230208071708.4048-1-namit@vmware.com/
> >
> > Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> > ---
> >   arch/x86/kernel/kprobes/core.c |   16 ++++++++--------
> >   1 file changed, 8 insertions(+), 8 deletions(-)
> >
> > diff --git a/arch/x86/kernel/kprobes/core.c b/arch/x86/kernel/kprobes/core.c
> > index 695873c0f50b..2440e736d0e4 100644
> > --- a/arch/x86/kernel/kprobes/core.c
> > +++ b/arch/x86/kernel/kprobes/core.c
> > @@ -607,19 +607,19 @@ static int prepare_emulation(struct kprobe *p, struct insn *insn)
> >   	case 0xe8:	/* near call relative */
> >   		p->ainsn.emulate_op = kprobe_emulate_call;
> >   		if (insn->immediate.nbytes == 2)
> > -			p->ainsn.rel32 = *(s16 *)&insn->immediate.value;
> > +			p->ainsn.rel32 = insn->immediate.value;
> >   		else
> > -			p->ainsn.rel32 = *(s32 *)&insn->immediate.value;
> > +			p->ainsn.rel32 = insn->immediate.value;
> 
> Hmm.. I don't get it. What the purpose of keeping the duplicated code 
> (after your change)?
> 

Oops, good catch! I think p->ainsn.rel32 can be set without any check in general.
Let me update it.

Thank you!


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>