[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <aJBrXwHESPRTpwYa@krava>
Date: Mon, 4 Aug 2025 10:12:15 +0200
From: Jiri Olsa <olsajiri@...il.com>
To: Oleg Nesterov <oleg@...hat.com>
Cc: Masami Hiramatsu <mhiramat@...nel.org>,
Peter Zijlstra <peterz@...radead.org>,
Andrii Nakryiko <andrii@...nel.org>, bpf@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-trace-kernel@...r.kernel.org,
x86@...nel.org, Song Liu <songliubraving@...com>,
Yonghong Song <yhs@...com>,
John Fastabend <john.fastabend@...il.com>,
Hao Luo <haoluo@...gle.com>, Steven Rostedt <rostedt@...dmis.org>,
Ingo Molnar <mingo@...nel.org>
Subject: Re: [RFC 1/4] uprobe: Do not emulate/sstep original instruction when
ip is changed
On Sat, Aug 02, 2025 at 12:34:27PM +0200, Oleg Nesterov wrote:
> On 08/01, Jiri Olsa wrote:
> >
> > If uprobe handler changes instruction pointer we still execute single
> > step) or emulate the original instruction and increment the (new) ip
> > with its length.
>
> Yes... but what if we there are multiple consumers? The 1st one changes
> instruction_pointer, the next is unaware. Or it may change regs->ip too...
right, and I think that's already bad in current code
how about we dd flag to the consumer that ensures it's the only consumer
on the uprobe.. and we would skip original instruction execution for such
uprobe if its consumer changes the regs->ip.. I'll try to come up with the
patch
jirka
Powered by blists - more mailing lists