[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241014145808.GA8567@redhat.com>
Date: Mon, 14 Oct 2024 16:58:09 +0200
From: Oleg Nesterov <oleg@...hat.com>
To: Ma Qiao <mqaio@...ux.alibaba.com>
Cc: linux-trace-kernel@...r.kernel.org, rostedt@...dmis.org,
mhiramat@...nel.org, mathieu.desnoyers@...icios.com,
namhyung.kim@....com, linux-kernel@...r.kernel.org,
bpf@...r.kernel.org
Subject: Re: [PATCH] uprobe: avoid out-of-bounds memory access of fetching
args
Sorry, currently I don't have time to even try to read this patch, just
one note below...
On 10/14, Ma Qiao wrote:
>
> @@ -979,6 +980,11 @@ static struct uprobe_cpu_buffer *prepare_uprobe_buffer(struct trace_uprobe *tu,
> ucb = uprobe_buffer_get();
> ucb->dsize = tu->tp.size + dsize;
>
> + if (WARN_ON_ONCE(ucb->dsize > MAX_UCB_BUFFER_SIZE)) {
> + ucb->dsize = MAX_UCB_BUFFER_SIZE;
> + dsize = MAX_UCB_BUFFER_SIZE - tu->tp.size;
> + }
> +
Then you can probably kill the
if (WARN_ON_ONCE(ucb->dsize > PAGE_SIZE))
check in __uprobe_trace_func(), no?
Oleg.
Powered by blists - more mailing lists