Re: [PATCH bpf-next v3 4/5] bpftool: Add option to wait for program stream output

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] bpf: bpftool deadlock in exit_stream signal handler
--

commit 482ed48b5d6910c6aec52e854a7926f6f16ce58a
Author: Kumar Kartikeya Dwivedi <memxor@gmail.com>

bpftool: Add option to wait for program stream output

bpftool prog tracelog { stdout | stderr } PROG dumps the output buffered
in a program stream and exits, which is all BPF_PROG_STREAM_READ_BY_FD
allows. Waiting for further output means rerunning the command.

> diff --git a/tools/bpf/bpftool/prog.c b/tools/bpf/bpftool/prog.c
> index 24e40dfab4690..5e27b22444f25 100644
> --- a/tools/bpf/bpftool/prog.c
> +++ b/tools/bpf/bpftool/prog.c
> @@ -1119,21 +1119,66 @@ enum prog_tracelog_mode {
>  	TRACE_STDERR,
>  };
>  
> +static void exit_stream(int signo)
> +{
> +	exit(0);
> +}
> +

[Severity: Medium]
Since exit_stream() is used as a signal handler for SIGINT, SIGHUP, and
SIGTERM, is it safe to call exit(0) here?

exit() is not async-signal-safe and attempts to flush open stdio streams.
If the signal interrupts the main loop in prog_tracelog_stream() while it
is executing fwrite() or fflush(), it could try to acquire the same
non-reentrant lock, resulting in a deadlock.

Could _exit(0) be used instead to avoid this deadlock?

[ ... ]

> +	/*
> +	 * Exit from the handler like the trace pipe variant does. A flag checked
> +	 * between reads would miss a signal that lands before read() blocks and
> +	 * only end the loop at the next print.
> +	 */
> +	for (i = 0; i < ARRAY_SIZE(signals); i++)
> +		sigaction(signals[i], &act, NULL);
> +	while ((ret = read(fd, buf, sizeof(buf))) > 0) {
> +		fwrite(buf, sizeof(buf[0]), ret, file);
> +		fflush(file);
> +	}
> +	if (ret < 0)
> +		p_err("failed to read stream: %s", strerror(errno));
> +	close(fd);
> +out:
>  	fflush(file);
>  	return ret ? -1 : 0;
>  }

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260925045536.1480933-1-memxor@gmail.com?part=4