Re: [RFC PATCH v9 11/11] trace-cmd: Add splice() recording from FIFO without additional pipe buffer

From: Steven Rostedt <rostedt@goodmis.org>
To: Slavomir Kaslev <kaslevs@vmware.com>
Cc: linux-trace-devel@vger.kernel.org, slavomir.kaslev@gmail.com
Subject: Re: [RFC PATCH v9 11/11] trace-cmd: Add splice() recording from FIFO without additional pipe buffer
Date: Wed, 3 Apr 2019 08:55:15 -0400	[thread overview]
Message-ID: <20190403085515.1652201a@gandalf.local.home> (raw)
In-Reply-To: <20190402134540.32321-12-kaslevs@vmware.com>

On Tue,  2 Apr 2019 16:45:40 +0300
Slavomir Kaslev <kaslevs@vmware.com> wrote:

> +/*
> + * Returns -1 on error.
> + *          or bytes of data read.
> + */
> +static long direct_splice_data(struct tracecmd_recorder *recorder)
> +{
> +	struct pollfd pfd = {
> +		.fd = recorder->trace_fd,
> +		.events = POLLIN,
> +	};
> +	long read;
> +	int ret;
> +
> +	ret = poll(&pfd, 1, POLL_TIMEOUT_MS);
> +	if (ret < 0)
> +		return -1;
> +
> +	if (!(pfd.revents | POLLIN))
> +		return 0;

Why the polling code, and not just block on splice?

-- Steve

> +
> +	read = splice(recorder->trace_fd, NULL, recorder->fd, NULL,
> +		      recorder->pipe_size, recorder->trace_fd_flags);
> +	if (read < 0) {
> +		if (errno == EAGAIN || errno == EINTR)
> +			return 0;
> +
> +		warning("recorder error in splice input");
> +		return -1;
> +	}
> +
> +	return read;
> +}
> +