[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <0b1b5224-b891-bf6d-8d6f-f5d236890127@linaro.org>
Date: Mon, 31 Jul 2023 10:20:54 +0200
From: Konrad Dybcio <konrad.dybcio@...aro.org>
To: Bjorn Andersson <quic_bjorande@...cinc.com>,
Bjorn Andersson <andersson@...nel.org>,
Chris Lew <quic_clew@...cinc.com>
Cc: Alex Elder <elder@...nel.org>,
"David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>,
Mathieu Poirier <mathieu.poirier@...aro.org>,
netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
linux-arm-msm@...r.kernel.org, linux-remoteproc@...r.kernel.org
Subject: Re: [PATCH 3/4] soc: qcom: aoss: Format string in qmp_send()
On 31.07.2023 06:10, Bjorn Andersson wrote:
> The majority of callers to qmp_send() composes the message dynamically
> using some form of sprintf(), resulting in unnecessary complication and
> stack usage.
>
> By changing the interface of qmp_send() to take a format string and
> arguments, the duplicated composition of the commands can be moved to a
> single location.
>
> Signed-off-by: Bjorn Andersson <quic_bjorande@...cinc.com>
> ---
[...]
> +int qmp_send(struct qmp *qmp, const char *fmt, ...)
> {
> long time_left;
> + va_list args;
> char buf[QMP_MSG_LEN];
> + int len;
> int ret;
>
> - if (WARN_ON(IS_ERR_OR_NULL(qmp) || !data))
> + if (WARN_ON(IS_ERR_OR_NULL(qmp) || !fmt))
> return -EINVAL;
>
> - if (WARN_ON(strlen(data) >= sizeof(buf)))
> - return -EINVAL;
> + memset(buf, 0, sizeof(buf));
Wouldn't initializing the array with = { 0 } be faster?
Otherwise, this looks very cool
Reviewed-by: Konrad Dybcio <konrad.dybcio@...aro.org>
Konrad
Powered by blists - more mailing lists