linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Bjorn Andersson <andersson@kernel.org>
To: Mukesh Ojha <quic_mojha@quicinc.com>
Cc: konrad.dybcio@linaro.org, linux-arm-msm@vger.kernel.org,
	 linux-kernel@vger.kernel.org, linus.walleij@linaro.org,
	linux-gpio@vger.kernel.org,
	 Srinivas Kandagatla <srinivas.kandagatla@linaro.org>,
	Kathiravan Thirumoorthy <quic_kathirav@quicinc.com>
Subject: Re: [PATCH v12 2/9] firmware: qcom: scm: provide a read-modify-write function
Date: Sat, 2 Mar 2024 13:09:39 -0600	[thread overview]
Message-ID: <xht25xxxzxmb24yobz4drmo5u4btlqo4akhscow7g5to7zyh3g@75bl5ddhib43> (raw)
In-Reply-To: <20240227155308.18395-3-quic_mojha@quicinc.com>

On Tue, Feb 27, 2024 at 09:23:01PM +0530, Mukesh Ojha wrote:
> It is possible that different bits of a secure register is used
> for different purpose and currently, there is no such available
> function from SCM driver to do that; one similar usage was pointed
> by Srinivas K. inside pinctrl-msm where interrupt configuration
> register lying in secure region and written via read-modify-write
> operation.
> 
> Export qcom_scm_io_rmw() to do read-modify-write operation on secure
> register and reuse it wherever applicable, also document scm_lock
> to convey its usage.
> 
> Suggested-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
> Signed-off-by: Mukesh Ojha <quic_mojha@quicinc.com>
> Tested-by: Kathiravan Thirumoorthy <quic_kathirav@quicinc.com> # IPQ9574 and IPQ5332
> ---
>  drivers/firmware/qcom/qcom_scm.c       | 33 ++++++++++++++++++++++++++
>  include/linux/firmware/qcom/qcom_scm.h |  1 +
>  2 files changed, 34 insertions(+)
> 
> diff --git a/drivers/firmware/qcom/qcom_scm.c b/drivers/firmware/qcom/qcom_scm.c
> index 2d0ba529cf56..8f766fce5f7c 100644
> --- a/drivers/firmware/qcom/qcom_scm.c
> +++ b/drivers/firmware/qcom/qcom_scm.c
> @@ -193,6 +193,11 @@ static void qcom_scm_bw_disable(void)
>  }
>  
>  enum qcom_scm_convention qcom_scm_convention = SMC_CONVENTION_UNKNOWN;
> +/*
> + * scm_lock to serialize call to query SMC convention and
> + * to atomically operate(read-modify-write) on different
> + * bits of secure register.
> + */
>  static DEFINE_SPINLOCK(scm_lock);
>  
>  static enum qcom_scm_convention __get_convention(void)
> @@ -481,6 +486,34 @@ static int qcom_scm_disable_sdi(void)
>  	return ret ? : res.result[0];
>  }
>  
> +int qcom_scm_io_rmw(phys_addr_t addr, unsigned int mask, unsigned int val)
> +{
> +	unsigned long flags;
> +	unsigned int old;
> +	unsigned int new;
> +	int ret;
> +
> +	if (!__scm)
> +		return -EPROBE_DEFER;
> +
> +	/*
> +	 * Lock to atomically do rmw operation on different bits
> +	 * of secure register
> +	 */

A spinlock does not make something globally atomic, all you have made
sure is that:
1) qcom_scm_io_rmw() can not happen concurrently with __get_convention()

The reuse of the lock makes me wonder what the use case you're having a
need to protect #1... When is rmw happening concurrently with convention
detection?

2) Two qcom_scm_io_rmw() can not happen concurrently

What happens if a separate process invokes qcom_scm_io_write() of the
same address concurrently with qcom_scm_io_rmw()?


Quite likely neither of these will happen in practice, and I'm guessing
that there will not be any caching issues etc among different calls to
qcom_scm_io_*(), and hence this spinlock serves just to complicate the
implementation.

Please add a kernel-doc comment to this function (and perhaps
qcom_scm_io_write()) and describe that we don't guarantee this operation
to happen atomically - or if you have a valid reason, document that and
it's exact limitations.


PS. I would have been perfectly happy with us not adding a rmw function.
You're adding 34 lines of code to save 2*3 lines of duplicated, easy to
understand, code.

Regards,
Bjorn

> +	spin_lock_irqsave(&scm_lock, flags);
> +	ret = qcom_scm_io_readl(addr, &old);
> +	if (ret)
> +		goto unlock;
> +
> +	new = (old & ~mask) | (val & mask);
> +
> +	ret = qcom_scm_io_writel(addr, new);
> +unlock:
> +	spin_unlock_irqrestore(&scm_lock, flags);
> +	return ret;
> +}
> +EXPORT_SYMBOL_GPL(qcom_scm_io_rmw);
> +
>  static int __qcom_scm_set_dload_mode(struct device *dev, bool enable)
>  {
>  	struct qcom_scm_desc desc = {
> diff --git a/include/linux/firmware/qcom/qcom_scm.h b/include/linux/firmware/qcom/qcom_scm.h
> index ccaf28846054..3a8bb2e603b3 100644
> --- a/include/linux/firmware/qcom/qcom_scm.h
> +++ b/include/linux/firmware/qcom/qcom_scm.h
> @@ -82,6 +82,7 @@ bool qcom_scm_pas_supported(u32 peripheral);
>  
>  int qcom_scm_io_readl(phys_addr_t addr, unsigned int *val);
>  int qcom_scm_io_writel(phys_addr_t addr, unsigned int val);
> +int qcom_scm_io_rmw(phys_addr_t addr, unsigned int mask, unsigned int val);
>  
>  bool qcom_scm_restore_sec_cfg_available(void);
>  int qcom_scm_restore_sec_cfg(u32 device_id, u32 spare);
> -- 
> 2.43.0.254.ga26002b62827
> 

  reply	other threads:[~2024-03-02 19:09 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-27 15:52 [PATCH v12 0/9] Misc SCM driver changes Mukesh Ojha
2024-02-27 15:53 ` [PATCH v12 1/9] firmware: qcom: scm: Rename scm_query_lock to scm_lock Mukesh Ojha
2024-03-02 19:10   ` Bjorn Andersson
2024-02-27 15:53 ` [PATCH v12 2/9] firmware: qcom: scm: provide a read-modify-write function Mukesh Ojha
2024-03-02 19:09   ` Bjorn Andersson [this message]
2024-03-05 10:39     ` Mukesh Ojha
2024-02-27 15:53 ` [PATCH v12 3/9] firmware: qcom: scm: Modify only the download bits in TCSR register Mukesh Ojha
2024-03-02 19:13   ` Bjorn Andersson
2024-03-05 10:43     ` Mukesh Ojha
2024-02-27 15:53 ` [PATCH v12 4/9] firmware: qcom: scm: Rework dload mode availability check Mukesh Ojha
2024-03-02 19:16   ` Bjorn Andersson
2024-03-05 10:54     ` Mukesh Ojha
2024-02-27 15:53 ` [PATCH v12 5/9] pinctrl: qcom: Use qcom_scm_io_rmw() function Mukesh Ojha
2024-02-29 13:56   ` Linus Walleij
2024-02-27 15:53 ` [PATCH v12 6/9] firmware: qcom: scm: Remove log reporting memory allocation failure Mukesh Ojha
2024-03-02 19:18   ` Bjorn Andersson
2024-02-27 15:53 ` [PATCH v12 7/9] firmware: qcom: scm: Fix __scm->dev assignement Mukesh Ojha
2024-03-02 19:25   ` Bjorn Andersson
2024-03-18 13:08     ` Mukesh Ojha
2024-03-19  1:17       ` Pavan Kondeti
2024-03-19 10:08         ` Mukesh Ojha
2024-03-19 10:22           ` Pavan Kondeti
2024-03-19 14:39             ` Mukesh Ojha
2024-02-27 15:53 ` [PATCH v12 8/9] firmware: qcom: scm: Add check to prevent Null pointer dereference Mukesh Ojha
2024-02-27 16:56   ` Elliot Berman
2024-02-28 15:08     ` Mukesh Ojha
2024-03-01 23:42   ` Konrad Dybcio
2024-03-02 18:57   ` Bjorn Andersson
2024-02-27 15:53 ` [PATCH v12 9/9] firmware: scm: Remove redundant scm argument from qcom_scm_waitq_wakeup() Mukesh Ojha
2024-03-02 19:23   ` Bjorn Andersson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=xht25xxxzxmb24yobz4drmo5u4btlqo4akhscow7g5to7zyh3g@75bl5ddhib43 \
    --to=andersson@kernel.org \
    --cc=konrad.dybcio@linaro.org \
    --cc=linus.walleij@linaro.org \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-gpio@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=quic_kathirav@quicinc.com \
    --cc=quic_mojha@quicinc.com \
    --cc=srinivas.kandagatla@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).