public inbox for linux-arm-kernel@lists.infradead.org
 help / color / mirror / Atom feed
From: Florian Fainelli <florian.fainelli@broadcom.com>
To: Sudeep Holla <sudeep.holla@arm.com>
Cc: linux-arm-kernel@lists.infradead.org,
	Rob Herring <robh@kernel.org>,
	Krzysztof Kozlowski <krzk+dt@kernel.org>,
	Conor Dooley <conor+dt@kernel.org>,
	Cristian Marussi <cristian.marussi@arm.com>,
	"open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS"
	<devicetree@vger.kernel.org>,
	open list <linux-kernel@vger.kernel.org>,
	"open list:SYSTEM CONTROL & POWER/MANAGEMENT INTERFACE"
	<arm-scmi@vger.kernel.org>,
	justin.chen@broadcom.com, opendmb@gmail.com,
	kapil.hali@broadcom.com, bcm-kernel-feedback-list@broadcom.com
Subject: Re: [PATCH v3 2/2] firmware: arm_scmi: Support 'reg-io-width' property for shared memory
Date: Thu, 22 Aug 2024 15:50:03 -0700	[thread overview]
Message-ID: <078bc5ee-e591-427b-a9fe-9386738a847f@broadcom.com> (raw)
In-Reply-To: <ZscW_E33YXF8Nx7r@bogus>

On 8/22/24 03:46, Sudeep Holla wrote:
> Hi Florian,
> 
> Sorry for getting late to this party, I wasn't able to review this before.
> Overall changes look correct. But my main concern is that is SCMI the right
> place to have such IO accessors. It is better to run it through Arnd if
> he is happy with it before I send him the pull request containing these.

Sure, would definitively want more eyes to review.

> 
> On Fri, Aug 16, 2024 at 03:42:21PM -0700, Florian Fainelli wrote:
>> Some shared memory areas might only support a certain access width,
>> such as 32-bit, which memcpy_{from,to}_io() does not adhere to at least
>> on ARM64 by making both 8-bit and 64-bit accesses to such memory.
>>
> 
> Is this limitation on the hardware for both read and writes ?

This applies to both reads and writes. We have to make accesses on a 4 
byte boundary and of exactly 4 bytes in size.

> The reason I ask is I see arm64 does have memcpy_toio_aligned() or
> __iowrite32_copy_full() for 32 bit aligned writes.
> 

That appears to work nicely on ARM64 and ARM 32-bit, thanks for the 
suggestion! One needs to be careful that __io{read,write}32_copy takes 
32-bit units, not bytes!

FWIW, here is my diff between v3 and v4:

diff --git a/drivers/firmware/arm_scmi/common.h 
b/drivers/firmware/arm_scmi/common.h
index 73bb496fac01..a13f79b37c99 100644
--- a/drivers/firmware/arm_scmi/common.h
+++ b/drivers/firmware/arm_scmi/common.h
@@ -319,9 +319,9 @@ enum scmi_bad_msg {
  /* Used for compactness and signature validation of the function 
pointers being
   * passed.
   */
-typedef void (*shmem_copy_toio_t)(volatile void __iomem *to, const void 
*from,
+typedef void (*shmem_copy_toio_t)(void __iomem *to, const void *from,
                                   size_t count);
-typedef void (*shmem_copy_fromio_t)(void *to, const volatile void 
__iomem *from,
+typedef void (*shmem_copy_fromio_t)(void *to, const void __iomem *from,
                                     size_t count);

  /**
diff --git a/drivers/firmware/arm_scmi/shmem.c 
b/drivers/firmware/arm_scmi/shmem.c
index aded5f1cd49f..e9f30ab671a8 100644
--- a/drivers/firmware/arm_scmi/shmem.c
+++ b/drivers/firmware/arm_scmi/shmem.c
@@ -35,33 +35,25 @@ struct scmi_shared_mem {
  };

  static inline void shmem_memcpy_fromio32(void *to,
-                                        const volatile void __iomem *from,
+                                        const void __iomem *from,
                                          size_t count)
  {
-       while (count) {
-               *(u32 *)to = __raw_readl(from);
-               from += 4;
-               to += 4;
-               count -= 4;
-       }
+       WARN_ON(!IS_ALIGNED((unsigned long)from, 4) ||
+               !IS_ALIGNED((unsigned long)to, 4) ||
+               count % 4);

-       /* Ensure all reads from I/O have completed */
-       rmb();
+       __ioread32_copy(to, from, count / 4);
  }

-static inline void shmem_memcpy_toio32(volatile void __iomem *to,
+static inline void shmem_memcpy_toio32(void __iomem *to,
                                        const void *from,
                                        size_t count)
  {
-       while (count) {
-               __raw_writel(*(u32 *)from, to);
-               from += 4;
-               to += 4;
-               count -= 4;
-       }
+       WARN_ON(!IS_ALIGNED((unsigned long)to, 4) ||
+               !IS_ALIGNED((unsigned long)from, 4) ||
+               count % 4);

-       /* Ensure all writes to I/O have completed */
-       wmb();
+       __iowrite32_copy(to, from, count / 4);
  }

  static struct scmi_shmem_io_ops shmem_io_ops32 = {
@@ -73,13 +65,13 @@ static struct scmi_shmem_io_ops shmem_io_ops32 = {
   * pre-processor.
   */
  static inline void shmem_memcpy_fromio(void *to,
-                                      const volatile void __iomem *from,
+                                      const void __iomem *from,
                                        size_t count)
  {
         memcpy_fromio(to, from, count);
  }

-static inline void shmem_memcpy_toio(volatile void __iomem *to,
+static inline void shmem_memcpy_toio(void __iomem *to,
                                      const void *from,
                                      size_t count)
  {

-- 
Florian



      reply	other threads:[~2024-08-22 22:55 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-16 22:42 [PATCH v3 0/2] Support for I/O width within ARM SCMI SHMEM Florian Fainelli
2024-08-16 22:42 ` [PATCH v3 1/2] dt-bindings: sram: Document reg-io-width property Florian Fainelli
2024-08-16 22:42 ` [PATCH v3 2/2] firmware: arm_scmi: Support 'reg-io-width' property for shared memory Florian Fainelli
2024-08-20  2:49   ` Peng Fan
2024-08-20 16:34     ` Florian Fainelli
2024-08-21  1:01       ` Peng Fan
2024-08-22 10:46   ` Sudeep Holla
2024-08-22 22:50     ` Florian Fainelli [this message]

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=078bc5ee-e591-427b-a9fe-9386738a847f@broadcom.com \
    --to=florian.fainelli@broadcom.com \
    --cc=arm-scmi@vger.kernel.org \
    --cc=bcm-kernel-feedback-list@broadcom.com \
    --cc=conor+dt@kernel.org \
    --cc=cristian.marussi@arm.com \
    --cc=devicetree@vger.kernel.org \
    --cc=justin.chen@broadcom.com \
    --cc=kapil.hali@broadcom.com \
    --cc=krzk+dt@kernel.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=opendmb@gmail.com \
    --cc=robh@kernel.org \
    --cc=sudeep.holla@arm.com \
    /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