Skip to contentSkip to conversation

add mailbox support for i.MX7D

PATCH Superseded Oleksij Rempel 6 participants 75 messages Jul 26, 2018 – Aug 9, 2018

Subsystems (4)

v7 of 10 Superseded Latest is v10 →

lore
Share & tools
lore
#0 lore raw

v7 cover add mailbox support for i.MX7D

20180626 changes v7:
- DT: add i.MX6SX and i.MX7S to the documentation.
- imx-mailbox: don't use devm_ functions for startup and shutdown.
- imx-mailbox: rename imx_mu_rmw to imx_mu_xcr_rmw and add locks
- imx-mailbox: pass of_property_read_bool directly to side_b

20180622 changes v6:
- include one more patch provided by Aisheng
- DT: add fall back compatible fsl,imx6sx-mu
- imx-mailbox: for now, use only fsl,imx6sx-mu

20180621 changes v5:
- DT: revert most of the changes from previous version
- imx-mailbox: remove struct imx_mu_cfg
- imx-mailbox: remove !! from imx_mu_last_tx_done()

20180618 changes v4:
- DT: change fsl,mu-side-a to fsl,mu-side-b
- DT: split the patches.
- DT: add all currently known SoCs.
- imx-mailbox: free allocated irq name on channel shutdown
- imx-mailbox: rename *_imx7 functions to *_generic

20180615 changes v3:
- DT: remove prosaic part of documentation. It describes software or firmware specific usage and not relevant for HW description.
- DT: use <soc>-mu instead of <soc>-mu-<mu side> and add fsl,mu-side-a parameter.
- DT: add most of know i.MX variants with MU
- imx-mailbox: use macros instead of precalculated bit index.
- imx-mailbox: remove warning message for clk.
- imx-mailbox: use imx_mu_chan[%idx] for devm_request_irq.
- imx-mailbox: depend on ARCH_MXC instead of SOX_IMX7

20180615 changes v2:
- DT: use mailbox@ instead of mu@
- DT: change interrupts description
- clk: use imx_clk_gate4 instead of imx_clk_gate2
- imx-mailbox: remove last_tx_done support
- imx-mailbox: fix module description

This patches are providing support for mailbox (Messaging Unit) for i.MX7D.
Functionality was tested on PHYTEC phyBOARD-Zeta i.MX7D with Linux running on all cores: ARM Cortex-A7 and ARM Cortex-M4.

Both parts of i.MX messaging Unit are visible for all CPUs available on i.MX7D. Communication worked independent of MU side in combination with CPU. For example MU-A used on ARM Cortex-A7 and MU-B used on ARM Cortex-M4 or other ways around.

Dong Aisheng (2):
dt-bindings: mailbox: allow mbox-cells to be equal to 0
dt-bindings: arm: fsl: add mu binding doc

Oleksij Rempel (4):
dt-bindings: mailbox: imx-mu: add generic MU channel support
dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.
ARM: dts: imx7s: add i.MX7 messaging unit support
mailbox: Add support for i.MX7D messaging unit

 .../devicetree/bindings/mailbox/fsl,mu.txt    |  41 +++
 .../devicetree/bindings/mailbox/mailbox.txt   |   3 +-
 arch/arm/boot/dts/imx7s.dtsi                  |  19 ++
 drivers/mailbox/Kconfig                       |   6 +
 drivers/mailbox/Makefile                      |   2 +
 drivers/mailbox/imx-mailbox.c                 | 276 ++++++++++++++++++
 6 files changed, 345 insertions(+), 2 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/mailbox/fsl,mu.txt
 create mode 100644 drivers/mailbox/imx-mailbox.c
Signature · 1 line
2.18.0
#1 lore raw

PATCH v7 1/6 dt-bindings: mailbox: allow mbox-cells to be equal to 0

From: Dong Aisheng <[email protected]>

Mailbox devices may have only one channel which means the mbox-cells at least 1 does not make sense for this type devices. Let's remove that limitation to allow the mbox-cells to be equal to 0.

Cc Mark Rutland <[email protected]>
Cc Sudeep Holla <[email protected]>
Cc devicetree at vger.kernel.org
Cc linux-kernel at vger.kernel.org
---
 Documentation/devicetree/bindings/mailbox/mailbox.txt | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
Documentation/devicetree/bindings/mailbox/mailbox.txt +1 -2
index af8ecee2ac68..c2fcd054141a 100644
@@ -6,8 +6,7 @@ assign appropriate mailbox channel to client drivers.
 * Mailbox Controller
 
 Required property:
-- #mbox-cells: Must be at least 1. Number of cells in a mailbox
-		specifier.
+- #mbox-cells: Number of cells in a mailbox specifier.
 
 Example:
 	mailbox: mailbox {
Signature · 1 line
2.18.0
maintainer
#2 lore raw

Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 6 lines · Oleksij Rempel

From: Dong Aisheng <[email protected]>

Mailbox devices may have only one channel which means the mbox-cells at least 1 does not make sense for this type devices. Let's remove that limitation to allow the mbox-cells to be equal to 0.

OK

But please realise that you'll have to write more code -- implement your own of_xlate() callback.
Most developers would prefer to call the only channel as channel number 0, and use the common of_xlate implementation.

-jassi

#3 lore raw

RE: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 24 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 5:42 PM
To: Oleksij Rempel <[email protected]>
Cc: Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; A.s. Dong <[email protected]>; Vladimir Zapolskiy <[email protected]>; Sudeep Holla <[email protected]>; Devicetree List <[email protected]>; Linux Kernel Mailing List <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 6 lines · Oleksij Rempel

From: Dong Aisheng <[email protected]>

Mailbox devices may have only one channel which means the mbox-cells at least 1 does not make sense for this type devices. Let's remove that limitation to allow the mbox-cells to be equal to 0.

Quoted 7 lines

OK

But please realise that you'll have to write more code -- implement your own of_xlate() callback.
Most developers would prefer to call the only channel as channel number 0, and use the common of_xlate implementation.

Yes, that's true because the common of_xlate assuming the mbox-cells is 1. And there's already such users:
arch/arm/boot/dts/bcm283x.dtsi

More comments will be on your reply of patch-3.

Regards
Dong Aisheng

Quoted 1 line

-jassi

maintainer
#4 lore raw

Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 32 lines · A.s. Dong
Quoted 24 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 5:42 PM
To: Oleksij Rempel <[email protected]>
Cc: Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; A.s. Dong <[email protected]>; Vladimir Zapolskiy <[email protected]>; Sudeep Holla <[email protected]>; Devicetree List <[email protected]>; Linux Kernel Mailing List <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 6 lines · Oleksij Rempel

From: Dong Aisheng <[email protected]>

Mailbox devices may have only one channel which means the mbox-cells at least 1 does not make sense for this type devices. Let's remove that limitation to allow the mbox-cells to be equal to 0.

Quoted 7 lines

OK

But please realise that you'll have to write more code -- implement your own of_xlate() callback.
Most developers would prefer to call the only channel as channel number 0, and use the common of_xlate implementation.

Quoted 4 lines

Yes, that's true because the common of_xlate assuming the mbox-cells is 1. And there's already such users:
arch/arm/boot/dts/bcm283x.dtsi

You mean that is a good thing, right?

#5 lore raw

RE: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 52 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 7:37 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; Sudeep Holla <[email protected]>; Devicetree List <[email protected]>; Linux Kernel Mailing List <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 35 lines · A.s. Dong
Quoted 34 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 5:42 PM
To: Oleksij Rempel <[email protected]>
Cc: Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; A.s. Dong <[email protected]>; Vladimir Zapolskiy <[email protected]>; Sudeep Holla <[email protected]>; Devicetree List <[email protected]>; Linux Kernel Mailing List <[email protected]>; , Sascha Hauer <[email protected]>; ,
linux-arm-kernel at lists.infradead.org, linux-
mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 6 lines · Oleksij Rempel

From: Dong Aisheng <[email protected]>

Mailbox devices may have only one channel which means the mbox-cells at least 1 does not make sense for this type devices. Let's remove that limitation to allow the mbox-cells to be equal to 0.

OK

But please realise that you'll have to write more code -- implement your own
of_xlate() callback.
Most developers would prefer to call the only channel as channel number 0, and use the common of_xlate implementation.

Quoted 5 lines
Quoted 4 lines

Yes, that's true because the common of_xlate assuming the mbox-cells is 1. And there's already such users:
arch/arm/boot/dts/bcm283x.dtsi

You mean that is a good thing, right?

No, I mean there's similar HW in kernel already that only has one channel. Are you suggesting even for one channel mailbox, we should still use Mbox-cells 1 and use 'mboxes = <&mailbox 0>' in devicetree?

Regards
Dong Aisheng

maintainer
#6 lore raw

Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 56 lines · A.s. Dong
Quoted 56 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 7:37 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; Sudeep Holla <[email protected]>; Devicetree List <[email protected]>; Linux Kernel Mailing List <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 39 lines · A.s. Dong
Quoted 26 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 5:42 PM
To: Oleksij Rempel <[email protected]>
Cc: Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; A.s. Dong <[email protected]>; Vladimir Zapolskiy <[email protected]>; Sudeep Holla <[email protected]>; Devicetree List <[email protected]>; Linux Kernel Mailing List <[email protected]>; , Sascha Hauer <[email protected]>; ,
linux-arm-kernel at lists.infradead.org, linux-
mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 1/6] dt-bindings: mailbox: allow mbox-cells to be equal to 0

Quoted 6 lines · Oleksij Rempel

From: Dong Aisheng <[email protected]>

Mailbox devices may have only one channel which means the mbox-cells at least 1 does not make sense for this type devices. Let's remove that limitation to allow the mbox-cells to be equal to 0.

Quoted 8 lines

OK

But please realise that you'll have to write more code -- implement your own
of_xlate() callback.
Most developers would prefer to call the only channel as channel number 0, and use the common of_xlate implementation.

Yes, that's true because the common of_xlate assuming the mbox-cells is 1. And there's already such users:
arch/arm/boot/dts/bcm283x.dtsi

Quoted 6 lines
Quoted 1 line

You mean that is a good thing, right?

No, I mean there's similar HW in kernel already that only has one channel. Are you suggesting even for one channel mailbox, we should still use Mbox-cells 1 and use 'mboxes = <&mailbox 0>' in devicetree?

As i said, I am ok with the mbox-cells to be 0.

But if you have two instances of the controller, one with 4 channels and the other with 1 channel. Then you can/should not have different bindings.

And, if the controller can have only 1 channel, then please feel free to have mbox-cells=0 and implement your own of_xlate().

#7 lore raw

PATCH v7 2/6 dt-bindings: arm: fsl: add mu binding doc

In reply to Oleksij Rempel

From: Dong Aisheng <[email protected]>

The Messaging Unit module enables two processors within the SoC to communicate and coordinate by passing messages (e.g. data, status and control) through the MU interface.

Cc Shawn Guo <[email protected]>
Cc Sascha Hauer <[email protected]>
Cc Fabio Estevam <[email protected]>
Cc Mark Rutland <[email protected]>
Cc devicetree at vger.kernel.org
---
 .../devicetree/bindings/mailbox/fsl,mu.txt    | 34 +++++++++++++++++++
 1 file changed, 34 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/mailbox/fsl,mu.txt
Documentation/devicetree/bindings/mailbox/fsl,mu.txt new file +34 -0
new file mode 100644
index 000000000000..90e4905dfc69
@@ -0,0 +1,34 @@
+NXP i.MX Messaging Unit (MU)
+--------------------------------------------------------------------
+
+The Messaging Unit module enables two processors within the SoC to
+communicate and coordinate by passing messages (e.g. data, status
+and control) through the MU interface. The MU also provides the ability
+for one processor to signal the other processor using interrupts.
+
+Because the MU manages the messaging between processors, the MU uses
+different clocks (from each side of the different peripheral buses).
+Therefore, the MU must synchronize the accesses from one side to the
+other. The MU accomplishes synchronization using two sets of matching
+registers (Processor A-facing, Processor B-facing).
+
+Messaging Unit Device Node:
+=============================
+
+Required properties:
+-------------------
+- compatible :	should be "fsl,<chip>-mu", the supported chips include
+		imx8qxp, imx8qm.
+- reg :		Should contain the registers location and length
+- interrupts :	Interrupt number. The interrupt specifier format depends
+		on the interrupt controller parent.
+- #mbox-cells:  Must be 0. Number of cells in a mailbox
+
+Examples:
+--------
+lsio_mu0: mailbox at 5d1b0000 {
+	compatible = "fsl,imx8qxp-mu";
+	reg = <0x0 0x5d1b0000 0x0 0x10000>;
+	interrupts = <GIC_SPI 176 IRQ_TYPE_LEVEL_HIGH>;
+	#mbox-cells = <0>;
+};
Signature · 1 line
2.18.0
maintainer
#8 lore raw

PATCH v7 2/6 dt-bindings: arm: fsl: add mu binding doc

Quoted 40 lines · Oleksij Rempel

From: Dong Aisheng <[email protected]>

The Messaging Unit module enables two processors within the SoC to communicate and coordinate by passing messages (e.g. data, status and control) through the MU interface.

Cc Shawn Guo <[email protected]>
Cc Sascha Hauer <[email protected]>
Cc Fabio Estevam <[email protected]>
Cc Mark Rutland <[email protected]>
Cc devicetree at vger.kernel.org
---
 .../devicetree/bindings/mailbox/fsl,mu.txt    | 34 +++++++++++++++++++
 1 file changed, 34 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/mailbox/fsl,mu.txt
Documentation/devicetree/bindings/mailbox/fsl,mu.txt new file +15 -0
new file mode 100644
index 000000000000..90e4905dfc69
@@ -0,0 +1,34 @@
+NXP i.MX Messaging Unit (MU)
+--------------------------------------------------------------------
+
+The Messaging Unit module enables two processors within the SoC to
+communicate and coordinate by passing messages (e.g. data, status
+and control) through the MU interface. The MU also provides the ability
+for one processor to signal the other processor using interrupts.
+
+Because the MU manages the messaging between processors, the MU uses
+different clocks (from each side of the different peripheral buses).
+Therefore, the MU must synchronize the accesses from one side to the
+other. The MU accomplishes synchronization using two sets of matching
+registers (Processor A-facing, Processor B-facing).
+
+Messaging Unit Device Node:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt line 25

Quoted 10 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt new file +10 -0
+=============================
+
+Required properties:
+-------------------
+- compatible : should be "fsl,<chip>-mu", the supported chips include
+               imx8qxp, imx8qm.
+- reg :                Should contain the registers location and length
+- interrupts : Interrupt number. The interrupt specifier format depends
+               on the interrupt controller parent.
+- #mbox-cells:  Must be 0. Number of cells in a mailbox

Please consider my reply to patch-3

#9 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Oleksij Rempel

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +8 -1
index 90e4905dfc69..113d6ab931ef 100644
@@ -22,7 +22,14 @@ Required properties:
 - reg :		Should contain the registers location and length
 - interrupts :	Interrupt number. The interrupt specifier format depends
 		on the interrupt controller parent.
-- #mbox-cells:  Must be 0. Number of cells in a mailbox
+- #mbox-cells:  Must be:
+		0 - for single channel mode. i.MX8* SCU protocol specific.
+		1 - for multichannel (generic) mode.
+
+Optional properties:
+-------------------
+- clocks :	phandle to the input clock.
+- fsl,mu-side-b : Should be set for side B MU.
 
 Examples:
 --------
Signature · 1 line
2.18.0
maintainer
#10 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 3 lines · Oleksij Rempel

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Required properties: · line 28

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

#11 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 13 lines · Jassi Brar
Quoted 3 lines · Oleksij Rempel

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Required properties: · line 28

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

We spend already some week to publicly talk about this..
The problem is, imx8 has multiple MUs, only some of them communicate directly with SCU. SCU has some well defined protocol and
MU should be used and configured in some SCU specific way. From NXP perspective, SCU firmware should not be replaceable and from Linux side seen as part of the HW.

@Dong, is it correct description?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180726/300c4146/attachment.sig>

maintainer
#12 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 16 lines · Oleksij Rempel
Quoted 13 lines · Jassi Brar
Quoted 3 lines · Oleksij Rempel

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 8 lines
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

We spend already some week to publicly talk about this..

You should have kept me in the loop.
I know, I am not that important, but for the sake of the protocol you should have CC'ed the mailbox maintainer :)

Quoted 2 lines

The problem is, imx8 has multiple MUs

You mean multiple instances of the MU. Some have 4 channels, some 1. Is that correct?

Quoted 3 lines

, only some of them communicate
directly with SCU.

OK

Quoted 5 lines

SCU has some well defined protocol and
MU should be used and configured in some SCU specific way. From NXP perspective, SCU firmware should not be replaceable and from Linux side seen as part of the HW.

Sure. But the behaviour (protocol) should be implemented in the client driver, not the controller driver.

#13 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Jassi Brar

Hi Jassi,

Quoted 28 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 5:50 PM
To: Oleksij Rempel <[email protected]>
Cc: Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; A.s. Dong <[email protected]>; Vladimir Zapolskiy <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 3 lines · Oleksij Rempel

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Required properties: · line 28

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

The reality is that MU hardware does not define channels in reference manual. However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each other.

Passing short messages
Transmit register(s) can be used to pass short messages
from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an interrupt at the receiver side).

The reference manual is at here: (Chapter 42 Messaging Unit (MU)
https://www.nxp.com/docs/en/reference-manual/IMX6ULRM.pdf

And SCU firmware does use MU in above way specified by reference manual. Even from HW point of view, it's still one channel only mailbox. That's why we change the mbox-cells to 0 accordingly to avoid confusing the multi channels users supported in this driver.

As per Sasha's request, we're going to support both protocol (SCU and generic M4) in one mailbox driver, that means we need deal with this two case property.

If we use mbox-cells 1 case for SCU, then the using would be like:
mboxes = <&mailbox 0> which is mixed with the normal multi channel users. And we also must state clearly in binding doc that for mbox-cells
1 case for SCU, the cell value must be 0 which is also confusing with
Multi channel users.

How do you suggest for such case?

Regards
Dong Aisheng

maintainer
#14 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 53 lines · A.s. Dong

Hi Jassi,

Quoted 28 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 5:50 PM
To: Oleksij Rempel <[email protected]>
Cc: Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; A.s. Dong <[email protected]>; Vladimir Zapolskiy <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 3 lines · Oleksij Rempel

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Required properties: · line 28

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

The reality is that MU hardware does not define channels in reference manual. However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each other.

Passing short messages
Transmit register(s) can be used to pass short messages
from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an interrupt at the receiver side).

Quoted 3 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)
https://www.nxp.com/docs/en/reference-manual/IMX6ULRM.pdf

Thanks. That makes life much easier.

Quoted 8 lines

And SCU firmware does use MU in above way specified by reference manual. Even from HW point of view, it's still one channel only mailbox. That's why we change the mbox-cells to 0 accordingly to avoid confusing the multi channels users supported in this driver.

As per Sasha's request, we're going to support both protocol (SCU and generic M4) in one mailbox driver, that means we need deal with this two case property.

Quoted 8 lines

If we use mbox-cells 1 case for SCU, then the using would be like:
mboxes = <&mailbox 0> which is mixed with the normal multi channel users. And we also must state clearly in binding doc that for mbox-cells
1 case for SCU, the cell value must be 0 which is also confusing with
Multi channel users.

How do you suggest for such case?

Let me read the spec and try to see/shine the light :)

Cheers!

#15 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 14 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 8:27 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

[...]

Quoted 17 lines
Quoted 1 line

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 6 lines

Even from HW point of view, it's still one channel only mailbox. That's why we change the mbox-cells to 0 accordingly to avoid confusing the multi channels users supported in this driver.

As per Sasha's request, we're going to support both protocol (SCU and generic M4) in one mailbox driver, that means we need deal with this two

case property.

Quoted 8 lines

If we use mbox-cells 1 case for SCU, then the using would be like:
mboxes = <&mailbox 0> which is mixed with the normal multi channel users. And we also must state clearly in binding doc that for mbox-cells
1 case for SCU, the cell value must be 0 which is also confusing with
Multi channel users.

Quoted 3 lines
Quoted 2 lines

How do you suggest for such case?

Let me read the spec and try to see/shine the light :)

Great, many thanks!

Regards
Dong Aisheng

Quoted 2 lines

Cheers!

#17 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 26 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 9:50 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 3 lines · A.s. Dong

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw
ww.nxp.com%2Fdocs%2Fen%2Freference- manual%2FIMX6ULRM.pdf&amp;data=02%7C01%7Caisheng.dong%40nxp.co m%7Ca1e872436611497795bb08d5f2fe9feb%7C686ea1d3bc2b4c6fa92cd99c5c 301635%7C0%7C0%7C636682097774333065&amp;sdata=Zqf9kgJrutPtQE8OP %2BDu1%2BvULXWSQcgHbszwIWkTNS0%3D&amp;reserved=0

Quoted 1 line
Quoted 4 lines

You seem to have shared a wrong link. It has
Chapter 42 Synchronous Audio Interface (SAI)

and finding "Messaging Unit" shows no result.

Sorry, it should be:
https://www.nxp.com/docs/en/reference-manual/IMX6SXRM.pdf

Regards
Dong Aisheng

maintainer
#18 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to A.s. Dong
Quoted 17 lines · A.s. Dong
Quoted 11 lines
Quoted 4 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Required properties: · line 28

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

Quoted 9 lines

The reality is that MU hardware does not define channels in reference manual. However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each other.

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write 4 word data over 4 rx/tx channels, instead of the virtually concocted one.

Quoted 8 lines

short messages
Transmit register(s) can be used to pass short messages
from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an interrupt at the receiver side).

Quoted 6 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)
https://www.nxp.com/docs/en/reference-manual/IMX6ULRM.pdf

And SCU firmware does use MU in above way specified by reference manual. Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

#19 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Hi Jassi,

Quoted 52 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 11:44 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 11 lines · A.s. Dong
Quoted 11 lines
Quoted 4 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 7 lines
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

The reality is that MU hardware does not define channels in reference

manual.

Quoted 7 lines

However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each

other.

Quoted 1 line

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

Quoted 7 lines

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

This looks like reasonable to me, theoretically.
Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

And AFAIK ARM MHU is doing the same way as MU which looks like also unidirectional channel.
http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0515b/CHDGBGIF.html
drivers/mailbox/arm_mhu.c

Sascha & Oleksij,
Any idea about this?

Quoted 7 lines

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

This is one of the recommended ways designed in HW reference manual and it allows to send frame information up to 4 words without using shared memory. SCU just follows it, so it's hard to believe it's a bad design.

Quoted 11 lines

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write 4 word data over 4 rx/tx channels, instead of the virtually concocted one.

It may work If SCU protocol data length is fixed to 4 words. However, it's length is flexible for different SVC service. That means this binding won't work for SCU. And it will introduce much complexities during the implementation.

Instead, we're using polling mode for both TX/RX and the data size is stored in the msg header and sending msgs using all 4 data tx/rx registers as a channel fifo.

Quoted 27 lines
Quoted 8 lines

short messages
Transmit register(s) can be used to pass short messages from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an

interrupt at the receiver side).

Quoted 3 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw
ww

Quoted 1 line

.nxp.com%2Fdocs%2Fen%2Freference-

manual%2FIMX6ULRM.pdf&amp;data=02%7C0

Quoted 1 line

1%7Caisheng.dong%40nxp.com%7Cef349d7c9caf46c4260008d5f30e9ef5%7C6 86ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636682166494860332&amp;sdat a=54rE

Quoted 1 line

iDm%2BGD6EY8NE64ck1LXVGoUCtWYrHXPoWqcujUo%3D&amp;reserved=0

Quoted 2 lines

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 8 lines
Quoted 2 lines

Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

First of all, really appreciate for your suggestions.
It may be hard to find a common binding with the same mbox-cells. It looks like we just need a property is distinguish how the Mailbox is used In one channel or multi channel mode.

Directly checking mbox-cells seems the most easy way and it is already Acked by Rob. Can't this way be Okay to you?

Regards
Dong Aisheng

maintainer
#20 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 44 lines · A.s. Dong
Quoted 36 lines
Quoted 12 lines
Quoted 12 lines
Quoted 5 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 7 lines
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

The reality is that MU hardware does not define channels in reference

manual.

Quoted 7 lines

However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each

other.

Quoted 1 line

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

Quoted 7 lines

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

Quoted 5 lines

This looks like reasonable to me, theoretically.
Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

It _is_ necessary to write controller driver independent of client drivers.

Quoted 4 lines

And AFAIK ARM MHU is doing the same way as MU which looks like also unidirectional channel.
http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0515b/CHDGBGIF.html
drivers/mailbox/arm_mhu.c

MHU driver is doing exactly what the data sheet says. I know because I wrote the driver.

Quoted 24 lines
Quoted 7 lines

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

This is one of the recommended ways designed in HW reference manual and it allows to send frame information up to 4 words without using shared memory. SCU just follows it, so it's hard to believe it's a bad design.

Quoted 11 lines

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write 4 word data over 4 rx/tx channels, instead of the virtually concocted one.

Quoted 7 lines

It may work If SCU protocol data length is fixed to 4 words. However, it's length is flexible for different SVC service. That means this binding won't work for SCU. And it will introduce much complexities during the implementation.

Instead, we're using polling mode for both TX/RX and the data size is stored in the msg header and sending msgs using all 4 data tx/rx registers as a channel fifo.

You first give me the "Passing short messages" usecase (quite a bad one) and ask how it can work. When I give you a solution, you effectively say "well, my usecase isn't even that". I feel I just wasted my time.

Quoted 33 lines
Quoted 33 lines
Quoted 8 lines

short messages
Transmit register(s) can be used to pass short messages from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an

interrupt at the receiver side).

Quoted 3 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw
ww

Quoted 1 line

.nxp.com%2Fdocs%2Fen%2Freference-

manual%2FIMX6ULRM.pdf&amp;data=02%7C0

Quoted 1 line

1%7Caisheng.dong%40nxp.com%7Cef349d7c9caf46c4260008d5f30e9ef5%7C6 86ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636682166494860332&amp;sdat a=54rE

Quoted 1 line

iDm%2BGD6EY8NE64ck1LXVGoUCtWYrHXPoWqcujUo%3D&amp;reserved=0

Quoted 2 lines

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 2 lines

Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

Quoted 8 lines
Quoted 2 lines

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

First of all, really appreciate for your suggestions.
It may be hard to find a common binding with the same mbox-cells. It looks like we just need a property is distinguish how the Mailbox is used In one channel or multi channel mode.

I get the idea you were ready to see the code merged in the coming window and be done with it. And now it feels lazy to change the code. I am sorry, but I can't allow controller drivers "emulating" some mode for a client driver. That is moving a part of client code into the controller driver.

Quoted 3 lines

Directly checking mbox-cells seems the most easy way and it is already Acked by Rob. Can't this way be Okay to you?

Rob is indeed far better than I am. But he also has to look into 100 other subsystems, whereas I am only looking into mailbox. I have time to dig into your datasheets. I believe Rob will point out anything wrong I suggest.

BTW, the submitted driver doesn't even support the SCU usecase. Why the bindings?

#21 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 62 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Friday, July 27, 2018 12:56 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 46 lines · A.s. Dong
Quoted 45 lines
Quoted 12 lines
Quoted 12 lines
Quoted 5 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 8 lines
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

The reality is that MU hardware does not define channels in reference

manual.

Quoted 8 lines

However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each

other.

Quoted 1 line

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

Quoted 7 lines
Quoted 5 lines

This looks like reasonable to me, theoretically.
Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

It _is_ necessary to write controller driver independent of client drivers.

Yes, that's true. What if we think we're writing driver against HW capabilities which support 4 pair of channel links(tx/rx)?
It looks like independent of client drivers and may make life easily. Does it make sense?

Quoted 15 lines
Quoted 2 lines

And AFAIK ARM MHU is doing the same way as MU which looks like also

unidirectional channel.

Quoted 1 line

https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finfo

Quoted 1 line

center.arm.com%2Fhelp%2Findex.jsp%3Ftopic%3D%2Fcom.arm.doc.ddi0515 b%2F

Quoted 1 line

CHDGBGIF.html&amp;data=02%7C01%7Caisheng.dong%40nxp.com%7Cb728 16362983

Quoted 1 line

4208f9e908d5f37d3e00%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0% 7C6366

Quoted 1 line
Quoted 8 lines

82641593785009&amp;sdata=VsG0oXxEObPRwT5KVss2eZthSTMTR2%2BMrv PqhDUpPGU

Quoted 3 lines

%3D&amp;reserved=0
drivers/mailbox/arm_mhu.c

MHU driver is doing exactly what the data sheet says. I know because I wrote the driver.

Hmm... Maybe I missed something, but seems no from what I see:
http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0922g/CCHHGIAH.html

Let's see the data sheet:
2.1.1. Physical channels
The CSS MHU peripheral provides physical channels that are used for communication between the SCP and the AP. These channels are called physical channels because their implementation is fixed in hardware. Each physical channel is unidirectional (SCP to AP or AP to SCP).

A physical channel comprises:
A 32-bit STAT (STATUS) register:
A 32-bit SET register:
A 32-bit CLEAR register:

The driver seems composes them into a pair of links with both tx and rx physical channel. drivers/mailbox/arm_mhu.c
#define MHU_CHANS 3

struct mhu_link {
        unsigned irq;
        void __iomem *tx_reg;
        void __iomem *rx_reg;
};
for (i = 0; i < MHU_CHANS; i++) {
        mhu->chan[i].con_priv = &mhu->mlink[i];
        mhu->mlink[i].irq = adev->irq[i];
        mhu->mlink[i].rx_reg = mhu->base + mhu_reg[i];
        mhu->mlink[i].tx_reg = mhu->mlink[i].rx_reg + TX_REG_OFFSET;
}

And the driver totally supports 3 links lp, hp and sec. Each links actually are using two unidirectional physical channels. Am I correct?

Quoted 31 lines
Quoted 11 lines
Quoted 8 lines

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

This is one of the recommended ways designed in HW reference manual and it allows to send frame information up to 4 words without using shared

memory.

Quoted 8 lines

SCU just follows it, so it's hard to believe it's a bad design.

Quoted 6 lines

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Quoted 8 lines
Quoted 5 lines

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write 4 word data over 4 rx/tx channels, instead of the virtually concocted one.

It may work If SCU protocol data length is fixed to 4 words. However, it's length is flexible for different SVC service. That means this binding

won't work for SCU.

Quoted 2 lines

And it will introduce much complexities during the implementation.

Quoted 8 lines
Quoted 2 lines

Instead, we're using polling mode for both TX/RX and the data size is stored in the msg header and sending msgs using all 4 data tx/rx registers

as a channel fifo.

Quoted 1 line

You first give me the "Passing short messages" usecase (quite a bad one) and ask how it can work. When I give you a solution, you effectively say "well, my usecase isn't even that". I feel I just wasted my time.

Sorry for may not clear, "Passing short message' usecase is to tell how the HW is working on one channel mode sending up to 4 words in one time As specified in reference manual.

SCU does work that way, the only difference is it's using polling mode rather than interrupt driven. The point is the data size may be different for each msg, so we can't simply know which data register interrupt should be enable from static data defined in device tree.

Quoted 42 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

short messages
Transmit register(s) can be used to pass short messages from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an

Quoted 5 lines
Quoted 5 lines

interrupt at the receiver side).

Quoted 3 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw

Quoted 5 lines
Quoted 5 lines

ww

Quoted 1 line

.nxp.com%2Fdocs%2Fen%2Freference-

manual%2FIMX6ULRM.pdf&amp;data=02%7C0

Quoted 1 line

1%7Caisheng.dong%40nxp.com%7Cef349d7c9caf46c4260008d5f30e9ef5%7C6

Quoted 3 lines
Quoted 3 lines

86ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636682166494860332&amp;sdat

Quoted 3 lines
Quoted 3 lines

a=54rE

Quoted 1 line

iDm%2BGD6EY8NE64ck1LXVGoUCtWYrHXPoWqcujUo%3D&amp;reserved=0

Quoted 12 lines
Quoted 11 lines
Quoted 2 lines

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 2 lines

Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

Quoted 7 lines
Quoted 5 lines

First of all, really appreciate for your suggestions.
It may be hard to find a common binding with the same mbox-cells. It looks like we just need a property is distinguish how the Mailbox is used In one channel or multi channel mode.

I get the idea you were ready to see the code merged in the coming window and be done with it. And now it feels lazy to change the code.

For me, I'm glad to change if we have a clear better solution. And I do appreciate your suggestion and review time.

Why I'm a bit hesitate now is because your suggestion may not work for SCU, (see above explanation), but it does work for generic M4 case. But we' re now going to support both protocol in one mailbox driver. Any suggestion on how to treat them properly if change the binding?

Quoted 3 lines

I am sorry, but I can't allow controller drivers "emulating" some mode for a client driver. That is moving a part of client code into the controller driver.

Okay, let's figure out it first.
Would you be more specific on what "emulating" did you mean in controller driver? Sending up to 4 words capability in one channel mode as specified in reference manual? That's I'm a bit confusing because I thought it's HW capability independent of client driver.

Or anything else?

Quoted 8 lines
Quoted 4 lines

Directly checking mbox-cells seems the most easy way and it is already Acked by Rob. Can't this way be Okay to you?

Rob is indeed far better than I am. But he also has to look into 100 other subsystems, whereas I am only looking into mailbox. I have time to dig into your datasheets. I believe Rob will point out anything wrong I suggest.

Yes, you're in the fair enough authority to judge it. Thanks for your effort.

Quoted 2 lines

BTW, the submitted driver doesn't even support the SCU usecase. Why the bindings?

Because that patch is firstly Acked by Rob. Others are reworked and ready to be sent out against this patch series. But it seems we still have unresolved issues now as you pointed out. We can first resolve them. Or do you need me to send out for your reference?

Regards
Dong Aisheng

maintainer
#22 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 67 lines · A.s. Dong
Quoted 67 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Friday, July 27, 2018 12:56 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 51 lines · A.s. Dong
Quoted 38 lines
Quoted 12 lines
Quoted 12 lines
Quoted 5 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

(quoted diff) +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 8 lines
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

The reality is that MU hardware does not define channels in reference

manual.

Quoted 8 lines

However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each

other.

Quoted 1 line

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

Quoted 7 lines

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

This looks like reasonable to me, theoretically.
Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

Quoted 8 lines
Quoted 2 lines

It _is_ necessary to write controller driver independent of client drivers.

Yes, that's true. What if we think we're writing driver against HW capabilities which support 4 pair of channel links(tx/rx)?
It looks like independent of client drivers and may make life easily. Does it make sense?

No, that would be emulation.
Why doesn't a usb device controller (udc) driver emulate FSG/HID etc, by "thinking" it has a hardware backed storage/keyboard? It doesn't because that is the job of upper protocol layer.

Quoted 20 lines
Quoted 20 lines
Quoted 2 lines

And AFAIK ARM MHU is doing the same way as MU which looks like also

unidirectional channel.

Quoted 1 line

https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finfo

Quoted 1 line

center.arm.com%2Fhelp%2Findex.jsp%3Ftopic%3D%2Fcom.arm.doc.ddi0515 b%2F

Quoted 1 line

CHDGBGIF.html&amp;data=02%7C01%7Caisheng.dong%40nxp.com%7Cb728 16362983

Quoted 1 line

4208f9e908d5f37d3e00%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0% 7C6366

Quoted 1 line

82641593785009&amp;sdata=VsG0oXxEObPRwT5KVss2eZthSTMTR2%2BMrv PqhDUpPGU

Quoted 3 lines

%3D&amp;reserved=0
drivers/mailbox/arm_mhu.c

Quoted 7 lines
Quoted 3 lines

MHU driver is doing exactly what the data sheet says. I know because I wrote the driver.

Hmm... Maybe I missed something, but seems no from what I see:
http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0922g/CCHHGIAH.html

That is not the MHU datasheet.

Read Section 3.6 at page-3-41 of
http://infocenter.arm.com/help/topic/com.arm.doc.ddi0515b/DDI0515B_juno_arm_development_platform_soc_trm.pdf

This is another reason I can't let a bad code (emulation code) through, because people start looking for examples to justify their implementation rather than fixing it.

Quoted 40 lines
Quoted 31 lines
Quoted 11 lines
Quoted 8 lines

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

This is one of the recommended ways designed in HW reference manual and it allows to send frame information up to 4 words without using shared

memory.

Quoted 8 lines

SCU just follows it, so it's hard to believe it's a bad design.

Quoted 6 lines

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Quoted 8 lines
Quoted 5 lines

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write 4 word data over 4 rx/tx channels, instead of the virtually concocted one.

It may work If SCU protocol data length is fixed to 4 words. However, it's length is flexible for different SVC service. That means this binding

won't work for SCU.

Quoted 2 lines

And it will introduce much complexities during the implementation.

Quoted 8 lines
Quoted 2 lines

Instead, we're using polling mode for both TX/RX and the data size is stored in the msg header and sending msgs using all 4 data tx/rx registers

as a channel fifo.

Quoted 1 line

You first give me the "Passing short messages" usecase (quite a bad one) and ask how it can work. When I give you a solution, you effectively say "well, my usecase isn't even that". I feel I just wasted my time.

Quoted 9 lines

Sorry for may not clear, "Passing short message' usecase is to tell how the HW is working on one channel mode sending up to 4 words in one time As specified in reference manual.

SCU does work that way, the only difference is it's using polling mode rather than interrupt driven. The point is the data size may be different for each msg, so we can't simply know which data register interrupt should be enable from static data defined in device tree.

And you think passing variable data through registers is a better idea than passing packets via shared-memory?

Quoted 50 lines
Quoted 42 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

short messages
Transmit register(s) can be used to pass short messages from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an

Quoted 5 lines
Quoted 5 lines

interrupt at the receiver side).

Quoted 3 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw

Quoted 5 lines
Quoted 5 lines

ww

Quoted 1 line

.nxp.com%2Fdocs%2Fen%2Freference-

manual%2FIMX6ULRM.pdf&amp;data=02%7C0

Quoted 1 line

1%7Caisheng.dong%40nxp.com%7Cef349d7c9caf46c4260008d5f30e9ef5%7C6

Quoted 3 lines
Quoted 3 lines

86ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636682166494860332&amp;sdat

Quoted 3 lines
Quoted 3 lines

a=54rE

Quoted 1 line

iDm%2BGD6EY8NE64ck1LXVGoUCtWYrHXPoWqcujUo%3D&amp;reserved=0

Quoted 12 lines
Quoted 11 lines
Quoted 2 lines

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 2 lines

Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

Quoted 7 lines
Quoted 5 lines

First of all, really appreciate for your suggestions.
It may be hard to find a common binding with the same mbox-cells. It looks like we just need a property is distinguish how the Mailbox is used In one channel or multi channel mode.

I get the idea you were ready to see the code merged in the coming window and be done with it. And now it feels lazy to change the code.

Quoted 8 lines

For me, I'm glad to change if we have a clear better solution. And I do appreciate your suggestion and review time.

Why I'm a bit hesitate now is because your suggestion may not work for SCU, (see above explanation), but it does work for generic M4 case. But we' re now going to support both protocol in one mailbox driver. Any suggestion on how to treat them properly if change the binding?

In your last post you said "This looks like reasonable to me, theoretically" My suggestion is the same because I don't see why it won't work.

Quoted 11 lines
Quoted 3 lines

I am sorry, but I can't allow controller drivers "emulating" some mode for a client driver. That is moving a part of client code into the controller driver.

Okay, let's figure out it first.
Would you be more specific on what "emulating" did you mean in controller driver? Sending up to 4 words capability in one channel mode as specified in reference manual? That's I'm a bit confusing because I thought it's HW capability independent of client driver.

Or anything else?

Emulation means pretending something that we are not.

The hardware has 8 unidirectional channels. But your protocol (SCU implementation) assumes there is one _virtual_ channel that has 4 registers and 1/0 irq --- which is not true. Instead of fixing the assumption in your protocol or emulating the virtual channel in the protocol level (user of a MU), you want to add code in the controller driver that ignores interrupts and club the 4 independent channels together.

There is no end to protocols and their kinky assumptions, adding "xyz-mode" support for each protocol isn't scalable.

Quoted 9 lines
Quoted 8 lines
Quoted 4 lines

Directly checking mbox-cells seems the most easy way and it is already Acked by Rob. Can't this way be Okay to you?

Rob is indeed far better than I am. But he also has to look into 100 other subsystems, whereas I am only looking into mailbox. I have time to dig into your datasheets. I believe Rob will point out anything wrong I suggest.

Quoted 9 lines

Yes, you're in the fair enough authority to judge it. Thanks for your effort.

Quoted 2 lines

BTW, the submitted driver doesn't even support the SCU usecase. Why the bindings?

Because that patch is firstly Acked by Rob. Others are reworked and ready to be sent out against this patch series. But it seems we still have unresolved issues now as you pointed out. We can first resolve them. Or do you need me to send out for your reference?

I am sure Rob took the best decision at the time with whatever information provided to him.
Now, after reading the datasheet, we have the option to avoid implementing consumer code in the provider driver.

Thanks.

#23 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 93 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Friday, July 27, 2018 2:47 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 35 lines · A.s. Dong
Quoted 35 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Friday, July 27, 2018 12:56 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; ,
linux-arm-kernel at lists.infradead.org,
linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 18 lines · A.s. Dong
Quoted 18 lines
Quoted 18 lines
Quoted 17 lines
Quoted 5 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Review

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

(quoted diff) +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

The reality is that MU hardware does not define channels in reference

manual.

Quoted 2 lines

However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current

driver.

Quoted 28 lines
Quoted 28 lines
Quoted 28 lines
Quoted 15 lines
Quoted 6 lines

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each

other.

Quoted 1 line

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

Quoted 7 lines

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

This looks like reasonable to me, theoretically.
Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

Quoted 8 lines
Quoted 2 lines

It _is_ necessary to write controller driver independent of client drivers.

Yes, that's true. What if we think we're writing driver against HW capabilities which support 4 pair of channel links(tx/rx)? It looks like independent of client drivers and may make life easily. Does it make sense?

Quoted 5 lines

No, that would be emulation.
Why doesn't a usb device controller (udc) driver emulate FSG/HID etc, by "thinking" it has a hardware backed storage/keyboard? It doesn't because that is the job of upper protocol layer.

Sorry I'm not quite familiar with USB device.
IMHO the HW supports many capabilities, but it does not mean
we need support them all. For MU case, a pair of channel link capable of both tx/rx seems like a better using. It's irrelevant of client drivers. It's simply that HW supports different kind of modes (one channel mode, one link mode, separate per register mode) and we just support link mode.

Another reason is I doubt that we may never use per register mode in a different register pair in the future. For example:
AP:

node {
	...
	// cell 0 meaning: 0: tx 1: rx	cell1 meaning: channel id
	Mboxes = <&mbox 0 1 &mbox 1 2>
	Mbox-names = "tx", "rx";
Quoted 1 line

M4:

node {
	...
	Mboxes = <&mbox 0 2 &mbox 1 1>
	Mbox-names = "tx", "rx";
Quoted 1 line

This make things complicated and error prone as I said before.

But that's just my understanding and may overlook something, if you still think we should do exactly as above, I will not against it because it does work for M4 case.

Then the left question is how we handle SCU case?

Quoted 58 lines
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

And AFAIK ARM MHU is doing the same way as MU which looks like also

unidirectional channel.

Quoted 1 line

https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finf

Quoted 3 lines
Quoted 3 lines

o

Quoted 1 line

center.arm.com%2Fhelp%2Findex.jsp%3Ftopic%3D%2Fcom.arm.doc.ddi0515

Quoted 3 lines
Quoted 3 lines

b%2F

Quoted 1 line

CHDGBGIF.html&amp;data=02%7C01%7Caisheng.dong%40nxp.com%7Cb728

Quoted 3 lines
Quoted 3 lines

16362983

Quoted 1 line

4208f9e908d5f37d3e00%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%

Quoted 3 lines
Quoted 3 lines

7C6366

Quoted 1 line

82641593785009&amp;sdata=VsG0oXxEObPRwT5KVss2eZthSTMTR2%2BMrv

Quoted 10 lines
Quoted 7 lines

PqhDUpPGU

Quoted 3 lines

%3D&amp;reserved=0
drivers/mailbox/arm_mhu.c

MHU driver is doing exactly what the data sheet says. I know because I wrote the driver.

Hmm... Maybe I missed something, but seems no from what I see:

https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finfo

Quoted 1 line

center.arm.com%2Fhelp%2Findex.jsp%3Ftopic%3D%2Fcom.arm.doc.dui0922 g%2F

Quoted 1 line

CCHHGIAH.html&amp;data=02%7C01%7Caisheng.dong%40nxp.com%7Cd2b1 23b89dae

Quoted 1 line

4a00cb7108d5f38cb611%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C1% 7C6366

Quoted 1 line

82708036553137&amp;sdata=4SMUpH%2FO9MWArC%2BjbPy%2BrbNAqUla o9IezKUi7UX

Quoted 2 lines

gIyQ%3D&amp;reserved=0

That is not the MHU datasheet.

Read Section 3.6 at page-3-41 of
https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finfo
center.arm.com%2Fhelp%2Ftopic%2Fcom.arm.doc.ddi0515b%2FDDI0515B_j uno_arm_development_platform_soc_trm.pdf&amp;data=02%7C01%7Cais heng.dong%40nxp.com%7Cd2b123b89dae4a00cb7108d5f38cb611%7C686ea1 d3bc2b4c6fa92cd99c5c301635%7C0%7C1%7C636682708036553137&amp;sdat a=eUd75VnwEs44ZoVV0n7R1G3UfAH%2B%2BpeoGns7Mq5UBN8%3D&amp; reserved=0

Quoted 4 lines

This is another reason I can't let a bad code (emulation code) through, because people start looking for examples to justify their implementation rather than fixing it.

I'm a bit confusing....
The section 3.6 you pointed is the MHU register description. It does not conflict with what I see from ARM doc center that each physical channel is unidirectional.

See below:
Chan 1:
0x000 SCP_INTR_L_STAT
0x008 SCP_INTR_L_SET
0x010 SCP _INTR_L_CLEAR

Chan 2:
0x020 SCP_INTR_H_STAT
0x028 SCP_INTR_H_SET
0x030 SCP _INTR_H_CLEAR

Chan 3:
0x100 CPU_INTR_L_STAT
0x108 CPU_INTR_L_SET
0x110 CPU_INTR_L_CLEAR

Chan 4:
0x120 CPU_INTR_H_STAT
0x128 CPU_INTR_H_SET
0x130 CPU_INTR_H_CLEAR

Chan 5:
0x200 SCP_INTR_S_STAT
0x208 SCP_INTR_S_SET
0x210 SCP _INTR_S_CLEAR

Chan 6:
0x300 CPU_INTR_S_STAT
0x308 CPU_INTR_S_SET
0x310 CPU_INTR_S_CLEAR

And the driver compose them into 3 channel links (lp, hp and sec). Am I wrong?

Quoted 52 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 9 lines

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

This is one of the recommended ways designed in HW reference

manual

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

and it allows to send frame information up to 4 words without using shared

memory.

Quoted 9 lines

SCU just follows it, so it's hard to believe it's a bad design.

Quoted 7 lines

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an
IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write 4 word data over 4 rx/tx channels, instead of the virtually concocted

one.

Quoted 8 lines
Quoted 8 lines
Quoted 5 lines
Quoted 1 line

It may work If SCU protocol data length is fixed to 4 words. However, it's length is flexible for different SVC service. That means this binding

won't work for SCU.

Quoted 2 lines

And it will introduce much complexities during the implementation.

Quoted 8 lines
Quoted 8 lines
Quoted 3 lines

Instead, we're using polling mode for both TX/RX and the data size is stored in the msg header and sending msgs using all 4 data tx/rx registers

as a channel fifo.

Quoted 1 line

You first give me the "Passing short messages" usecase (quite a bad one) and ask how it can work. When I give you a solution, you effectively say "well, my usecase isn't even that". I feel I just wasted my

time.

Quoted 6 lines
Quoted 1 line

Sorry for may not clear, "Passing short message' usecase is to tell how the HW is working on one channel mode sending up to 4 words in one time As specified in reference manual.

Quoted 8 lines
Quoted 5 lines

SCU does work that way, the only difference is it's using polling mode rather than interrupt driven. The point is the data size may be different for each msg, so we can't simply know which data register interrupt should be enable from static data defined in device tree.

And you think passing variable data through registers is a better idea than passing packets via shared-memory?

You got me. :-)
I've no idea about which one is better. The problem is SCU firmware is already there passing packets through data registers, we have no way to change it.

Quoted 58 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines

short messages
Transmit register(s) can be used to pass short messages from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines

interrupt at the receiver side).

Quoted 3 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines

ww

Quoted 1 line

.nxp.com%2Fdocs%2Fen%2Freference-

manual%2FIMX6ULRM.pdf&amp;data=02%7C0

Quoted 1 line

1%7Caisheng.dong%40nxp.com%7Cef349d7c9caf46c4260008d5f30e9ef5%7C6

Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

86ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636682166494860332&amp;sdat

Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

a=54rE

Quoted 1 line

iDm%2BGD6EY8NE64ck1LXVGoUCtWYrHXPoWqcujUo%3D&amp;reserved=0

Quoted 23 lines
Quoted 12 lines
Quoted 12 lines
Quoted 11 lines
Quoted 2 lines

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 2 lines

Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

Quoted 7 lines
Quoted 5 lines

First of all, really appreciate for your suggestions. It may be hard to find a common binding with the same mbox-cells. It looks like we just need a property is distinguish how the Mailbox is used In one channel or multi channel mode.

I get the idea you were ready to see the code merged in the coming window and be done with it. And now it feels lazy to change the code.

For me, I'm glad to change if we have a clear better solution. And I do appreciate your suggestion and review time.

Quoted 8 lines
Quoted 5 lines

Why I'm a bit hesitate now is because your suggestion may not work for SCU, (see above explanation), but it does work for generic M4 case. But we' re now going to support both protocol in one mailbox driver. Any suggestion on how to treat them properly if change the binding?

In your last post you said "This looks like reasonable to me, theoretically" My suggestion is the same because I don't see why it won't work.

Sorry for not clear. I mean it's reasonable for M4 generic using case. But not for SCU case.

Quoted 16 lines
Quoted 3 lines
Quoted 3 lines

I am sorry, but I can't allow controller drivers "emulating" some mode for a client driver. That is moving a part of client code into the

controller driver.

Quoted 4 lines
Quoted 1 line

Okay, let's figure out it first.
Would you be more specific on what "emulating" did you mean in

controller driver?

Quoted 1 line

Sending up to 4 words capability in one channel mode as specified in

reference manual?

Quoted 1 line

That's I'm a bit confusing because I thought it's HW capability independent

of client driver.

Quoted 3 lines

Or anything else?

Quoted 9 lines

Emulation means pretending something that we are not.

The hardware has 8 unidirectional channels. But your protocol (SCU implementation) assumes there is one _virtual_ channel that has 4 registers and 1/0 irq --- which is not true. Instead of fixing the assumption in your protocol or emulating the virtual channel in the protocol level (user of a MU), you want to add code in the controller driver that ignores interrupts and club the 4 independent channels together.

This stucks me. This is how the hardware is designed and suggested to use in hardware reference manual. And now you're telling me this is wrong and we should not use the design in reference manual...

Quoted 3 lines

There is no end to protocols and their kinky assumptions, adding "xyz-mode" support for each protocol isn't scalable.

This is also how Sascha suggested me to move to mailbox to support both protocols and handle them differently in mailbox driver.

Honestly I was hesitated to do that before because I doubt the value we can gain if switching to mailbox besides the unnecessary complexity introduced and performance downgrading (extra execution of a few unnecessary code in mailbox API and It's even worse if use 8 separate channels for SCU, comparing to original only tens of lines of library API way), but Sascha insisted...

The mailbox itself is somehow protocol specific (doorbell, signal, data packet and etc) It's hard for us to find a common way to support two totally different protocols. So the controller needs to know different protocol it is transferring.

Quoted 19 lines
Quoted 8 lines
Quoted 8 lines
Quoted 5 lines

Directly checking mbox-cells seems the most easy way and it is already Acked by Rob. Can't this way be Okay to you?

Rob is indeed far better than I am. But he also has to look into 100 other subsystems, whereas I am only looking into mailbox. I have time to dig into your datasheets. I believe Rob will point out anything wrong I

suggest.

Quoted 10 lines
Quoted 1 line

Yes, you're in the fair enough authority to judge it. Thanks for your effort.

Quoted 2 lines

BTW, the submitted driver doesn't even support the SCU usecase. Why the bindings?

Because that patch is firstly Acked by Rob. Others are reworked and ready to be sent out against this patch series. But it seems we still have unresolved issues now as you pointed out. We can first resolve them.

Quoted 6 lines

Or do you need me to send out for your reference?

Quoted 1 line

I am sure Rob took the best decision at the time with whatever information provided to him.
Now, after reading the datasheet, we have the option to avoid implementing consumer code in the provider driver.

I have one doubt, irrelevant of SCU protocol, from the datasheet, the hardware does support transfer message up to 4 words in one channel mode and it is the recommended way in datasheet for less than 4 words frame transferring, why switching to mailbox framework, we can't use it now?
And makes us lose the HW capability.

If that, what's meaning for us to switch to mailbox framework?

Regards
Dong Aisheng

Quoted 2 lines

Thanks.

#24 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 493 lines

-----Original Message-----
From: A.s. Dong
Sent: Friday, July 27, 2018 4:42 PM
To: Jassi Brar <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: RE: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 16 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Friday, July 27, 2018 2:47 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; ,
linux-arm-kernel at lists.infradead.org,
linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 8 lines
Quoted 7 lines · A.s. Dong
Quoted 7 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Friday, July 27, 2018 12:56 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring

<[email protected]>;

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; ,
linux-arm-kernel at lists.infradead.org,
linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

On Fri, Jul 27, 2018 at 9:32 AM, A.s. Dong <[email protected]>

Quoted 10 lines · wrote:
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 5 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Quoted 3 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox

+- #mbox-cells:  Must be:
+               0 - for single channel mode. i.MX8* SCU protocol

specific.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+               1 - for multichannel (generic) mode.
+

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in. Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

Quoted 28 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

The reality is that MU hardware does not define channels in reference

manual.

Quoted 3 lines

However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current

driver.

Quoted 21 lines
Quoted 21 lines
Quoted 21 lines
Quoted 21 lines
Quoted 6 lines

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each

other.

Quoted 1 line

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the

controller is!

Quoted 15 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 1 line

This looks like reasonable to me, theoretically. Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

Quoted 8 lines
Quoted 2 lines

It _is_ necessary to write controller driver independent of client drivers.

Yes, that's true. What if we think we're writing driver against HW capabilities which support 4 pair of channel links(tx/rx)? It looks like independent of client drivers and may make life easily. Does it make sense?

Quoted 5 lines

No, that would be emulation.
Why doesn't a usb device controller (udc) driver emulate FSG/HID etc, by "thinking" it has a hardware backed storage/keyboard? It doesn't because that is the job of upper protocol layer.

Sorry I'm not quite familiar with USB device.
IMHO the HW supports many capabilities, but it does not mean we need support them all. For MU case, a pair of channel link capable of both tx/rx seems like a better using. It's irrelevant of client drivers. It's simply that HW supports different kind of modes (one channel mode, one link mode, separate per register mode) and we just support link mode.

Another reason is I doubt that we may never use per register mode in a different register pair in the future. For example:
AP:

node {
	...
	// cell 0 meaning: 0: tx 1: rx	cell1 meaning: channel id
	Mboxes = <&mbox 0 1 &mbox 1 2>
	Mbox-names = "tx", "rx";
Quoted 1 line

M4:

node {
	...
	Mboxes = <&mbox 0 2 &mbox 1 1>
	Mbox-names = "tx", "rx";
Quoted 1 line

This make things complicated and error prone as I said before.

But that's just my understanding and may overlook something, if you still think we should do exactly as above, I will not against it because it does work for M4 case.

Then the left question is how we handle SCU case?

Quoted 11 lines
Quoted 6 lines
Quoted 6 lines
Quoted 3 lines

And AFAIK ARM MHU is doing the same way as MU which looks like also

unidirectional channel.

Quoted 1 line

https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finf

Quoted 3 lines
Quoted 3 lines

o

Quoted 1 line

center.arm.com%2Fhelp%2Findex.jsp%3Ftopic%3D%2Fcom.arm.doc.ddi0515

Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

b%2F

Quoted 1 line

CHDGBGIF.html&amp;data=02%7C01%7Caisheng.dong%40nxp.com%7Cb728

Quoted 8 lines
Quoted 3 lines
Quoted 3 lines

16362983

Quoted 1 line

4208f9e908d5f37d3e00%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%

Quoted 3 lines
Quoted 3 lines

7C6366

Quoted 1 line

82641593785009&amp;sdata=VsG0oXxEObPRwT5KVss2eZthSTMTR2%2BMrv

Quoted 11 lines
Quoted 10 lines
Quoted 7 lines

PqhDUpPGU

Quoted 3 lines

%3D&amp;reserved=0
drivers/mailbox/arm_mhu.c

MHU driver is doing exactly what the data sheet says. I know because I wrote the driver.

Hmm... Maybe I missed something, but seems no from what I see:

https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finfo

Quoted 2 lines
Quoted 1 line

center.arm.com%2Fhelp%2Findex.jsp%3Ftopic%3D%2Fcom.arm.doc.dui0922

Quoted 3 lines

g%2F

Quoted 1 line

CCHHGIAH.html&amp;data=02%7C01%7Caisheng.dong%40nxp.com%7Cd2b1

Quoted 6 lines

23b89dae

Quoted 1 line

4a00cb7108d5f38cb611%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C1% 7C6366

Quoted 1 line

82708036553137&amp;sdata=4SMUpH%2FO9MWArC%2BjbPy%2BrbNAqUla

Quoted 7 lines

o9IezKUi7UX

Quoted 2 lines

gIyQ%3D&amp;reserved=0

That is not the MHU datasheet.

Read Section 3.6 at page-3-41 of

https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Finfo

Quoted 1 line

center.arm.com%2Fhelp%2Ftopic%2Fcom.arm.doc.ddi0515b%2FDDI0515B_j

Quoted 1 line

uno_arm_development_platform_soc_trm.pdf&amp;data=02%7C01%7Cais

Quoted 1 line

heng.dong%40nxp.com%7Cd2b123b89dae4a00cb7108d5f38cb611%7C686ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C1%7C636682708036553137&amp;sdat

Quoted 1 line

a=eUd75VnwEs44ZoVV0n7R1G3UfAH%2B%2BpeoGns7Mq5UBN8%3D&amp;

Quoted 6 lines

reserved=0

This is another reason I can't let a bad code (emulation code) through, because people start looking for examples to justify their implementation rather than fixing it.

I'm a bit confusing....
The section 3.6 you pointed is the MHU register description. It does not conflict with what I see from ARM doc center that each physical channel is unidirectional.

See below:
Chan 1:
0x000 SCP_INTR_L_STAT
0x008 SCP_INTR_L_SET
0x010 SCP _INTR_L_CLEAR

Chan 2:
0x020 SCP_INTR_H_STAT
0x028 SCP_INTR_H_SET
0x030 SCP _INTR_H_CLEAR

Chan 3:
0x100 CPU_INTR_L_STAT
0x108 CPU_INTR_L_SET
0x110 CPU_INTR_L_CLEAR

Chan 4:
0x120 CPU_INTR_H_STAT
0x128 CPU_INTR_H_SET
0x130 CPU_INTR_H_CLEAR

Chan 5:
0x200 SCP_INTR_S_STAT
0x208 SCP_INTR_S_SET
0x210 SCP _INTR_S_CLEAR

Chan 6:
0x300 CPU_INTR_S_STAT
0x308 CPU_INTR_S_SET
0x310 CPU_INTR_S_CLEAR

And the driver compose them into 3 channel links (lp, hp and sec). Am I wrong?

Quoted 55 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 9 lines

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

This is one of the recommended ways designed in HW reference

manual

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

and it allows to send frame information up to 4 words without using shared

memory.

Quoted 9 lines

SCU just follows it, so it's hard to believe it's a bad design.

Quoted 7 lines

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an
IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write
4 word data over 4 rx/tx channels, instead of the virtually concocted

one.

Quoted 11 lines
Quoted 11 lines
Quoted 5 lines
Quoted 1 line

It may work If SCU protocol data length is fixed to 4 words. However, it's length is flexible for different SVC service. That means this binding

won't work for SCU.

Quoted 5 lines

And it will introduce much complexities during the implementation.

Instead, we're using polling mode for both TX/RX and the data size is stored in the msg header and sending msgs using all 4 data tx/rx registers

Quoted 6 lines
Quoted 6 lines

as a channel fifo.

Quoted 1 line

You first give me the "Passing short messages" usecase (quite a bad one) and ask how it can work. When I give you a solution, you effectively say "well, my usecase isn't even that". I feel I just wasted my

time.

Quoted 6 lines
Quoted 1 line

Sorry for may not clear, "Passing short message' usecase is to tell how the HW is working on one channel mode sending up to 4 words in one time As specified in reference manual.

Quoted 8 lines
Quoted 5 lines

SCU does work that way, the only difference is it's using polling mode rather than interrupt driven. The point is the data size may be different for each msg, so we can't simply know which data register interrupt should be enable from static data defined in device tree.

And you think passing variable data through registers is a better idea than passing packets via shared-memory?

You got me. :-)
I've no idea about which one is better. The problem is SCU firmware is already there passing packets through data registers, we have no way to change it.

Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines

short messages
Transmit register(s) can be used to pass short messages from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an

Quoted 8 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines

interrupt at the receiver side).

Quoted 4 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw

Quoted 7 lines
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines

ww

Quoted 1 line

.nxp.com%2Fdocs%2Fen%2Freference-

manual%2FIMX6ULRM.pdf&amp;data=02%7C0

Quoted 1 line

1%7Caisheng.dong%40nxp.com%7Cef349d7c9caf46c4260008d5f30e9ef5%7C6

Quoted 5 lines
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

86ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636682166494860332&amp;sdat

Quoted 5 lines
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

a=54rE

Quoted 1 line

iDm%2BGD6EY8NE64ck1LXVGoUCtWYrHXPoWqcujUo%3D&amp;reserved=0

Quoted 24 lines
Quoted 24 lines
Quoted 13 lines
Quoted 13 lines
Quoted 12 lines
Quoted 3 lines

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 2 lines

Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

Quoted 7 lines
Quoted 5 lines

First of all, really appreciate for your suggestions. It may be hard to find a common binding with the same mbox-cells. It looks like we just need a property is distinguish how the Mailbox is used In one channel or multi channel mode.

I get the idea you were ready to see the code merged in the coming window and be done with it. And now it feels lazy to change the code.

For me, I'm glad to change if we have a clear better solution. And I do appreciate your suggestion and review time.

Quoted 8 lines
Quoted 5 lines

Why I'm a bit hesitate now is because your suggestion may not work for SCU, (see above explanation), but it does work for generic M4 case. But we' re now going to support both protocol in one mailbox driver. Any suggestion on how to treat them properly if change the binding?

In your last post you said "This looks like reasonable to me, theoretically" My suggestion is the same because I don't see why it won't work.

Sorry for not clear. I mean it's reasonable for M4 generic using case. But not for SCU case.

Quoted 18 lines
Quoted 4 lines
Quoted 4 lines

I am sorry, but I can't allow controller drivers "emulating" some mode for a client driver. That is moving a part of client code into the

controller driver.

Quoted 4 lines
Quoted 1 line

Okay, let's figure out it first.
Would you be more specific on what "emulating" did you mean in

controller driver?

Quoted 1 line

Sending up to 4 words capability in one channel mode as specified in

reference manual?

Quoted 2 lines

That's I'm a bit confusing because I thought it's HW capability independent

of client driver.

Quoted 3 lines

Or anything else?

Quoted 8 lines

Emulation means pretending something that we are not.

The hardware has 8 unidirectional channels. But your protocol (SCU implementation) assumes there is one _virtual_ channel that has 4 registers and 1/0 irq --- which is not true. Instead of fixing the assumption in your protocol or emulating the virtual channel in the protocol level (user of a MU), you want to add code in the controller driver that ignores interrupts and club the 4 independent channels

together.

Quoted 1 line

This stucks me. This is how the hardware is designed and suggested to use in hardware reference manual. And now you're telling me this is wrong and we should not use the design in reference manual...

Quoted 1 line

There is no end to protocols and their kinky assumptions, adding "xyz-

mode"

Quoted 2 lines

support for each protocol isn't scalable.

This is also how Sascha suggested me to move to mailbox to support both protocols and handle them differently in mailbox driver.

Honestly I was hesitated to do that before because I doubt the value we can gain if switching to mailbox besides the unnecessary complexity introduced and performance downgrading (extra execution of a few unnecessary code in mailbox API and It's even worse if use 8 separate channels for SCU, comparing to original only tens of lines of library API way), but Sascha insisted...

The mailbox itself is somehow protocol specific (doorbell, signal, data packet and etc) It's hard for us to find a common way to support two totally different protocols.
So the controller needs to know different protocol it is transferring.

Quoted 20 lines
Quoted 9 lines
Quoted 9 lines
Quoted 5 lines

Directly checking mbox-cells seems the most easy way and it is already Acked by Rob. Can't this way be Okay to you?

Rob is indeed far better than I am. But he also has to look into 100 other subsystems, whereas I am only looking into mailbox. I have time to dig into your datasheets. I believe Rob will point out anything wrong I

suggest.

Quoted 10 lines
Quoted 1 line

Yes, you're in the fair enough authority to judge it. Thanks for your effort.

Quoted 2 lines

BTW, the submitted driver doesn't even support the SCU usecase. Why the bindings?

Because that patch is firstly Acked by Rob. Others are reworked and ready to be sent out against this patch series. But it seems we still have unresolved issues now as you pointed out. We can first resolve

them.

Quoted 6 lines

Or do you need me to send out for your reference?

Quoted 1 line

I am sure Rob took the best decision at the time with whatever information provided to him.
Now, after reading the datasheet, we have the option to avoid implementing consumer code in the provider driver.

Quoted 6 lines

I have one doubt, irrelevant of SCU protocol, from the datasheet, the hardware does support transfer message up to 4 words in one channel mode and it is the recommended way in datasheet for less than 4 words frame transferring, why switching to mailbox framework, we can't use it now? And makes us lose the HW capability.

See TI mailbox does in a quite similar way as SCU.
The difference is TI mailbox use TX/Rx buffer while i.MX MU uses 4 TX/RX regsiters. And TI mailbox support send up to 8 words in TX/RX buffer in one time while i.MX MU supports Send up to 4 words in one time.

truct ti_msgmgr_message {
        size_t len;
        u8 *buf;
};

drivers/mailbox/ti-msgmgr.c: ti_msgmgr_send_data()

Regards
Dong Aisheng

Quoted 7 lines

If that, what's meaning for us to switch to mailbox framework?

Regards
Dong Aisheng

Quoted 2 lines

Thanks.

maintainer
#25 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to A.s. Dong
Quoted 26 lines · A.s. Dong
Quoted 26 lines
Quoted 26 lines
Quoted 13 lines
Quoted 13 lines
Quoted 12 lines

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

Quoted 7 lines
Quoted 5 lines

This looks like reasonable to me, theoretically.
Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

It _is_ necessary to write controller driver independent of client drivers.

Yes, that's true. What if we think we're writing driver against HW capabilities which support 4 pair of channel links(tx/rx)? It looks like independent of client drivers and may make life easily. Does it make sense?

Quoted 8 lines
Quoted 5 lines

No, that would be emulation.
Why doesn't a usb device controller (udc) driver emulate FSG/HID etc, by "thinking" it has a hardware backed storage/keyboard? It doesn't because that is the job of upper protocol layer.

Sorry I'm not quite familiar with USB device.

Which subsystem are you familiar with? Every stack is designed like that -- controller drivers reflect the physical hardware, any emulation/protocol is implemented in the user layer.

Quoted 3 lines

Another reason is I doubt that we may never use per register mode in a different register pair in the future.

You may never use, but if the driver reflects the controller as precisely as it is, linux could support any usecase that some baremetal code could support on your platform.
For example, you could support "normal" and "scu" mode over the 3-cell mechanism I suggested. But your original proposal/assumption of "scu-mode", failed immediately, hence Oleksij had to imlpement the "normal" mode.

Quoted 9 lines

For example:
AP:

node {
        ...
        // cell 0 meaning: 0: tx 1: rx  cell1 meaning: channel id
        Mboxes = <&mbox 0 1 &mbox 1 2>
        Mbox-names = "tx", "rx";
Quoted 1 line
Quoted 8 lines

M4:

node {
        ...
        Mboxes = <&mbox 0 2 &mbox 1 1>
        Mbox-names = "tx", "rx";
Quoted 1 line

This make things complicated and error prone as I said before.

I don't see how.

Quoted 3 lines

But that's just my understanding and may overlook something, if you still think we should do exactly as above, I will not against it because it does work for M4 case.

Cool, done.

Quoted 2 lines

Then the left question is how we handle SCU case?

Please point me to the code that you worry might not work.

Quoted 31 lines

I'm a bit confusing....
The section 3.6 you pointed is the MHU register description. It does not conflict with what I see from ARM doc center that each physical channel is unidirectional.

See below:
Chan 1:
0x000 SCP_INTR_L_STAT
0x008 SCP_INTR_L_SET
0x010 SCP _INTR_L_CLEAR

Chan 2:
0x020 SCP_INTR_H_STAT
0x028 SCP_INTR_H_SET
0x030 SCP _INTR_H_CLEAR

Chan 3:
0x100 CPU_INTR_L_STAT
0x108 CPU_INTR_L_SET
0x110 CPU_INTR_L_CLEAR

Chan 4:
0x120 CPU_INTR_H_STAT
0x128 CPU_INTR_H_SET
0x130 CPU_INTR_H_CLEAR

Chan 5:
0x200 SCP_INTR_S_STAT
0x208 SCP_INTR_S_SET
0x210 SCP _INTR_S_CLEAR

Quoted 8 lines

Chan 6:
0x300 CPU_INTR_S_STAT
0x308 CPU_INTR_S_SET
0x310 CPU_INTR_S_CLEAR

And the driver compose them into 3 channel links (lp, hp and sec). Am I wrong?

The RX and TX are not independent of each other, their priorities are tied together in hardware. So it makes more sense (as compared to declaring them independent) to club them together as one channel. Whatever example you take - MHU or TI - you won't find a driver that implements a virtual mode like "scu-mode".

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

Sorry for may not clear, "Passing short message' usecase is to tell how the HW is working on one channel mode sending up to 4 words in one time As specified in reference manual.

SCU does work that way, the only difference is it's using polling mode rather than interrupt driven. The point is the data size may be different for each msg, so we can't simply know which data register interrupt should be enable from static data defined in device tree.

Quoted 7 lines
Quoted 3 lines

And you think passing variable data through registers is a better idea than passing packets via shared-memory?

You got me. :-)
I've no idea about which one is better.

Passing data directly via controller makes sense only when your protocol defines fixed length packets that fit into registers (usually FIFOs) or there is no shared memory between the processors. Otherwise, you write a data packet at some located in shmem, and provide its start and size along with the code of expected operation on it, over the mailbox. Please refer to "Passing frame information" para of page-2743 of your manual.
That is another reason the SCU implementation is broken - it has variable length packets and yet use registers to pass them around.

Quoted 3 lines

The problem is SCU firmware is already
there passing packets through data registers, we have no way to change it.

OK. But what is SCU exactly? Part of ATF? Do you get some binary from third party? If the last, you shouldn't be paying hsit for such a design.

Quoted 9 lines
Quoted 8 lines

The hardware has 8 unidirectional channels. But your protocol (SCU implementation) assumes there is one _virtual_ channel that has 4 registers and 1/0 irq --- which is not true. Instead of fixing the assumption in your protocol or emulating the virtual channel in the protocol level (user of a MU), you want to add code in the controller driver that ignores interrupts and club the 4 independent channels together.

Quoted 4 lines

This stucks me. This is how the hardware is designed and suggested to use in hardware reference manual. And now you're telling me this is wrong and we should not use the design in reference manual...

What you call "suggested to use in hardware reference manual" is just 1/5 examples of usecase.
And SCU is not even doing that correctly (it uses variable length packet, unlike the fixed 4-words suggestion).

A manual can suggest multiple ways of implementing a usecase. It is our job to chose the best one.

#26 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

Quoted 180 lines · Jassi Brar
Quoted 26 lines · A.s. Dong
Quoted 26 lines
Quoted 26 lines
Quoted 13 lines
Quoted 13 lines
Quoted 12 lines

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

Quoted 7 lines
Quoted 5 lines

This looks like reasonable to me, theoretically.
Just not sure whether it's necessary to support it because we probably will never use like that in reality, then it might become meaningless complicity introduced and error prone.

It _is_ necessary to write controller driver independent of client drivers.

Yes, that's true. What if we think we're writing driver against HW capabilities which support 4 pair of channel links(tx/rx)? It looks like independent of client drivers and may make life easily. Does it make sense?

Quoted 8 lines
Quoted 5 lines

No, that would be emulation.
Why doesn't a usb device controller (udc) driver emulate FSG/HID etc, by "thinking" it has a hardware backed storage/keyboard? It doesn't because that is the job of upper protocol layer.

Sorry I'm not quite familiar with USB device.

Which subsystem are you familiar with? Every stack is designed like that -- controller drivers reflect the physical hardware, any emulation/protocol is implemented in the user layer.

Quoted 3 lines

Another reason is I doubt that we may never use per register mode in a different register pair in the future.

You may never use, but if the driver reflects the controller as precisely as it is, linux could support any usecase that some baremetal code could support on your platform.
For example, you could support "normal" and "scu" mode over the 3-cell mechanism I suggested. But your original proposal/assumption of "scu-mode", failed immediately, hence Oleksij had to imlpement the "normal" mode.

Quoted 9 lines

For example:
AP:

node {
        ...
        // cell 0 meaning: 0: tx 1: rx  cell1 meaning: channel id
        Mboxes = <&mbox 0 1 &mbox 1 2>
        Mbox-names = "tx", "rx";
Quoted 1 line
Quoted 8 lines

M4:

node {
        ...
        Mboxes = <&mbox 0 2 &mbox 1 1>
        Mbox-names = "tx", "rx";
Quoted 1 line

This make things complicated and error prone as I said before.

I don't see how.

Quoted 3 lines

But that's just my understanding and may overlook something, if you still think we should do exactly as above, I will not against it because it does work for M4 case.

Cool, done.

Quoted 2 lines

Then the left question is how we handle SCU case?

Please point me to the code that you worry might not work.

Quoted 31 lines

I'm a bit confusing....
The section 3.6 you pointed is the MHU register description. It does not conflict with what I see from ARM doc center that each physical channel is unidirectional.

See below:
Chan 1:
0x000 SCP_INTR_L_STAT
0x008 SCP_INTR_L_SET
0x010 SCP _INTR_L_CLEAR

Chan 2:
0x020 SCP_INTR_H_STAT
0x028 SCP_INTR_H_SET
0x030 SCP _INTR_H_CLEAR

Chan 3:
0x100 CPU_INTR_L_STAT
0x108 CPU_INTR_L_SET
0x110 CPU_INTR_L_CLEAR

Chan 4:
0x120 CPU_INTR_H_STAT
0x128 CPU_INTR_H_SET
0x130 CPU_INTR_H_CLEAR

Chan 5:
0x200 SCP_INTR_S_STAT
0x208 SCP_INTR_S_SET
0x210 SCP _INTR_S_CLEAR

Quoted 8 lines

Chan 6:
0x300 CPU_INTR_S_STAT
0x308 CPU_INTR_S_SET
0x310 CPU_INTR_S_CLEAR

And the driver compose them into 3 channel links (lp, hp and sec). Am I wrong?

The RX and TX are not independent of each other, their priorities are tied together in hardware. So it makes more sense (as compared to declaring them independent) to club them together as one channel. Whatever example you take - MHU or TI - you won't find a driver that implements a virtual mode like "scu-mode".

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

Sorry for may not clear, "Passing short message' usecase is to tell how the HW is working on one channel mode sending up to 4 words in one time As specified in reference manual.

SCU does work that way, the only difference is it's using polling mode rather than interrupt driven. The point is the data size may be different for each msg, so we can't simply know which data register interrupt should be enable from static data defined in device tree.

Quoted 7 lines
Quoted 3 lines

And you think passing variable data through registers is a better idea than passing packets via shared-memory?

You got me. :-)
I've no idea about which one is better.

Passing data directly via controller makes sense only when your protocol defines fixed length packets that fit into registers (usually FIFOs) or there is no shared memory between the processors. Otherwise, you write a data packet at some located in shmem, and provide its start and size along with the code of expected operation on it, over the mailbox. Please refer to "Passing frame information" para of page-2743 of your manual.
That is another reason the SCU implementation is broken - it has variable length packets and yet use registers to pass them around.

Quoted 3 lines

The problem is SCU firmware is already
there passing packets through data registers, we have no way to change it.

OK. But what is SCU exactly? Part of ATF? Do you get some binary from third party? If the last, you shouldn't be paying hsit for such a design.

Quoted 9 lines
Quoted 8 lines

The hardware has 8 unidirectional channels. But your protocol (SCU implementation) assumes there is one _virtual_ channel that has 4 registers and 1/0 irq --- which is not true. Instead of fixing the assumption in your protocol or emulating the virtual channel in the protocol level (user of a MU), you want to add code in the controller driver that ignores interrupts and club the 4 independent channels together.

Quoted 4 lines

This stucks me. This is how the hardware is designed and suggested to use in hardware reference manual. And now you're telling me this is wrong and we should not use the design in reference manual...

Quoted 8 lines

What you call "suggested to use in hardware reference manual" is just 1/5 examples of usecase.
And SCU is not even doing that correctly (it uses variable length packet, unlike the fixed 4-words suggestion).

A manual can suggest multiple ways of implementing a usecase. It is our job to chose the best one.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180730/79bc18db/attachment.sig>

#27 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Hi Oleksij,

Quoted 15 lines

-----Original Message-----
From: Oleksij Rempel [mailto:o.rempel at pengutronix.de]
Sent: Monday, July 30, 2018 3:36 PM
To: Jassi Brar <[email protected]>; A.s. Dong <[email protected]>
Cc: , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; ", linux-arm- kernel"@lists.infradead.org; linux-mediatek at lists.infradead.org; srv_heupstream <[email protected]>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 5 lines

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I just send you an email of Jassie's early suggestion.

According to the suggestion, the binding needs to be changed into 8 separate TX/RX channels. You can quickly start from there.

Regards
Dong Aisheng

maintainer
#28 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Oleksij Rempel
Quoted 5 lines · Oleksij Rempel

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

thnx.

#29 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 21 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Monday, July 30, 2018 9:05 PM
To: Oleksij Rempel <[email protected]>
Cc: A.s. Dong <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; ", linux-arm- kernel"@lists.infradead.org; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-
mediatek at lists.infradead.org>; srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 2 lines · Oleksij Rempel

Hi Aishen, Jassie,

Quoted 8 lines
Quoted 3 lines

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

Do you think implement "Passing short messages" defined in reference manual in controller driver is protocol specific?
That seems like our most argument before.

Quoted 4 lines

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

That's out of our control.
SCU firmware is not using Linux and there's no mailbox framework. It's simply follow "passing short messages" way to send multi words with 4 data Registers.

Quoted 3 lines

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

You can refer to this patch for the user code of "scu-mode". [RFC PATCH v6 2/2] mailbox: imx-mailbox: add scu protocol support
https://www.spinics.net/lists/arm-kernel/msg665135.html

And I do admit that there're two protocol specific bits in the driver.
1. Unlike TI message header, SCU size is in the second u8. Ti message header:

struct ti_msgmgr_message {
        size_t len;
        u8 *buf;
};

Controller get the size from the first u32.

SCU message header:

struct sc_rpc_msg {
        uint8_t ver;
        uint8_t size;
        uint8_t svc; 
        uint8_t func;
};

Controller get the size from the second u8.

And a full SCU message is like:

struct imx_sc_msg_req_set_clock_rate {
        struct sc_rpc_msg hdr;
        u32 rate;
        u16 resource;
        u8 clk;
} __packed;

2. There's no synchronous read function in mailbox framework, So we use mbox_client_peek_data to read data which can't return the read data. In order to save a mem copy from read data to the data buffer, the controller driver saves the tx data buffer in itself. (Anyway to improve? Maybe we should implement a synchronous mbox_client_read_data()?)

And for SCU IPC client, it's something like:
int sc_call_rpc(sc_ipc_t ipc, void *msg, bool no_resp)

{
        ...
        mbox_send_message(sc_ipc->chan, msg);

        if (!no_resp) 
                ret = mbox_client_peek_data(sc_ipc->chan);

        mbox_client_txdone(sc_ipc->chan, ret);
        ...
}

int sc_ipc_open(sc_ipc_t *ipc, struct device *dev)

{
        sc_ipc->cl.dev = dev; 
        sc_ipc->cl.tx_block = false;
        sc_ipc->cl.knows_txdone = true;

        sc_ipc->chan = mbox_request_channel(&sc_ipc->cl, 0);
        ...
}

Any better ideas?

Regards
Dong Aisheng

Quoted 2 lines

thnx.

#30 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

[...]

Quoted 62 lines
Quoted 3 lines

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

You can refer to this patch for the user code of "scu-mode". [RFC PATCH v6 2/2] mailbox: imx-mailbox: add scu protocol support
https://www.spinics.net/lists/arm-kernel/msg665135.html

And I do admit that there're two protocol specific bits in the driver.
1. Unlike TI message header, SCU size is in the second u8. Ti message header:

struct ti_msgmgr_message {
        size_t len;
        u8 *buf;
};

Controller get the size from the first u32.

SCU message header:

struct sc_rpc_msg {
        uint8_t ver;
        uint8_t size;
        uint8_t svc;
        uint8_t func;
};

Controller get the size from the second u8.

And a full SCU message is like:

struct imx_sc_msg_req_set_clock_rate {
        struct sc_rpc_msg hdr;
        u32 rate;
        u16 resource;
        u8 clk;
} __packed;

2. There's no synchronous read function in mailbox framework, So we use mbox_client_peek_data to read data which can't return the read data. In order to save a mem copy from read data to the data buffer, the controller driver saves the tx data buffer in itself.
(Anyway to improve? Maybe we should implement a synchronous mbox_client_read_data()?)

And for SCU IPC client, it's something like:

int sc_call_rpc(sc_ipc_t ipc, void *msg, bool no_resp) {
        ...
        mbox_send_message(sc_ipc->chan, msg);

        if (!no_resp)
                ret = mbox_client_peek_data(sc_ipc->chan);

        mbox_client_txdone(sc_ipc->chan, ret);
        ...
}
int sc_ipc_open(sc_ipc_t *ipc, struct device *dev) {
        sc_ipc->cl.dev = dev;
        sc_ipc->cl.tx_block = false;
        sc_ipc->cl.knows_txdone = true;

        sc_ipc->chan = mbox_request_channel(&sc_ipc->cl, 0);
        ...
}
Quoted 2 lines

Any better ideas?

And this the original implementation used by NXP internally without mailbox. As it's quite simple and suppose more easy to help you understand the using of SCU.

void mu_send_msg(struct mu_priv *priv, uint32_t index, uint32_t msg)

{
        ...
        /* Wait TX register to be empty. */ 
        while (!(readl_relaxed(priv->base + MU_ASR) & mask))
                ;
        writel_relaxed(msg, priv->base + MU_ATR0  + (index * 4));
}

/*
* Wait to receive message from the other core.
*/
void mu_receive_msg(struct mu_priv *priv, uint32_t index, uint32_t *msg)

{
        ...
        /* Wait RX register to be full. */
        while (!(readl_relaxed(priv->base + MU_ASR) & mask))
                ;
        *msg = readl_relaxed(priv->base + MU_ARR0 + (index * 4));
}

static void sc_ipc_read(struct sc_ipc *sc_ipc, void *data)

{
        sc_rpc_msg_t *msg = (sc_rpc_msg_t *)data;
 
        /* Read first word */
        mu_receive_msg(sc_ipc->mu, 0, (uint32_t *)msg);
        count++;
 
        /* Read remaining words */
        while (count < msg->size) {
                mu_receive_msg(sc_ipc->mu, count % MU_RR_COUNT,
                               &(msg->DATA.u32[count - 1]));
                count++;
        }
}

static void sc_ipc_write(struct sc_ipc *sc_ipc, void *data)

{
        sc_rpc_msg_t *msg = (sc_rpc_msg_t *)data;
        uint8_t count = 0;

        /* Write first word */
        mu_send_msg(sc_ipc->mu, 0, *((uint32_t *)msg));
        count++;

        while (count < msg->size) {
                mu_send_msg(sc_ipc->mu, count % MU_TR_COUNT,
                            msg->DATA.u32[count - 1]);
                count++;
        }
}

void sc_call_rpc(sc_ipc_t ipc, sc_rpc_msg_t *msg, bool no_resp)

{
        sc_ipc_write(sc_ipc, msg);
        if (!no_resp)
                sc_ipc_read(sc_ipc, msg);
}

For details, see:
https://www.spinics.net/lists/arm-kernel/msg659138.html

Regards
Dong Aisheng

Quoted 5 lines

Regards
Dong Aisheng

Quoted 2 lines

thnx.

#31 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Jassi Brar
Quoted 29 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Monday, July 30, 2018 9:05 PM
To: Oleksij Rempel <[email protected]>
Cc: A.s. Dong <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; ", linux-arm- kernel"@lists.infradead.org; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-
mediatek at lists.infradead.org>; srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 5 lines · Oleksij Rempel

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

Quoted 7 lines

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

The initial idea is using mbox-cells value to distinguish which MU mode is used (single chan or multi chans).

Regards
Dong Aisheng

Quoted 2 lines

thnx.

#32 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Jassi Brar
Quoted 11 lines · Jassi Brar
Quoted 5 lines · Oleksij Rempel

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

Quoted 7 lines

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

Tomorrow I have a time slot to address your generic iMX MU suggestions. So, what is better, uni- or bi-directional channels? Should I implement *all* (4TX+FIFO, 4RX+FIFO, 4TX-simple, 4RX-simple) channels in this run? If yes, how should it be reflected in DT?
- 1 cell: &mu [0-15]
- 2 cells: &mu [0-7] [TX,RX]
- 3 cells: &mu [0-4] [FIFO,NOFIFO] [TX,RX]

Signature · 11 lines
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180730/0ff71b90/attachment.sig>
maintainer
#33 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 16 lines · Oleksij Rempel
Quoted 15 lines · Jassi Brar
Quoted 5 lines · Oleksij Rempel

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

Quoted 7 lines
Quoted 3 lines

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

Tomorrow I have a time slot to address your generic iMX MU suggestions. So, what is better, uni- or bi-directional channels?

The datasheet indicates there are 4 tx and 4 rx channels. So 8 uni-directional channels (which allow more fine-grained/efficient resource allocation btw).

Quoted 4 lines

Should I implement
*all* (4TX+FIFO, 4RX+FIFO, 4TX-simple, 4RX-simple) channels in this run?

From datasheet, each of 8 channels should be defined as signal+data

i.e, IRQ + TX/RX_Reg.
The rest 4 GP channels are doorbells (irq only).

So we can have 2-cells.

    First cell is 0->Tx, 1->RX, 2->Doorbell
    Second cell is index of the channel {0,3}

Now you may implement only RX+TX, and leave 'doorbell' out for future. Thats ok, because we wouldn't have to change bindings then.

However, if SCU (in its current form) must be supported. We may need to add the third cell (irq enable or not) or some better way, right now.

Quoted 5 lines

If yes, how should it be reflected in DT?
- 1 cell: &mu [0-15]
- 2 cells: &mu [0-7] [TX,RX]
- 3 cells: &mu [0-4] [FIFO,NOFIFO] [TX,RX]

It may be possible for a device binding to have variable cells, but I haven't seen any. And I am not sure that would look neat either.

#34 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Hi Jassi,

Quoted 7 lines
Quoted 5 lines

If yes, how should it be reflected in DT?
- 1 cell: &mu [0-15]
- 2 cells: &mu [0-7] [TX,RX]
- 3 cells: &mu [0-4] [FIFO,NOFIFO] [TX,RX]

It may be possible for a device binding to have variable cells, but I haven't seen any. And I am not sure that would look neat either.

Please refer to Documentation/devicetree/bindings/interrupt-controller/arm,gic-v3.txt
- #interrupt-cells : Specifies the number of cells needed to encode an interrupt source. Must be a single cell with a value of at least 3. If the system requires describing PPI affinity, then the value must be at least 4.

Regards
Dong Aisheng

maintainer
#35 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Jassi Brar
Quoted 39 lines · Jassi Brar
Quoted 16 lines · Oleksij Rempel
Quoted 15 lines · Jassi Brar
Quoted 5 lines · Oleksij Rempel

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

Quoted 7 lines
Quoted 3 lines

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

Tomorrow I have a time slot to address your generic iMX MU suggestions. So, what is better, uni- or bi-directional channels?

The datasheet indicates there are 4 tx and 4 rx channels. So 8 uni-directional channels (which allow more fine-grained/efficient resource allocation btw).

Quoted 3 lines

Should I implement
*all* (4TX+FIFO, 4RX+FIFO, 4TX-simple, 4RX-simple) channels in this run?

From datasheet, each of 8 channels should be defined as signal+data i.e, IRQ + TX/RX_Reg.
The rest 4 GP channels are doorbells (irq only).

So we can have 2-cells.

    First cell is 0->Tx, 1->RX, 2->Doorbell
    Second cell is index of the channel {0,3}
Quoted 7 lines

Now you may implement only RX+TX, and leave 'doorbell' out for future. Thats ok, because we wouldn't have to change bindings then.

However, if SCU (in its current form) must be supported. We may need to add the third cell (irq enable or not) or some better way, right now.

Looking at imx_mu_scu_send_data(), which simply polls on the tx, I think we don't even need third cell for scu client. A simple 2-cell, 8 uni-dir channel setup should work.
If I see the scu client driver, I could confirm how it would work.

So, lets please do the 8 uni-directional, 2 cells implementation.

Regards.

#36 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 47 lines · Jassi Brar
Quoted 39 lines · Jassi Brar
Quoted 16 lines · Oleksij Rempel
Quoted 15 lines · Jassi Brar
Quoted 5 lines · Oleksij Rempel

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

Quoted 7 lines
Quoted 3 lines

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

Tomorrow I have a time slot to address your generic iMX MU suggestions. So, what is better, uni- or bi-directional channels?

The datasheet indicates there are 4 tx and 4 rx channels. So 8 uni-directional channels (which allow more fine-grained/efficient resource allocation btw).

Quoted 3 lines

Should I implement
*all* (4TX+FIFO, 4RX+FIFO, 4TX-simple, 4RX-simple) channels in this run?

From datasheet, each of 8 channels should be defined as signal+data i.e, IRQ + TX/RX_Reg.
The rest 4 GP channels are doorbells (irq only).

So we can have 2-cells.

    First cell is 0->Tx, 1->RX, 2->Doorbell
    Second cell is index of the channel {0,3}
Quoted 7 lines

Now you may implement only RX+TX, and leave 'doorbell' out for future. Thats ok, because we wouldn't have to change bindings then.

However, if SCU (in its current form) must be supported. We may need to add the third cell (irq enable or not) or some better way, right now.

Quoted 6 lines

Looking at imx_mu_scu_send_data(), which simply polls on the tx, I think we don't even need third cell for scu client. A simple 2-cell, 8 uni-dir channel setup should work.
If I see the scu client driver, I could confirm how it would work.

So, lets please do the 8 uni-directional, 2 cells implementation.

Just in case.. to avoid confusion (or add more...). There are 4 GIP (General Interrupt Request n Pending bits) or RX-doorbell and 4 GIR (General Interrupt Request bits)
or TX-doorbell.

Signature · 11 lines
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180730/d16bec26/attachment.sig>
maintainer
#37 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 53 lines · Oleksij Rempel
Quoted 52 lines · Jassi Brar
Quoted 39 lines · Jassi Brar
Quoted 16 lines · Oleksij Rempel
Quoted 15 lines · Jassi Brar
Quoted 5 lines · Oleksij Rempel

Hi Aishen, Jassie,

i'm lost in this discussion. Please, as soon as I need to add some changes to my patches, notify me. Preferably on top for email.

I am ok with however you choose to implement, 8 unidirectional or 4 bidirectional channels or whatever.

We just can't have protocol specific s/w modes in the controller drivers.

The best solution is to fix the SCU firmware. If that is _really_ impossible, I provided a solution (3 cells work around). If you have a better idea please feel free to propose and implement that.

Quoted 7 lines
Quoted 3 lines

It will also help if you could share the user code of "scu-mode". If there is no such code (and we know the driver doesn't respect the "scu-mode" property) why do we even have that binding? Maybe drop it.

Tomorrow I have a time slot to address your generic iMX MU suggestions. So, what is better, uni- or bi-directional channels?

The datasheet indicates there are 4 tx and 4 rx channels. So 8 uni-directional channels (which allow more fine-grained/efficient resource allocation btw).

Quoted 3 lines

Should I implement
*all* (4TX+FIFO, 4RX+FIFO, 4TX-simple, 4RX-simple) channels in this run?

From datasheet, each of 8 channels should be defined as signal+data i.e, IRQ + TX/RX_Reg.
The rest 4 GP channels are doorbells (irq only).

So we can have 2-cells.

    First cell is 0->Tx, 1->RX, 2->Doorbell
    Second cell is index of the channel {0,3}
Quoted 7 lines

Now you may implement only RX+TX, and leave 'doorbell' out for future. Thats ok, because we wouldn't have to change bindings then.

However, if SCU (in its current form) must be supported. We may need to add the third cell (irq enable or not) or some better way, right now.

Looking at imx_mu_scu_send_data(), which simply polls on the tx, I think we don't even need third cell for scu client. A simple 2-cell, 8 uni-dir channel setup should work.
If I see the scu client driver, I could confirm how it would work.

Quoted 6 lines
Quoted 1 line

So, lets please do the 8 uni-directional, 2 cells implementation.

Just in case.. to avoid confusion (or add more...). There are 4 GIP (General Interrupt Request n Pending bits) or RX-doorbell and 4 GIR (General Interrupt Request bits)
or TX-doorbell.

Ok. But the approach remains the same.

thanks.

#38 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Jassi Brar

Hi Jassi,

Quoted 7 lines
Quoted 4 lines

However, if SCU (in its current form) must be supported. We may need to add the third cell (irq enable or not) or some better way, right now.

Looking at imx_mu_scu_send_data(), which simply polls on the tx, I think we don't even need third cell for scu client. A simple 2-cell, 8 uni-dir channel setup should work.

How would you suggest to use for SCU in 2 -cells?

Quoted 2 lines

If I see the scu client driver, I could confirm how it would work.

I've already given a general overview from here:
https://www.spinics.net/lists/arm-kernel/msg669202.html

Do you need me to send the full scu client driver patch for the reference? If yes, please feel free to let me know.

Regards
Dong Aisheng

Quoted 3 lines

So, lets please do the 8 uni-directional, 2 cells implementation.

Regards.

maintainer
#39 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 10 lines · A.s. Dong

Hi Jassi,

Quoted 7 lines
Quoted 4 lines

However, if SCU (in its current form) must be supported. We may need to add the third cell (irq enable or not) or some better way, right now.

Looking at imx_mu_scu_send_data(), which simply polls on the tx, I think we don't even need third cell for scu client. A simple 2-cell, 8 uni-dir channel setup should work.

Quoted 8 lines

How would you suggest to use for SCU in 2 -cells?

Quoted 2 lines

If I see the scu client driver, I could confirm how it would work.

I've already given a general overview from here:
https://www.spinics.net/lists/arm-kernel/msg669202.html

No, that is controller side code.

Quoted 3 lines

Do you need me to send the full scu client driver patch for the reference? If yes, please feel free to let me know.

Yes, that'll make it clearer.

maintainer
#40 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 19 lines · Jassi Brar
Quoted 10 lines · A.s. Dong

Hi Jassi,

Quoted 7 lines
Quoted 4 lines

However, if SCU (in its current form) must be supported. We may need to add the third cell (irq enable or not) or some better way, right now.

Looking at imx_mu_scu_send_data(), which simply polls on the tx, I think we don't even need third cell for scu client. A simple 2-cell, 8 uni-dir channel setup should work.

Quoted 8 lines

How would you suggest to use for SCU in 2 -cells?

Quoted 2 lines

If I see the scu client driver, I could confirm how it would work.

I've already given a general overview from here:
https://www.spinics.net/lists/arm-kernel/msg669202.html

Quoted 7 lines

No, that is controller side code.

Quoted 3 lines

Do you need me to send the full scu client driver patch for the reference? If yes, please feel free to let me know.

Yes, that'll make it clearer.

Thanks for sharing the patch offline.

I don't see why it can't be adapted to work with the simple 2-cell implementation. Instead of one virtual, ask for 4 physical channels and send data in u32 chunks sequentially over them.
In fact, that should make it even better by removing the tight-loop (zero delay!!) polling in send_data() -- MU does support interrupts, why not use them to avoid polling? If you find it hard, I can make a patch on top of your client patchset, once Oleksij is done with the MU driver.

I have some more suggestions for the client driver. But since I don't have to live with them (you do), I can't demand you make those changes. If you like, please feel free to cc me on the next revision of the client driver.

#41 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Hi Jassi,

Quoted 4 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Tuesday, July 31, 2018 8:42 PM
To: A.s. Dong <[email protected]>

[...]

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

Do you need me to send the full scu client driver patch for the reference? If yes, please feel free to let me know.

Yes, that'll make it clearer.

Thanks for sharing the patch offline.

Quoted 8 lines

I don't see why it can't be adapted to work with the simple 2-cell implementation. Instead of one virtual, ask for 4 physical channels and send data in u32 chunks sequentially over them.
In fact, that should make it even better by removing the tight-loop (zero delay!!) polling in send_data() -- MU does support interrupts, why not use them to avoid polling? If you find it hard, I can make a patch on top of your client patchset, once Oleksij is done with the MU driver.

As you insist, I will try to implement multi channels mode, then we can do a clear comparison to see which one is better for SCU mode rather than assumptions.

First of all, using interrupt for each chan will likely to terribly slow down the SCU Message handling speed. SCU Messages usually are handled by SCU firmware very quickly in a few microseconds. (< 10us). That's why we use polling mode for SCU message handling.

If we break down a SCU message into a few separate MU channel interrupts driven, There will be a lot of unnecessary and meaningless interrupt handling. And the whole flow may look like:

Sending Word 0 ->
Chan 0 interrupt -> (meaningless)
Sending Word 1 ->
Chan1 interrupt -> (meaningless)
Sending Word 2 ->
Chan2 interrupt -> (meaningless)
Sending Word 3 ->
Chan3 interrupt-> (meaningless)

Sending Word 4 on Chan 0 again ->
Wait for Chan 0 interrupt (meaningless) ->
...

Then waiting for Rx:

Chan 0 Rx interrupt ->
Read Word 0 ->
(Then we know response size here)

Chan 1 Rx interrupt ->
Read Word 1 ->
Chan 2 Rx interrupt ->
Read Word 2 ->
Chan 3 Rx interrupt ->
Read Word 3 ->

(If msg size > 4 words)
Chan 0 Rx interrupt ->
Read Word 4

...
And we must use a complicated way to map the received data of each channel into one SCU msg buffer and need handle the sequence properly for size > 4 words case

Is this what you intended?

Per my understanding, there're several issues for this follow:
1) Many unnecessary and meaningless interrupts hanllding of Tx and RX
2) Performance downgrading
A 4 word SCU MSG may need 8 interrupts to handle which terribly slow down the speed. (Interrupt latency will be terrible if many)
3) Complex RX flow in order to align with framework design

Any comments about these issues?

Generally, from what I see, the mailbox framework really is not designed for SCU type devices that handling msg sending or receiving via multiple channels. Including the ring buffer supported for handling several msgs on the same channel and rich completion mechanism, none of them are suitable for SCU MU in multiple channel mode.

What we really actually needed is a quite simple way to handle MSG Tx and Tx synchronously.

int sc_call_rpc(sc_ipc_t ipc, void *msg, bool no_resp)

{       
        ...
        ret = mbox_send_message(sc_ipc->chan, msg);
        if (!no_resp) 
                ret = mbox_receive_message(sc_ipc->chan, msg);
        ...
        return ret;
}

For me, I don't think the above way driven by interrupt is suitable for SCU. But I will still keep investigating if any better way to handle it in multi chan mode as you wish.

How would you suggest for such situation in order to support SCU properly?

Quoted 3 lines

I have some more suggestions for the client driver. But since I don't have to live with them (you do), I can't demand you make those changes. If you like, please feel free to cc me on the next revision of the client driver.

I certainly appreciate any clear and valuable improvement suggestions. Just feel free to let me know if any better idea.

Regards
Dong Aisheng

#42 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Hi Jassi,

Do you have any comments about this reply?

Sascha,
What's your option?

Regards
Dong Aisheng

Quoted 136 lines

-----Original Message-----
From: A.s. Dong
Sent: Thursday, August 2, 2018 5:24 PM
To: 'Jassi Brar' <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-
mediatek at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: RE: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Hi Jassi,

Quoted 4 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Tuesday, July 31, 2018 8:42 PM
To: A.s. Dong <[email protected]>

[...]

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

Do you need me to send the full scu client driver patch for the reference? If yes, please feel free to let me know.

Yes, that'll make it clearer.

Thanks for sharing the patch offline.

Quoted 9 lines

I don't see why it can't be adapted to work with the simple 2-cell implementation. Instead of one virtual, ask for 4 physical channels and send data in u32 chunks sequentially over them.
In fact, that should make it even better by removing the tight-loop (zero
delay!!) polling in send_data() -- MU does support interrupts, why not use them to avoid polling? If you find it hard, I can make a patch on top of your client patchset, once Oleksij is done with the MU driver.

As you insist, I will try to implement multi channels mode, then we can do a clear comparison to see which one is better for SCU mode rather than assumptions.

First of all, using interrupt for each chan will likely to terribly slow down the SCU Message handling speed. SCU Messages usually are handled by SCU firmware very quickly in a few microseconds. (< 10us). That's why we use polling mode for SCU message handling.

If we break down a SCU message into a few separate MU channel interrupts driven, There will be a lot of unnecessary and meaningless interrupt handling. And the whole flow may look like:

Sending Word 0 ->
Chan 0 interrupt -> (meaningless)
Sending Word 1 ->
Chan1 interrupt -> (meaningless)
Sending Word 2 ->
Chan2 interrupt -> (meaningless)
Sending Word 3 ->
Chan3 interrupt-> (meaningless)

Sending Word 4 on Chan 0 again ->
Wait for Chan 0 interrupt (meaningless) -> ...

Then waiting for Rx:

Chan 0 Rx interrupt ->
Read Word 0 ->
(Then we know response size here)

Chan 1 Rx interrupt ->
Read Word 1 ->
Chan 2 Rx interrupt ->
Read Word 2 ->
Chan 3 Rx interrupt ->
Read Word 3 ->

(If msg size > 4 words)
Chan 0 Rx interrupt ->
Read Word 4

...
And we must use a complicated way to map the received data of each channel into one SCU msg buffer and need handle the sequence properly for size > 4 words case

Is this what you intended?

Per my understanding, there're several issues for this follow:
1) Many unnecessary and meaningless interrupts hanllding of Tx and RX
2) Performance downgrading
A 4 word SCU MSG may need 8 interrupts to handle which terribly slow down the speed. (Interrupt latency will be terrible if many)
3) Complex RX flow in order to align with framework design

Any comments about these issues?

Generally, from what I see, the mailbox framework really is not designed for SCU type devices that handling msg sending or receiving via multiple channels. Including the ring buffer supported for handling several msgs on the same channel and rich completion mechanism, none of them are suitable for SCU MU in multiple channel mode.

What we really actually needed is a quite simple way to handle MSG Tx and Tx synchronously.

int sc_call_rpc(sc_ipc_t ipc, void *msg, bool no_resp)

{
        ...
        ret = mbox_send_message(sc_ipc->chan, msg);
        if (!no_resp)
                ret = mbox_receive_message(sc_ipc->chan, msg);
        ...
        return ret;
}

For me, I don't think the above way driven by interrupt is suitable for SCU. But I will still keep investigating if any better way to handle it in multi chan mode as you wish.

How would you suggest for such situation in order to support SCU properly?

Quoted 3 lines

I have some more suggestions for the client driver. But since I don't have to live with them (you do), I can't demand you make those changes. If you like, please feel free to cc me on the next revision of the

client driver.

I certainly appreciate any clear and valuable improvement suggestions. Just feel free to let me know if any better idea.

Regards
Dong Aisheng

maintainer
#43 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 4 lines · A.s. Dong

Hi Jassi,

Do you have any comments about this reply?

I did read your post. And my opinion isn't changed. If mailbox is new to you, please try to see how other subsystems works, especially SPI, I2C and DMA.

Thanks.

#44 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 3 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, August 9, 2018 10:56 AM

[...]

Quoted 8 lines · A.s. Dong
Quoted 4 lines

Hi Jassi,

Do you have any comments about this reply?

I did read your post. And my opinion isn't changed. If mailbox is new to you, please try to see how other subsystems works, especially SPI, I2C and DMA.

Okay.
What's your comments about the known issues if doing that way?
1) Many unnecessary and meaningless interrupts handling of Tx and RX
2) Performance downgrading
A 4 word SCU MSG may need 8 interrupts to handle which terribly slow down the speed. (Interrupt latency will be terrible if many). SCU msg are mostly handled within 10 us.
3) Complex RX flow in order to align with framework design

Are you referring to them as not a real issue?
Or even they're issues/limitations, you would still prefer multi channels way In order to be consistent with mailbox subsystem by sacrificing the performance drop?

And to be clear, the working flow you want is exactly as follows, right? Sending Word 0 ->
Chan 0 interrupt -> (meaningless)
Sending Word 1 ->
Chan1 interrupt -> (meaningless)
Sending Word 2 ->
Chan2 interrupt -> (meaningless)
Sending Word 3 ->
Chan3 interrupt-> (meaningless)

Sending Word 4 on Chan 0 again ->
Wait for Chan 0 interrupt (meaningless) -> ...

Then waiting for Rx:

Chan 0 Rx interrupt ->
Read Word 0 ->
(Then we know response size here)

Chan 1 Rx interrupt ->
Read Word 1 ->
Chan 2 Rx interrupt ->
Read Word 2 ->
Chan 3 Rx interrupt ->
Read Word 3 ->

(If msg size > 4 words)
Chan 0 Rx interrupt ->
Read Word 4

Regards
Dong Aisheng

Quoted 1 line

Thanks.

#45 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Jassi Brar
Quoted 15 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Saturday, July 28, 2018 9:10 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; , Sascha Hauer <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux-mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

[...]

Quoted 15 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

Sorry for may not clear, "Passing short message' usecase is to tell how the HW is working on one channel mode sending up to 4 words in one time As specified in reference manual.

SCU does work that way, the only difference is it's using polling mode rather than interrupt driven. The point is the data size may be different for each msg, so we can't simply know which data register interrupt should be enable from static data defined in device

tree.

Quoted 5 lines
Quoted 4 lines
Quoted 1 line

And you think passing variable data through registers is a better idea than passing packets via shared-memory?

Quoted 6 lines
Quoted 3 lines

You got me. :-)
I've no idea about which one is better.

Passing data directly via controller makes sense only when your protocol defines fixed length packets that fit into registers (usually FIFOs) or there is no shared memory between the processors.

Don't have to be fixed length packets. Every data packets less than 4 words can be transferred in one time by controller perfectly well as it has 4 data registers. No shared memory need.

Quoted 7 lines

Otherwise, you write a data packet at some located in shmem, and provide its start and size along with the code of expected operation on it, over the mailbox. Please refer to "Passing frame information"
para of page-2743 of your manual.
That is another reason the SCU implementation is broken - it has variable length packets and yet use registers to pass them around.

" Passing frame information" is used for long message transfer via shared memory. But almost all the SCU data frames are less than 4 words. (only quite a few of them, 2~3, over 4 words) For these types of messages, it's obviously more efficient to transfer them in "Passing short message' way as the MU hardware supports well.

So I believe our SCU firmware has already done the best choice in a most efficient way (register write is obviously faster than memory copies). Thus, no shared memory need to be used for SCU protocol case.

Quoted 7 lines
Quoted 4 lines

The problem is SCU firmware is already there passing packets through data registers, we have no way to change it.

OK. But what is SCU exactly? Part of ATF? Do you get some binary from third party? If the last, you shouldn't be paying hsit for such a design.

The System Controller Firmware (SCFW) is a low-level system function which runs on a dedicated Cortex-M core to provide power, clock, and resource management. It exists on some i.MX8 processors. e.g. i.MX8QM (QM, QP), and i.MX8QX (QXP, DX).

It's developed by NXP internally.

Quoted 11 lines
Quoted 8 lines
Quoted 8 lines

The hardware has 8 unidirectional channels. But your protocol (SCU implementation) assumes there is one _virtual_ channel that has 4 registers and 1/0 irq --- which is not true. Instead of fixing the assumption in your protocol or emulating the virtual channel in the protocol level (user of a MU), you want to add code in the controller driver that ignores interrupts and club the 4 independent channels

together.

Quoted 2 lines
Quoted 1 line
Quoted 8 lines
Quoted 4 lines

This stucks me. This is how the hardware is designed and suggested to use in hardware reference manual. And now you're telling me this is wrong and we should not use the design in reference manual...

What you call "suggested to use in hardware reference manual" is just 1/5 examples of usecase.
And SCU is not even doing that correctly (it uses variable length packet, unlike the fixed 4-words suggestion).

SCU is doing well. Please see my above reply.

Quoted 3 lines

A manual can suggest multiple ways of implementing a usecase. It is our job to chose the best one.

Well, great. That is also our orientation.
Because we thought the way " Passing short message " is the most suitable way for SCU and is the most efficient way as the most SCU Messages are less than 4 words and MU hardware supports transfer them well. We could not think out a reason to not use it. That's why we choose that.

A simple objection that it uses variable length packet seems not quite convincible. Unless we have more convincible reasons...

How do you think?

For example, what's the real benefits we can get if we drop the HW capability of "passing Short messages" for SCU in one channel way and use separate 8 Tx/Rx channel Instead to send them each time? What about the Performance downgrading? Is it valuable comparing to the complexity introduced?

Regards
Dong Aisheng

#46 lore raw

PATCH v7 3/6 dt-bindings: mailbox: imx-mu: add generic MU channel support

In reply to Jassi Brar

Hi Oleksij & Sascha,

Quoted 72 lines

-----Original Message-----
From: Jassi Brar [mailto:jassisinghbrar at gmail.com]
Sent: Thursday, July 26, 2018 11:44 PM
To: A.s. Dong <[email protected]>
Cc: Oleksij Rempel <[email protected]>; Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; Vladimir Zapolskiy <[email protected]>; , Sascha Hauer <[email protected]>; , linux-arm-kernel at lists.infradead.org, linux- mediatek at lists.infradead.org, srv_heupstream <linux-arm- kernel at lists.infradead.org>; Devicetree List <[email protected]>; dl-linux-imx <[email protected]>
Subject: Re: [PATCH v7 3/6] dt-bindings: mailbox: imx-mu: add generic MU channel support

Quoted 11 lines · A.s. Dong
Quoted 11 lines
Quoted 4 lines

Each MU has four pairs of rx/tx data register with four rx/tx interrupts which can also be used as a separate channel.

So the hardware actually supports 4 channels.

Quoted 5 lines

-- #mbox-cells: Must be 0. Number of cells in a mailbox +- #mbox-cells: Must be:

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +3 -0
+               0 - for single channel mode. i.MX8* SCU protocol specific.
+               1 - for multichannel (generic) mode.
+
Quoted 7 lines
Quoted 5 lines

No, please.
DT bindings should reflect the real hardware, and not the software mode we want the driver to work in.
Please define mbox-cells=1 and have the i.MX8* platform always ask for channel-0.

The reality is that MU hardware does not define channels in reference

manual.

Quoted 7 lines

However, it does have four separate data tx/rx register which can be used as 'virtual' channels which is supported by this current driver.

See below HW description from the reference manual:
For messaging, the MU has four, 32-bit write-only transmit registers and four, 32-bit read-only receive registers on the Processor B and Processor A-sides. These registers are used for sending messages to each

other.

Quoted 1 line

For a while please forget the protocol(user) level usage, and consider only what your h/w is.

MU has 4 pairs of TX_Reg + TX_IRQ, and 4 pairs of RX_Reg + RX_IRQ. (MU also has 4 "doorbell" type channels that it calls GP, but those are not managed here, so lets not worry atm).

By definition, a mailbox channel is simply a signal, optionally with data attached. So, MU has 4 TX and 4 RX channels.

The MU driver should populate 8 unidirectional (4 Tx and 4 RX) channels and set each tx/rx operation to trigger the corresponding interrupt. This is not my whim, this is how the controller is!

The SCU is poorly implemented as it ignores 3 irqs and club all 4 register together (there are many other cons of this approach but lets not get into that).
Personally, I would push-back on such a bad design. But if you claim you have no choice but to support SCU as such, the work around could be simpler than defining a new "scu mode" altogether.

#mbox-cells: Must be 3.
First cell is 1 for TX and 0 for RX channel Second cell is index of the channel [0,1,2 or 3] Third cell is 1 if the channel triggers an IRQ,
0 if not. That is ACR.RIE/TIE bits are set or not.

Quoted 5 lines

Normal clients would always request a channel with irqs enabled. The SCU client would request all 8 channels -- TX/RX[0,1,2] with irqs disabled, TX/RX[3] with irqs enabled. And SCU will read/write 4 word data over 4 rx/tx channels, instead of the virtually concocted one.

Sascha & Oleksij,
Do you have any comments on this new approach suggested by Jassi? This looks like may work for M4 case.

But I'm still discussing with Jassi on whether it's suitable for SCU as well? As that means we drop the MU Hardware capability to send multi words in one time but send one word on each channel one time which seems like less efficiency and more complexity introduced for SCU case.

Regards
Dong Aisheng

Quoted 35 lines
Quoted 8 lines

short messages
Transmit register(s) can be used to pass short messages from one to four words in length. For example, when a four-word message is desired, only one of the registers needs to have its corresponding interrupt enable bit set at the receiver side; the message?s first three words are written to the registers whose interrupt is masked, and the fourth word is written to the other register (which triggers an

interrupt at the receiver side).

Quoted 3 lines

The reference manual is at here: (Chapter 42 Messaging Unit (MU)

https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fw
ww

Quoted 1 line

.nxp.com%2Fdocs%2Fen%2Freference-

manual%2FIMX6ULRM.pdf&amp;data=02%7C0

Quoted 1 line

1%7Caisheng.dong%40nxp.com%7Cef349d7c9caf46c4260008d5f30e9ef5%7C6 86ea1

Quoted 1 line

d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636682166494860332&amp;sdat a=54rE

Quoted 1 line

iDm%2BGD6EY8NE64ck1LXVGoUCtWYrHXPoWqcujUo%3D&amp;reserved=0

Quoted 2 lines

And SCU firmware does use MU in above way specified by reference

manual.

Quoted 2 lines

Even from HW point of view, it's still one channel only mailbox.

Please realise that any manual is written by a mere mortal afterall. How the controller works is set in stone, but how the controller can be used ... is just someones suggestion.

The approach I suggest above, conforms to the api and prevents a provider dancing to the tunes of a user.

#47 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

In reply to Oleksij Rempel

This are currently tested SoCs with imx-mailbox driver.

---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
index 113d6ab931ef..5616d2afca45 100644
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible :	should be "fsl,<chip>-mu", the supported chips include
-		imx8qxp, imx8qm.
+		imx6sx, imx7s, imx8qxp, imx8qm.
 - reg :		Should contain the registers location and length
 - interrupts :	Interrupt number. The interrupt specifier format depends
 		on the interrupt controller parent.
Signature · 1 line
2.18.0
#48 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 14 lines

-----Original Message-----
From: Oleksij Rempel [mailto:o.rempel at pengutronix.de]
Sent: Thursday, July 26, 2018 2:53 PM
To: Shawn Guo <[email protected]>; Fabio Estevam <[email protected]>; Rob Herring <[email protected]>; Mark Rutland <[email protected]>; A.s. Dong <[email protected]>; Vladimir Zapolskiy <[email protected]>; Jassi Brar <[email protected]>
Cc: Oleksij Rempel <[email protected]>; kernel at pengutronix.de; linux-arm-kernel at lists.infradead.org; devicetree at vger.kernel.org; dl-linux- imx <[email protected]>
Subject: [PATCH v7 4/6] dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 3 lines

This are currently tested SoCs with imx-mailbox driver.

Regards
Dong Aisheng

maintainer
#49 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

In reply to Oleksij Rempel
Quoted 12 lines · Oleksij Rempel

This are currently tested SoCs with imx-mailbox driver.

---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 6 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller? I think the controller name, if it has one or create a new one like 'imx-mu', should be used here.
So,
compatible : should be "fsl,imx-mu"

BTW, the driver doesn't even probe anything other than 'fsl,imx6sx-mu', unlike what the binding says.

#50 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 9 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 8 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line

---
?Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +- ?1 file changed, 1 insertion(+), 1 deletion(-)

Quoted 13 lines
Quoted 11 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
index 113d6ab931ef..5616d2afca45 100644
@@ -18,7 +18,7 @@ Messaging Unit Device Node:

?Required properties:
?-------------------
?- compatible : should be "fsl,<chip>-mu", the supported chips include -???????????????imx8qxp, imx8qm.
+???????????????imx6sx, imx7s, imx8qxp, imx8qm.

This is not scalable. Do we add every new SoC that contains the same controller?

Yes, we do. This is a policy direction from the DT maintainers. If we ever going to want to validate DTs against the binding, all compatibles used in the DTs must be specified in the binding.

As we can't really tell if the controller is exactly the same or even has some SoC integration bugs, we generally add a new compatible for each SoC to key off any workarounds necessary in the driver without the need to change the DTs, breaking compatibility.

Regards,
Lucas

maintainer
#51 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 26 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 9 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 8 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 13 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 11 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
index 113d6ab931ef..5616d2afca45 100644
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 9 lines

If we
ever going to want to validate DTs against the binding, all compatibles used in the DTs must be specified in the binding.

As we can't really tell if the controller is exactly the same or even has some SoC integration bugs, we generally add a new compatible for each SoC to key off any workarounds necessary in the driver without the need to change the DTs, breaking compatibility.

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

#52 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Am Donnerstag, den 26.07.2018, 16:45 +0530 schrieb Jassi Brar:

Quoted 58 lines
Quoted 31 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 19 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 18 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line

---
?Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +- ?1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Documentation/devicetree/bindings/mailbox/fsl,mu.txt b/Documentation/devicetree/bindings/mailbox/fsl,mu.txt index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
@@ -18,7 +18,7 @@ Messaging Unit Device Node:

?Required properties:
?-------------------

Quoted 8 lines
Quoted 5 lines

?- compatible : should be "fsl,<chip>-mu", the supported chips include
-???????????????imx8qxp, imx8qm. +???????????????imx6sx, imx7s, imx8qxp, imx8qm.

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 5 lines

If we
ever going to want to validate DTs against the binding, all compatibles
used in the DTs must be specified in the binding.

Quoted 8 lines

As we can't really tell if the controller is exactly the same or even
has some SoC integration bugs, we generally add a new compatible for
each SoC to key off any workarounds necessary in the driver without the
need to change the DTs, breaking compatibility.

Quoted 4 lines

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

Even if it is the same IP core, the SoC integration might have bugs that need different behavior from the driver. We've already had that case with the i.MX6 SPI controller.

Quoted 3 lines

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

That's way too late. If we want to keep DTs stable the DT must include forward looking compatibles, to allow the driver to key any required behavior changes from already present compatibles. Otherwise you require the DT to be updated in step-lock with the kernel or need much more awkward detection in the drivers. IP integration bugs might be found way after the initial DT has been written.

Regards,
Lucas

maintainer
#53 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 59 lines · Lucas Stach

Am Donnerstag, den 26.07.2018, 16:45 +0530 schrieb Jassi Brar:

Quoted 58 lines
Quoted 31 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 19 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 18 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Documentation/devicetree/bindings/mailbox/fsl,mu.txt b/Documentation/devicetree/bindings/mailbox/fsl,mu.txt index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
Quoted 8 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 21

Quoted 5 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
 - compatible : should be "fsl,<chip>-mu", the supported chips

include

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 13 lines

If we
ever going to want to validate DTs against the binding, all compatibles
used in the DTs must be specified in the binding.

As we can't really tell if the controller is exactly the same or even
has some SoC integration bugs, we generally add a new compatible for
each SoC to key off any workarounds necessary in the driver without the
need to change the DTs, breaking compatibility.

Quoted 9 lines
Quoted 4 lines

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

Even if it is the same IP core, the SoC integration might have bugs that need different behavior from the driver. We've already had that case with the i.MX6 SPI controller.

For h/w quirks/bugs, a new "has-that-bug" property makes better sense. Or, if you insist, a new compatible based on the first soc that has the buggy block.

Quoted 6 lines
Quoted 3 lines

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

That's way too late. If we want to keep DTs stable

How do you keep the DT stable by explicitly defining every new SoC to the compatible list in DT, _add_ to the driver.... only to have the driver absolutely not care which SoC is it?
Which is the situation right now with this patchset.

thnx.

#54 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 65 lines · Jassi Brar
Quoted 64 lines · Lucas Stach

Am Donnerstag, den 26.07.2018, 16:45 +0530 schrieb Jassi Brar:

Quoted 62 lines
Quoted 31 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 19 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 18 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Documentation/devicetree/bindings/mailbox/fsl,mu.txt b/Documentation/devicetree/bindings/mailbox/fsl,mu.txt index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
Quoted 8 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 21

Quoted 5 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
 - compatible : should be "fsl,<chip>-mu", the supported chips

include

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 5 lines

If we
ever going to want to validate DTs against the binding, all compatibles
used in the DTs must be specified in the binding.

Quoted 8 lines

As we can't really tell if the controller is exactly the same or even
has some SoC integration bugs, we generally add a new compatible for
each SoC to key off any workarounds necessary in the driver without the
need to change the DTs, breaking compatibility.

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

Quoted 8 lines
Quoted 4 lines

Even if it is the same IP core, the SoC integration might have bugs that need different behavior from the driver. We've already had that case with the i.MX6 SPI controller.

For h/w quirks/bugs, a new "has-that-bug" property makes better sense. Or, if you insist, a new compatible based on the first soc that has the buggy block.

How do you propose to handle this property in the driver, if once flashed DTB does not contain it?

Quoted 11 lines
Quoted 6 lines
Quoted 3 lines

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

That's way too late. If we want to keep DTs stable

How do you keep the DT stable by explicitly defining every new SoC to the compatible list in DT, _add_ to the driver.... only to have the driver absolutely not care which SoC is it?
Which is the situation right now with this patchset.

For sake of simplicity you can assume that a driver can be changed in future, and SoC/board device tree bindings can not be changed ever.

--
Best wishes,
Vladimir

maintainer
#55 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 70 lines · Vladimir Zapolskiy
Quoted 69 lines · Jassi Brar
Quoted 68 lines · Lucas Stach

Am Donnerstag, den 26.07.2018, 16:45 +0530 schrieb Jassi Brar:

Quoted 58 lines
Quoted 31 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 19 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 18 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Documentation/devicetree/bindings/mailbox/fsl,mu.txt b/Documentation/devicetree/bindings/mailbox/fsl,mu.txt index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
Quoted 8 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt

Quoted 5 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
 - compatible : should be "fsl,<chip>-mu", the supported chips

include

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 5 lines

If we
ever going to want to validate DTs against the binding, all compatibles
used in the DTs must be specified in the binding.

Quoted 8 lines

As we can't really tell if the controller is exactly the same or even
has some SoC integration bugs, we generally add a new compatible for
each SoC to key off any workarounds necessary in the driver without the
need to change the DTs, breaking compatibility.

Quoted 4 lines

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

Even if it is the same IP core, the SoC integration might have bugs that need different behavior from the driver. We've already had that case with the i.MX6 SPI controller.

Quoted 8 lines
Quoted 4 lines

For h/w quirks/bugs, a new "has-that-bug" property makes better sense. Or, if you insist, a new compatible based on the first soc that has the buggy block.

How do you propose to handle this property in the driver, if once flashed DTB does not contain it?

I see, your point is - what if a bug appears only in a particular SoC ? Well, if the controller is same the bug should appear to all SoCs. Do you think these SoCs have different versions of MU ?

#56 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 77 lines · Jassi Brar
Quoted 75 lines · Vladimir Zapolskiy
Quoted 73 lines · Jassi Brar
Quoted 59 lines · Lucas Stach

Am Donnerstag, den 26.07.2018, 16:45 +0530 schrieb Jassi Brar:

Quoted 58 lines
Quoted 31 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 19 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 18 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Documentation/devicetree/bindings/mailbox/fsl,mu.txt b/Documentation/devicetree/bindings/mailbox/fsl,mu.txt index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
Quoted 8 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt

Quoted 5 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
 - compatible : should be "fsl,<chip>-mu", the supported chips

include

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 13 lines

If we
ever going to want to validate DTs against the binding, all compatibles
used in the DTs must be specified in the binding.

As we can't really tell if the controller is exactly the same or even
has some SoC integration bugs, we generally add a new compatible for
each SoC to key off any workarounds necessary in the driver without the
need to change the DTs, breaking compatibility.

Quoted 9 lines
Quoted 4 lines

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

Even if it is the same IP core, the SoC integration might have bugs that need different behavior from the driver. We've already had that case with the i.MX6 SPI controller.

For h/w quirks/bugs, a new "has-that-bug" property makes better sense. Or, if you insist, a new compatible based on the first soc that has the buggy block.

Quoted 7 lines
Quoted 3 lines

How do you propose to handle this property in the driver, if once flashed DTB does not contain it?

I see, your point is - what if a bug appears only in a particular SoC ? Well, if the controller is same the bug should appear to all SoCs. Do you think these SoCs have different versions of MU ?

I'm not a hardware developer, but I'm quite sure that strictly speaking IPs are different on different SoCs, the diversity of differences can be negligible, or it can be discovered in future and considered as significant, i.e. two IPs may be found as incompatible ones. Note that the DTBs flashed on boards are already prepared for the worst scenario.

But to some extend, and in today's knowledge, the IPs are compatible, let's say they are compatible to "fsl,imx6sx-mu" version, the one which is found in the published driver.

--
Best wishes,
Vladimir

#57 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

In reply to Jassi Brar

Am Donnerstag, den 26.07.2018, 17:16 +0530 schrieb Jassi Brar:

Quoted 65 lines
Quoted 65 lines · Lucas Stach

Am Donnerstag, den 26.07.2018, 16:45 +0530 schrieb Jassi Brar:

Quoted 62 lines
Quoted 31 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 19 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 18 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

<[email protected]> wrote:

This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line

---
?Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +- ?1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Documentation/devicetree/bindings/mailbox/fsl,mu.txt b/Documentation/devicetree/bindings/mailbox/fsl,mu.txt index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
@@ -18,7 +18,7 @@ Messaging Unit Device Node:

?Required properties:
?-------------------

Quoted 8 lines
Quoted 5 lines

?- compatible : should be "fsl,<chip>-mu", the supported chips include
-???????????????imx8qxp, imx8qm. +???????????????imx6sx, imx7s, imx8qxp, imx8qm.

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 5 lines

If we
ever going to want to validate DTs against the binding, all compatibles
used in the DTs must be specified in the binding.

Quoted 8 lines

As we can't really tell if the controller is exactly the same or even
has some SoC integration bugs, we generally add a new compatible for
each SoC to key off any workarounds necessary in the driver without the
need to change the DTs, breaking compatibility.

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

Quoted 9 lines
Quoted 4 lines

Even if it is the same IP core, the SoC integration might have bugs that need different behavior from the driver. We've already had that case with the i.MX6 SPI controller.

For h/w quirks/bugs, a new "has-that-bug" property makes better sense. Or, if you insist, a new compatible based on the first soc that has the buggy block.

Hardware bugs might not be known at the the time the DT is deployed. The DT is not part of the kernel, but might be part of a fixed device firmware.

Quoted 11 lines
Quoted 6 lines
Quoted 3 lines

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

That's way too late. If we want to keep DTs stable

How do you keep the DT stable by explicitly defining every new SoC to the compatible list in DT, _add_ to the driver.... only to have the driver absolutely not care which SoC is it?
Which is the situation right now with this patchset.

It means we can later, in a newer kernel change the driver to use the compatible to change behavior without the need to update the DT at that point.

Regards,
Lucas

#58 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

In reply to Jassi Brar
Quoted 27 lines · Jassi Brar
Quoted 26 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 22 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 13 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 6 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 3 lines
Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Some notes are found in Documentation/devicetree/bindings/ABI.txt

To guarantee "a stable binding" of a compatible property, to avoid unintentionally added incompatibilities and to fix a list of supported compatibles in the device driver code there should be a SoC specific compatible value in the list of 'compatible' property elements.

Quoted 15 lines

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 9 lines

If we
ever going to want to validate DTs against the binding, all compatibles used in the DTs must be specified in the binding.

As we can't really tell if the controller is exactly the same or even has some SoC integration bugs, we generally add a new compatible for each SoC to key off any workarounds necessary in the driver without the need to change the DTs, breaking compatibility.

Quoted 9 lines

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

Unfortunately it does not work well this way due to limited possibilities to distinugush different device IPs on different SoCs, if identical compatibles are given in both cases, and often it is not obvious that two IPs are different.

--
Best wishes,
Vladimir

maintainer
#59 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 28 lines · Vladimir Zapolskiy
Quoted 27 lines · Jassi Brar
Quoted 26 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 22 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 13 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 6 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 6 lines
Quoted 3 lines
Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Some notes are found in Documentation/devicetree/bindings/ABI.txt

Thanks. I have been through the ABI.txt.

Quoted 30 lines

To guarantee "a stable binding" of a compatible property, to avoid unintentionally added incompatibilities and to fix a list of supported compatibles in the device driver code there should be a SoC specific compatible value in the list of 'compatible' property elements.

Quoted 15 lines

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 9 lines

If we
ever going to want to validate DTs against the binding, all compatibles used in the DTs must be specified in the binding.

As we can't really tell if the controller is exactly the same or even has some SoC integration bugs, we generally add a new compatible for each SoC to key off any workarounds necessary in the driver without the need to change the DTs, breaking compatibility.

Quoted 9 lines

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

Quoted 5 lines

Unfortunately it does not work well this way due to limited possibilities to distinugush different device IPs on different SoCs, if identical compatibles are given in both cases, and often it is not obvious that two IPs are different.

Please note the submitted driver absolutely don't care which of the five SoCs it is.
In other words, all these SoCs have the same controller. So its about have just one compatible right now, and add more if some new SoC comes with a variation of the controller.

#60 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 68 lines · Jassi Brar
Quoted 28 lines · Vladimir Zapolskiy
Quoted 27 lines · Jassi Brar
Quoted 26 lines · Lucas Stach

Hi Jassi,

Am Donnerstag, den 26.07.2018, 15:25 +0530 schrieb Jassi Brar:

Quoted 22 lines

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel

Quoted 13 lines

<[email protected]> wrote:
This are currently tested SoCs with imx-mailbox driver.

Quoted 1 line
Quoted 1 line
---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +0 -0
index 113d6ab931ef..5616d2afca45 100644

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 6 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller?

Quoted 6 lines
Quoted 3 lines
Quoted 2 lines

Yes, we do. This is a policy direction from the DT maintainers.

I would love to read the post/documentation.

Some notes are found in Documentation/devicetree/bindings/ABI.txt

Thanks. I have been through the ABI.txt.

Quoted 30 lines

To guarantee "a stable binding" of a compatible property, to avoid unintentionally added incompatibilities and to fix a list of supported compatibles in the device driver code there should be a SoC specific compatible value in the list of 'compatible' property elements.

Quoted 24 lines

Consider the same h/w - controller and platforms, but only the the MU chapter said the controller name is, say, 'MU121'. I am sure now you will see it correct to call it "fsl,mu121" compatible. What changed? just the name, right?

Quoted 9 lines

If we
ever going to want to validate DTs against the binding, all compatibles used in the DTs must be specified in the binding.

As we can't really tell if the controller is exactly the same or even has some SoC integration bugs, we generally add a new compatible for each SoC to key off any workarounds necessary in the driver without the need to change the DTs, breaking compatibility.

I think if the h/w resources and behaviour remain the same and the documentation does not call it by a different name -- it is safe to assume its the same IP. Especially when the driver is absolutely indifferent to the 5 SoC names.

If/when we find the controller changes, we could revisit the binding and add another compatible option and modify the driver to catch that and adapt.

Quoted 7 lines
Quoted 5 lines

Unfortunately it does not work well this way due to limited possibilities to distinugush different device IPs on different SoCs, if identical compatibles are given in both cases, and often it is not obvious that two IPs are different.

Please note the submitted driver absolutely don't care which of the five SoCs it is.

True.

Quoted 1 line

In other words, all these SoCs have the same controller.

False :)

Quoted 3 lines

So its about have just one compatible right now, and add more if some new SoC comes with a variation of the controller.

True. The driver will be changed in this case, unfortunately the bindings are not so volatile.

--
Best wishes,
Vladimir

maintainer
#61 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 8 lines · Vladimir Zapolskiy
Quoted 6 lines · Jassi Brar
Quoted 1 line · Vladimir Zapolskiy

Please note the submitted driver absolutely don't care which of the five SoCs it is.

Quoted 7 lines

True.

Quoted 1 line

In other words, all these SoCs have the same controller.

False :)

OK, so the controllers are not identical, but same enough to have a common driver?
Is the driver not tested enough or are you planning to add more features?

Quoted 7 lines
Quoted 3 lines

So its about have just one compatible right now, and add more if some new SoC comes with a variation of the controller.

True. The driver will be changed in this case, unfortunately the bindings are not so volatile.

The volatility of the bindings will be same when you add a new SoC compatible ;) And you must add to bindings+driver, or the same h/w and same driver won't work for your new platform.

#62 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 10 lines · Jassi Brar
Quoted 8 lines · Vladimir Zapolskiy
Quoted 6 lines · Jassi Brar
Quoted 1 line · Vladimir Zapolskiy

Please note the submitted driver absolutely don't care which of the five SoCs it is.

Quoted 7 lines
Quoted 7 lines

True.

Quoted 1 line

In other words, all these SoCs have the same controller.

False :)

To be more precice, the statement itself may be true or false, but the implication ("In other words, ...") is definitely false.

Quoted 2 lines

OK, so the controllers are not identical, but same enough to have a common driver?

Right, the controllers are compatible, but likely they are non-identical.

Quoted 2 lines

Is the driver not tested enough or are you planning to add more features?

Test results can not serve as a formal eternal proof, but they are good as a hint. Also I'm not aware of any pending features to be added to the driver.

Quoted 11 lines
Quoted 7 lines
Quoted 3 lines

So its about have just one compatible right now, and add more if some new SoC comes with a variation of the controller.

True. The driver will be changed in this case, unfortunately the bindings are not so volatile.

The volatility of the bindings will be same when you add a new SoC compatible ;) And you must add to bindings+driver, or the same h/w and same driver won't work for your new platform.

The (non-)volatilitily of the bindings is important only in retrospective view.

Adding a new compatible to the documentation and unrelated board DTBs is irrelevant, and it is quite predictable that it will happen, when a new SoC with the same "compatible" IP is released.

--
Best wishes,
Vladimir

maintainer
#63 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

In reply to Jassi Brar
Quoted 9 lines · Jassi Brar
Quoted 7 lines · Oleksij Rempel

This are currently tested SoCs with imx-mailbox driver.

---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 16 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 11 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
index 113d6ab931ef..5616d2afca45 100644
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller? I think the controller name, if it has one or create a new one like 'imx-mu', should be used here.
So,
compatible : should be "fsl,imx-mu"

No, for all the reasons already correctly explained in this thread.

Quoted 2 lines

BTW, the driver doesn't even probe anything other than 'fsl,imx6sx-mu', unlike what the binding says.

This aspect should be documented here though. The documentation should clearly define what are valid combinations of compatibles.

Rob

maintainer
#64 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

Quoted 27 lines · Rob Herring
Quoted 9 lines · Jassi Brar
Quoted 7 lines · Oleksij Rempel

This are currently tested SoCs with imx-mailbox driver.

---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 16 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 11 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
index 113d6ab931ef..5616d2afca45 100644
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller? I think the controller name, if it has one or create a new one like 'imx-mu', should be used here.
So,
compatible : should be "fsl,imx-mu"

Quoted 2 lines

No, for all the reasons already correctly explained in this thread.

OK. I curious

Quoted 7 lines
Quoted 2 lines

BTW, the driver doesn't even probe anything other than 'fsl,imx6sx-mu', unlike what the binding says.

This aspect should be documented here though. The documentation should clearly define what are valid combinations of compatibles.

Rob

maintainer
#65 lore raw

PATCH v7 4/6 dt-bindings: mailbox: imx-mu: add i.MX6SX and i.MX7S SoCs.

In reply to Rob Herring
Quoted 27 lines · Rob Herring
Quoted 9 lines · Jassi Brar
Quoted 7 lines · Oleksij Rempel

This are currently tested SoCs with imx-mailbox driver.

---
 Documentation/devicetree/bindings/mailbox/fsl,mu.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 16 lines

Documentation/devicetree/bindings/mailbox/fsl,mu.txt Messaging Unit Device Node: · line 22

Quoted 11 lines
Documentation/devicetree/bindings/mailbox/fsl,mu.txt +1 -1
index 113d6ab931ef..5616d2afca45 100644
@@ -18,7 +18,7 @@ Messaging Unit Device Node:
 Required properties:
 -------------------
 - compatible : should be "fsl,<chip>-mu", the supported chips include
-               imx8qxp, imx8qm.
+               imx6sx, imx7s, imx8qxp, imx8qm.
 

This is not scalable. Do we add every new SoC that contains the same controller? I think the controller name, if it has one or create a new one like 'imx-mu', should be used here.
So,
compatible : should be "fsl,imx-mu"

Quoted 2 lines

No, for all the reasons already correctly explained in this thread.

OK. I am curious to see what soc specific fixes we need in future.

Thanks.

#66 lore raw

PATCH v7 5/6 ARM: dts: imx7s: add i.MX7 messaging unit support

In reply to Oleksij Rempel

Define the Messaging Unit (MU) for i.MX7 in the processor's dtsi. The respective driver is added in the next commit.

---
 arch/arm/boot/dts/imx7s.dtsi | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
arch/arm/boot/dts/imx7s.dtsi +19 -0
index ce85b3ca1a55..191a0286fa3b 100644
@@ -1001,6 +1001,25 @@
 				status = "disabled";
 			};
 
+			mu0a: mailbox at 30aa0000 {
+				compatible = "fsl,imx7s-mu", "fsl,imx6sx-mu";
+				reg = <0x30aa0000 0x10000>;
+				interrupts = <GIC_SPI 88 IRQ_TYPE_LEVEL_HIGH>;
+				clocks = <&clks IMX7D_MU_ROOT_CLK>;
+				#mbox-cells = <1>;
+				status = "disabled";
+			};
+
+			mu0b: mailbox at 30ab0000 {
+				compatible = "fsl,imx7s-mu", "fsl,imx6sx-mu";
+				reg = <0x30ab0000 0x10000>;
+				interrupts = <GIC_SPI 97 IRQ_TYPE_LEVEL_HIGH>;
+				clocks = <&clks IMX7D_MU_ROOT_CLK>;
+				#mbox-cells = <1>;
+				fsl,mu-side-b;
+				status = "disabled";
+			};
+
 			usbotg1: usb at 30b10000 {
 				compatible = "fsl,imx7d-usb", "fsl,imx27-usb";
 				reg = <0x30b10000 0x200>;
Signature · 1 line
2.18.0
#67 lore raw

PATCH v7 5/6 ARM: dts: imx7s: add i.MX7 messaging unit support

Am Donnerstag, den 26.07.2018, 08:53 +0200 schrieb Oleksij Rempel:

Quoted 5 lines

Define the Messaging Unit (MU) for i.MX7 in the processor's dtsi. The respective driver is added in the next commit.

Quoted 1 line
Quoted 34 lines

---
?arch/arm/boot/dts/imx7s.dtsi | 19 +++++++++++++++++++ ?1 file changed, 19 insertions(+)

arch/arm/boot/dts/imx7s.dtsi +0 -0
index ce85b3ca1a55..191a0286fa3b 100644
@@ -1001,6 +1001,25 @@
 > ?				status = "disabled";
 > ?			};

?

arch/arm/boot/dts/imx7s.dtsi line 1011

Quoted 8 lines
Quoted 1 line
arch/arm/boot/dts/imx7s.dtsi +1 -0
+			mu0a: mailbox at 30aa0000 {
arch/arm/boot/dts/imx7s.dtsi +7 -0
+				compatible = "fsl,imx7s-mu", "fsl,imx6sx-mu";
+				reg = <0x30aa0000 0x10000>;
+				interrupts = <GIC_SPI 88 IRQ_TYPE_LEVEL_HIGH>;
+				clocks = <&clks IMX7D_MU_ROOT_CLK>;
+				#mbox-cells = <1>;
+				status = "disabled";
+			};

+

arch/arm/boot/dts/imx7s.dtsi line 1021

Quoted 9 lines
Quoted 1 line
arch/arm/boot/dts/imx7s.dtsi +1 -0
+			mu0b: mailbox at 30ab0000 {
arch/arm/boot/dts/imx7s.dtsi +8 -0
+				compatible = "fsl,imx7s-mu", "fsl,imx6sx-mu";
+				reg = <0x30ab0000 0x10000>;
+				interrupts = <GIC_SPI 97 IRQ_TYPE_LEVEL_HIGH>;
+				clocks = <&clks IMX7D_MU_ROOT_CLK>;
+				#mbox-cells = <1>;
+				fsl,mu-side-b;
+				status = "disabled";
+			};

+

Quoted 3 lines
Quoted 1 line

? usbotg1: usb at 30b10000 {

? compatible = "fsl,imx7d-usb", "fsl,imx27-usb";
? reg = <0x30b10000 0x200>;

#68 lore raw

PATCH v7 6/6 mailbox: Add support for i.MX7D messaging unit

In reply to Oleksij Rempel

The Mailbox controller is able to send messages (up to 4 32 bit words) between the endpoints.

This driver was tested using the mailbox-test driver sending messages between the Cortex-A7 and the Cortex-M4.

---
 drivers/mailbox/Kconfig       |   6 +
 drivers/mailbox/Makefile      |   2 +
 drivers/mailbox/imx-mailbox.c | 276 ++++++++++++++++++++++++++++++++++
 3 files changed, 284 insertions(+)
 create mode 100644 drivers/mailbox/imx-mailbox.c
drivers/mailbox/Kconfig +6 -0
index a2bb27446dce..79060ddc380d 100644
@@ -15,6 +15,12 @@ config ARM_MHU
 	  The controller has 3 mailbox channels, the last of which can be
 	  used in Secure mode only.
 
+config IMX_MBOX
+	tristate "i.MX Mailbox"
+	depends on ARCH_MXC || COMPILE_TEST
+	help
+	  Mailbox implementation for i.MX Messaging Unit (MU).
+
 config PLATFORM_MHU
 	tristate "Platform MHU Mailbox"
 	depends on OF
drivers/mailbox/Makefile +2 -0
index cc23c3a43fcd..ba2fe1b6dd62 100644
@@ -7,6 +7,8 @@ obj-$(CONFIG_MAILBOX_TEST)	+= mailbox-test.o
 
 obj-$(CONFIG_ARM_MHU)	+= arm_mhu.o
 
+obj-$(CONFIG_IMX_MBOX)	+= imx-mailbox.o
+
 obj-$(CONFIG_PLATFORM_MHU)	+= platform_mhu.o
 
 obj-$(CONFIG_PL320_MBOX)	+= pl320-ipc.o
drivers/mailbox/imx-mailbox.c new file +276 -0

276 lines not on this page · raw message

Signature · 1 line
2.18.0
#69 lore raw

PATCH v7 6/6 mailbox: Add support for i.MX7D messaging unit

Am Donnerstag, den 26.07.2018, 08:53 +0200 schrieb Oleksij Rempel:

Quoted 9 lines

The Mailbox controller is able to send messages (up to 4 32 bit words) between the endpoints.

This driver was tested using the mailbox-test driver sending messages between the Cortex-A7 and the Cortex-M4.

Quoted 2 lines
Quoted 319 lines

---

?drivers/mailbox/Kconfig???????|???6 +
?drivers/mailbox/Makefile??????|???2 +
?drivers/mailbox/imx-mailbox.c | 276 ++++++++++++++++++++++++++++++++++

?3 files changed, 284 insertions(+)
?create mode 100644 drivers/mailbox/imx-mailbox.c

drivers/mailbox/Kconfig +0 -0
index a2bb27446dce..79060ddc380d 100644
@@ -15,6 +15,12 @@ config ARM_MHU
Quoted 2 lines

? ??The controller has 3 mailbox channels, the last of which can be
? ??used in Secure mode only.

?
+config IMX_MBOX

drivers/mailbox/Kconfig config ARM_MHU

Quoted 4 lines
drivers/mailbox/Kconfig +4 -0
+	tristate "i.MX Mailbox"
+	depends on ARCH_MXC || COMPILE_TEST
+	help
+	??Mailbox implementation for i.MX Messaging Unit (MU).

+
?config PLATFORM_MHU

Quoted 2 lines

? tristate "Platform MHU Mailbox"
? depends on OF

drivers/mailbox/Makefile +0 -0
index cc23c3a43fcd..ba2fe1b6dd62 100644

Review obj-$(CONFIG_MAILBOX_TEST) += mailbox-test.o

Quoted 1 line
(quoted diff) +0 -0
@@ -7,6 +7,8 @@ obj-$(CONFIG_MAILBOX_TEST)	+= mailbox-test.o

?

Quoted 1 line

?obj-$(CONFIG_ARM_MHU) += arm_mhu.o

?

Quoted 1 line

+obj-$(CONFIG_IMX_MBOX) += imx-mailbox.o

+

Quoted 1 line

?obj-$(CONFIG_PLATFORM_MHU) += platform_mhu.o

?

Quoted 1 line

?obj-$(CONFIG_PL320_MBOX) += pl320-ipc.o

drivers/mailbox/imx-mailbox.c new file +15 -0
new file mode 100644
index 000000000000..ecfddca5e31c
@@ -0,0 +1,276 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2018 Pengutronix, Oleksij Rempel <[email protected]>
+ */
+
+#include <linux/clk.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/mailbox_controller.h>
+#include <linux/module.h>
+#include <linux/of_device.h>
+#include <linux/slab.h>
+
+/* Transmit Register */
Quoted 1 line

+#define IMX_MU_xTRn(x) (0x00 + 4 * (x))

+/* Receive Register */

Quoted 1 line

+#define IMX_MU_xRRn(x) (0x10 + 4 * (x))

+/* Status Register */

Quoted 4 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+#define IMX_MU_xSR		0x20
+#define IMX_MU_xSR_TEn(x)	BIT(20 + (3 - (x)))
+#define IMX_MU_xSR_RFn(x)	BIT(24 + (3 - (x)))
+#define IMX_MU_xSR_BRDIP	BIT(9)
drivers/mailbox/imx-mailbox.c new file +2 -0
+
+/* Control Register */
Quoted 1 line

+#define IMX_MU_xCR 0x24

+/* Transmit Interrupt Enable */

Quoted 1 line

+#define IMX_MU_xCR_TIEn(x) BIT(20 + (3 - (x)))

+/* Receive Interrupt Enable */

Quoted 1 line

+#define IMX_MU_xCR_RIEn(x) BIT(24 + (3 - (x)))

+

Quoted 1 line

+#define IMX_MU_CHANS 4u

drivers/mailbox/imx-mailbox.c new file +2 -0
+
+struct imx_mu_con_priv {
Quoted 3 lines
Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	int			irq;
+	unsigned int		idx;
+	char			*irq_desc;
drivers/mailbox/imx-mailbox.c new file +3 -0
+};
+
+struct imx_mu_priv {
Quoted 3 lines

drivers/mailbox/imx-mailbox.c line 43

Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	struct device		*dev;
+	void __iomem		*base;
+	spinlock_t		xcr_lock; /* control register lock */

+

Quoted 2 lines

drivers/mailbox/imx-mailbox.c line 46

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	struct mbox_controller	mbox;
+	struct mbox_chan	mbox_chans[IMX_MU_CHANS];

+

Quoted 2 lines
(quoted diff) +1 -0
+	struct imx_mu_con_priv??con_priv[IMX_MU_CHANS];

drivers/mailbox/imx-mailbox.c line 49

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	struct clk		*clk;

+

Quoted 1 line
Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	bool			side_b;
drivers/mailbox/imx-mailbox.c new file +4 -0
+};
+
+static struct imx_mu_priv *to_imx_mu_priv(struct mbox_controller *mbox)
+{
Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	return container_of(mbox, struct imx_mu_priv, mbox);
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static void imx_mu_write(struct imx_mu_priv *priv, u32 val, u32 offs)
+{
Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	iowrite32(val, priv->base + offs);
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static u32 imx_mu_read(struct imx_mu_priv *priv, u32 offs)
+{
Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	return ioread32(priv->base + offs);
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static u32 imx_mu_xcr_rmw(struct imx_mu_priv *priv, u32 set, u32 clr)
+{

drivers/mailbox/imx-mailbox.c line 72

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	unsigned long flags;
+	u32 val;

+

drivers/mailbox/imx-mailbox.c line 79

Quoted 6 lines
drivers/mailbox/imx-mailbox.c new file +6 -0
+	spin_lock_irqsave(&priv->xcr_lock, flags);
+	val = imx_mu_read(priv, IMX_MU_xCR);
+	val &= ~clr;
+	val |= set;
+	imx_mu_write(priv, val, IMX_MU_xCR);
+	spin_unlock_irqrestore(&priv->xcr_lock, flags);

+

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	return val;
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static irqreturn_t imx_mu_isr(int irq, void *p)
+{

drivers/mailbox/imx-mailbox.c line 89

Quoted 4 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+	struct mbox_chan *chan = p;
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+	u32 val, ctrl, dat;

+

drivers/mailbox/imx-mailbox.c line 96

Quoted 6 lines
drivers/mailbox/imx-mailbox.c new file +6 -0
+	ctrl = imx_mu_read(priv, IMX_MU_xCR);
+	val = imx_mu_read(priv, IMX_MU_xSR);
+	val &= IMX_MU_xSR_TEn(cp->idx) | IMX_MU_xSR_RFn(cp->idx);
+	val &= ctrl & (IMX_MU_xCR_TIEn(cp->idx) | IMX_MU_xCR_RIEn(cp->idx));
+	if (!val)
+		return IRQ_NONE;

+

drivers/mailbox/imx-mailbox.c line 101

Quoted 4 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+	if (val & IMX_MU_xSR_TEn(cp->idx)) {
+		imx_mu_xcr_rmw(priv, 0, IMX_MU_xCR_TIEn(cp->idx));
+		mbox_chan_txdone(chan, 0);
+	}

+

drivers/mailbox/imx-mailbox.c line 106

Quoted 4 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+	if (val & IMX_MU_xSR_RFn(cp->idx)) {
+		dat = imx_mu_read(priv, IMX_MU_xRRn(cp->idx));
+		mbox_chan_received_data(chan, (void *)&dat);
+	}

+

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	return IRQ_HANDLED;
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static bool imx_mu_last_tx_done(struct mbox_chan *chan)
+{

Review

Quoted 2 lines
(quoted diff) +2 -0
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;

+

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	/* test if transmit register is empty */
+	return imx_mu_read(priv, IMX_MU_xSR) & IMX_MU_xSR_TEn(cp->idx);
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static int imx_mu_send_data(struct mbox_chan *chan, void *data)
+{

drivers/mailbox/imx-mailbox.c line 124

Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+	u32 *arg = data;

+

drivers/mailbox/imx-mailbox.c line 127

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	if (!imx_mu_last_tx_done(chan))
+		return -EBUSY;

+

drivers/mailbox/imx-mailbox.c line 130

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	imx_mu_write(priv, *arg, IMX_MU_xTRn(cp->idx));
+	imx_mu_xcr_rmw(priv, IMX_MU_xSR_TEn(cp->idx), 0);

+

Quoted 1 line
(quoted diff) +1 -0
+	return 0;
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static int imx_mu_startup(struct mbox_chan *chan)
+{

drivers/mailbox/imx-mailbox.c line 139

Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+	int ret;

+

drivers/mailbox/imx-mailbox.c line 143

Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	cp->irq_desc = kasprintf(GFP_KERNEL, "imx_mu_chan[%i]", cp->idx);
+	if (!cp->irq_desc)
+		return -ENOMEM;

+

drivers/mailbox/imx-mailbox.c line 151

Quoted 7 lines
drivers/mailbox/imx-mailbox.c new file +7 -0
+	ret = request_irq(cp->irq, imx_mu_isr, IRQF_SHARED, cp->irq_desc,
+			??chan);
+	if (ret) {
+		dev_err(priv->dev,
+			"Unable to acquire IRQ %d\n", cp->irq);
+		return ret;
+	}

+

drivers/mailbox/imx-mailbox.c line 153

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	imx_mu_xcr_rmw(priv, IMX_MU_xCR_RIEn(cp->idx), 0);

+

Quoted 1 line
(quoted diff) +1 -0
+	return 0;
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static void imx_mu_shutdown(struct mbox_chan *chan)
+{

Review

Quoted 2 lines
(quoted diff) +2 -0
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;

+

Review

Quoted 2 lines
(quoted diff) +2 -0
+	imx_mu_xcr_rmw(priv, 0,
+		???IMX_MU_xCR_TIEn(cp->idx) | IMX_MU_xCR_RIEn(cp->idx));

+

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	free_irq(cp->irq, chan);
+	kfree(cp->irq_desc);
drivers/mailbox/imx-mailbox.c new file +3 -0
+}
+
+static const struct mbox_chan_ops imx_mu_ops = {
Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	.send_data = imx_mu_send_data,
+	.startup = imx_mu_startup,
+	.shutdown = imx_mu_shutdown,
drivers/mailbox/imx-mailbox.c new file +4 -0
+};
+
+static void imx_mu_init_generic(struct imx_mu_priv *priv)
+{

drivers/mailbox/imx-mailbox.c line 179

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	if (priv->side_b)
+		return;

+

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	/* Set default MU configuration */
+	imx_mu_write(priv, 0, IMX_MU_xCR);
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static int imx_mu_probe(struct platform_device *pdev)
+{

drivers/mailbox/imx-mailbox.c line 192

Quoted 6 lines
drivers/mailbox/imx-mailbox.c new file +6 -0
+	struct device *dev = &pdev->dev;
+	struct device_node *np = dev->of_node;
+	struct resource *iomem;
+	struct imx_mu_priv *priv;
+	unsigned int i;
+	int irq, ret;

+

drivers/mailbox/imx-mailbox.c line 196

Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;

+

drivers/mailbox/imx-mailbox.c line 198

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	priv->dev = dev;

+

drivers/mailbox/imx-mailbox.c line 203

Quoted 4 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+	iomem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	priv->base = devm_ioremap_resource(&pdev->dev, iomem);
+	if (IS_ERR(priv->base))
+		return PTR_ERR(priv->base);

+

drivers/mailbox/imx-mailbox.c line 207

Quoted 3 lines
drivers/mailbox/imx-mailbox.c new file +3 -0
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0)
+		return irq;

+

drivers/mailbox/imx-mailbox.c line 212

Quoted 4 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+	priv->clk = devm_clk_get(dev, NULL);
+	if (IS_ERR(priv->clk)) {
+		if (PTR_ERR(priv->clk) != -ENOENT)
+			return PTR_ERR(priv->clk);

+

Review

Quoted 2 lines
(quoted diff) +2 -0
+		priv->clk = NULL;
+	}

+

drivers/mailbox/imx-mailbox.c line 221

Quoted 5 lines
drivers/mailbox/imx-mailbox.c new file +5 -0
+	ret = clk_prepare_enable(priv->clk);
+	if (ret) {
+		dev_err(dev, "Failed to enable clock\n");
+		return ret;
+	}

+

drivers/mailbox/imx-mailbox.c line 224

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	for (i = 0; i < IMX_MU_CHANS; i++) {
+		struct imx_mu_con_priv *cp = &priv->con_priv[i];

+

drivers/mailbox/imx-mailbox.c line 229

Quoted 4 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+		cp->idx = i;
+		cp->irq = irq;
+		priv->mbox_chans[i].con_priv = cp;
+	}

+

drivers/mailbox/imx-mailbox.c line 231

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	priv->side_b = of_property_read_bool(np, "fsl,mu-side-b");

+

drivers/mailbox/imx-mailbox.c line 233

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	spin_lock_init(&priv->xcr_lock);

+

drivers/mailbox/imx-mailbox.c line 239

Quoted 5 lines
drivers/mailbox/imx-mailbox.c new file +5 -0
+	priv->mbox.dev = dev;
+	priv->mbox.ops = &imx_mu_ops;
+	priv->mbox.chans = priv->mbox_chans;
+	priv->mbox.num_chans = IMX_MU_CHANS;
+	priv->mbox.txdone_irq = true;

+

drivers/mailbox/imx-mailbox.c line 241

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	platform_set_drvdata(pdev, priv);

+

drivers/mailbox/imx-mailbox.c line 243

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	imx_mu_init_generic(priv);

+

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	return mbox_controller_register(&priv->mbox);
drivers/mailbox/imx-mailbox.c new file +4 -0
+}
+
+static int imx_mu_remove(struct platform_device *pdev)
+{

drivers/mailbox/imx-mailbox.c line 250

Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+	struct imx_mu_priv *priv = platform_get_drvdata(pdev);

+

drivers/mailbox/imx-mailbox.c line 253

Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	mbox_controller_unregister(&priv->mbox);
+	clk_disable_unprepare(priv->clk);

+

Quoted 1 line
(quoted diff) +1 -0
+	return 0;
drivers/mailbox/imx-mailbox.c new file +3 -0
+}
+
+static const struct of_device_id imx_mu_dt_ids[] = {
Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	{ .compatible = "fsl,imx6sx-mu" },
+	{ },
drivers/mailbox/imx-mailbox.c new file +4 -0
+};
+MODULE_DEVICE_TABLE(of, imx_mu_dt_ids);
+
+static struct platform_driver imx_mu_driver = {
Quoted 6 lines
Quoted 2 lines
drivers/mailbox/imx-mailbox.c new file +2 -0
+	.probe		= imx_mu_probe,
+	.remove		= imx_mu_remove,
drivers/mailbox/imx-mailbox.c new file +1 -0
+	.driver = {
Quoted 1 line
drivers/mailbox/imx-mailbox.c new file +1 -0
+		.name	= "imx_mu",
drivers/mailbox/imx-mailbox.c new file +2 -0
+		.of_match_table = imx_mu_dt_ids,
+	},
drivers/mailbox/imx-mailbox.c new file +3 -0
+};
+module_platform_driver(imx_mu_driver);
+
Quoted 1 line

+MODULE_AUTHOR("Oleksij Rempel <[email protected]>");

drivers/mailbox/imx-mailbox.c new file +2 -0
+MODULE_DESCRIPTION("Message Unit driver for i.MX");
+MODULE_LICENSE("GPL v2");
maintainer
#70 lore raw

PATCH v7 6/6 mailbox: Add support for i.MX7D messaging unit

In reply to Oleksij Rempel

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel
<[email protected]> wrote:

.....

drivers/mailbox/imx-mailbox.c line 33

Quoted 10 lines
drivers/mailbox/imx-mailbox.c new file +10 -0
+
+/* Control Register */
+#define IMX_MU_xCR             0x24
+/* Transmit Interrupt Enable */
+#define IMX_MU_xCR_TIEn(x)     BIT(20 + (3 - (x)))
+/* Receive Interrupt Enable */
+#define IMX_MU_xCR_RIEn(x)     BIT(24 + (3 - (x)))
+
+#define IMX_MU_CHANS   4u
+

The 'u' unnecessarily refreshes my knowledge of C :)

drivers/mailbox/imx-mailbox.c line 37

Quoted 5 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+struct imx_mu_con_priv {
+       int                     irq;
+       unsigned int            idx;
+       char                    *irq_desc;

Having char irq_desc[16]; seems simpler than free/kfree upon
each startup/shutdown. Simply use snprintf() in probe().

The rest of the driver seems fine.

Cheers!

#71 lore raw

PATCH v7 6/6 mailbox: Add support for i.MX7D messaging unit

Quoted 5 lines · Jassi Brar

On Thu, Jul 26, 2018 at 12:23 PM, Oleksij Rempel
<[email protected]> wrote:

.....

Quoted 11 lines

drivers/mailbox/imx-mailbox.c line 33

Quoted 10 lines
drivers/mailbox/imx-mailbox.c new file +10 -0
+
+/* Control Register */
+#define IMX_MU_xCR             0x24
+/* Transmit Interrupt Enable */
+#define IMX_MU_xCR_TIEn(x)     BIT(20 + (3 - (x)))
+/* Receive Interrupt Enable */
+#define IMX_MU_xCR_RIEn(x)     BIT(24 + (3 - (x)))
+
+#define IMX_MU_CHANS   4u
+

The 'u' unnecessarily refreshes my knowledge of C :)

ok

Quoted 8 lines

drivers/mailbox/imx-mailbox.c line 37

Quoted 6 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+struct imx_mu_con_priv {
+       int                     irq;
+       unsigned int            idx;
+       char                    *irq_desc;

Having char irq_desc[16]; seems simpler than free/kfree upon
each startup/shutdown. Simply use snprintf() in probe().

Currently all known configurations use one or two channels. I don't see any sense to allocate all at the probe.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180726/6f9f1e09/attachment.sig>

maintainer
#72 lore raw

PATCH v7 6/6 mailbox: Add support for i.MX7D messaging unit

Quoted 6 lines · Oleksij Rempel
Quoted 6 lines
Quoted 6 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+struct imx_mu_con_priv {
+       int                     irq;
+       unsigned int            idx;
+       char                    *irq_desc;
Quoted 6 lines
Quoted 2 lines

Having char irq_desc[16]; seems simpler than free/kfree upon
each startup/shutdown. Simply use snprintf() in probe().

Currently all known configurations use one or two channels. I don't see any sense to allocate all at the probe.

Just because only 2 out of 4 channels are being used, it doesn't make sense?

What is the concern? "wasting" 16x2 bytes of memory? How many bytes, LOC and execution time does kasprintf() waste?

#73 lore raw

PATCH v7 6/6 mailbox: Add support for i.MX7D messaging unit

Quoted 11 lines · Jassi Brar
Quoted 9 lines · Oleksij Rempel
Quoted 8 lines

drivers/mailbox/imx-mailbox.c line 37

Quoted 6 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+struct imx_mu_con_priv {
+       int                     irq;
+       unsigned int            idx;
+       char                    *irq_desc;

Having char irq_desc[16]; seems simpler than free/kfree upon
each startup/shutdown. Simply use snprintf() in probe().

Quoted 7 lines
Quoted 3 lines

Currently all known configurations use one or two channels. I don't see any sense to allocate all at the probe.

Just because only 2 out of 4 channels are being used, it doesn't make sense?

What is the concern? "wasting" 16x2 bytes of memory? How many bytes, LOC and execution time does kasprintf() waste?

Startup or shutdown is not speed critical path in this driver, or do i miss some thing? Speed vs size is not valid argumentation here.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20180726/825beecf/attachment.sig>

maintainer
#74 lore raw

PATCH v7 6/6 mailbox: Add support for i.MX7D messaging unit

Quoted 17 lines · Oleksij Rempel
Quoted 14 lines · Jassi Brar
Quoted 12 lines · Oleksij Rempel
Quoted 8 lines

drivers/mailbox/imx-mailbox.c line 37

Quoted 6 lines
drivers/mailbox/imx-mailbox.c new file +4 -0
+struct imx_mu_con_priv {
+       int                     irq;
+       unsigned int            idx;
+       char                    *irq_desc;

Having char irq_desc[16]; seems simpler than free/kfree upon
each startup/shutdown. Simply use snprintf() in probe().

Currently all known configurations use one or two channels. I don't see any sense to allocate all at the probe.

Quoted 8 lines
Quoted 4 lines

Just because only 2 out of 4 channels are being used, it doesn't make sense?

What is the concern? "wasting" 16x2 bytes of memory? How many bytes, LOC and execution time does kasprintf() waste?

Startup or shutdown is not speed critical path in this driver, or do i miss some thing?

No they are not.

Quoted 2 lines

Speed vs size is not valid argumentation here.

If the code looks cleaner and smaller and lesser expensive execution wise, at least I see that as a valid enough argument.

But OK, if you insist. If someone sends a "cleanup" patch tomorrow, I am going to thank and apply.