MCP23017 interrupt broke on 6.6.y overlays #5677

hftsai256 · 2023-10-24T02:59:16Z

Describe the bug

GPIO interrupt is not configured correctly when setting up mcp23017 with devicetree overlay in config.txt with kernel branch rpi-6.6.y.

dtoverlay=mcp23017,addr=0x23,gpiopin=23

If I roll back to the latest Raspberry OS Lite image (2023-10-10-raspios-bookworm-arm64-lite), the interrupt will then be configured as expected and can show up under /proc/interrupts:

 42:     418457          0          0          0  pinctrl-bcm2835  23 Level     1-0023

Steps to reproduce the behaviour

Start with the latest Raspberry OS Lite release (2023-10-10).
Cross-compile the kernel from rpi-6.6.y branch, commit f275f8f: Typo in overlays README

KERNEL=kernel8
make ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- bcm2711_defconfig
make ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- menuconfig
make ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- -j8 Image modules dtbs

Install kernel image to SD card/mounted eMMC:

cp arch/arm64/boot/Image $BOOTFS/$KERNEL.img

Install kernel modules:

sudo env PATH=$PATH make ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- INSTALL_MOD_PATH=$ROOTFS modules_install

Install Broadcom devicetree blobs.

cp arch/arm64/boot/dts/broadcom/*.dtb $BOOTFS/

Install overlays (breaking step)

cp arch/arm64/boot/dts/overlays/*.dtb* $BOOTFS/overlays/

Device (s)

Raspberry Pi 4 Mod. B

System

~$ cat /etc/rpi-issue
Raspberry Pi reference 2023-10-10
Generated using pi-gen, https://github.com/RPi-Distro/pi-gen, 962bf483c8f326405794827cce8c0313fd5880a8, stage2

~$ vcgencmd version
Aug 10 2023 15:33:38
Copyright (c) 2012 Broadcom
version 03dc77429335caee083e22ddc8eec09c07f12a7a (clean) (release) (start)

~$ uname -a
Linux pidev 6.6.0-rc6-v8+ #5 SMP PREEMPT Mon Oct 23 16:14:25 EDT 2023 aarch64 GNU/Linux

Logs

dmesg at step 5 (working): http://0x0.st/HJDN.txt
dmesg at step 6 (broken): http://0x0.st/HJDK.txt

Additional context

I dumped the devicetree base between step 5 and step 6. A diff quickly unveils the culprit:

--- step6_broken.dts
+++ step5_working.diff
@@ -248,12 +248,17 @@
 			phandle = <0x48>;
 
 			mcp@23 {
+				microchip,irq-mirror;
 				gpio-controller;
+				interrupts = <0x17 0x02>;
+				interrupt-parent = <0x07>;
 				compatible = "microchip,mcp23017";
+				#interrupt-cells = <0x02>;
 				status = "okay";
 				reg = <0x23>;
-				phandle = <0xf1>;
+				phandle = <0xf0>;
 				#gpio-cells = <0x02>;
+				interrupt-controller;
 			};
 		};
 
@@ -1170,7 +1175,7 @@
 
 			mcp23017_pins@23 {
 				brcm,function = <0x00>;
-				phandle = <0xf2>;
+				phandle = <0xf1>;
 				brcm,pins = <0x17>;
 			};

Not only the interrupt-related properties are missing from step 6, but the phandle addresses are also not consistent across the whole tree.

As a side note, I'm under the impression that this was working on 6.6.0-rc2. I noticed this problem when I updated the kernel to rc-6 recently.

The text was updated successfully, but these errors were encountered:

hftsai256 · 2023-10-24T03:25:46Z

I reverted mcp23017-overlay.dts back to commit 06e36e8 seems resolve the problem. Please review the breaking change on 27f0e0e.

The incorrect fragment order (*) caused broke the interrupt usage, and while it was being fixed the lack of a reference to the pinctrl declaration was noticed. See: raspberrypi#5677 Signed-off-by: Phil Elwell <[email protected]> (*) Ideally all fragments would appear in the file in the order in which they should be merged, but that is easy to forget and can be awkward, so the firmware merges all "intra" fragments (those that target other fragments in the overlay) before "inter" fragments (those that target the base DTB). However, intra fragments that target other intra fragments is a level of nesting too far for this logic to cope, so they must appear before the fragments they target.

pelwell · 2023-10-24T09:19:51Z

This is caused by the ease of getting the fragment order wrong, something I'd like to detect and automatically warn about, and actually affects rpi-6.1.y as well. If you have time it would be helpful it you could install a trial kernel - wait about an hour until the build checks (https://github.com/raspberrypi/linux/pull/5678/checks) complete then run sudo rpi-update pulls/5678.

pelwell · 2023-10-24T10:32:59Z

run sudo rpi-update pulls/5678

This is ready to test now, if you are able to.

The incorrect fragment order (*) caused broke the interrupt usage, and while it was being fixed the lack of a reference to the pinctrl declaration was noticed. See: #5677 Signed-off-by: Phil Elwell <[email protected]> (*) Ideally all fragments would appear in the file in the order in which they should be merged, but that is easy to forget and can be awkward, so the firmware merges all "intra" fragments (those that target other fragments in the overlay) before "inter" fragments (those that target the base DTB). However, intra fragments that target other intra fragments is a level of nesting too far for this logic to cope, so they must appear before the fragments they target.

hftsai256 · 2023-10-24T14:03:53Z

The pull rolled it to kernel 6.1.58-v8+. The targeted GPIO pin 23 is marked as unused, and there is no entry found in /proc/interrupts. The interrupt-related properties are now populated in mcp@23, but the phandle is still inconsistent.

                         mcp@23 {
                                microchip,irq-mirror;
                                pinctrl-name = "default";
                                pinctrl-0 = <0x104>;
                                gpio-controller;
                                interrupts = <0x17 0x02>;
                                interrupt-parent = <0x07>;
                                compatible = "microchip,mcp23017";
                                #interrupt-cells = <0x02>;
                                status = "okay";
                                reg = <0x23>;
                                phandle = <0x103>;
                                #gpio-cells = <0x02>;
                                interrupt-controller;
                        };

...

                        mcp23017_pins@23 {
                                brcm,function = <0x00>;
                                phandle = <0x104>;
                                brcm,pins = <0x17>;
                        };

pelwell · 2023-10-24T14:09:29Z

the phandle is still inconsistent.

Inconsistent or different? phandles are allocated by the dtc compiler and firmware, and depend on the entire content of the dts/dtb file. What matters is that if in one branch node A uses phandle X and node B has the property phandle = <X>;, that in other branches node A uses phandle Y and node B has the property phandle = <Y>;, where X and Y may be different values.

The incorrect fragment order (*) caused broke the interrupt usage, and while it was being fixed the lack of a reference to the pinctrl declaration was noticed. See: #5677 Signed-off-by: Phil Elwell <[email protected]> (*) Ideally all fragments would appear in the file in the order in which they should be merged, but that is easy to forget and can be awkward, so the firmware merges all "intra" fragments (those that target other fragments in the overlay) before "inter" fragments (those that target the base DTB). However, intra fragments that target other intra fragments is a level of nesting too far for this logic to cope, so they must appear before the fragments they target.

hftsai256 · 2023-10-24T14:14:51Z

Thanks for the clarification. "Different" may be the proper term in this case. Are there other places in your mind to investigate?

pelwell · 2023-10-24T14:24:00Z

Are there any errors or warnings in the kernel log? I don't have an MCP23017 to test.

hftsai256 · 2023-10-24T14:40:32Z

The problem was due to a short on one of the i2c addressing pins. The interrupt is now adequately configured with your new patch. Thanks.

pelwell · 2023-10-24T14:42:00Z

Thank goodness for that - you had me worried.
The changes are now in 6.5 and 6.6.

pelwell mentioned this issue Oct 24, 2023

fixup! overlays: mcp23017: allow specification of the i2c bus #5678

Merged

pelwell closed this as completed Nov 17, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

MCP23017 interrupt broke on 6.6.y overlays #5677

MCP23017 interrupt broke on 6.6.y overlays #5677

hftsai256 commented Oct 24, 2023 •

edited

Loading

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023 •

edited

Loading

Uh oh!

pelwell commented Oct 24, 2023

Uh oh!

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023 •

edited

Loading

Uh oh!

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023

Uh oh!

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023

Uh oh!

MCP23017 interrupt broke on 6.6.y overlays #5677

MCP23017 interrupt broke on 6.6.y overlays #5677

Comments

hftsai256 commented Oct 24, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Describe the bug

Steps to reproduce the behaviour

Device (s)

System

Logs

Additional context

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pelwell commented Oct 24, 2023

Uh oh!

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023

Uh oh!

hftsai256 commented Oct 24, 2023

Uh oh!

pelwell commented Oct 24, 2023

Uh oh!

hftsai256 commented Oct 24, 2023 •

edited

Loading

pelwell commented Oct 24, 2023 •

edited

Loading

pelwell commented Oct 24, 2023 •

edited

Loading