Skip to contentSkip to conversation

gpiolib: ensure that fwnode is properly set

RFC PATCH In review Brian Masney 10 participants 14 messages Nov 14, 2022 – Nov 16, 2022

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 gpiolib: ensure that fwnode is properly set

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

---
 drivers/gpio/gpiolib.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +1 -1
index 11fb7ec883e9..8bec66008869 100644
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
 	 * Assign fwnode depending on the result of the previous calls,
 	 * if none of them succeed, assign it to the parent's one.
 	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
 
 	gdev->id = ida_alloc(&gpio_ida, GFP_KERNEL);
 	if (gdev->id < 0) {
Signature · 1 line
2.38.1
#1 lore raw
Quoted 15 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Quoted 3 lines

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

Hi, the following patch should fix it for you, I have hit the same issue on IPQ8074.

https://patchwork.ozlabs.org/project/linux-gpio/patch/[email protected]/

Regards,
Robert

Quoted 19 lines
---
  drivers/gpio/gpiolib.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +1 -1
index 11fb7ec883e9..8bec66008869 100644
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
  	 * Assign fwnode depending on the result of the previous calls,
  	 * if none of them succeed, assign it to the parent's one.
  	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
 
  	gdev->id = ida_alloc(&gpio_ida, GFP_KERNEL);
  	if (gdev->id < 0) {
#3 lore raw
In reply to Robert Marko
Quoted 17 lines · Robert Marko
Quoted 15 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Quoted 9 lines
Quoted 3 lines

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

Hi, the following patch should fix it for you, I have hit the same issue on IPQ8074.

https://patchwork.ozlabs.org/project/linux-gpio/patch/[email protected]/

This fixes the issue I was facing for sa8540p (sc8280xp). Thanks Robert.

Shazad

Quoted 22 lines

Regards,
Robert

Quoted 19 lines

---
  drivers/gpio/gpiolib.c | 2 +-

  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +1 -1
index 11fb7ec883e9..8bec66008869 100644
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip 
 *gc, void *data,
        * Assign fwnode depending on the result of the previous calls,
        * if none of them succeed, assign it to the parent's one.
        */
-    gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+    gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
       gdev->id = ida_alloc(&gpio_ida, GFP_KERNEL);
       if (gdev->id < 0) {
Tested-by on v1 1/1
#4 lore raw
In reply to Brian Masney
Quoted 15 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Quoted 5 lines

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

Thanks, this fixes the following abort I'm observing on multiple Qualcomm platforms (sdm630, Sony Nile, and msm8956, Sony Loire):

    [    0.391439] Internal error: synchronous external abort: 0000000096000210 [#1] PREEMPT SMP
    [    0.391526] Modules linked in:
    [    0.398678] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 6.1.0-rc5-next-20221114-07647-gb3ed2836a8f7 #38
    [    0.401642] Hardware name: Sony Xperia XA2 Ultra (DT)
    [    0.410879] pstate: 00400005 (nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
    [    0.415957] pc : msm_gpio_get_direction+0x40/0x80
    [    0.422680] lr : msm_gpio_get_direction+0x18/0x80
    [    0.427544] sp : ffff80000805b750
    [    0.432270] x29: ffff80000805b750 x28: ffff739180b48c00 x27: ffff739180b50180
    [    0.435537] x26: 0000000000000008 x25: 0000000000000002 x24: ffffdcf976192e98
    [    0.442695] x23: ffff739180b31890 x22: 0000000000000000 x21: ffff739180b48c08
    [    0.449814] x20: ffffdcf975ff4148 x19: 0000000000000008 x18: 0000000000000000
    [    0.456891] x17: 64656c62616e655f x16: 7469647561206465 x15: 0000000000000002
    [    0.464049] x14: 0000000000000001 x13: 006c7274636e6970 x12: 2e30303030303133
    [    0.471166] x11: ffffdcf9760c98d8 x10: 000000000042c70 x9 : 0000000000000004
    [    0.478243] x8 : 0101010101010101 x7 : 0073656d616e2d65 x6 : 0911040aadece9ee
    [    0.485405] x5 : 6e696c2d0a041109 x4 : 0000000000000180 x3 : 0000000000008000
    [    0.492483] x2 : 0000000000000000 x1 : ffffdcf975628db0 x0 : ffff800008808000
    [    0.499642] Call trace:
    [    0.506703]  msm_gpio_get_direction+0x40/0x80
    [    0.509008]  gpiochip_add_data_with_key+0x638/0xed0
    [    0.513481]  msm_pinctrl_probe+0x430/0x580
    [    0.518168]  sdm660_pinctrl_probe+0x18/0x30
    [    0.522376]  platform_probe+0x68/0xc0
    [    0.526413]  really_probe+0xc0/0x3dc
    [    0.530234]  __driver_probe_device+0x7c/0x190
    [    0.533922]  driver_probe_device+0x3c/0x110
    [    0.538132]  __device_attach_driver+0xbc/0x160
    [    0.542168]  bus_for_each_drv+0x7c/0xd4
    [    0.546637]  __device_attach+0x9c/0x1c0
    [    0.550370]  device_initial_probe+0x14/0x20
    [    0.554231]  bus_probe_device+0x9c/0xa4
    [    0.558358]  device_add+0x3ac/0x8b0
    [    0.562175]  of_device_add+0x54/0x64
    [    0.565689]  of_platform_device_create_pdata+0x90/0x124
    [    0.569470]  of_platform_bus_create+0x18c/0x510
    [    0.574416]  of_platform_bus_create+0x1ec/0x510
    [    0.578971]  of_platform_populate+0x60/0x150
    [    0.583445]  of_platform_default_populate_init+0xe4/0x104
    [    0.588002]  do_one_initcall+0x64/0x1dc
    [    0.593253]  kernel_init_freeable+0x1b8/0x220
    [    0.596900]  kernel_init+0x24/0x130
    [    0.601453]  ret_from_fork+0x10/0x20
    [    0.604715] Code: d37d0442 8b020000 f941b400 8b030000 (b9400000)
    [    0.608583] ---[ end trace 0000000000000000 ]---
    [    0.619572] Kernel panic - not syncing: Attempted to kill init! exitcode=0x0000000b

This wasn't an issue on -next 20221109.

Quoted 20 lines
---
 drivers/gpio/gpiolib.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +1 -1
index 11fb7ec883e9..8bec66008869 100644
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
 	 * Assign fwnode depending on the result of the previous calls,
 	 * if none of them succeed, assign it to the parent's one.
 	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
 
 	gdev->id = ida_alloc(&gpio_ida, GFP_KERNEL);
 	if (gdev->id < 0) {
Signature · 1 line
2.38.1
Tested-by on v1 1/1
#5 lore raw
In reply to Brian Masney
Quoted 15 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Quoted 6 lines

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

---

Fixes boot on 8450 on next-20221115

Konrad

Quoted 16 lines
  drivers/gpio/gpiolib.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +1 -1
index 11fb7ec883e9..8bec66008869 100644
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
  	 * Assign fwnode depending on the result of the previous calls,
  	 * if none of them succeed, assign it to the parent's one.
  	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
 
  	gdev->id = ida_alloc(&gpio_ida, GFP_KERNEL);
  	if (gdev->id < 0) {
#6 lore raw
In reply to Brian Masney
Quoted 29 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

---
  drivers/gpio/gpiolib.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +0 -0
index 11fb7ec883e9..8bec66008869 100644

drivers/gpio/gpiolib.c int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data, · line 684

Quoted 8 lines
drivers/gpio/gpiolib.c +1 -1
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
  	 * Assign fwnode depending on the result of the previous calls,
  	 * if none of them succeed, assign it to the parent's one.
  	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
 
  	gdev->id = ida_alloc(&gpio_ida, GFP_KERNEL);
  	if (gdev->id < 0) {

Fixes repeated failure to get GPIO despite the controller having successfully probed.

[    6.151075] ov9282 21-0060: failed to get reset gpio -517
[    6.160599] ov9282 21-0060: HW configuration is not supported
[    6.160970] hub 1-1:1.0: USB hub found
[    6.168054] imx412 22-001a: failed to get reset gpio -517
[    6.170710] hub 1-1:1.0: 4 ports detected
[    6.175904] imx412 22-001a: HW configuration is not supported
#7 lore raw

Hi,

Others in the thread pointed to Thierry's patch, but on the Lenovo Thinkpad X13s, that patch did *not* fix the issue, and with it applied, the X13s still immediately reboots as soon as exiting EFI services. With this patch applied, the X13s does *not* reboot after exiting EFI services, so thank you for this patch.

#8 lore raw
Quoted 10 lines · Steev Klimaszewski

Hi,

Others in the thread pointed to Thierry's patch, but on the Lenovo Thinkpad X13s, that patch did *not* fix the issue, and with it applied, the X13s still immediately reboots as soon as exiting EFI services. With this patch applied, the X13s does *not* reboot after exiting EFI services, so thank you for this patch.

My bad, I was under the impression that Thierry's patch fixed this issue too. Now applied.

Bart

#9 lore raw
In reply to Steev Klimaszewski
Quoted 9 lines · Steev Klimaszewski

Hi,

Others in the thread pointed to Thierry's patch, but on the Lenovo Thinkpad X13s, that patch did *not* fix the issue, and with it applied, the X13s still immediately reboots as soon as exiting EFI services. With this patch applied, the X13s does *not* reboot after exiting EFI services, so thank you for this patch.

Do you happen to know what driver the X13s is using? As mentioned in another email, I don't think overriding gc->fwnode at this point is the right thing to do, so I suspect this field might be used unexpectedly in some other place, so I'd like to take a closer look at that.

Thierry

Tested-by on v1 1/1
#10 lore raw
In reply to Brian Masney
Quoted 29 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

---
  drivers/gpio/gpiolib.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +0 -0
index 11fb7ec883e9..8bec66008869 100644

drivers/gpio/gpiolib.c int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data, · line 684

Quoted 8 lines
drivers/gpio/gpiolib.c +1 -1
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
  	 * Assign fwnode depending on the result of the previous calls,
  	 * if none of them succeed, assign it to the parent's one.
  	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
 
  	gdev->id = ida_alloc(&gpio_ida, GFP_KERNEL);
  	if (gdev->id < 0) {

Fixes boot on 8550 on next-20221115

Thanks,
Neil

#11 lore raw
In reply to Brian Masney
Quoted 29 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

---
 drivers/gpio/gpiolib.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +0 -0
index 11fb7ec883e9..8bec66008869 100644

drivers/gpio/gpiolib.c int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data, · line 681

Quoted 5 lines
drivers/gpio/gpiolib.c +1 -1
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
 	 * Assign fwnode depending on the result of the previous calls,
 	 * if none of them succeed, assign it to the parent's one.
 	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;

This doesn't look right to me. Looking at the documentation of gc->fwnode and how it is used, the purpose of this is to allow explicitly overriding the fwnode that the GPIO chip will use.

So really this should not be used beyond the initial registration in gpiochip_add_data_with_key(). If the above patch fixes anything, then I suspect somebody is using gc->fwnode outside of this registration.

Looking at gpiolib, the only remaining place that seems to do this is the gpio-reserved-ranges handling code, in which case, the below on top of my initial patch might fix that. That might explain why MSM is still seeing issues.

--- >8 ---

drivers/gpio/gpiolib.c +4 -2
index 11fb7ec883e9..d692ad5c5a27 100644
@@ -447,10 +447,11 @@ static unsigned long *gpiochip_allocate_mask(struct gpio_chip *gc)
 
 static unsigned int gpiochip_count_reserved_ranges(struct gpio_chip *gc)
 {
+	struct fwnode_handle *fwnode = dev_fwnode(&gc->gpiodev->dev);
 	int size;
 
 	/* Format is "start, count, ..." */
-	size = fwnode_property_count_u32(gc->fwnode, "gpio-reserved-ranges");
+	size = fwnode_property_count_u32(fwnode, "gpio-reserved-ranges");
 	if (size > 0 && size % 2 == 0)
 		return size;
 
@@ -471,6 +472,7 @@ static int gpiochip_alloc_valid_mask(struct gpio_chip *gc)
 
 static int gpiochip_apply_reserved_ranges(struct gpio_chip *gc)
 {
+	struct fwnode_handle *fwnode = dev_fwnode(&gc->gpiodev->dev);
 	unsigned int size;
 	u32 *ranges;
 	int ret;
@@ -483,7 +485,7 @@ static int gpiochip_apply_reserved_ranges(struct gpio_chip *gc)
 	if (!ranges)
 		return -ENOMEM;
 
-	ret = fwnode_property_read_u32_array(gc->fwnode, "gpio-reserved-ranges", ranges, size);
+	ret = fwnode_property_read_u32_array(fwnode, "gpio-reserved-ranges", ranges, size);
 	if (ret) {
 		kfree(ranges);
 		return ret;

--- >8 ---

I don't have a good idea about the Lenovo X13 issue, though, but I haven't looked at ACPI at all since I don't have any hardware to test on.

Thierry

#12 lore raw
Quoted 87 lines · Thierry Reding
Quoted 29 lines · Brian Masney

Note that this is a RFC patch and not meant to be merged. I looked into a problem with linux-next-20221110 on the Qualcomm SA8540P automotive board (sc8280xp) where the UFS host controller would fail to probe due to repeated probe deferrals when trying to get reset-gpios via devm_gpiod_get_optional().

of_get_named_gpiod_flags() returns -EPROBE_DEFER, which is caused by of_gpiochip_match_node_and_xlate() returning 0 since the of_xlate function pointer is not set for the qcom,sc8280xp-tlmm pinctrl driver. The pinctrl driver doesn't define one, so of_gpiochip_add() should automatically setup of_gpio_simple_xlate() on it's behalf. This doesn't happen since the fwnode member on the struct gpiochip is set to null when of_gpiochip_add() is called. Let's work around this by ensuring that it's set if available.

Note that this broke sometime within the last few weeks within linux-next and I haven't bisected this. I'm posting this in the hopes that someone may know offhand which patch(es) may have broken this.

---
 drivers/gpio/gpiolib.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib.c +0 -0
index 11fb7ec883e9..8bec66008869 100644

drivers/gpio/gpiolib.c int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data, · line 681

Quoted 5 lines
drivers/gpio/gpiolib.c +1 -1
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
 	 * Assign fwnode depending on the result of the previous calls,
 	 * if none of them succeed, assign it to the parent's one.
 	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;

This doesn't look right to me. Looking at the documentation of gc->fwnode and how it is used, the purpose of this is to allow explicitly overriding the fwnode that the GPIO chip will use.

So really this should not be used beyond the initial registration in gpiochip_add_data_with_key(). If the above patch fixes anything, then I suspect somebody is using gc->fwnode outside of this registration.

Looking at gpiolib, the only remaining place that seems to do this is the gpio-reserved-ranges handling code, in which case, the below on top of my initial patch might fix that. That might explain why MSM is still seeing issues.

--- >8 ---

drivers/gpio/gpiolib.c +4 -3
index 11fb7ec883e9..d692ad5c5a27 100644
@@ -447,10 +447,11 @@ static unsigned long *gpiochip_allocate_mask(struct gpio_chip *gc)
 
 static unsigned int gpiochip_count_reserved_ranges(struct gpio_chip *gc)
 {
+	struct fwnode_handle *fwnode = dev_fwnode(&gc->gpiodev->dev);
 	int size;
 
 	/* Format is "start, count, ..." */
-	size = fwnode_property_count_u32(gc->fwnode, "gpio-reserved-ranges");
+	size = fwnode_property_count_u32(fwnode, "gpio-reserved-ranges");
 	if (size > 0 && size % 2 == 0)
 		return size;
 
@@ -471,6 +472,7 @@ static int gpiochip_alloc_valid_mask(struct gpio_chip *gc)
 
 static int gpiochip_apply_reserved_ranges(struct gpio_chip *gc)
 {
+	struct fwnode_handle *fwnode = dev_fwnode(&gc->gpiodev->dev);
 	unsigned int size;
 	u32 *ranges;
 	int ret;
@@ -483,7 +485,7 @@ static int gpiochip_apply_reserved_ranges(struct gpio_chip *gc)
 	if (!ranges)
 		return -ENOMEM;
 
-	ret = fwnode_property_read_u32_array(gc->fwnode, "gpio-reserved-ranges", ranges, size);
+	ret = fwnode_property_read_u32_array(fwnode, "gpio-reserved-ranges", ranges, size);
 	if (ret) {
 		kfree(ranges);
 		return ret;
--- >8 ---
Quoted 3 lines

I don't have a good idea about the Lenovo X13 issue, though, but I haven't looked at ACPI at all since I don't have any hardware to test on.

Ah... looks like that device was actually a Thinkpad X13*s*, which is based on a Qualcomm chip, so maybe this patch fixes that one, too. It does use gpio-reserved-ranges, so seems at least likely.

Thierry

#13 lore raw
In reply to Thierry Reding
Quoted 21 lines · Thierry Reding

drivers/gpio/gpiolib.c int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data, · line 681

Quoted 10 lines · Brian Masney
drivers/gpio/gpiolib.c +1 -1
index 11fb7ec883e9..8bec66008869 100644
@@ -678,7 +678,7 @@ int gpiochip_add_data_with_key(struct gpio_chip *gc, void *data,
 	 * Assign fwnode depending on the result of the previous calls,
 	 * if none of them succeed, assign it to the parent's one.
 	 */
-	gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;
+	gc->fwnode = gdev->dev.fwnode = dev_fwnode(&gdev->dev) ?: fwnode;

This doesn't look right to me. Looking at the documentation of gc->fwnode and how it is used, the purpose of this is to allow explicitly overriding the fwnode that the GPIO chip will use.

So really this should not be used beyond the initial registration in gpiochip_add_data_with_key(). If the above patch fixes anything, then I suspect somebody is using gc->fwnode outside of this registration.

Quoted 4 lines

Looking at gpiolib, the only remaining place that seems to do this is the gpio-reserved-ranges handling code, in which case, the below on top of my initial patch might fix that. That might explain why MSM is still seeing issues.

That is correct. The Thinkpad x13s laptop uses the driver drivers/pinctrl/qcom/pinctrl-sc8280xp.c and arch/arm64/boot/dts/qcom/sc8280xp-lenovo-thinkpad-x13s.dts defines a gpio-reserved-ranges property. The SA8540p automotive board has the same SoC, however the DTS we are using doesn't use the gpio-reserved-ranges property, and why only your original patch fixed the issue for this board.

I think my patch should be removed from the GPIO tree if Thierry's two patches work for everyone.

Brian