Skip to contentSkip to conversation

2.6.14-rc4-mm1: udev/sysfs wierdness

DISCUSSION Aaron Gyes 3 participants 10 messages Oct 18, 2005 – Oct 20, 2005 Input drivers

lore
Share & tools
lore

Aaron Gyes

author
#0 lore raw

For some reason this rule stopped working:

KERNEL=="event*", SYSFS{manufacturer}="Logitech", SYSFS{product}="USB Receiver", NAME="input/mx1000", MODE="0644"

Did stuff in /sys/ change? Do I need to change all my rules to make up for this? udevs fault? I do have the correct /dev/input/event0 node.

Aaron Gyes

#1 lore raw
Quoted 7 lines · Aaron Gyes

For some reason this rule stopped working:

KERNEL=="event*", SYSFS{manufacturer}="Logitech", SYSFS{product}="USB Receiver", NAME="input/mx1000", MODE="0644"

Did stuff in /sys/ change? Do I need to change all my rules to make up for this? udevs fault? I do have the correct /dev/input/event0 node.

You have that node? That's a good start :)

I think the "name" might have changed, it looks like I messed that up somehow. What does:
udevinfo -p /sys/class/input/input0/event0/ -a

show (or whatever that sysfs path is.)

Oops, heh, that dies on my box too. Ok, I think that's the issue, sorry. I'm working on it...

thanks,

greg k-h

#2 lore raw
Quoted 11 lines · Greg KH
Quoted 7 lines · Aaron Gyes

For some reason this rule stopped working:

KERNEL=="event*", SYSFS{manufacturer}="Logitech", SYSFS{product}="USB Receiver", NAME="input/mx1000", MODE="0644"

Did stuff in /sys/ change? Do I need to change all my rules to make up for this? udevs fault? I do have the correct /dev/input/event0 node.

You have that node? That's a good start :)

Quoted 8 lines

I think the "name" might have changed, it looks like I messed that up somehow. What does:
udevinfo -p /sys/class/input/input0/event0/ -a

show (or whatever that sysfs path is.)

Oops, heh, that dies on my box too. Ok, I think that's the issue, sorry. I'm working on it...

Can you try the patch below to see if that fixes the issue? That should keep udevinfo from dieing.

thanks,

greg k-h

---
 drivers/input/input.c |   19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/input/input.c +12 -7
@@ -522,17 +522,22 @@ static ssize_t input_dev_show_##name(str
 	up(&input_dev->sem);							\
 										\
 	return retval;								\
-}
+}										\
+static CLASS_DEVICE_ATTR(name, S_IRUGO, input_dev_show_##name, NULL);
 
 INPUT_DEV_STRING_ATTR_SHOW(name);
 INPUT_DEV_STRING_ATTR_SHOW(phys);
 INPUT_DEV_STRING_ATTR_SHOW(uniq);
 
-static struct class_device_attribute input_dev_attrs[] = {
-	__ATTR(name, S_IRUGO, input_dev_show_name, NULL),
-	__ATTR(phys, S_IRUGO, input_dev_show_phys, NULL),
-	__ATTR(uniq, S_IRUGO, input_dev_show_uniq, NULL),
-	__ATTR_NULL
+static struct attribute *input_dev_attrs[] = {
+	&class_device_attr_name.attr,
+	&class_device_attr_phys.attr,
+	&class_device_attr_uniq.attr,
+	NULL
+};
+
+static struct attribute_group input_dev_group = {
+	.attrs	= input_dev_attrs,
 };
 
 #define INPUT_DEV_ID_ATTR(name)							\
@@ -694,7 +699,6 @@ struct class input_class = {
 	.name			= "input",
 	.release		= input_dev_release,
 	.hotplug		= input_dev_hotplug,
-	.class_dev_attrs	= input_dev_attrs,
 };
 
 struct input_dev *input_allocate_device(void)
@@ -732,6 +736,7 @@ static void input_register_classdevice(s
 	kfree(path);
 
 	class_device_add(&dev->cdev);
+	sysfs_create_group(&dev->cdev.kobj, &input_dev_group);
 	sysfs_create_group(&dev->cdev.kobj, &input_dev_id_attr_group);
 	sysfs_create_group(&dev->cdev.kobj, &input_dev_caps_attr_group);
 }
#3 lore raw

Greg KH <[email protected]> disait dernièrement que :

Quoted 16 lines · Greg KH
Quoted 15 lines · Aaron Gyes
Quoted 7 lines

For some reason this rule stopped working:

KERNEL=="event*", SYSFS{manufacturer}="Logitech", SYSFS{product}="USB Receiver", NAME="input/mx1000", MODE="0644"

Did stuff in /sys/ change? Do I need to change all my rules to make up for this? udevs fault? I do have the correct /dev/input/event0 node.

You have that node? That's a good start :)

I think the "name" might have changed, it looks like I messed that up somehow. What does:
udevinfo -p /sys/class/input/input0/event0/ -a

Quoted 7 lines
Quoted 4 lines

show (or whatever that sysfs path is.)

Oops, heh, that dies on my box too. Ok, I think that's the issue, sorry. I'm working on it...

Can you try the patch below to see if that fixes the issue? That should keep udevinfo from dieing.

huh, it doesn't apply to 2.6.14-rc4-mm1.... for which tree is it exactly ?

patching file drivers/input/input.c
Hunk #1 succeeded at 641 (offset 119 lines).
Hunk #2 FAILED at 818.
Hunk #3 succeeded at 769 (offset 33 lines).
1 out of 3 hunks FAILED -- saving rejects to file drivers/input/input.c.rej

best regards,

Signature · 1 line
Mathieu
#4 lore raw
Quoted 26 lines · Mathieu Segaud

Greg KH <[email protected]> disait derni??rement que :

Quoted 16 lines · Greg KH
Quoted 15 lines · Aaron Gyes
Quoted 7 lines

For some reason this rule stopped working:

KERNEL=="event*", SYSFS{manufacturer}="Logitech", SYSFS{product}="USB Receiver", NAME="input/mx1000", MODE="0644"

Did stuff in /sys/ change? Do I need to change all my rules to make up for this? udevs fault? I do have the correct /dev/input/event0 node.

You have that node? That's a good start :)

I think the "name" might have changed, it looks like I messed that up somehow. What does:
udevinfo -p /sys/class/input/input0/event0/ -a

Quoted 7 lines
Quoted 4 lines

show (or whatever that sysfs path is.)

Oops, heh, that dies on my box too. Ok, I think that's the issue, sorry. I'm working on it...

Can you try the patch below to see if that fixes the issue? That should keep udevinfo from dieing.

Quoted 8 lines

huh, it doesn't apply to 2.6.14-rc4-mm1.... for which tree is it exactly ?

patching file drivers/input/input.c
Hunk #1 succeeded at 641 (offset 119 lines).
Hunk #2 FAILED at 818.
Hunk #3 succeeded at 769 (offset 33 lines).
1 out of 3 hunks FAILED -- saving rejects to file drivers/input/input.c.rej

I was against my latest tree, which is on kernel.org. Someone already posted an updated patch on lkml if you can't get that second hunk to apply.

thanks,

greg k-h

author
#5 lore raw
Quoted 3 lines · Greg KH

I was against my latest tree, which is on kernel.org. Someone already posted an updated patch on lkml if you can't get that second hunk to apply.

I applied that, and I still don't see the node being created.

Here's the udevinfo output you originally asked for:

floam@agorastome ~ $ udevinfo -p /sys/class/input/event0/ -a

udevinfo starts with the device the node belongs to and then walks up the
device chain, to print for every device found, all possibly useful attributes
in the udev key format.
Only attributes within one device section may be used together in one rule,
to match the device for which the node will be created.

device '/sys/class/input/event0' has major:minor 13:64
looking at class device '/sys/class/input/event0':

    KERNEL=="event0"
    SUBSYSTEM=="input"
    SYSFS{dev}=="13:64"
#6 lore raw
Quoted 6 lines · Aaron Gyes
Quoted 3 lines · Greg KH

I was against my latest tree, which is on kernel.org. Someone already posted an updated patch on lkml if you can't get that second hunk to apply.

I applied that, and I still don't see the node being created.

Yeah, sorry.

You need a udev update. 071 has been released which should fix this problem (fixes it for me.) Can you try that out?

thanks,

greg k-h

author
#7 lore raw
Quoted 6 lines · Greg KH

You need a udev update. 071 has been released which should fix this problem (fixes it for me.) Can you try that out?

thanks,

greg k-h

Excuse my igorance, but for the life of me I can't find udev 071. Obviously it exists somewhere since my distribution (x86_64 Gentoo) seems to have a package for it, but it's failing to be able to grab it, and I'm unable to find a tarball anywhere at kernel.org

Aaron Gyes

#8 lore raw
Quoted 4 lines · Aaron Gyes
Quoted 3 lines · Greg KH

You need a udev update. 071 has been released which should fix this problem (fixes it for me.) Can you try that out?

Quoted 8 lines
Quoted 3 lines

thanks,

greg k-h

Excuse my igorance, but for the life of me I can't find udev 071. Obviously it exists somewhere since my distribution (x86_64 Gentoo) seems to have a package for it, but it's failing to be able to grab it, and I'm unable to find a tarball anywhere at kernel.org

It looks like the kernel.org mirror system got stuck again, the tarball is sitting on the upload server, but hasn't been mirrored out to the public side yet. Give it a few hours...

thanks,

greg k-h

#9 lore raw
In reply to Greg KH

Greg KH <[email protected]> disait dernièrement que :

Quoted 36 lines · Mathieu Segaud
Quoted 26 lines

Greg KH <[email protected]> disait derni??rement que :

Quoted 16 lines · Greg KH
Quoted 15 lines · Aaron Gyes
Quoted 7 lines

For some reason this rule stopped working:

KERNEL=="event*", SYSFS{manufacturer}="Logitech", SYSFS{product}="USB Receiver", NAME="input/mx1000", MODE="0644"

Did stuff in /sys/ change? Do I need to change all my rules to make up for this? udevs fault? I do have the correct /dev/input/event0 node.

You have that node? That's a good start :)

I think the "name" might have changed, it looks like I messed that up somehow. What does:
udevinfo -p /sys/class/input/input0/event0/ -a

Quoted 7 lines
Quoted 4 lines

show (or whatever that sysfs path is.)

Oops, heh, that dies on my box too. Ok, I think that's the issue, sorry. I'm working on it...

Can you try the patch below to see if that fixes the issue? That should keep udevinfo from dieing.

Quoted 8 lines

huh, it doesn't apply to 2.6.14-rc4-mm1.... for which tree is it exactly ?

patching file drivers/input/input.c
Hunk #1 succeeded at 641 (offset 119 lines).
Hunk #2 FAILED at 818.
Hunk #3 succeeded at 769 (offset 33 lines).
1 out of 3 hunks FAILED -- saving rejects to file drivers/input/input.c.rej

Quoted 3 lines

I was against my latest tree, which is on kernel.org. Someone already posted an updated patch on lkml if you can't get that second hunk to apply.

I made it apply :)
thanks, I was just wondering if some bot had put an email into the chaos-bin. It solved the problems with udevinfo as expected. Thanks a lot.

Signature · 1 line
Mathieu