remove detach_state from udev's wait_for_sysfs

remove detach_state from udev's wait_for_sysfs

[Kay Sievers]

Change udev's built-in wait-for-sysfs to wait for the "power"
directory instead of the "detached_state" file which no longer
exists with kernel 2.6.12. Without that patch udev will delay
events for some subsystems by 5 seconds.

Signed-off-by: Kay Sievers <kay.sievers@suse.de>
---

diff --git a/udev_sysfs.c b/udev_sysfs.c
--- a/udev_sysfs.c
+++ b/udev_sysfs.c
@@ -297,38 +297,38 @@ int wait_for_devices_device(struct sysfs
 		{ .bus = "usb",		.file = "idVendor" },
 		{ .bus = "usb",		.file = "iInterface" },
 		{ .bus = "usb",		.file = "bNumEndpoints" },
-		{ .bus = "usb-serial",	.file = "detach_state" },
-		{ .bus = "ide",		.file = "detach_state" },
+		{ .bus = "usb-serial",	.file = "power" },
+		{ .bus = "ide",		.file = "power" },
 		{ .bus = "pci",		.file = "vendor" },
-		{ .bus = "platform",	.file = "detach_state" },
-		{ .bus = "pcmcia",	.file = "detach_state" },
-		{ .bus = "i2c",		.file = "detach_state" },
+		{ .bus = "platform",	.file = "power" },
+		{ .bus = "pcmcia",	.file = "power" },
+		{ .bus = "i2c",		.file = "power" },
 		{ .bus = "ieee1394",	.file = "node_count" },
 		{ .bus = "ieee1394",	.file = "nodeid" },
 		{ .bus = "ieee1394",	.file = "address" },
 		{ .bus = "bttv-sub",	.file = NULL },
-		{ .bus = "pnp",		.file = "detach_state" },
-		{ .bus = "eisa",	.file = "detach_state" },
-		{ .bus = "serio",	.file = "detach_state" },
-		{ .bus = "pseudo",	.file = "detach_state" },
-		{ .bus = "mmc",		.file = "detach_state" },
-		{ .bus = "macio",	.file = "detach_state" },
-		{ .bus = "of_platform",	.file = "detach_state" },
-		{ .bus = "vio",		.file = "detach_state" },
-		{ .bus = "ecard",	.file = "detach_state" },
-		{ .bus = "sa1111-rab",	.file = "detach_state" },
-		{ .bus = "amba",	.file = "detach_state" },
-		{ .bus = "locomo-bus",	.file = "detach_state" },
-		{ .bus = "logicmodule",	.file = "detach_state" },
-		{ .bus = "parisc",	.file = "detach_state" },
-		{ .bus = "ocp",		.file = "detach_state" },
-		{ .bus = "dio",		.file = "detach_state" },
-		{ .bus = "MCA",		.file = "detach_state" },
-		{ .bus = "wl",		.file = "detach_state" },
-		{ .bus = "ccwgroup",	.file = "detach_state" },
-		{ .bus = "css",		.file = "detach_state" },
-		{ .bus = "ccw",		.file = "detach_state" },
-		{ .bus = "iucv",	.file = "detach_state" },
+		{ .bus = "pnp",		.file = "power" },
+		{ .bus = "eisa",	.file = "power" },
+		{ .bus = "serio",	.file = "power" },
+		{ .bus = "pseudo",	.file = "power" },
+		{ .bus = "mmc",		.file = "power" },
+		{ .bus = "macio",	.file = "power" },
+		{ .bus = "of_platform",	.file = "power" },
+		{ .bus = "vio",		.file = "power" },
+		{ .bus = "ecard",	.file = "power" },
+		{ .bus = "sa1111-rab",	.file = "power" },
+		{ .bus = "amba",	.file = "power" },
+		{ .bus = "locomo-bus",	.file = "power" },
+		{ .bus = "logicmodule",	.file = "power" },
+		{ .bus = "parisc",	.file = "power" },
+		{ .bus = "ocp",		.file = "power" },
+		{ .bus = "dio",		.file = "power" },
+		{ .bus = "MCA",		.file = "power" },
+		{ .bus = "wl",		.file = "power" },
+		{ .bus = "ccwgroup",	.file = "power" },
+		{ .bus = "css",		.file = "power" },
+		{ .bus = "ccw",		.file = "power" },
+		{ .bus = "iucv",	.file = "power" },
 		{ NULL, NULL }
 	};
 	const struct device_file *devicefile = NULL;



-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id\x16492&op=click
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: remove detach_state from udev's wait_for_sysfs

[Ananth N Mavinakayanahalli]

On Mon, Jun 20, 2005 at 01:01:26AM +0200, Kay Sievers wrote:
> Change udev's built-in wait-for-sysfs to wait for the "power"
> directory instead of the "detached_state" file which no longer
> exists with kernel 2.6.12. Without that patch udev will delay
> events for some subsystems by 5 seconds.

Kay,

The presence of the "power" directory is contingent to the kernel being
compiled with power management support. Wouldn't a "bus" link check be
a better idea?

Ananth


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id\x16492&op=click
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: remove detach_state from udev's wait_for_sysfs

[Kay Sievers]

On Sun, Jun 19, 2005 at 09:19:35PM -0400, Ananth N Mavinakayanahalli wrote:
> On Mon, Jun 20, 2005 at 01:01:26AM +0200, Kay Sievers wrote:
> > Change udev's built-in wait-for-sysfs to wait for the "power"
> > directory instead of the "detached_state" file which no longer
> > exists with kernel 2.6.12. Without that patch udev will delay
> > events for some subsystems by 5 seconds.
> 
> Kay,
> 
> The presence of the "power" directory is contingent to the kernel being
> compiled with power management support. Wouldn't a "bus" link check be
> a better idea?

We depend on kernel 2.6.11 with the "bus"-link. I'm fine with dropping support
for older kernels if that's ok.

Kay


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id\x16492&op=click
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: remove detach_state from udev's wait_for_sysfs

[Greg KH]

On Mon, Jun 20, 2005 at 02:15:16PM +0200, Kay Sievers wrote:
> On Sun, Jun 19, 2005 at 09:19:35PM -0400, Ananth N Mavinakayanahalli wrote:
> > On Mon, Jun 20, 2005 at 01:01:26AM +0200, Kay Sievers wrote:
> > > Change udev's built-in wait-for-sysfs to wait for the "power"
> > > directory instead of the "detached_state" file which no longer
> > > exists with kernel 2.6.12. Without that patch udev will delay
> > > events for some subsystems by 5 seconds.
> > 
> > Kay,
> > 
> > The presence of the "power" directory is contingent to the kernel being
> > compiled with power management support. Wouldn't a "bus" link check be
> > a better idea?
> 
> We depend on kernel 2.6.11 with the "bus"-link. I'm fine with dropping support
> for older kernels if that's ok.

As this is needed for 2.6.12, I don't have a problem dropping older
kernel support.

thanks,

greg k-h


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id\x16492&op=click
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: remove detach_state from udev's wait_for_sysfs

[Marco d'Itri]

[-- Attachment #1: Type: text/plain, Size: 356 bytes --]

On Jun 20, Greg KH <greg@kroah.com> wrote:

> > We depend on kernel 2.6.11 with the "bus"-link. I'm fine with dropping support
> > for older kernels if that's ok.
> As this is needed for 2.6.12, I don't have a problem dropping older
> kernel support.
Can you clarify this?
udev releases >= 059 will require a kernel >= 2.6.12?

-- 
ciao,
Marco

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: remove detach_state from udev's wait_for_sysfs

[Greg KH]

On Mon, Jun 20, 2005 at 07:45:41PM +0200, Marco d'Itri wrote:
> On Jun 20, Greg KH <greg@kroah.com> wrote:
> 
> > > We depend on kernel 2.6.11 with the "bus"-link. I'm fine with dropping support
> > > for older kernels if that's ok.
> > As this is needed for 2.6.12, I don't have a problem dropping older
> > kernel support.
> Can you clarify this?
> udev releases >= 059 will require a kernel >= 2.6.12?

Yes, would that be a problem?

thanks,

greg k-h


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id\x16492&op=click
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: remove detach_state from udev's wait_for_sysfs

[Marco d'Itri]

[-- Attachment #1: Type: text/plain, Size: 542 bytes --]

On Jun 20, Greg KH <greg@kroah.com> wrote:

> > Can you clarify this?
> > udev releases >= 059 will require a kernel >= 2.6.12?
> Yes, would that be a problem?
Yes, thinking again about this I think it could be very annoying: users
would need to upgrade both the kernel and udev at the same time, but it
would not be possible because the udev package should abort the
installation if an old kernel is detected. Users would need to upgrade
the kernel, reboot to a partially working system and then upgrade udev.

-- 
ciao,
Marco

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: remove detach_state from udev's wait_for_sysfs

[Greg KH]

On Mon, Jun 20, 2005 at 08:21:46PM +0200, Marco d'Itri wrote:
> On Jun 20, Greg KH <greg@kroah.com> wrote:
> 
> > > Can you clarify this?
> > > udev releases >= 059 will require a kernel >= 2.6.12?
> > Yes, would that be a problem?
> Yes, thinking again about this I think it could be very annoying: users
> would need to upgrade both the kernel and udev at the same time, but it
> would not be possible because the udev package should abort the
> installation if an old kernel is detected. Users would need to upgrade
> the kernel, reboot to a partially working system and then upgrade udev.

No, this doesn't "break" support on older kernels, just makes new
device discover a little slow.  It will all still work just fine.  So a
user can upgrade udev, and then their kernel.

thanks,

greg k-h


-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id\x16492&op=click
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: remove detach_state from udev's wait_for_sysfs

[Marco d'Itri]

[-- Attachment #1: Type: text/plain, Size: 876 bytes --]

On Jun 20, Greg KH <greg@kroah.com> wrote:

> > > > Can you clarify this?
> > > > udev releases >= 059 will require a kernel >= 2.6.12?
> > > Yes, would that be a problem?
> > Yes, thinking again about this I think it could be very annoying: users
> > would need to upgrade both the kernel and udev at the same time, but it
> > would not be possible because the udev package should abort the
> > installation if an old kernel is detected. Users would need to upgrade
> > the kernel, reboot to a partially working system and then upgrade udev.
> No, this doesn't "break" support on older kernels, just makes new
> device discover a little slow.  It will all still work just fine.  So a
> user can upgrade udev, and then their kernel.
OK, then newer kernels are not actually required.
I will have the init script print a big warning instead.

-- 
ciao,
Marco

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]