Skip to contentSkip to conversation

Driver-Core: devtmpfs - remove EXPERIMENTAL and enable it by default

PATCH Kay Sievers 6 participants 14 messages Jan 14, 2010 – Jan 21, 2010 Driver core

lore
Share & tools
lore
#0 lore raw

From: Kay Sievers <[email protected]>
Subject: Driver-Core: devtmpfs - remove EXPERIMENTAL and enable it by default

All major distros enable devtmpfs on recent systems, so remove the EXPERIMENTAL flag, and enable it by default to reflect how it is used today.

---
 drivers/base/Kconfig |   45 ++++++++++++++++++++++++++++-----------------
 1 file changed, 28 insertions(+), 17 deletions(-)
drivers/base/Kconfig +28 -17
@@ -9,29 +9,40 @@ config UEVENT_HELPER_PATH
 	  every uevent.
 
 config DEVTMPFS
-	bool "Create a kernel maintained /dev tmpfs (EXPERIMENTAL)"
+	bool "Maintain a devtmpfs filesystem to mount at /dev"
 	depends on HOTPLUG && SHMEM && TMPFS
+	default y
 	help
-	  This creates a tmpfs filesystem, and mounts it at bootup
-	  and mounts it at /dev. The kernel driver core creates device
-	  nodes for all registered devices in that filesystem. All device
-	  nodes are owned by root and have the default mode of 0600.
-	  Userspace can add and delete the nodes as needed. This is
-	  intended to simplify bootup, and make it possible to delay
-	  the initial coldplug at bootup done by udev in userspace.
-	  It should also provide a simpler way for rescue systems
-	  to bring up a kernel with dynamic major/minor numbers.
-	  Meaningful symlinks, permissions and device ownership must
-	  still be handled by userspace.
-	  If unsure, say N here.
+	  This creates a tmpfs filesystem instance early at bootup.
+	  In this filesystem, the kernel driver core maintains device
+	  nodes with their default names and permissions for all
+	  registered devices with an assigned major/minor number.
+	  Userspace can modify the filesystem content as needed, add
+	  symlinks, and apply needed permissions.
+	  It provides a fully functional /dev directory, where usually
+	  udev runs on top, managing permissions and adding meaningful
+	  symlinks.
+	  In very limited environments, it may provide a sufficient
+	  functional /dev without any further help. It also allows simple
+	  rescue systems, and reliably handles dynamic major/minor numbers.
+	  If unsure, say Y here.
 
 config DEVTMPFS_MOUNT
-	bool "Automount devtmpfs at /dev"
+	bool "Automount devtmpfs at /dev, after the kernel mouted the rootfs"
 	depends on DEVTMPFS
+	default y
 	help
-	  This will mount devtmpfs at /dev if the kernel mounts the root
-	  filesystem. It will not affect initramfs based mounting.
-	  If unsure, say N here.
+	  This will instruct the kernel to automatically mount the
+	  devtmpfs filesystem at /dev, directly after the kernel has
+	  mounted the root filesystem. The behavior can be overridden
+	  with the commandline parameter: devtmpfs.mount=0|1.
+	  This option does not affect initramfs based booting, here
+	  the devtmpfs filesystem always needs to be mouted manually
+	  after the roots is mounted.
+	  With this option enabled, it allows to bring up a system in
+	  rescue mode with init=/bin/sh, even when the /dev directory
+	  on the rootfs is completely empty.
+	  If unsure, say Y here.
 
 config STANDALONE
 	bool "Select only drivers that don't need compile-time external firmware" if EXPERIMENTAL
#1 lore raw

drivers/base/Kconfig config UEVENT_HELPER_PATH · line 31

Quoted 3 lines · Kay Sievers

config DEVTMPFS_MOUNT

drivers/base/Kconfig +1 -1
-	bool "Automount devtmpfs at /dev"
+	bool "Automount devtmpfs at /dev, after the kernel mouted the rootfs"

mouted -> mounted

drivers/base/Kconfig config UEVENT_HELPER_PATH · line 40

Quoted 12 lines
drivers/base/Kconfig +7 -3
 	depends on DEVTMPFS
+	default y
 	help
-	  This will mount devtmpfs at /dev if the kernel mounts the root
-	  filesystem. It will not affect initramfs based mounting.
-	  If unsure, say N here.
+	  This will instruct the kernel to automatically mount the
+	  devtmpfs filesystem at /dev, directly after the kernel has
+	  mounted the root filesystem. The behavior can be overridden
+	  with the commandline parameter: devtmpfs.mount=0|1.
+	  This option does not affect initramfs based booting, here
+	  the devtmpfs filesystem always needs to be mouted manually

Here too.

Signature · 3 lines
Stefan Richter
-=====-==-=- ---= -====
http://arcgraph.de/sr/
#2 lore raw
Quoted 7 lines · Stefan Richter

drivers/base/Kconfig config UEVENT_HELPER_PATH · line 31

Quoted 3 lines · Kay Sievers

config DEVTMPFS_MOUNT

drivers/base/Kconfig +1 -1
-	bool "Automount devtmpfs at /dev"
+	bool "Automount devtmpfs at /dev, after the kernel mouted the rootfs"

mouted -> mounted

Quoted 14 lines

drivers/base/Kconfig config UEVENT_HELPER_PATH · line 40

Quoted 12 lines
drivers/base/Kconfig +7 -3
 	depends on DEVTMPFS
+	default y
 	help
-	  This will mount devtmpfs at /dev if the kernel mounts the root
-	  filesystem. It will not affect initramfs based mounting.
-	  If unsure, say N here.
+	  This will instruct the kernel to automatically mount the
+	  devtmpfs filesystem at /dev, directly after the kernel has
+	  mounted the root filesystem. The behavior can be overridden
+	  with the commandline parameter: devtmpfs.mount=0|1.
+	  This option does not affect initramfs based booting, here
+	  the devtmpfs filesystem always needs to be mouted manually

Here too.

Thanks for the review, I've edited the patch to include these spelling changes.

greg k-h

#3 lore raw
In reply to Kay Sievers
Quoted 7 lines · On Thu, 14 Jan 2010 22:47:57 +0100 Kay Sievers <[email protected]> wrote:

From: Kay Sievers <[email protected]>
Subject: Driver-Core: devtmpfs - remove EXPERIMENTAL and enable it by default

All major distros enable devtmpfs on recent systems, so remove the EXPERIMENTAL flag, and enable it by default to reflect how it is used today.

removing experimental is fine... but setting it by default is a bit over the top and very inconsistent with how the 'default' option is used.

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#4 lore raw
Quoted 7 lines · Arjan van de Ven
Quoted 4 lines · On Thu, 14 Jan 2010 22:47:57 +0100 Kay Sievers <[email protected]> wrote:

From: Kay Sievers <[email protected]>
Subject: Driver-Core: devtmpfs - remove EXPERIMENTAL and enable it by default

Quoted 7 lines
Quoted 3 lines

All major distros enable devtmpfs on recent systems, so remove the EXPERIMENTAL flag, and enable it by default to reflect how it is used today.

removing experimental is fine... but setting it by default is a bit over the top and very inconsistent with how the 'default' option is used.

Why? All major distros need this at boot time, and as we have been through a release cycle with the config option, anyone who is incrementally updating will continue with their existing value. But new people coming in, will get the option as it is most likely required to boot properly.

Makes sense to me.

thanks,

greg k-h

#5 lore raw
Quoted 1 line

Why? All major distros need this at boot time, and as we have been

No they don't.

Centos doesn't (in fact I suspect it'll break) Fedora 11 doesn't
Fedora 12 doesn't seem to (but seems to be willing to use it) Ditto all the older SuSE, Ubuntu etc releases that are *still* active/supported/maintained

Quoted 6 lines

through a release cycle with the config option, anyone who is incrementally updating will continue with their existing value. But new people coming in, will get the option as it is most likely required to boot properly.

Makes sense to me.

That may well be true - in which case it needs to be very clearly documented when to enable it.

#6 lore raw
Quoted 9 lines · Alan Cox
Quoted 1 line

Why?  All major distros need this at boot time, and as we have been

No they don't.

Centos doesn't (in fact I suspect it'll break) Fedora 11 doesn't
Fedora 12 doesn't seem to (but seems to be willing to use it) Ditto all the older SuSE, Ubuntu etc releases that are *still* active/supported/maintained

Fedora, SUSE, Ubuntu, ... All releases currently in development enable and use it.

Quoted 9 lines
Quoted 6 lines

through a release cycle with the config option, anyone who is incrementally updating will continue with their existing value.  But new people coming in, will get the option as it is most likely required to boot properly.

Makes sense to me.

That may well be true - in which case it needs to be very clearly documented when to enable it.

For now, the systems will bootup fine without it, but I expect over time, it will get a requirement. Udev does not care about it, but the boot scripts/initramfs tools will probably get trimmed down, because there is much less to do with the kernel provided nodes, than what is needed today to bootstrap /dev.

And it should not harm any old system if it is enabled. If initramfs is used it's completely invisible, if a custom kernel with kernel-mounted rootfs is used, the udev boot script usually over-mounts the devtmpfs at /dev with an empty tmpfs, like it has always done it before.

The only thing it should change is that we now should be able to bootup a box without an initramfs, without manually filling /dev with static nodes beforehand. Current distro installations do not put a single file in /dev on the rootfs, and trying to boot a kernel without an initramfs usually just fails without putting some nodes there in preparation.

Thanks,
Kay

#7 lore raw
Quoted 5 lines · On Sat, 16 Jan 2010 16:51:23 +0100 Kay Sievers <[email protected]> wrote:
Quoted 4 lines · Alan Cox
Quoted 1 line

Why?  All major distros need this at boot time, and as we have been

No they don't.

Quoted 8 lines
Quoted 5 lines

Centos doesn't (in fact I suspect it'll break) Fedora 11 doesn't
Fedora 12 doesn't seem to (but seems to be willing to use it) Ditto all the older SuSE, Ubuntu etc releases that are *still* active/supported/maintained

Fedora, SUSE, Ubuntu, ... All releases currently in development enable and use it.

Yes - in future your statement may become true (or devtmpfs get obsoleted by the next best thing. Thats why it needs to be documented if it is set that way - I've no problem with it being set that way at all.

#8 lore raw
In reply to Kay Sievers
Quoted 2 lines · On Sat, 16 Jan 2010 16:51:23 +0100 Kay Sievers <[email protected]> wrote:

Current distro installations do not put a
single file in /dev on the rootfs,

any distro that does that is terminally broken; the minimum needed is /dev/console.
I guess this can be counted as a bugreport against your distro ;-)

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#9 lore raw
In reply to Greg KH
Quoted 16 lines · Greg KH
Quoted 7 lines · Arjan van de Ven
Quoted 4 lines · On Thu, 14 Jan 2010 22:47:57 +0100 Kay Sievers <[email protected]> wrote:

From: Kay Sievers <[email protected]>
Subject: Driver-Core: devtmpfs - remove EXPERIMENTAL and enable it by default

Quoted 7 lines
Quoted 3 lines

All major distros enable devtmpfs on recent systems, so remove the EXPERIMENTAL flag, and enable it by default to reflect how it is used today.

removing experimental is fine... but setting it by default is a bit over the top and very inconsistent with how the 'default' option is used.

Quoted 5 lines

Why? All major distros need this at boot time, and as we have been through a release cycle with the config option, anyone who is incrementally updating will continue with their existing value. But new people coming in, will get the option as it is most likely required to boot properly.

Defaults should be back-compatible. This is not... people should be still able to skip releases when upgrading.

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#10 lore raw
In reply to Greg KH
Quoted 5 lines · On Fri, 15 Jan 2010 10:03:41 -0800 Greg KH <[email protected]> wrote:
Quoted 3 lines

removing experimental is fine... but setting it by default is a bit over the top and very inconsistent with how the 'default' option is used.

Why?

because the convention is that we use "default y" only for those things that used to be on, and are now turned into a config option.

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#11 lore raw
Quoted 10 lines · Arjan van de Ven
Quoted 5 lines · On Fri, 15 Jan 2010 10:03:41 -0800 Greg KH <[email protected]> wrote:
Quoted 3 lines

removing experimental is fine... but setting it by default is a bit over the top and very inconsistent with how the 'default' option is used.

Why?

because the convention is that we use "default y" only for those things that used to be on, and are now turned into a config option.

Oh, I never heard of such a convention. Sure, we can remove that "y", if this is the way it should be used, not be used.

Looking at the current Kconfig files many "default y" looks more like:
"stuff that should be in a kernel to work in a common setup". I see new functionality often switched on with that flag.

As said, I don't mind dropping that "y", if there is such a convention.

Thanks,
Kay

#12 lore raw
Quoted 8 lines · On Wed, 20 Jan 2010 18:53:49 +0100 Kay Sievers <[email protected]> wrote:
Quoted 6 lines · Arjan van de Ven
Quoted 4 lines · On Fri, 15 Jan 2010 10:03:41 -0800 Greg KH <[email protected]> wrote:
Quoted 3 lines

removing experimental is fine... but setting it by default is a bit over the top and very inconsistent with how the 'default' option is used.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

Why?

because the convention is that we use "default y" only for those things that used to be on, and are now turned into a config option.

Oh, I never heard of such a convention. Sure, we can remove that "y", if this is the way it should be used, not be used.

For example in (this is just the first one google found, there's been a bunch of discussion at various times)

http://article.gmane.org/gmane.linux.kernel/706458

Linus was rather explicit:

The rule of thumb should be:

NO NEW FEATURES SHOULD _EVER_ DEFAULT TO 'ON'!

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#13 lore raw
Quoted 22 lines · Arjan van de Ven
Quoted 8 lines · On Wed, 20 Jan 2010 18:53:49 +0100 Kay Sievers <[email protected]> wrote:
Quoted 6 lines · Arjan van de Ven
Quoted 4 lines · On Fri, 15 Jan 2010 10:03:41 -0800 Greg KH <[email protected]> wrote:
Quoted 3 lines

removing experimental is fine... but setting it by default is a bit over the top and very inconsistent with how the 'default' option is used.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

Why?

because the convention is that we use "default y" only for those things that used to be on, and are now turned into a config option.

Oh, I never heard of such a convention. Sure, we can remove that "y", if this is the way it should be used, not be used.

For example in (this is just the first one google found, there's been a bunch of discussion at various times)

Quoted 7 lines

http://article.gmane.org/gmane.linux.kernel/706458

Linus was rather explicit:

The rule of thumb should be:

NO NEW FEATURES SHOULD _EVER_ DEFAULT TO 'ON'!

But this is not a "new" feature, it's been around for 6+ months by the time it hits a release. Does this imply that we can never change anything to 'on' in the Kconfig files?

thanks,

greg k-h