Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-20 Thread pk
On 2011-11-15 07:21, waltd...@waltdnes.org wrote:

   The purpose of this email is to ask adventurous people here to beta
 test my approach to a udev-less Gentoo.  If we don't find any
 showstopper problems, we can think about requesting Gentoo developers to
 support an mdev-based profile.  It would help the cause if a large
 number of testers can report that it works for them.  The instructions
 for a udev-ectomy follow below.  Thanks to Zac Medico and others on the
 Gentoo developers' list for their helpful hints and pointers on how to
 do this.  I couldn't have figured this out by myself.

I apologize for not having replied sooner. This sounds very interesting
and I will try this out, when I can find the time to tinker. Thanks
for doing this!

Best regards

Peter K



Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-16 Thread Pandu Poluan
On Wed, Nov 16, 2011 at 07:52, Pandu Poluan pa...@poluan.info wrote:

 On Nov 16, 2011 3:21 AM, waltd...@waltdnes.org wrote:

  The more scenarios we can test, the better.  mdev might shave a second
  or two off the VM's bootup time, versus udev.


 Okay, I have two staging VMs on XenServer and one on VMware. I'm going to
 report back what happens. If anything bad happens, *should* be an easy
 rollback to the previous snapshot.


1st Report: amd64-hardened on XenServer, PV-mode

Booted okay. There's 1 (one) red asterisk but it booted so fast I
can't see what that asterisk was about. `less /var/log/rc.log` did not
show anything...

... but I realized that I can scroll up on XenServer console, so I saw
that the asterisk said:

* Error: fopen(/lib64/rc/init.d/rc.log) failed: No such file or directory

Nothing serious though. I think it was because /lib64/rc/init.d got
overlaid by rc-svcdir during boot, so rc.log disappeared.

(Doing `mount -o bind / /mnt/rooot  ls /mnt/rooot/lib64/rc/init.d`
showed that rc.log existed in the non-overlaid /lib64/rc/init.d)

/dev/xvdb* (which was not created statically in the non-overlaid /dev) appeared.

/dev/xvdd appeared when I attached an ISO image to the VM.

mount /dev/xvdd /mnt/cd worked. I could `ls /mnt/cd` and see the
contents of the CD.

Ejecting the ISO image made /dev/xvdd disappear (as it should).

Networking is okay.

The Xen virtual console (hvc0) works okay.

`rc-update | grep dev` showed `udev-postmount` still part of default,
so I did `rc-update del udev-postmount default` and rebooted, and all
were still well (and still a glint of red asterisk).

Unmerging udev went well.

All in all, rebooting after switching to udev *seems* to be faster.

Rgds,
-- 
FdS Pandu E Poluan
~ IT Optimizer ~

 • LOPSA Member #15248
 • Blog : http://pepoluan.tumblr.com
 • Linked-In : http://id.linkedin.com/in/pepoluan



Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-16 Thread Pandu Poluan
On Wed, Nov 16, 2011 at 07:52, Pandu Poluan pa...@poluan.info wrote:

 On Nov 16, 2011 3:21 AM, waltd...@waltdnes.org wrote:

  The more scenarios we can test, the better.  mdev might shave a second
  or two off the VM's bootup time, versus udev.


 Okay, I have two staging VMs on XenServer and one on VMware. I'm going to
 report back what happens. If anything bad happens, *should* be an easy
 rollback to the previous snapshot.


2nd Report: amd64-hardened on VMware, using VMware PVSCSI for hard
disk, but e1000 for network.

Booted okay, and similar to the report on XenServer. But this time. I
got 2 (two) red asterisks during boot. The 1st one seems to say Error
... read-only file system. The second starts with Warning temp file
left behind ... (or something similar)

`rc-update del udev-postmount default  reboot` ... no problem.

Unmerged udev  reboot ... no problem.


There. No problem with XenServer and/or VMware. Except for the 1 or 2
red asterisks during boot (which I'm not sure caused by udev--mdev
switch or something else).


I'll experiment with VirtualBox (on Windows) tomorrow. This might be
much more interesting :-)


Rgds,
-- 
FdS Pandu E Poluan
~ IT Optimizer ~

 • LOPSA Member #15248
 • Blog : http://pepoluan.tumblr.com
 • Linked-In : http://id.linkedin.com/in/pepoluan



Re: CRTs and EDID (was: Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-16 Thread waltdnes
On Tue, Nov 15, 2011 at 09:53:47AM -0500, Michael Mol wrote

 Just an FYI, EDID blocks have been part of CRT tech since the mid to
 late 90s; it's the basis of plug  play monitors.
 
 IIRC, the EDID block is transported via DDC, which is essentially I2C
 implemented on top of your VGA cable. I've got three EDID-supporting,
 19 1600x1200 CRTs staring me in the face right now.
 
 https://plus.google.com/108080062547354628132/posts/ZLLw66eL4We

  Maybe X has learned how to read them without udev's help.  The xorg
logfile shows the EDID block, max/min horizontal/vertical scan rates,
supported modes, etc.

-- 
Walter Dnes waltd...@waltdnes.org



Re: CRTs and EDID (was: Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-16 Thread Michael Mol
On 11/16/2011 06:20 PM, waltd...@waltdnes.org wrote:
 On Tue, Nov 15, 2011 at 09:53:47AM -0500, Michael Mol wrote

 Just an FYI, EDID blocks have been part of CRT tech since the mid to
 late 90s; it's the basis of plug  play monitors.

 IIRC, the EDID block is transported via DDC, which is essentially I2C
 implemented on top of your VGA cable. I've got three EDID-supporting,
 19 1600x1200 CRTs staring me in the face right now.

 https://plus.google.com/108080062547354628132/posts/ZLLw66eL4We
   Maybe X has learned how to read them without udev's help.  The xorg
 logfile shows the EDID block, max/min horizontal/vertical scan rates,
 supported modes, etc.

Xorg's been around since 2004, and udev since 2003, so it's possible
that it's depended on udev for displays.

It seems unlikely, though; I remember XFree86 4.x having EDID support,
just prior to people migrating over to Xorg. It was a royal pain,
actually; my Thinkpad 760XL's LVDA (I think it was LVDA. It wasn't a
common laptop video connection yet) display didn't support EDID, and I
never did manage to get any version of XFree86 newer than 3.3.6 working
on it.

Point is, Xorg autoconfiguration has worked find for *most* setups for
almost a decade. At the very least, it's worked fine since the first
graphical Knoppix and Ubuntu live CDs.



CRTs and EDID (was: Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread Michael Mol
On Tue, Nov 15, 2011 at 1:21 AM,  waltd...@waltdnes.org wrote:
 Contrary to the FUD I've heard, X works just fine, thank you, without an
 xorg.conf.  Modern flatscreens with EDID info are set up automatically.  I
 suppose that old CRT monitors without EDID info might require xorg.conf,
 but that's exotic hardware nowadays.

Just an FYI, EDID blocks have been part of CRT tech since the mid to
late 90s; it's the basis of plug  play monitors.

IIRC, the EDID block is transported via DDC, which is essentially I2C
implemented on top of your VGA cable. I've got three EDID-supporting,
19 1600x1200 CRTs staring me in the face right now.

https://plus.google.com/108080062547354628132/posts/ZLLw66eL4We

-- 
:wq



Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread Alan McKinnon
On Tue, 15 Nov 2011 14:44:58 +0700
Pandu Poluan pa...@poluan.info wrote:

   Create the file if it doesn't already exist.  You now have a
  totally udev-free machine
 
 
 Sounds nice!
 
 However, my Gentoo systems are all virtual servers (DomU VMs on
 XenServer). So, the hardware devices are static. Will switching over
 to mdev give any benefits?
 
 I even am toying around with the idea of having a completely
 static /dev, but still can't find any guide/pointers yet.
 
 (Apologies if my email is OOT)
 

A VM can be surprisingly useful for this. If you can emulate different
hardware you can generate useful testing scenarios quickly. The tests
won't be conclusive (emulated hardware is not the same thing as real
hardware) but you *can* test to a standard. 

-- 
Alan McKinnnon
alan.mckin...@gmail.com



Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread Albert W. Hopkins
On Tue, 2011-11-15 at 14:44 +0700, Pandu Poluan wrote:
 However, my Gentoo systems are all virtual servers (DomU VMs on
 XenServer).
 So, the hardware devices are static. Will switching over to mdev give
 any
 benefits?
 
 I even am toying around with the idea of having a completely
 static /dev,
 but still can't find any guide/pointers yet.

I have a completely static /dev/ on my VMs.. I just disable the udev
devfs services and create whatever device nodes needed manually.





Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread Pandu Poluan
On Nov 15, 2011 11:19 PM, Alan McKinnon alan.mckin...@gmail.com wrote:

 On Tue, 15 Nov 2011 14:44:58 +0700
 Pandu Poluan pa...@poluan.info wrote:

Create the file if it doesn't already exist.  You now have a
   totally udev-free machine
  
 
  Sounds nice!
 
  However, my Gentoo systems are all virtual servers (DomU VMs on
  XenServer). So, the hardware devices are static. Will switching over
  to mdev give any benefits?
 
  I even am toying around with the idea of having a completely
  static /dev, but still can't find any guide/pointers yet.
 
  (Apologies if my email is OOT)
 

 A VM can be surprisingly useful for this. If you can emulate different
 hardware you can generate useful testing scenarios quickly. The tests
 won't be conclusive (emulated hardware is not the same thing as real
 hardware) but you *can* test to a standard.


True. Unfortunately, I don't have 'exotic' hardware to test mdev against,
and USB pass-through is not yet supported on XenServer 5.6 (which I'm using
right now).

I can try it inside VirtualBox on my Windows workstation though. Will that
help?

Rgds,


Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread Pandu Poluan
On Nov 15, 2011 11:43 PM, Albert W. Hopkins mar...@letterboxes.org
wrote:

 On Tue, 2011-11-15 at 14:44 +0700, Pandu Poluan wrote:
  However, my Gentoo systems are all virtual servers (DomU VMs on
  XenServer).
  So, the hardware devices are static. Will switching over to mdev give
  any
  benefits?
 
  I even am toying around with the idea of having a completely
  static /dev,
  but still can't find any guide/pointers yet.

 I have a completely static /dev/ on my VMs.. I just disable the udev
 devfs services and create whatever device nodes needed manually.


Ah, thanks! Now I have more confidence to experiment, knowing someone else
have successfully went the static /dev road :-)

Rgds,


Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread waltdnes
On Tue, Nov 15, 2011 at 02:44:58PM +0700, Pandu Poluan wrote

 However, my Gentoo systems are all virtual servers (DomU VMs on XenServer).
 So, the hardware devices are static. Will switching over to mdev give any
 benefits?
 
 I even am toying around with the idea of having a completely static /dev,
 but still can't find any guide/pointers yet.
 
 (Apologies if my email is OOT)

  The more scenarios we can test, the better.  mdev might shave a second
  or two off the VM's bootup time, versus udev.

-- 
Walter Dnes waltd...@waltdnes.org



Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread Pandu Poluan
On Nov 16, 2011 3:21 AM, waltd...@waltdnes.org wrote:

 On Tue, Nov 15, 2011 at 02:44:58PM +0700, Pandu Poluan wrote

  However, my Gentoo systems are all virtual servers (DomU VMs on
XenServer).
  So, the hardware devices are static. Will switching over to mdev give
any
  benefits?
 
  I even am toying around with the idea of having a completely static
/dev,
  but still can't find any guide/pointers yet.
 
  (Apologies if my email is OOT)

  The more scenarios we can test, the better.  mdev might shave a second
  or two off the VM's bootup time, versus udev.


Okay, I have two staging VMs on XenServer and one on VMware. I'm going to
report back what happens. If anything bad happens, *should* be an easy
rollback to the previous snapshot.

Rgds,


Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread Pandu Poluan
Plus, I'm feeling adventurous and will experiment with VirtualBox also ;)

Rgds,
 On Nov 16, 2011 7:52 AM, Pandu Poluan pa...@poluan.info wrote:


 On Nov 16, 2011 3:21 AM, waltd...@waltdnes.org wrote:
 
  On Tue, Nov 15, 2011 at 02:44:58PM +0700, Pandu Poluan wrote
 
   However, my Gentoo systems are all virtual servers (DomU VMs on
 XenServer).
   So, the hardware devices are static. Will switching over to mdev give
 any
   benefits?
  
   I even am toying around with the idea of having a completely static
 /dev,
   but still can't find any guide/pointers yet.
  
   (Apologies if my email is OOT)
 
   The more scenarios we can test, the better.  mdev might shave a second
   or two off the VM's bootup time, versus udev.
 

 Okay, I have two staging VMs on XenServer and one on VMware. I'm going to
 report back what happens. If anything bad happens, *should* be an easy
 rollback to the previous snapshot.

 Rgds,



Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-15 Thread yegle
I'm using systemd as init. Currently there's no .service file for mdev.
Hope someone on this list can provide one :-)

On Wed, Nov 16, 2011 at 9:41 AM, Pandu Poluan pa...@poluan.info wrote:

 Plus, I'm feeling adventurous and will experiment with VirtualBox also ;)

 Rgds,
  On Nov 16, 2011 7:52 AM, Pandu Poluan pa...@poluan.info wrote:


 On Nov 16, 2011 3:21 AM, waltd...@waltdnes.org wrote:
 
  On Tue, Nov 15, 2011 at 02:44:58PM +0700, Pandu Poluan wrote
 
   However, my Gentoo systems are all virtual servers (DomU VMs on
 XenServer).
   So, the hardware devices are static. Will switching over to mdev give
 any
   benefits?
  
   I even am toying around with the idea of having a completely static
 /dev,
   but still can't find any guide/pointers yet.
  
   (Apologies if my email is OOT)
 
   The more scenarios we can test, the better.  mdev might shave a second
   or two off the VM's bootup time, versus udev.
 

 Okay, I have two staging VMs on XenServer and one on VMware. I'm going to
 report back what happens. If anything bad happens, *should* be an easy
 rollback to the previous snapshot.

 Rgds,




[gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-14 Thread waltdnes
  After a recent thread, about udev developers wanting /usr on the same
partition as / (or else requiring initramfs), it was pretty obvious
that 90%+ of the users here strongly disliked the idea.  I went around
asking on various lists if it was possible to run Gentoo without udev.
After some research, and various unrelated delays, I've come up with a
working Gentoo without udev.  It turns out that busybox's mdev
implementation is sufficient for my needs.  I do the usual email, web
surfing, including Youtube.  I'm listening to Live365.com as I type this
email, so Flash works just fine.  Contrary to the FUD I've heard, X
works just fine, thank you, without an xorg.conf.  Modern flatscreens
with EDID info are set up automatically.  I suppose that old CRT
monitors without EDID info might require xorg.conf, but that's exotic
hardware nowadays.  The only change I notice is somewhat faster bootup.

  The purpose of this email is to ask adventurous people here to beta
test my approach to a udev-less Gentoo.  If we don't find any
showstopper problems, we can think about requesting Gentoo developers to
support an mdev-based profile.  It would help the cause if a large
number of testers can report that it works for them.  The instructions
for a udev-ectomy follow below.  Thanks to Zac Medico and others on the
Gentoo developers' list for their helpful hints and pointers on how to
do this.  I couldn't have figured this out by myself.

  The usual warnings apply...
* this is a beta
* use a spare test machine
* if you don't follow the instructions correctly, the result might be
  an unbootable linux
* even if you do follow instructions, the result might be an unbootable
  linux


1) Set up your kernel to support and automount a devtmpfs filesystem at
   /dev

* If you prefer to edit .config directly, set
  CONFIG_DEVTMPFS=y and CONFIG_DEVTMPFS_MOUNT=y

* If you prefer make menuconfig, the route is as shown below.  Note
  that the Autount devtmpfs... option won't appear until you enable
  Maintain a devtmpf... option.

make menuconfig
  Device Drivers  ---
Generic Driver Options  ---
  [*] Maintain a devtmpfs filesystem to mount at /dev
  [*]   Automount devtmpfs at /dev, after the kernel mounted the rootfs

  Once you've made the changes, rebuild the kernel.


2) Set up for emerging busybox, there are 2 items to change

A) It appears that there may be an mdev bug in older versions of
   busybox.  To avoid that bug, keyword busybox-1.19.2 in
   /etc/portage/package.keywords  E.g. if you're using 32-bit Gentoo on
   Intel, the incantation is...

=sys-apps/busybox-1.19.2 ~x86

   Change the ~x86 to reflect your architecture, etc.

B) busybox requires the mdev flag in this situation.  The static
flag is probably also a good idea.  In file /etc/portage/package.use
add the line

sys-apps/busybox static mdev

   Now, emerge busybox


3) In the bootloader append line, include init=/sbin/linuxrc where
   the file /sbin/linuxrc consists of *AT LEAST*...

#!/sbin/busybox ash
mount -t proc proc /proc
mount -t sysfs sysfs /sys
exec /sbin/init

   This should be enough for most users.  If you have an unusual setup,
   you may need additional stuff in there.  If you're using lilo remember
   to re-run lilo to implement the changes.

4) Remove udev from the services list, and replace it with mdev.  Type
   the following 2 commands at the command line
rc-update del udev sysinit
rc-update add mdev sysinit


5) reboot to your new kernel.  You're now running without using udev.


6) ***THIS STEP IS OPTIONAL***  This is only to alay any suspicion that
   udev is still in use.  udev is pulled in by virtual/dev-manager,
   which in turn is pulled in by the kernel.
* cd /usr/portage/virtual/dev-manager
* Make a backup copy of dev-manager-0.ebuild
* Edit dev-manager-0.ebuild to include sys-apps/busybox as one option
  in RDEPEND, like so...

RDEPEND=|| ( sys-fs/udev
sys-fs/devfsd
sys-apps/busybox
sys-fs/static-dev
sys-freebsd/freebsd-sbin )

  I had really wanted to use sys-apps/busybox[mdev], but an EAPI-0
  ebuild can't handle that syntax.

* execute the following 3 commands at the commandline
ebuild dev-manager-0.ebuild digest
emerge -1 dev-manager
emerge --unmerge sys-fs/udev

* In file /atc/portage/package.mask, append the line
sys-fs/udev
  Create the file if it doesn't already exist.  You now have a totally
  udev-free machine

-- 
Walter Dnes waltd...@waltdnes.org



Re: [gentoo-user] Anybody want to beta test Gentoo with mdev instead of udev?

2011-11-14 Thread Pandu Poluan
On Nov 15, 2011 1:24 PM, waltd...@waltdnes.org wrote:

  After a recent thread, about udev developers wanting /usr on the same
 partition as / (or else requiring initramfs), it was pretty obvious
 that 90%+ of the users here strongly disliked the idea.  I went around
 asking on various lists if it was possible to run Gentoo without udev.
 After some research, and various unrelated delays, I've come up with a
 working Gentoo without udev.  It turns out that busybox's mdev
 implementation is sufficient for my needs.  I do the usual email, web
 surfing, including Youtube.  I'm listening to Live365.com as I type this
 email, so Flash works just fine.  Contrary to the FUD I've heard, X
 works just fine, thank you, without an xorg.conf.  Modern flatscreens
 with EDID info are set up automatically.  I suppose that old CRT
 monitors without EDID info might require xorg.conf, but that's exotic
 hardware nowadays.  The only change I notice is somewhat faster bootup.

  The purpose of this email is to ask adventurous people here to beta
 test my approach to a udev-less Gentoo.  If we don't find any
 showstopper problems, we can think about requesting Gentoo developers to
 support an mdev-based profile.  It would help the cause if a large
 number of testers can report that it works for them.  The instructions
 for a udev-ectomy follow below.  Thanks to Zac Medico and others on the
 Gentoo developers' list for their helpful hints and pointers on how to
 do this.  I couldn't have figured this out by myself.

  The usual warnings apply...
 * this is a beta
 * use a spare test machine
 * if you don't follow the instructions correctly, the result might be
  an unbootable linux
 * even if you do follow instructions, the result might be an unbootable
  linux


 1) Set up your kernel to support and automount a devtmpfs filesystem at
   /dev

 * If you prefer to edit .config directly, set
  CONFIG_DEVTMPFS=y and CONFIG_DEVTMPFS_MOUNT=y

 * If you prefer make menuconfig, the route is as shown below.  Note
  that the Autount devtmpfs... option won't appear until you enable
  Maintain a devtmpf... option.

 make menuconfig
  Device Drivers  ---
Generic Driver Options  ---
  [*] Maintain a devtmpfs filesystem to mount at /dev
  [*]   Automount devtmpfs at /dev, after the kernel mounted the rootfs

  Once you've made the changes, rebuild the kernel.


 2) Set up for emerging busybox, there are 2 items to change

 A) It appears that there may be an mdev bug in older versions of
   busybox.  To avoid that bug, keyword busybox-1.19.2 in
   /etc/portage/package.keywords  E.g. if you're using 32-bit Gentoo on
   Intel, the incantation is...

 =sys-apps/busybox-1.19.2 ~x86

   Change the ~x86 to reflect your architecture, etc.

 B) busybox requires the mdev flag in this situation.  The static
 flag is probably also a good idea.  In file /etc/portage/package.use
 add the line

 sys-apps/busybox static mdev

   Now, emerge busybox


 3) In the bootloader append line, include init=/sbin/linuxrc where
   the file /sbin/linuxrc consists of *AT LEAST*...

 #!/sbin/busybox ash
 mount -t proc proc /proc
 mount -t sysfs sysfs /sys
 exec /sbin/init

   This should be enough for most users.  If you have an unusual setup,
   you may need additional stuff in there.  If you're using lilo remember
   to re-run lilo to implement the changes.

 4) Remove udev from the services list, and replace it with mdev.  Type
   the following 2 commands at the command line
 rc-update del udev sysinit
 rc-update add mdev sysinit


 5) reboot to your new kernel.  You're now running without using udev.


 6) ***THIS STEP IS OPTIONAL***  This is only to alay any suspicion that
   udev is still in use.  udev is pulled in by virtual/dev-manager,
   which in turn is pulled in by the kernel.
 * cd /usr/portage/virtual/dev-manager
 * Make a backup copy of dev-manager-0.ebuild
 * Edit dev-manager-0.ebuild to include sys-apps/busybox as one option
  in RDEPEND, like so...

 RDEPEND=|| ( sys-fs/udev
sys-fs/devfsd
sys-apps/busybox
sys-fs/static-dev
sys-freebsd/freebsd-sbin )

  I had really wanted to use sys-apps/busybox[mdev], but an EAPI-0
  ebuild can't handle that syntax.

 * execute the following 3 commands at the commandline
 ebuild dev-manager-0.ebuild digest
 emerge -1 dev-manager
 emerge --unmerge sys-fs/udev

 * In file /atc/portage/package.mask, append the line
 sys-fs/udev
  Create the file if it doesn't already exist.  You now have a totally
  udev-free machine


Sounds nice!

However, my Gentoo systems are all virtual servers (DomU VMs on XenServer).
So, the hardware devices are static. Will switching over to mdev give any
benefits?

I even am toying around with the idea of having a completely static /dev,
but still can't find any guide/pointers yet.

(Apologies if my email is OOT)

Rgds,