Re: [gentoo-user] getting somewhere -- was about a bug

2006-05-22 Thread Richard Fish

On 5/19/06, maxim wexler [EMAIL PROTECTED] wrote:

ls -l /dev/ttyS*:

lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS0 -
tts/0
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS1 -
tts/1
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS2 -
tts/2
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS3 -
tts/3
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS4 -
tts/4
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS5 -
tts/5
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS6 -
tts/6
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS7 -
tts/7

grep ttyS
/etc/udev/rules.d/*:

/etc/udev/rules.d/50-udev.rules:KERNEL==ttyS[0-9]*,
NAME=%k, SYMLINK=tts/%n, GROUP=tty


Sorry about the slow response here...

Those /dev/ttyS* entries should not be symlinks according to the udev
rule.  The udev rule will try to create actual nodes of /dev/ttyS, and
then symlinks to those in /dev/tts/.

Do you have RC_DEVICE_TARBALL set in /etc/conf.d/rc?  That is the only
way I can think of that this would occur.  If so, try setting it to
no, to allow udev to completely manage /dev.  RC_DEVICE_TARBALL is
rarely needed now...


/usr/sbin/pppd: Couldn't stat /dev/ttyS0: Too many
levels of symbolic links.


Yeah, /dev/ttyS0 - /dev/tts/0 - /dev/ttyS0 - /dev/tts/0 - ...

-Richard

--
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] getting somewhere -- was about a bug

2006-05-19 Thread maxim wexler


--- Richard Fish [EMAIL PROTECTED] wrote:

 On 5/18/06, maxim wexler [EMAIL PROTECTED] wrote:
  Just one little problem left. From the boot
 console:
 
  udevd-event[3787]: udev_make_node:
 mknod(/dev/ttyS0,
  020660, 4, 64)
  failed: File exists
 
 Hmm, odd.  What do ls -l /dev/ttyS* and grep ttyS
 /etc/udev/rules.d/* produce?

This is weird. As root using the --color option the
links are displayed in normal turquoise. But as user
the links are flashing white on red, indicating
missing links(?). 

ls -l /dev/ttyS*:

lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS0 -
tts/0
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS1 -
tts/1
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS2 -
tts/2
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS3 -
tts/3
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS4 -
tts/4
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS5 -
tts/5
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS6 -
tts/6
lrwxrwxrwx 1 root root 5 May 19 10:07 /dev/ttyS7 -
tts/7

grep ttyS
/etc/udev/rules.d/*:

/etc/udev/rules.d/50-udev.rules:KERNEL==ttyS[0-9]*,
NAME=%k, SYMLINK=tts/%n, GROUP=tty


More weirdness: seems my provider file is wiped out
every time I reboot. Have to re-run pppconfig over and
over.

But it still won't dial-out as root or user, error is:

/usr/sbin/pppd: Couldn't stat /dev/ttyS0: Too many
levels of symbolic links. 

Over to you Richard.

-Maxim

 
 -Richard
 
 -- 
 gentoo-user@gentoo.org mailing list
 
 



__
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
-- 
gentoo-user@gentoo.org mailing list



[gentoo-user] getting somewhere -- was about a bug

2006-05-18 Thread maxim wexler
Hi all,

A bunch of things came together in the mother of all
screw-ups.

One, somehow the /boot line in fstab got set to
noauto. How that happened have no idea.

Two, at some point I must have compiled a kernel after
chrooting without making sure /dev/boot was mounted
on /boot. Then, when I copied over bzImage it landed
under the empty /boot dir on the root device.

Three, when I was told to save .config somewhere then
clean the source then cp .config back and make
oldconfig, I couldn't find it because I looked in the
wrong dir!

Four, I made another kernel but this time neglected to
include the proper SATA drivers. When I said no
warning before the panic, there was actually lots but
I took them for neutral messages.

So that's sorted.

Just one little problem left. From the boot console:

udevd-event[3787]: udev_make_node: mknod(/dev/ttyS0,
020660, 4, 64) 
failed: File exists
udevd-event[3788]: udev_make_node: mknod(/dev/ttyS1,
020660, 4, 65) 
failed: File exists
udevd-event[3789]: udev_make_node: mknod(/dev/ttyS2,
020660, 4, 66) 
failed: File exists
udevd-event[3790]: udev_make_node: mknod(/dev/ttyS3,
020660, 4, 67) 
failed: File exists   
   
[ !! ] *

This must have something to do with the end of console
boot just before login where it says:

/usr/sbin/ppd/: Couldn't stat /dev/ttyS0: Too many
levels of symbolic links

So in the end, I still can't dial out, on that box
anyway.

-mw

__
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] getting somewhere -- was about a bug

2006-05-18 Thread Neil Bothwick
On Thu, 18 May 2006 12:43:55 -0700 (PDT), maxim wexler wrote:

 One, somehow the /boot line in fstab got set to
 noauto. How that happened have no idea.

It is supposed to have noauto, because /boot does not need to be mounted
in the normal course of events. GRUB doesn't use /etc/fstab, it uses
grub.conf to find the kernel. The only time you need to mount /boot is
when installing a new kernel.


-- 
Neil Bothwick

I have a mind like a steel...uh...thingamajig...


signature.asc
Description: PGP signature


Re: [gentoo-user] getting somewhere -- was about a bug

2006-05-18 Thread Richard Fish

On 5/18/06, maxim wexler [EMAIL PROTECTED] wrote:

Just one little problem left. From the boot console:

udevd-event[3787]: udev_make_node: mknod(/dev/ttyS0,
020660, 4, 64)
failed: File exists


Hmm, odd.  What do ls -l /dev/ttyS* and grep ttyS
/etc/udev/rules.d/* produce?

-Richard

--
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] getting somewhere -- was about a bug

2006-05-18 Thread Richard Fish

On 5/18/06, Neil Bothwick [EMAIL PROTECTED] wrote:

It is supposed to have noauto, because /boot does not need to be mounted
in the normal course of events. GRUB doesn't use /etc/fstab, it uses
grub.conf to find the kernel. The only time you need to mount /boot is
when installing a new kernel.


I disagree that it is 'supposed' to have noauto.  This could make the
system more secure, but so could mounting it read-only.   Users do
forget to mount it before updating the kernel, and they get confused
about why the system isn't booting from their freshly compiled kernel.
I am sure Maxim is not the only one to do this...

In fact, the current handbook says:

---
Code Listing 2: An example /boot line for /etc/fstab
 /dev/hda1   /boot ext2defaults1 2

  Some users don't want their /boot partition to be mounted
automatically to improve their system's security. Those people should
substitute defaults with noauto. This does mean that you need to
manually mount this partition every time you want to use it.
---

So it seems that 'noauto' is an option for specific users.  Not for
general use...

-Richard

--
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] getting somewhere -- was about a bug

2006-05-18 Thread Mike Williams
On Thursday 18 May 2006 21:34, Richard Fish wrote:
  It is supposed to have noauto, because /boot does not need to be mounted
  in the normal course of events. GRUB doesn't use /etc/fstab, it uses
  grub.conf to find the kernel. The only time you need to mount /boot is
  when installing a new kernel.

 I disagree that it is 'supposed' to have noauto.  This could make the
 system more secure, but so could mounting it read-only.   Users do
 forget to mount it before updating the kernel, and they get confused
 about why the system isn't booting from their freshly compiled kernel.
  I am sure Maxim is not the only one to do this...

noauto was the default.
One accident a couple of years ago soon made me change to mounting /boot 
read-only (and successfully submitting a patch to genkernel to handle that).
You can format an un-mounted filesystem. That's bad when it's /dev/sda1, and 
not in fact the /dev/sdb1 which you actually wanted to format.

-- 
Mike Williams

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] getting somewhere -- was about a bug

2006-05-18 Thread Neil Bothwick
On Thu, 18 May 2006 13:34:55 -0700, Richard Fish wrote:

 In fact, the current handbook says:
 
 ---
 Code Listing 2: An example /boot line for /etc/fstab
   /dev/hda1   /boot ext2defaults1 2
 
Some users don't want their /boot partition to be mounted
 automatically to improve their system's security. Those people should
 substitute defaults with noauto. This does mean that you need to
 manually mount this partition every time you want to use it.
 ---
 
 So it seems that 'noauto' is an option for specific users.  Not for
 general use...

Things have changed in the installation handbook then, one of the
disadvantages of running a system that never needs re-installing, you
remember out of date information.

noauto *was* the default.


-- 
Neil Bothwick

People who eat natural foods die from natural causes.


signature.asc
Description: PGP signature