Trouble compiling new 2.6.21.5 on Slack 12

raypen · 06-22-2008, 04:09 PM

In compiling a new kernel on Slackware 12, I've run into
a few problems. Upon reboot, I have encountered several
error messages as follows:

Code:

Jun 22 13:09:49 chipper udevd-event[1804]: node_symlink: symlink(pty/m127, /dev/ptywf) failed: File exists
Jun 22 13:09:50 chipper udevd-event[1806]: node_symlink: symlink(pty/m129, /dev/ptyx1) failed: File exists
Jun 22 13:09:50 chipper udevd-event[1807]: node_symlink: symlink(pty/m130, /dev/ptyx2) failed: File exists
Jun 22 13:09:50 chipper udevd-event[1808]: node_symlink: symlink(pty/m131, /dev/ptyx3) failed: File exists
Jun 22 13:09:50 chipper udevd-event[1811]: node_symlink: symlink(pty/m133, /dev/ptyx5) failed: File exists
Jun 22 13:09:52 chipper udevd-event[1805]: node_symlink: symlink(pty/m128, /dev/ptyx0) failed: File exists
Jun 22 13:09:52 chipper udevd-event[1814]: node_symlink: symlink(pty/m136, /dev/ptyx8) failed: File exists
Jun 22 13:09:52 chipper udevd-event[1812]: node_symlink: symlink(pty/m134, /dev/ptyx6) failed: File exists
Jun 22 13:09:52 chipper udevd-event[1815]: node_symlink: symlink(pty/m137, /dev/ptyx9) failed: File exists
Jun 22 13:09:52 chipper udevd-event[1816]: node_symlink: symlink(pty/m138, /dev/ptyxa) failed: File exists
Jun 22 13:09:52 chipper udevd-event[1809]: node_symlink: symlink(pty/m132, /dev/ptyx4) failed: File exists
Jun 22 13:09:52 chipper udevd-event[1818]: node_symlink: symlink(pty/m140, /dev/ptyxc) failed: File exists
....

These appeared in syslog on my first reboot. I reran the compile, only adding support for device-mapper
without adding LVM. The messages continue to appear on the boot screen but are now absent from syslog.
Nothing like this appears in dmesg, meassages or debug.

I know something is wrong with my .config file since the generic-2.6.25.1 works fine. I have tried to
compare the two, but there are so many differences, it is impossible to home-in on the specifics
causing the above problems.

The machine I am running is a rather inexpensive pcChips motherboard, with the following hardware:

AMD Athlon XP 1800+ processor
S3 ProSavageDDR K4M266 Integrated Video
VT82C586/B/686/A/B PIPC Bus Master IDE
VT8375 [KM266] Host Bridge
VT8633 [Apollo Pro266 AGP] PCI Bridge

The new 2.6 kernel is proving to be somewhat of a mystery to me. I have never had as many problems trying
to recompile a kernel. I would appreciate any help in this matter, expecially a link to good treatise on
the interaction between the various new components; udev, LVM, device-mapper, sysfs etc. I find the usual
help pages within menuconfig are not as helpful as they once were.

storkus · 06-23-2008, 04:14 AM

You *DO* realize that 2.6 isn't that new anymore, right? The current version is 2.6.25.7 with 2.6.26 around the corner.

As for your problem, it looks easy: you compiled in /dev/pty support--disable it. It is incompatible with udev.

Mike

raypen · 06-23-2008, 10:07 AM

When I say I compiled a new kernel I only meant a new kernel based
on the 2.6.21.5 kernel source upon which Slack 12 is based. Almost
any distribution will necessarily provide a kernel which is not the
latest available.

Oddly, the configuration of the generic-2.6.21.5 includes pty support
without any type of error similar to the above. There were additional
errors which I did not show in the previous post due to space limitations:

Code:

Jun 22 13:09:56 chipper udevd-event[1878]: node_symlink: symlink(vc/21, /dev/tty21) failed: File exists
Jun 22 13:09:56 chipper udevd-event[1879]: node_symlink: symlink(vc/22, /dev/tty22) failed: File exists
Jun 22 13:09:56 chipper udevd-event[1880]: node_symlink: symlink(vc/23, /dev/tty23) failed: File exists
Jun 22 13:09:56 chipper udevd-event[1881]: node_symlink: symlink(vc/24, /dev/tty24) failed: File exists
Jun 22 13:09:56 chipper udevd-event[1874]: node_symlink: symlink(vc/19, /dev/tty19) failed: File exists
Jun 22 13:09:56 chipper udevd-event[1883]: node_symlink: symlink(vc/26, /dev/tty26) failed: File exists
...........

I assume vc means virtual console.

I compiled once again with device-mapper/multi-device support, but
am still receiving the same messages.

In my latest compile I did remove Legacy PTY (CONFIG_LEGACY_PTYS is not
set) while retaining UNIX 98 PTY support. This did not make any
difference. It seems to me that you do need PTY support for some
functions and can't eliminate it completely.

I'll continue to play around with it unless you have any further
useful suggestions.

raypen · 06-25-2008, 09:44 AM

Transferred a known good .config file from another AMD type machine,
altered it slightly in make menuconfig to properly recognize different
NIC's and add support for a SCSI card and rebooted. The new kernel
booted perfectly, with no error messages. PTY support is built in.

After comparing the good/bad .config files, I noticed that the bad
configuration had the following set:

Code:

CONFIG_TICK_ONESHOT=y
CONFIG_NO_HZ=y
CONFIG_HIGH_RES_TIMERS=y

After searching the net, I found that this setup causes several
problems with many installations.

In booting the bad config, several lines at the top of dmesg referred
to remapping of some sort. I vaguely recall something in menuconfig
in the processor section about remapping code to the high memory
area to avoid congestion of the low memory area. These messages
were gone in the good kernel.

While I am not certain these entries are the only culprits, it seems
likely. Another benefit of this change is that X-windows is now
working properly with video acceleration, whereas before, exiting
XWindows caused the system to hang.

onebuck · 06-25-2008, 11:03 AM

Hi,

Quote:

Originally Posted by raypen

<snip> Another benefit of this change is that X-windows is now
working properly with video acceleration, whereas before, exiting
XWindows caused the system to hang.

It is not 'XWindows' or 'X-windows' but common as 'X or X11'. If you refer to 'X Window System' then that is proper. I know a lot of people fail to use the proper term for 'X'.

raypen · 07-03-2008, 09:25 AM

Well, I am terribly embarrassed. Forgive me for using improper
terminology when referring to the "X Window System". I am so
delighted that, instead of suggesting a solution to my problem,
you saw fit to correct my use of terminology.

BTW: To correct the problem, make sure 'tmpfs' is included in
the configuration.

onebuck · 07-03-2008, 10:03 PM

Hi,

Well I'll Be? Spank me for trying to provide the proper terms to enable communication. This is a volunteer forum and we can and do provide information. Not on demand!

I'll make sure to note this for future references. I really don't remember your problem or why I didn't complete my post within the thread. Not that important too me. Especially now in the manner that you addressed my post. Nothing wrong with my post BTW, if you cannot take constructive information or criticisms then you won't go to far in life. Oh my that's a bit sage of me. Not PC, spank me again!

Wow, You found a answer, under a rock somewhere? You didn't complete the solution or description so others may garnish from your big problem. But more bandwidth to point out that I made a mistake to correct something that was indeed wrong. Enough said and to much time spent on nothing.

raypen · 07-06-2008, 10:45 PM

Methinks the lady doth protest too much!