Re: [gentoo-amd64] emerge -eav system failed with ati-drivers

[David Relson <relson@osagesoftware.com>]

On Sun, 21 Jun 2009 15:19:53 -0400
John P. Burkett wrote:

> David Relson wrote:
> > On Sun, 21 Jun 2009 14:54:36 -0400
> > John P. Burkett wrote:
> > 
> >> David Relson wrote:
> >>> On Sun, 21 Jun 2009 13:55:10 -0400
> >>> Drake Donahue wrote:
> >>>
> >>>> <snip>
> >>>>
> >>>> I'm guessing you did an "emerge --depclean" sometime after
> >>>> compiling the 2.6.22-r2 kernel and after emerging a new
> >>>> gentoo-sources that erased your make files, This is something
> >>>> depclean loves to do while leaving 99% of the older kernel source
> >>>> files intact. I think depclean leaves your .config file in place.
> >>>> Been a few weeks since I burned myself last with this one. 
> >>>> If you were following directions back in 2007 you should have a
> >>>> stored 2.6.22-r2 config in the /boot directory.
> >>>>
> >>>> Try "emerge =sys-kernel/gentoo-sources-2.6.22-r2 to re-emerge the
> >>>> old package and regain the make files. What the osmp is about
> >>>> beats me.
> >>>>
> >>>> or
> >>>>
> >>>> Emerge a new gentoo-sources, eselect it, configure, compile,
> >>>> install a new (current) kernel.
> >>> My understanding is that ati-drivers expects /usr/src/linux to
> >>> contain the source of the currently running kernel.
> >>>
> >>> I'm running a 2.6.28 kernel.  I have downloaded 2.6.30 and built
> >>> that but have not yet rebooted.  "emerge ati-drivers" failed with
> >>> 2.6.30 in /usr/src/linux.  Changing directory names allowed
> >>> ati-drivers to build happily.
> >>>
> >> David, thank you for sharing your experience. On my system, doing
> >> "uname -r" gets the response "2.6.22-gentoo-r2-osmp".  In
> >> my /usr/src/linux directory I currently have a .config file
> >> starting with the lines # Automatically generated make config:
> >> don't edit # Linux kernel version: 2.6.22-gentoo-r2
> >> # Mon Sep  3 21:48:37 2007
> >> Can you suggest how to change directory names to allow the
> >> ati-drivers to build?
> >>
> >> Best regards,
> >> John
> > 
> > John,
> > 
> > /usr/src/linux is what I'd want to see.  Do you have just a few
> > files (like .config) or the full kernel source? 
> Thanks for your continuing help with the ati-driver problem.  Doing
> "ls -al" in my /usr/src/linux directory produces the following list:
> total 28451
> drwxr-xr-x 19 root root    1088 Jun  8 01:16 .
> drwxr-xr-x  3 root root     136 Jun  8 01:19 ..
> -rw-r--r--  1 root root     440 Sep  4  2007 ..tmp_kallsyms1.o.cmd
> -rw-r--r--  1 root root     440 Sep  4  2007 ..tmp_kallsyms2.o.cmd
> -rw-r--r--  1 root root     636 Sep  4  2007 ..tmp_vmlinux1.cmd
> -rw-r--r--  1 root root     652 Sep  4  2007 ..tmp_vmlinux2.cmd
> -rw-r--r--  1 root root   47220 Jun 21 13:34 .config
> -rw-r--r--  1 root root   47220 Sep  3  2007 .config.old
> -rw-r--r--  1 root root     365 Sep  4  2007 .missing-syscalls.d
> -rw-r--r--  1 root root 1418737 Sep  4  2007 .tmp_System.map
> -rw-r--r--  1 root root 1578817 Sep  4  2007 .tmp_kallsyms1.S
> -rw-r--r--  1 root root  735072 Sep  4  2007 .tmp_kallsyms1.o
> -rw-r--r--  1 root root 1578817 Sep  4  2007 .tmp_kallsyms2.S
> -rw-r--r--  1 root root  735072 Sep  4  2007 .tmp_kallsyms2.o
> drwxr-xr-x  2 root root   10176 Sep  4  2007 .tmp_versions
> -rwxr-xr-x  1 root root 9069262 Sep  4  2007 .tmp_vmlinux1
> -rwxr-xr-x  1 root root 9388791 Sep  4  2007 .tmp_vmlinux2
> -rw-r--r--  1 root root       2 Sep  4  2007 .version
> -rw-r--r--  1 root root     640 Sep  4  2007 .vmlinux.cmd
> -rw-r--r--  1 root root  252456 Sep  4  2007 Module.symvers
> -rw-r--r--  1 root root 1418737 Sep  4  2007 System.map
> drwxr-xr-x  4 root root      96 Jun  8 01:16 arch
> drwxr-xr-x  2 root root     752 Jun  8 01:16 block
> drwxr-xr-x  2 root root    6640 Jun  8 01:16 crypto
> drwxr-xr-x 35 root root     968 Jun  8 01:16 drivers
> drwxr-xr-x 35 root root    4936 Jun  8 01:16 fs
> drwxr-xr-x  5 root root     152 Jun  8 01:16 include
> drwxr-xr-x  2 root root     736 Jun  8 01:16 init
> drwxr-xr-x  2 root root     680 Jun  8 01:16 ipc
> drwxr-xr-x  5 root root    3832 Jun  8 01:16 kernel
> drwxr-xr-x  4 root root    4512 Jun  8 01:16 lib
> drwxr-xr-x  2 root root    2616 Jun  8 01:16 mm
> drwxr-xr-x 23 root root     856 Jun  8 01:16 net
> drwxr-xr-x  6 root root     320 Jun  8 01:16 scripts
> drwxr-xr-x  2 root root     208 Jun  8 01:16 security
> drwxr-xr-x 16 root root     904 Jun  8 01:16 sound
> drwxr-xr-x  2 root root     424 Jun  8 01:16 usr
> -rwxr-xr-x  1 root root 9388791 Sep  4  2007 vmlinux
> 
> Does that appear to have what ati-drivers should need?
> 
> Best regards,
> John

John,

That looks like what I've got.  If you're up to some debugging, here's
some more info:

When I deliberating have the wrong kernel source in /usr/src/linux, my
build fails the last few lines printed are:

 If you need support, post the topmost build error, and the call stack
 if relevant.
 * A complete build log is located at
 '/var/log/portage/x11-drivers:ati-drivers-8.552-r2:20090621-200106.log'.
 * The ebuild environment file is located at
   '/var/tmp/portage/x11-drivers/ati-drivers-8.552-r2/temp/environment'.


Looking at file .../temp/environment I see it's a bash script and
contains line 2536 is

    get_version || die "Unable to calculate Linux Kernel version";

which is the message in your original post.  Evidently function
get_version() is failing and causing the "die" command to be executed.
 I have function get_version() at line 2326 of .../environment.  Take a
 look at that and see if you can figure out where it might be dying.
 Running the script in debug mode, i.e. using "sh -x environment", will
 give you detailed information about the script is finding and doing.


HTH,

David