Symbol versioning problem

Symbol versioning problem

[Remco Treffkorn]

I am compiling a recent benh kernel. After the build the depmod against
the System.map does not complain, but then, the symbols in it are not
versioned.

Booting the new kernel and depmodding against /proc/ksyms gives me
lots of unresolved symbols like:

depmod: *** Unresolved symbols in /lib/modules/2.4.19-rc1-ben0/kernel/fs/hfs/hfs.o
depmod:         __out_of_line_bug
depmod: *** Unresolved symbols in /lib/modules/2.4.19-rc1-ben0/kernel/fs/smbfs/smbfs.o
depmod:         zone_table
depmod:         flush_dcache_page

Looking those up in /proc/ksyms gives the following:

c00168e4 __out_of_line_bug_R__ver___out_of_line_bug
c02b3e94 zone_table_R__ver_zone_table
c0011994 flush_dcache_page_R__ver_flush_dcache_page

So, instead of the hash we have the symbol name again.
Only very few symbols have this problem.
A quick look at the include files for versioning gave me a head ache ;-)

I don't have this problem with i386, I have not tried another gcc version, but upped
modutils to the latest. No dice

Here is what I have:

Linux io 2.4.19-rc1-ben0 #59 Wed Jul 10 12:06:33 PDT 2002 ppc unknown

Gnu C                  2.95.4
Gnu make               3.79.1
util-linux             2.11f
mount                  2.11g
modutils               2.4.16
e2fsprogs              1.26
reiserfsprogs          3.x.1a
PPP                    2.4.1
Linux C Library        2.2.5
Dynamic linker (ldd)   2.2.5
Procps                 2.0.7
Net-tools              1.60
Console-tools          0.3.3
Sh-utils               2.0.11
Modules Loaded         dmasound_pmac i2c-keywest i2c-core dmasound_core soundcore

Cheers,
Remco


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Symbol versioning problem

[Stefan Eletzhofer]

On Wed, Jul 10, 2002 at 02:29:39PM -0700, Remco Treffkorn wrote:
>
> I am compiling a recent benh kernel. After the build the depmod against
> the System.map does not complain, but then, the symbols in it are not
> versioned.
>
> Booting the new kernel and depmodding against /proc/ksyms gives me
> lots of unresolved symbols like:
>
> depmod: *** Unresolved symbols in /lib/modules/2.4.19-rc1-ben0/kernel/fs/hfs/hfs.o
> depmod:         __out_of_line_bug
> depmod: *** Unresolved symbols in /lib/modules/2.4.19-rc1-ben0/kernel/fs/smbfs/smbfs.o
> depmod:         zone_table
> depmod:         flush_dcache_page
>
> Looking those up in /proc/ksyms gives the following:
>
> c00168e4 __out_of_line_bug_R__ver___out_of_line_bug
> c02b3e94 zone_table_R__ver_zone_table
> c0011994 flush_dcache_page_R__ver_flush_dcache_page
>
> So, instead of the hash we have the symbol name again.
> Only very few symbols have this problem.
> A quick look at the include files for versioning gave me a head ache ;-)
>
> I don't have this problem with i386, I have not tried another gcc version, but upped
> modutils to the latest. No dice

Hmm, compiled this kernel with no problems for my PowerBook. My gcc is a 2.4.95, depmod is 2.4.15.
And it a grep of /proc/ksyms reveals
c00180f8 __out_of_line_bug_R8b0fd3c5

Wild guess: do you have
CONFIG_MODVERSIONS=y
in your .config?

just my .02$ anyway :)

>
> Here is what I have:
>
> Linux io 2.4.19-rc1-ben0 #59 Wed Jul 10 12:06:33 PDT 2002 ppc unknown
>
> Gnu C                  2.95.4
> Gnu make               3.79.1
> util-linux             2.11f
> mount                  2.11g
> modutils               2.4.16
> e2fsprogs              1.26
> reiserfsprogs          3.x.1a
> PPP                    2.4.1
> Linux C Library        2.2.5
> Dynamic linker (ldd)   2.2.5
> Procps                 2.0.7
> Net-tools              1.60
> Console-tools          0.3.3
> Sh-utils               2.0.11
> Modules Loaded         dmasound_pmac i2c-keywest i2c-core dmasound_core soundcore
>
> Cheers,
> Remco
>
>

--
Eletztrick Computing - Customized Linux Development
Stefan Eletzhofer
Gottlieb-Daimler-Strasse 10
88214 Ravensburg
GERMANY
http://www.eletztrick.de
mailto://stefan.eletzhofer@eletztrick.de
+49 751 35 44 112
+49 171 23 24 529

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Symbol versioning problem

[Remco Treffkorn]

On Monday 15 July 2002 02:04, Stefan Eletzhofer wrote:
...
> Hmm, compiled this kernel with no problems for my PowerBook. My gcc is a
> 2.4.95, depmod is 2.4.15. And it a grep of /proc/ksyms reveals
> c00180f8 __out_of_line_bug_R8b0fd3c5
>
> Wild guess: do you have
> CONFIG_MODVERSIONS=y
> in your .config?
>
> just my .02$ anyway :)
>

Yes. I even checked, to be sure :-)

Since there was no storm of "me too"s, I assume that it's my particular
system. I'll spend some time to figure it out, and will let you know...

Cheers,
Remco

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Symbol versioning problem

[Tom Rini]

On Tue, Jul 16, 2002 at 11:21:50AM -0700, Remco Treffkorn wrote:
>
> On Monday 15 July 2002 02:04, Stefan Eletzhofer wrote:
> ...
> > Hmm, compiled this kernel with no problems for my PowerBook. My gcc is a
> > 2.4.95, depmod is 2.4.15. And it a grep of /proc/ksyms reveals
> > c00180f8 __out_of_line_bug_R8b0fd3c5
> >
> > Wild guess: do you have
> > CONFIG_MODVERSIONS=y
> > in your .config?
> >
> > just my .02$ anyway :)
> >
>
> Yes. I even checked, to be sure :-)
>
> Since there was no storm of "me too"s, I assume that it's my particular
> system. I'll spend some time to figure it out, and will let you know...

Well, have you tried backing up your .config, running 'make distclean',
moving your config back (and re running your favorite config program)
and trying again?

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Symbol versioning problem

[Remco Treffkorn]

On Tuesday 16 July 2002 12:12, Tom Rini wrote:
...
> Well, have you tried backing up your .config, running 'make distclean',
> moving your config back (and re running your favorite config program)
> and trying again?

I did 'clean', not 'distclean', but your suggestion just sounded right.
The re-build after the distclean did not have the problem. Apparently some
crud stays around through 'make clean'. Maybe a corrupt .ver file...

Good catch, Tom.

Cheers,
Remco

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/