Looking at arch/i386/config.in
if [ "$CONFIG_SMP" = "y" -a "$CONFIG_X86_CMPXCHG" = "y" ]; then
define_bool CONFIG_HAVE_DEC_LOCK y
fi
endmenu
and arch/i386/lib/Makefile
.obj-$(CONFIG_HAVE_DEC_LOCK) += dec_and_lock.o
We need to have SMP set inorder to use dec_and_lock. The file fs/dcache.c
in function dput uses a function atomic_dec_and_lock function, which is
defined in dec_and_lock.c. I think the SMP dependency should go
If this patch is ok, please apply
--- config.in.org Thu Nov 8 18:43:49 2001
+++ config.in Thu Nov 8 18:44:04 2001
@@ -183,7 +183,7 @@
bool 'Multiquad NUMA system' CONFIG_MULTIQUAD
fi
-if [ "$CONFIG_SMP" = "y" -a "$CONFIG_X86_CMPXCHG" = "y" ]; then
+if [ "$CONFIG_X86_CMPXCHG" = "y" ]; then
define_bool CONFIG_HAVE_DEC_LOCK y
fi
On Thu, 08 Nov 2001 18:49:27 +0530,
"BALBIR SINGH" <[email protected]> wrote:
>Looking at arch/i386/config.in
>
>if [ "$CONFIG_SMP" = "y" -a "$CONFIG_X86_CMPXCHG" = "y" ]; then
> define_bool CONFIG_HAVE_DEC_LOCK y
>fi
>endmenu
>
>and arch/i386/lib/Makefile
>
>.obj-$(CONFIG_HAVE_DEC_LOCK) += dec_and_lock.o
>
>We need to have SMP set inorder to use dec_and_lock. The file fs/dcache.c
>in function dput uses a function atomic_dec_and_lock function, which is
>defined in dec_and_lock.c.
There is a #define for atomic_dec_and_lock when compiling for UP. I
suspect that you have been bitten by the broken makefiles, otherwise
everybody would be reporting problems with 2.4.14. Before you send any
more patches (and before your existing patches are used), please follow
the steps in http://www.tux.org/lkml/#s8-8.
If you still get problems after that, mail the error messages and your
.config instead of sending patches to "fix" bugs which nobody else is
seeing. You should include the errors you are getting, otherwise we
cannot tell if the patch makes sense or not.
>
>
>
>There is a #define for atomic_dec_and_lock when compiling for UP. I
>suspect that you have been bitten by the broken makefiles, otherwise
>everybody would be reporting problems with 2.4.14. Before you send any
>more patches (and before your existing patches are used), please follow
>the steps in http://www.tux.org/lkml/#s8-8.
>
>If you still get problems after that, mail the error messages and your
>.config instead of sending patches to "fix" bugs which nobody else is
>seeing. You should include the errors you are getting, otherwise we
>cannot tell if the patch makes sense or not.
>
Agreed and understood, that was my fault. I saw that dec_and_lock.c did not
compiled. So, I began to think that was the case.
Balbir