How about: #if defined(CONFIG_ARCH_S390) || defined (CONFIG_ARCH_S390X) .. #endif
? (note the ||, not |)
- Alex
Post, Mark K wrote:
Well, it turns out this won't work as written (even if I fix the syntax error in the #if statement.) :( There's so much I don't understand about kernels it's not even funny. Sigh.
Maybe someone from Boeblingen can help out here? This is a start: #if (defined(CONFIG_ARCH_S390) | (CONFIG_ARCH_S390X)) EXPORT_SYMBOL(show_trace); #endif
but, it also requires that the show_trace routine defintion be included somehow. Unfortunately, the routine is declared in ./arch/s390/kernel/traps.c, and not in a header file, which is how most of the other symbols are covered. Changing the mod to: #if (defined(CONFIG_ARCH_S390) | (CONFIG_ARCH_S390X)) extern void show_trace; EXPORT_SYMBOL(show_trace); #endif
works, but is it the correct thing to do? So many questions, so few @#$% answers.
Mark Post
-----Original Message----- From: Post, Mark K Sent: Monday, February 02, 2004 1:47 PM To: 'Linux on 390 Port' Subject: RE: Kernel 2.4.24 qeth.o - unresolved symdol show_trace
You can try inserting this into linux/kernel/ksyms.c: EXPORT_SYMBOL(show_trace);
Or, if you want to be more rigorous about it: #if defined (CONFIG_ARCH_S390 || CONFIG_ARCH_S390X) EXPORT_SYMBOL(show_trace); #endif
It might make your problem go away.
Mark Post
-----Original Message----- From: Linux on 390 Port [mailto:[EMAIL PROTECTED] Behalf Of Kasza Karoly Sent: Wednesday, January 28, 2004 2:40 AM To: [EMAIL PROTECTED] Subject: Kernel 2.4.24 qeth.o - unresolved symdol show_trace
Hi all.
Im trying to compile kernel 2.4.24 for z800 in 31bit more. Currently i have applied ibm june 2003 experimental patches (i know they are for 2.4.23, but - as mentioned on this list - they patch 2.4.24 also perfectly), and a patch for devfs from some ibm folk (i dont remember his name, i found the patch somewhere in google in a mailing list (maybe on this?)). Now the kernel and the modules compile, but qeth.o - which is the most important module, we can say that i think - says that he got an unresolved symbol: show_trace. If i am correct, this issue has been fixed in somewhere 2.5.73, but i didn't find that is has been fixed back in 2.4. Now does anybody knows anything about this unresolved symdol thingie, or will i have to backport the bugfix from 2.5.73 to 2.4.24? Thank you for any help.
Kasza Karoly
P.S.: it is kind of weird that a factory kernel cannot be compiled on the 390 architecture, and one have to collect patches and fixes from all over the net, dont you think?