[Devel] 2.6.22-ovz002 on powerpc

[Pavel Emelyanov]

Christian Kaiser2 wrote:
> Hi all,
> 
> I tried my best today at your new development kernel based on a 2.6.22
> Linux Kernel. My first question is: why did you use the 2.6.22 Kernel as
> base for your work and not the latest 2.6.22.6 that already includes some
> bugfixes. The first error I got when I tried to compile on a powerpc64
> machine was:
> 
> In file included from include/asm/pgalloc.h:6,
>                  from arch/powerpc/mm/mem.c:36:
> include/asm/pgalloc-64.h: In function ‘do_pte_alloc’:
> include/asm/pgalloc-64.h:87: error: ‘PTE_CACHE_NUM’ undeclared (first use
> in this function)
> include/asm/pgalloc-64.h:87: error: (Each undeclared identifier is reported
> only once
> include/asm/pgalloc-64.h:87: error: for each function it appears in.)
> make[1]: *** [arch/powerpc/mm/mem.o] Error 1
> make: *** [arch/powerpc/mm] Error 2
> make: *** Waiting for unfinished jobs....
> 
> So I took the updated version of "include/asm/pgalloc-64.h" from 2.6.22.6
> and that error was solved.
>  Beside this I realized that you forgot to update the powerpc system tabe.
> So I added the fairsched specific system calls in
> "include/asm-powerpc/systbl.h" and "include/asm-powerpc/unistd.h".
> 
> But I am still getting another error:
> 
>   Building modules, stage 2.
>   MODPOST 143 modules
>   WRAP    arch/powerpc/boot/zImage.pseries
> ERROR: "nf_ct_icmp_timeout" [net/netfilter/nf_conntrack.ko] undefined!
> make[1]: *** [__modpost] Error 1
> make: *** [modules] Error 2
> make: *** Waiting for unfinished jobs....
> 
> These conntrack things are confusing me... so I added the
> "nf_ct_icmp_timeout" variable at the beginning of
> "net/netfilter/nf_conntrack_standalone.c". Without CONFIG_NF_CONNTRACK=m it
> fails building somewhere else.

We haven't fixed the netfilter compilation yet. But we're going to have
them in ovz004 kernel I think.

> Booting the Kernel on a QS20 blade ends up in this loop of exceptions:
> 
> Memory: 1020352k/1048576k available (4736k kernel code, 28224k reserved,
> 768k data, 590k bss, 384k in)
> Mount-cache hash table entries: 4096
> Page beancounter hash is 8192 entries.
> Processor 1 found.
> Processor 2 found.
> Processor 3 found.
> Brought up 4 CPUs
> ------------[ cut here ]------------
> Badness at kernel/sched.c:7495

Oh yes! This is a known thing as well. I believe this will be fixed in ovz004
kernel as well.

All of th above is unlikely to appear in ovz003 as we have another critical
BUG with VE stop, so ovz003 kernel will come with this fix along with the
2.6.22 stable update, tickless kernel support and SLUB allocator. The ppc
issues will be delayed up to ovz004.

> Call Trace:
> [c00000003ef03550] [c00000000000f3d4] .show_stack+0x68/0x1b0 (unreliable)
> [c00000003ef035f0] [c0000000001ad700] .report_bug+0x94/0xe8
> [c00000003ef03680] [c0000000000246f4] .program_check_exception+0x178/0x61c
> [c00000003ef03730] [c000000000004f0c] program_check_common+0x10c/0x180
> --- Exception: 700 at .measure_one+0xc4/0x16c
>     LR = .measure_one+0xac/0x16c
> [c00000003ef03ae0] [c00000000004bbac] .build_sched_domains+0x950/0xda0
> [c00000003ef03d70] [c00000000004c030] .arch_init_sched_domains+0x34/0x70
> [c00000003ef03e10] [c000000000457038] .sched_init_smp+0x30/0xcc
> [c00000003ef03ea0] [c000000000440b30] .kernel_init+0x14c/0x3c4
> [c00000003ef03f90] [c00000000002613c] .ppc_kernel_thread+0x4c/0x68
> ------------[ cut here ]------------
> Badness at kernel/sched.c:7495
> Call Trace:
> [c00000003ef03550] [c00000000000f3d4] .show_stack+0x68/0x1b0 (unreliable)
> [c00000003ef035f0] [c0000000001ad700] .report_bug+0x94/0xe8
> [c00000003ef03680] [c0000000000246f4] .program_check_exception+0x178/0x61c
> [c00000003ef03730] [c000000000004f0c] program_check_common+0x10c/0x180
> --- Exception: 700 at .measure_one+0xc4/0x16c
>     LR = .measure_one+0xac/0x16c
> [...]
> 
> Mit freundlichen Grüßen / Best Regards
> Christian Kaiser
> --
> IBM Deutschland Entwicklung GmbH
> Open Systems Firmware Development
> mail: ckaiser2 at de.ibm.com
> 
> IBM Deutschland Entwicklung GmbH
> Vorsitzender des Aufsichtsrats: Martin Jetter
> Geschäftsführung: Herbert Kircher
> Sitz der Gesellschaft: Böblingen
> Registergericht: Amtsgericht Stuttgart, HRB 243294
> 
> _______________________________________________
> Devel mailing list
> Devel at openvz.org
> https://openvz.org/mailman/listinfo/devel
>