[mipsel-manda-01.debian.org] lemote-3a-itx-a1101 kernel boot issues

[Hector Oron <hector.oron@gmail.com>]

Hello,

  As some of you might know, there was some issue preventing lemote
machine (Loongson-3) boot from Debian kernel. After some
investigation, I have been able to boot a (patched) jessie kernel on
that machine, however the status quo is not that great!

  Disabling the following patches from 3.16.7-ckt2-1 Debian linux kernel:
    ### features/mips/MIPS-Add-NUMA-support-for-Loongson-3.patch
    ### features/mips/MIPS-increase-MAX-PHYSMEM-BITS-on-Loongson-3-only.patch
    ### features/mips/MIPS-Add-Loongson-3B-support.patch

  made the kernel boot again. However I get a trace at boot time (but
machine boots fine to login):

[    0.088000] ------------[ cut here ]------------
[    0.088000] WARNING: CPU: 0 PID: 0 at
/build/numa/linux-3.16.7-ckt2/kernel/smp.c:391
smp_call_function_many+0x130/0x3b0()
[    0.088000] Modules linked in:
[    0.088000] CPU: 0 PID: 0 Comm: swapper/0 Not tainted
3.16.0-4-loongson-3 #1 Debian 3.16.7-ckt2-1
[    0.088000] Stack : 0000000000000004 ffffffff80294bf4
ffffffff80ad4598 ffffffff80ad0000
[    0.088000]    0000000000000000 0000000000000000 0000000000000000
0000000000000000
[    0.088000]    ffffffff808dfd48 ffffffff809ab9c7 ffffffff80ace068
ffffffff809abfb0
[    0.088000]    0000000000000000 0000000000000000 0000000000000000
0000000000000001
[    0.088000]    ffffffff809a26c0 ffffffff802554c8 0000000000000000
ffffffff8098fc48
[    0.088000]    ffffffff809a26c0 ffffffff80296734 ffffffff809abae0
ffffffff808dfd48
[    0.088000]    0000000000000000 0000000000000000 0000000000000000
0000000000000000
[    0.088000]    0000000000000000 ffffffff8098fb90 0000000000000000
ffffffff807e46c0
[    0.088000]    0000000000000000 0000000000000000 0000000000000000
db4b7fde7785ab00
[    0.088000]    0000000000000000 ffffffff8020bc50 ffffffff802b87a8
ffffffff807e46c0
[    0.088000]    ...
[    0.088000] Call Trace:
[    0.088000] [<ffffffff8020bc50>] show_stack+0x78/0x90
[    0.088000] [<ffffffff807e46c0>] dump_stack+0x88/0xe0
[    0.088000] [<ffffffff802323f4>] warn_slowpath_common+0x94/0xc8
[    0.088000] [<ffffffff802b87a8>] smp_call_function_many+0x130/0x3b0
[    0.088000] [<ffffffff802b8a74>] smp_call_function+0x4c/0xa0
[    0.088000] [<ffffffff8022867c>] r4k_flush_icache_range+0x44/0x150
[    0.088000] [<ffffffff8020fc50>] ftrace_modify_code+0x30/0x50
[    0.088000] [<ffffffff80a41300>] ftrace_dyn_arch_init+0x90/0xa8
[    0.088000] [<ffffffff80a4a20c>] ftrace_init+0x28/0x1b0
[    0.088000] [<ffffffff80a3ccf0>] start_kernel+0x600/0x624
[    0.088000]
[    0.088000] ---[ end trace 955ad6e4cf2a041d ]---

  Someone on #mipslinux IRC channel hinted the following:

 < amalon> zumbi_: i asked around, james cowgill said it might be because
                "PMON in those machines hangs when it tries to load a kernel
                with a large bss section"
 < amalon> zumbi_: apparently it appears after
                c46173183657bbdbe0d54a981c28807581648422
 < amalon> zumbi_: it's not in v3.16.7, but is in debian's 3.16 kernel
                apparently. Could that be the problem?
 < amalon> (that commit was introduced to mainline in v3.17-rc1)

  That guy seemed to be on track!

  Could you porters take this issue upstream and try to improve things
on future kernels?

  TIA

Regards,
-- 
 Héctor Orón  -.. . -... .. .- -.   -.. . ...- . .-.. --- .--. . .-.