Hi, ralf and all,
I found that the latest upstream kernel has some problems on
Loongson-3, I want to know whether other MIPS has similar problems.
1, When configure with SMP but no NUMA, it complaims "schedule while atomic":
[ 0.000000] allocated 2097152 bytes of page_cgroup
[ 0.000000] please try 'cgroup_disable=memory' option if you don't
want memory cgroups
[ 0.023437] Calibrating delay loop... 717.28 BogoMIPS (lpj=1400832)
[ 0.054687] pid_max: default: 32768 minimum: 301
[ 0.058593] Mount-cache hash table entries: 1024
[ 0.062500] Initializing cgroup subsys memory
[ 0.066406] ------------[ cut here ]------------
[ 0.070312] WARNING: at lib/idr.c:423 idr_preload+0x5c/0x174()
[ 0.074218] Modules linked in:
[ 0.078125] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.10.0-rc6+ #1271
[ 0.082031] Stack : 0000000000000800 ffffffff80428ec4
ffffffff8095fcb8 ffffffff8120d348
ffffffff81210000 ffffffff81210000 ffffffff808c7608 ffffffff809771f7
00000000000001a7 ffffffff808e30c0 00000000000000d0 0000000000000000
0000000000000000 0000000000000000 0000000000000000 ffffffff80759bc8
ffffffff8095fcb8 ffffffff8022b8e4 0000000000000000 ffffffff8022f17c
ffffffff80977660 ffffffff808c7608 0000000000000000 0000000000000000
ffffffff80977600 0000000000000000 0000000000000000 0000000000000000
0000000000000000 ffffffff8095fc10 0000000000000000 ffffffff8022b9dc
0000000000000000 0000000000000000 00000000000001a7 0000000000000009
ffffffff8041e0b4 ffffffff8020977c ffffffff808e30c0 ffffffff8022b9dc
...
[ 0.152343] Call Trace:
[ 0.152343] [<ffffffff8020977c>] show_stack+0x48/0x70
[ 0.156250] [<ffffffff8022b9dc>] warn_slowpath_common+0x78/0xa4
[ 0.160156] [<ffffffff8041e0b4>] idr_preload+0x5c/0x174
[ 0.164062] [<ffffffff8028632c>] get_new_cssid+0x50/0xec
[ 0.167968] [<ffffffff80288788>] cgroup_init_idr+0x30/0x6c
[ 0.171875] [<ffffffff811c7954>] cgroup_init+0x90/0x1dc
[ 0.175781] [<ffffffff811bca40>] start_kernel+0x4f4/0x554
[ 0.179687]
[ 0.179687] ---[ end trace 554a547e6319cb64 ]---
[ 0.183593] Initializing cgroup subsys blkio
[ 0.187500] Checking for the daddi bug... no.
[ 0.191406] BUG: scheduling while atomic: swapper/0/0/0xffc60201
[ 0.195312] Modules linked in:
[ 0.195312] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G W
3.10.0-rc6+ #1271
[ 0.199218] Stack : 0000000000000800 ffffffff80428ec4
ffffffff8095fcc8 ffffffff8120d348
ffffffff81210000 ffffffff81210000 ffffffff808c7608 ffffffff809771f7
0000000000000001 ffffffff811c0000 0000000000000001 ffffffff808cc820
ffffffff811c0000 ffffffff811fa420 98000000059c6420 ffffffff80759bc8
ffffffff8095fcc8 ffffffff8022b85c 0000000000000000 ffffffff8022f17c
ffffffff80977660 ffffffff808c7608 0000000000000000 0000000000000000
ffffffff80977600 0000000000000000 0000000000000000 0000000000000000
0000000000000000 ffffffff8095fc20 0000000000000000 ffffffff80256010
0000000000000000 0000000000000000 000000001400cce3 ffffffff811fa420
ffffffff80972fe8 ffffffff8020977c ffffffff811c0000 ffffffff80256010
...
[ 0.269531] Call Trace:
[ 0.269531] [<ffffffff8020977c>] show_stack+0x48/0x70
[ 0.273437] [<ffffffff80256010>] __schedule_bug+0x44/0x64
[ 0.277343] [<ffffffff8075b7d0>] __schedule+0xb8/0x6dc
[ 0.281250] [<ffffffff8075c248>] schedule_preempt_disabled+0x18/0x30
[ 0.285156] [<ffffffff80749e98>] rest_init+0x7c/0x8c
[ 0.289062] [<ffffffff811bca84>] start_kernel+0x538/0x554
2, when both SMP and NUMA configured, it complaims "early out of memory":
[ 0.000000] CPU revision is: 00006305 (ICT Loongson-3A)
[ 0.000000] FPU revision is: 00770501
[ 0.000000] Checking for the multiply/shift bug... no.
[ 0.000000] Checking for the daddiu bug... no.
[ 0.000000] Determined physical RAM map:
[ 0.000000] memory: 000000000fe00000 @ 0000000000200000 (usable)
[ 0.000000] memory: 0000000070000000 @ 0000000110000000 (usable)
[ 0.000000] Initrd not found or empty - disabling initrd
[ 0.000000] software IO TLB [mem 0x03a60000-0x07a60000] (64MB)
mapped at [9800000003a60000-9800000007a5ffff]
[ 0.000000] Zone ranges:
[ 0.000000] DMA32 [mem 0x01000000-0xffffffff]
[ 0.000000] Normal [mem 0x100000000-0x17fffffff]
[ 0.000000] Movable zone start for each node
[ 0.000000] Early memory node ranges
[ 0.000000] node 0: [mem 0x01000000-0x0fffffff]
[ 0.000000] node 0: [mem 0x110000000-0x17fffffff]
[ 0.000000] Detected 3 available secondary CPU(s)
[ 0.000000] Primary instruction cache 64kB, VIPT, 4-way, linesize 32 bytes.
[ 0.000000] Primary data cache 64kB, 4-way, VIPT, no aliases,
linesize 32 bytes
[ 0.000000] Unified secondary cache 4096kB 4-way, linesize 32 bytes.
[ 0.000000] PERCPU: Embedded 3 pages/cpu @9800000007adc000 s11968
r8192 d28992 u49152
[ 0.000000] Built 1 zonelists in Zone order, mobility grouping on.
Total pages: 129603
[ 0.000000] Policy zone: Normal
[ 0.000000] Kernel command line: console=tty console=ttyS0,115200
root=/dev/sda5 PMON_VER=A1101-3.1.8 EC_VER=undefined
machtype=lemote-3a-itx-a1101 autoplug=off
[ 0.000000] PID hash table entries: 4096 (order: 1, 32768 bytes)
[ 0.000000] total ram pages initialed 0
[ 0.000000] total ram pages are 123261
[ 0.000000] Kernel panic - not syncing: Oh boy, that early out of memory?
3, If I revert the commit ad04c2e9 (MIPS: Use inline function to
access current thread pointer), everything is OK.
If other MIPS have no problem, could anyone give me some suggestions?
Kernel code of Loongson-3 is here:
http://dev.lemote.com/cgit/linux-official.git/
Thanks,
Huacai Chen
