[U-Boot] Odroid n1- booting hangs if we add memtest=4 to kernel command line.

Wed Sep 18 10:15:04 UTC 2019

Hi Kevrer,

On Wed, 18 Sep 2019 at 12:02, Kever Yang <kever.yang at rock-chips.com> wrote:
>
> Hi Anand,
>
>
> On 2019/9/16 下午7:03, Anand Moon wrote:
>
> Hi Jagan / Kever,
>
> I have with me Odroid - N1 per-released development board,
> It's almost same form factor to "Firefly-RK3399 Board" with
> rk3399-sdram-ddr3-1600.dtsi ram configuration.
>
> I have configure Odroid - N1 board to boot using TPL / SPL configuration
> using latest kernel and latest u-boot, but I am facing some issue with not
> able to upgrade my distro packages, It's literally hangs on heavy I/O
> operations.
>
> So I would like to check if the memory is working correctly, but below
> command failed.
>
> # sudo memtester 1024 5
> memtester version 4.3.0 (64-bit)
> Copyright (C) 2001-2012 Charles Cazabon.
> Licensed under the GNU General Public License version 2 (only).
>
> pagesize is 4096
> pagesizemask is 0xfffffffffffff000
> want 1024MB (1073741824 bytes)
> got  1024MB (1073741824 bytes), trying mlock ...
>
> Then I would like to enable *memtest=4* via bootargs ie kernel command
> line and it failed to boot up.
>
>
> I'm not able to understand well about your issue, does it:
> - boots fine with odroid-N1 but hang at memtester ?
> - hang at "Starting kernel..." with kernel command like update
> Since it boots OK, the DRAM should works OK, you can replace the TPL with rockchip binary ddr[0]
> to double check if they have the same output.
> I have no idea about the bootargs "memtest=4", but I suppose this should not affect the kernel
> boot process, maybe you have do something wrong and not update the bootargs as what you
> expect, you can print it in U-Boot to check the actually bootargs.
>
> Thanks,
> - Kever
> [0] https://github.com/rockchip-linux/rkbin/blob/master/bin/rk33/rk3399_ddr_800MHz_v1.23.bin
>

Ok thanks for your input. I will give this a try.

Here is the scenario I faced  I am using Archlinux using latest u-boot
and 5.13 kernel.
First issue that during upgrade on distro packages it suddenly stop
with not kernel panic.
So I could not understand the root caused.

1 > Then I tried some fio test with large file mapping it also failed
with no kernel panic.
2 > Then I tired memtester it also failed with no kernel panic.
3 > Finally I tried the memtest kernel memory testing feature at boot
time, but it hangs.

So I could only conclude that it's not able to map large file, because
memory is not tuned correctly.

Well I have tried to rebuild the u-boot using below image

[0] https://github.com/rockchip-linux/rkbin/blob/master/bin/rk33/rk3399_ddr_800MHz_v1.23.bin

Here is the boot log of above and kernel panic using fio command below.

[1] https://pastebin.com/Qe0Lpqss

fio --randrepeat=1 --ioengine=libaio --direct=1 --gtod_reduce=1
--name=test --filename=test --bs=4k --iodepth=64 --size=4G
--readwrite=randrw --rwmixread=75

Again memtester is not able to test memory, it's not able to map large file map.

[root at alarm ~]# memtester 4GB
memtester version 4.3.0 (64-bit)
Copyright (C) 2001-2012 Charles Cazabon.
Licensed under the GNU General Public License version 2 (only).

pagesize is 4096
pagesizemask is 0xfffffffffffff000
want 4096MB (4294967296 bytes)
got  3866MB (4054790144 bytes), trying mlock ...[  110.647775] SError
Interrupt on CPU5, code 0xbf000000 -- SError
[  110.647776] CPU: 5 PID: 372 Comm: memtester Not tainted 5.3.0-xn1ml1 #2
[  110.647777] Hardware name: Hardkernel ODROID-N1 (DT)
[  110.647778] pstate: 00000005 (nzcv daif -PAN -UAO)
[  110.647779] pc : clear_page+0x10/0x24
[  110.647780] lr : __cpu_clear_user_page+0xc/0x18
[  110.647781] sp : ffff000011bcbaf0
[  110.647781] x29: ffff000011bcbaf0 x28: ffff8000eda6ccc0
[  110.647783] x27: ffff8000e5620000 x26: ffff8000e562c3c0
[  110.647785] x25: ffff7e0003971dc0 x24: 0000000000000000
[  110.647786] x23: ffff7e0000270000 x22: 0000000000001000
[  110.647788] x21: 0000ffff66000000 x20: 0000000000000000
[  110.647790] x19: ffff8000e5620000 x18: 0000000000000000
[  110.647791] x17: 0000000000000010 x16: 0000000000000010
[  110.647793] x15: 0000000000000000 x14: 000000000000009e
[  110.647795] x13: ffff7e0000270000 x12: ffff7e0003ba8500
[  110.647796] x11: 0000000000000000 x10: ffff000011bcb9c0
[  110.647798] x9 : 0000000000000001 x8 : ffff8000efafc230
[  110.647799] x7 : 0000000000001665 x6 : 0000000000000000
[  110.647801] x5 : 0ffff00000000000 x4 : 0000000000000001
[  110.647803] x3 : 0000000000007fc0 x2 : 0000000000000004
[  110.647804] x1 : 0000000000000040 x0 : ffff800009dff4c0
[  110.647806] Kernel panic - not syncing: Asynchronous SError Interrupt
[  110.647807] CPU: 5 PID: 372 Comm: memtester Not tainted 5.3.0-xn1ml1 #2
[  110.647808] Hardware name: Hardkernel ODROID-N1 (DT)
[  110.647809] Call trace:
[  110.647809]  dump_backtrace+0x0/0x140
[  110.647810]  show_stack+0x14/0x20
[  110.647811]  dump_stack+0x98/0xbc
[  110.647811]  panic+0x130/0x2fc
[  110.647812]  __stack_chk_fail+0x0/0x18
[  110.647813]  arm64_serror_panic+0x74/0x80
[  110.647814]  is_valid_bugaddr+0x0/0x8
[  110.647814]  el1_error+0x84/0xf8
[  110.647815]  clear_page+0x10/0x24
[  110.647815]  clear_subpage+0x4c/0x88
[  110.647816]  clear_huge_page+0x74/0x220
[  110.647817]  do_huge_pmd_anonymous_page+0x2c8/0x6f0
[  110.647818]  __handle_mm_fault+0xb90/0x1090
[  110.647818]  handle_mm_fault+0x100/0x1c0
[  110.647819]  __get_user_pages+0x248/0x410
[  110.647820]  populate_vma_page_range+0x64/0x70
[  110.647821]  __mm_populate+0xb8/0x170
[  110.647821]  do_mlock+0xb4/0x220
[  110.647822]  __arm64_sys_mlock+0x14/0x20
[  110.647823]  el0_svc_common.constprop.0+0x68/0x160
[  110.647823]  el0_svc_handler+0x28/0x78
[  110.647824]  el0_svc+0x8/0xc
[  110.647950] SMP: stopping secondary CPUs
[  110.647950] Kernel Offset: disabled
[  110.647951] CPU features: 0x0002,2000600c
[  110.647952] Memory Limit: none

So what possible I could be doing wrong.
Is this kernel bug or memory tuning failed I do not understand.

Best Regards
-Anand