Bug 194915

Summary: commit c2febafc6773 in 4.11-rc2 blocks some ARM64 platforms from properly booting
Product: Memory Management Reporter: Icenowy Zheng (icenowy)
Component: OtherAssignee: Andrew Morton (akpm)
Status: RESOLVED CODE_FIX    
Severity: blocking CC: regressions
Priority: P1    
Hardware: ARM   
OS: Linux   
Kernel Version: 4.11-rc2 Subsystem:
Regression: Yes Bisected commit-id:
Attachments: my kernel config
example hang log (this time a lot of info is dumped)

Description Icenowy Zheng 2017-03-18 06:26:11 UTC
I tried to run 4.11-rc2 on my Pine64+ (with arch/arm64/boot/dts/allwinner/sun50i-a64-pine64-plus.dtb)

It succeeded to enter systemd, however, continously fail at ~5s (according to the printk timer).

An error like "Unable to handle kernel paging request at virtual address 128001b8550005a1" will appear (the address is random), then after random time between 0s to 1s the system will hang (and printk will stop even if one sentence is not fully printed; sometimes it's able to dump a full stack trace, sometimes only the Unable to handle kernel paging error info is printed)

By bisecting the commit c2febafc6773 ("mm: convert generic code to 5-level paging") is problematic, and with it reverted the system is booted again.
Comment 1 Icenowy Zheng 2017-03-18 06:26:49 UTC
Created attachment 255325 [details]
my kernel config
Comment 2 Icenowy Zheng 2017-03-18 06:27:15 UTC
Created attachment 255327 [details]
example hang log (this time a lot of info is dumped)
Comment 3 Icenowy Zheng 2017-03-18 07:13:19 UTC
On qemu-system-aarch64 -M virt, kernel with this commit will hang when booting (during systemd), but no error message available.

Just replace it with a kernel with this commit revert, the boot process can finish and I can get "login:".
Comment 4 The Linux kernel's regression tracker (Thorsten Leemhuis) 2017-03-27 15:36:55 UTC
Does this problem still show up in rc4? There was at least one fix reg. c2febafc6773 after rc2 was relaesed.

JFYI: I added this report to the list of regressions for Linux 4.11.
I'll watch this place for further updates on this issue to document
progress in my weekly reports, hence to allow me to keep track please send a quick note here in case this issue gets discussed somewhere else.
Comment 5 Icenowy Zheng 2017-03-27 16:37:28 UTC
(In reply to Thorsten Leemhuis from comment #4)
> Does this problem still show up in rc4? There was at least one fix reg.
> c2febafc6773 after rc2 was relaesed.

Yes it's already fixed, thanks!

> 
> JFYI: I added this report to the list of regressions for Linux 4.11.
> I'll watch this place for further updates on this issue to document
> progress in my weekly reports, hence to allow me to keep track please send a
> quick note here in case this issue gets discussed somewhere else.