Hi Heinrich,

On Tue, 22 Dec 2020 at 01:40, Heinrich Schuchardt xypron.glpk@gmx.de wrote:

On 12/22/20 9:14 AM, Heinrich Schuchardt wrote:

...

On 12/22/20 6:00 AM, Bin Meng wrote:

...

Hi,

Please check the log: https://dev.azure.com/bmeng/GitHub/_build/results?buildId=302&view=logs&...

Re-run the pipeline with the same commit, the tests pass. https://dev.azure.com/bmeng/GitHub/_build/results?buildId=304&view=logs&...

There might be some instability on qemu-x86 with efi_selftest. Any ideas?

Regards, Bin

Hello Bin,

The failing routine is

  u_boot_console.restart_uboot()

after the UEFI tests were completed successfully.

This is the function:

 def restart_uboot(self):
     """Shut down and restart U-Boot."""
     self.cleanup_spawn()
     self.ensure_spawned()

In ensure_spawned() we get the exception:

 Exception: Bad pattern found on console: main_signon

ensure_spawned() complains because it sees the main_signon message a second time.

This is the console output:

 U-Boot 2021.01-rc4-23431-g6ef1b75087 (Dec 22 2020 - 03:07:46 +0000)

 CPU:   QEMU Virtual CPU version 2.5+
 DRAM:  128 MiB
 unable to find cpus device
 Warning: MP init failure

 U-Boot 2021.01-rc4-23431-g6ef1b75087 (Dec 22 2020 - 03:07:46 +0000)

So obviously U-Boot is crashing during the boot process after x86_mp_init().

The same error has shown up on Gitlab as reported by Simon. Cf.

https://gitlab.denx.de/u-boot/custodians/u-boot-dm/-/jobs/190746/raw

I suggest that you add

 #define DEBUG 1

at the top of common.h to close in on the location of the crash.

Best regards

Heinrich

Here is a log with DEBUG set: https://gitlab.denx.de/u-boot/custodians/u-boot-efi/-/jobs/192520/raw

This is the interesting part:

+u-boot-test-reset qemu-x86 na initcall: fff6ef31 U-Boot 2021.01-rc4-00037-ge528c1d3e3 (Dec 22 2020 - 08:16:58 +0000) initcall: fff150c7 U-Boot code: FFF00000 -> FFFA880C BSS: -> FFFB5404

...

VBE: Setting VESA mode 0x144 <<<<CRASH occurs here >>>> initcall: fff6ef31 U-Boot 2021.01-rc4-00037-ge528c1d3e3 (Dec 22 2020 - 08:16:58 +0000)

On my local system I see:

VBE: framebuffer: d0000000 VBE: Setting VESA mode 0x144 Final vesa mode 0x144 base = d0000000, copy_base = 0

So the crash occurs before pci_rom_probe() is finished.

This may be related to the MP changes I made recently. If I recall correctly, MP init isn't really needed on qemu, but I'm not sure.

Regards, Simon