A problem with the system freeze during the initialization of the radeon driver after a hot reboot.

[Дмитрий Терехин]

Hellow collegues!

I am one of the developers of the Alt OS distribution
https://en.altlinux.org
https://en.wikipedia.org/wiki/ALT_Linux
for the computer BFK3.1
https://www.baikalelectronics.ru/products/bfk31/
https://medium.com/@malafeev/first-independent-tests-of-baikal-t1-processor-and-bfk-3-1-evaluation-board-bc7c1db12046
The computer is based on the processor BE-T1000 (Baikal-T1) https://en.wikipedia.org/wiki/Baikal_CPU

I use AMD SAPPHIRE Radeon R5 230 / HD 6450 (Caicos) graphics card with BFK3.1 and
Linux kernel from Baikal Electronics https://github.com/baikalelectronics/Linux-kernel.4.4.xx
Video driver drivers/gpu/drm/radeon

There is a problem with the system freeze during the initialization of the radeon driver after a hot reboot.
After a reboot with a complete power off, no freeze occurs.

A hang occurs during the initialization of the radeon driver, when manipulating the graphics card registers through
MMIO mechanism.

I assume that the hang may be due to some kind of improper state of the video card
at the time of driver initialization.

After a hot reboot, I observed the following variants for the softreset GPU:

1. In this case, no hang occurs.
radeon 0000:01:00.0: GPU softreset: 0x00000008
radeon 0000:01:00.0:   GRBM_STATUS               = 0xA0003828
radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0x00000007
radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0x00000007
radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x00010100
radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00020180
radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x80038042
radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57
radeon 0000:01:00.0: GRBM_SOFT_RESET=0x00004001
radeon 0000:01:00.0: SRBM_SOFT_RESET=0x00000100
radeon 0000:01:00.0:   GRBM_STATUS               = 0x00003828
radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0x00000007
radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0x00000007
radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x00000000
radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00000000
radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x00000000
radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57

2. After this, it hangs, usually when manipulating the graphics card registers
in the evergreen_init_golden_registers function
radeon 0000:01:00.0: GPU softreset: 0x00000004
radeon 0000:01:00.0:   GRBM_STATUS               = 0x00003828
radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0x00000007
radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0x00000007
radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x00000000
radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00000000
radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x00000000
radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83146
radeon 0000:01:00.0: SRBM_SOFT_RESET=0x00100000
radeon 0000:01:00.0:   GRBM_STATUS               = 0x00003828
radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0x00000007
radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0x00000007
radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x00000000
radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00000000
radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x00000000
radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57

3. The softreset GPU does not execute at all, no messages are issued about this.
After this, it hangs, usually when manipulating the graphics card registers
in the evergreen_init_golden_registers function

I think that to workaround the problem it is need to perform a complete restart of the video card.
I do not know how to do that.

BFK3.1 does not allow disconnecting power on the PCI slot from Linux.
The following commands
echo "1" > /sys/bus/pci/devices/0000:01:00.0/remove
echo "1" > /sys/bus/pci/rescan
do not restart of the video card.

I tried unconditionally calling evergreen_gpu_pci_config_reset from evergreen_asic_reset.
Got a permanent hang when initializing the driver.

I ask for your help.
Tell me, please, how to properly restart the video card when initializing the driver.

Dmitry Terekhin
jqt4 at basealt.ru
jqt4 at altlinux.org