Hi Krzysztof, all,
FYI bisecting showed that reset happens on 86588296acbf ("fdt: Properly handle
"no-map" field in the memory region"). Obviously some memory mapping is wrong,
on angler the same commit also caused reset, some time ago I fixed it in
0e5ded926f2a ("arm64: dts: qcom: msm8994-angler: Disable cont_splash_mem").
Therefore I'll try to find problematic mapping in bullhead dts.
Bullhead also contains oops in most of the commits (no matter if phone
resets during initramfs or keeps running):
[ 2.001953] Call trace:
[ 2.007244] clk_branch_wait+0x140/0x160
[ 2.009415] clk_branch2_disable+0x2c/0x40
[ 2.013581] clk_core_disable+0x60/0xc0
[ 2.017486] clk_disable+0x34/0x50
[ 2.021219] dwc3_qcom_probe+0xbec/0xf6c
[ 2.024691] platform_probe+0x68/0xc0
[ 2.028772] really_probe+0xe8/0x4a0
[ 2.032330] driver_probe_device+0x58/0xc0
[ 2.035976] device_driver_attach+0xc0/0xcc
[ 2.039883] __driver_attach+0x84/0x114
[ 2.043962] bus_for_each_dev+0x70/0xd0
[ 2.047782] driver_attach+0x24/0x30
[ 2.051600] bus_add_driver+0x104/0x1ec
[ 2.055421] driver_register+0x78/0x130
[ 2.058980] __platform_driver_register+0x28/0x34
[ 2.062801] dwc3_qcom_driver_init+0x1c/0x28
[ 2.067661] do_one_initcall+0x50/0x1b0
[ 2.072000] kernel_init_freeable+0x1d4/0x23c
[ 2.075562] kernel_init+0x14/0x118
[ 2.080073] ret_from_fork+0x10/0x34
[ 2.083379] ---[ end trace fdaf735b91bb478f ]---
If I remember correctly angler didn't have this oops.
Kind regards,
Petr
