v6.11-rc4 amdgpu regression from v6.10.0

[Andrew Worsley]

I did a git bisect on from v6.10 .. v6.11-rc4 and found the exact
change when the problem appeared (see below)

The problem is introduced by this

commit f9d8c5c7855d
Author: Jack Xiao <Jack.Xiao at amd.com>
Date:   Thu Apr 25 15:28:48 2024 +0800

    drm/amdgpu/gfx: enable mes to map legacy queue support

    Enable mes to map legacy queue support.

    v2: kiq_set_resources is required.

    Signed-off-by: Jack Xiao <Jack.Xiao at amd.com>
    Reviewed-by: Hawking Zhang <Hawking.Zhang at amd.com>
    Reviewed-by: Alex Deucher <alexander.deucher at amd.com>
    Signed-off-by: Alex Deucher <alexander.deucher at amd.com>

M    drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c


On Tue, 20 Aug 2024 at 00:13, Alex Deucher <alexdeucher at gmail.com> wrote:
>
> On Mon, Aug 19, 2024 at 9:55 AM Andrew Worsley <amworsley at gmail.com> wrote:
> >
> > The v6.11-rc4 linux hangs during amdgpu start up where as the v6.10.0
> > is fine. I had to take a photo of the screen (see attachment) from
> > which I generated
> > the following summary:
> >
> >     Booting linux v6.11-rc4 :
> > ...
> > amdgpu: Virtual CRAT table created for CPU
> > amdgpu: Topology: Add CPU node
> > initializing kernel modesetting (IP DISCOVERY 0x1002:0x15BF 0xF111:0x0005 0xC2).
> > register mmio base: 0x90500000
> > register mmio size: 524288
> > add ip block number 0 <soc21_common>
> > add ip block number 1 <gmc_v11_0>
> > add ip block number 2 <ih_v6_0>
> > add ip block number 3 <psp>
> > add ip block number 4 <smu>
> > add ip block number 5 <dm>
> > add ip block number 6 <gfx_v11_0>
> > add ip block number 7 <sdma_v6_0>
> > add ip block number 8 <vcn_v4_0>
> > add ip block number 9 <jpeg_v4_0>
> > add ip block number 10 <mes_v11_0>
> > amdgpu 0000:c1:00.0: amdgpu: Fetched VBIOS from VFCT
> > amdgpu: ATOM BIOS: 113-PHXGENERIC-001
> > amdgpu 0000:c1:00.0: Direct firmware load for
> > amdgpu/gc_11_0_1_mes_2.bin failed with error -2
> > amdgpu 0000:c1:00.0: amdgpu: try to fall back to amdgpu/gc_11_0_1_mes.bin

The full git bisect history is

git bisect start
# status: waiting for both good and bad commits
# good: [0c3836482481200ead7b416ca80c68a29cfdaabd] Linux 6.10
git bisect good 0c3836482481200ead7b416ca80c68a29cfdaabd
# status: waiting for bad commit, 1 good commit known
# bad: [47ac09b91befbb6a235ab620c32af719f8208399] Linux 6.11-rc4
git bisect bad 47ac09b91befbb6a235ab620c32af719f8208399
# bad: [de9c2c66ad8e787abec7c9d7eff4f8c3cdd28aed] Linux 6.11-rc2
git bisect bad de9c2c66ad8e787abec7c9d7eff4f8c3cdd28aed
# good: [280e36f0d5b997173d014c07484c03a7f7750668] nsfs: use cleanup guard
git bisect good 280e36f0d5b997173d014c07484c03a7f7750668
# bad: [a4f9285520584977127946a22eab2adfbc87d1bf] Merge tag
'clk-for-linus' of
git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux
git bisect bad a4f9285520584977127946a22eab2adfbc87d1bf
# bad: [365aa9f573995b46ca14a24165d85e31160e47b9] Merge tag
'amd-drm-next-6.11-2024-06-22' of
https://gitlab.freedesktop.org/agd5f/linux into drm-next
git bisect bad 365aa9f573995b46ca14a24165d85e31160e47b9
# bad: [1ddaaa244021aba8496536a6627b4ad2bc0f936a] Merge tag
'amd-drm-next-6.11-2024-06-07' of
https://gitlab.freedesktop.org/agd5f/linux into drm-next
git bisect bad 1ddaaa244021aba8496536a6627b4ad2bc0f936a
# bad: [b72fa761fc6be316bbf3ef9af4b7f145a208b1f3] drm/amdgpu: fix
documentation errors in sdma v7.0
git bisect bad b72fa761fc6be316bbf3ef9af4b7f145a208b1f3
# bad: [03f4b8c3ca7ad34d812d5760f0712e5c126c64c6] drm/amdgpu/mes12:
disable logging output
git bisect bad 03f4b8c3ca7ad34d812d5760f0712e5c126c64c6
# bad: [17e3bea65cdc453695b2fe4ff26d25d17f5339e9] drm/amd/pm: fix
warning using uninitialized value of max_vid_step
git bisect bad 17e3bea65cdc453695b2fe4ff26d25d17f5339e9
# good: [7e0357bef402875425de0296800c34c41842ba82] drm/amdgpu: remove
unused MCA driver codes
git bisect good 7e0357bef402875425de0296800c34c41842ba82
# good: [523ee6f4419bbeccbe43399d999f37381b5958c7] drm/amd/display:
For FPO + Vactive check that all pipes support VA
git bisect good 523ee6f4419bbeccbe43399d999f37381b5958c7
# good: [8fb20d9551368f0b1ef5b31fa0b4634f4be37157] drm/amdgpu: add
amdgpu MCA bank dispatch function support
git bisect good 8fb20d9551368f0b1ef5b31fa0b4634f4be37157
# good: [579f0c21baec9e7506b6bb3f60f0a9b6d07693b4] drm/amdgpu/pm:
Check the return value of smum_send_msg_to_smc
git bisect good 579f0c21baec9e7506b6bb3f60f0a9b6d07693b4
# good: [d53ce02352058639cacccc88cdbdbf48f1ee3b5a] drm/amdkfd: Evict
BO itself for contiguous allocation
git bisect good d53ce02352058639cacccc88cdbdbf48f1ee3b5a

Andrew