ACK/Cmt: [PATCH 0/4][SRU][Noble] Unexpected system reboot at loading GUI session on some AMD platforms

[Thibault Ferrante]

Acked-by: Thibault Ferrante <thibault.ferrante at canonical.com>


On 05/06/2025 20:35, You-Sheng Yang wrote:
> BugLink: https://bugs.launchpad.net/bugs/2112462
> 
> [ Impact ]
> 
> AMD Strix Halo devices experience unexpected reboots while starting GUI
> session every boot.
> 
> This is caused by:
> * 689275140cb8e drm/amdgpu/hdp7.0: do a posting read when flushing HDP
> * abe1cbaec6cfe drm/amdgpu/hdp6.0: do a posting read when flushing HDP
> * f756dbac1ce1d drm/amdgpu/hdp5.2: do a posting read when flushing HDP
> * cf424020e040b drm/amdgpu/hdp5.0: do a posting read when flushing HDP
> * c9b8dcabb52af drm/amdgpu/hdp4.0: do a posting read when flushing HDP
> 
> And fixed by:
> * 5a11a27677311 drm/amdgpu/hdp7: use memcfg register to post the write for HDP flush
> * ca28e80abe421 drm/amdgpu/hdp6: use memcfg register to post the write for HDP flush
> * dbc988c689333 drm/amdgpu/hdp5.2: use memcfg register to post the write for HDP flush
> * 0e33e0f339b91 drm/amdgpu/hdp5: use memcfg register to post the write for HDP flush
> * f690e3974755a drm/amdgpu/hdp4: use memcfg register to post the write for HDP flush
> 
> [ Test Plan ]
> 
> 1. Install -proposed kernel
> 2. Reboot to the installed kernel and check if the device may finish boot
>     process to GUI session.
> 
> [ Where problems could occur ]
> 
> No. This fix a problem when reading back the whole mapped
> KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL register by reading another register that
> also triggers the flush.
As described https://documentation.ubuntu.com/sru/en/latest/reference/bug-template/ : "This must never be "None" or "Low", or entirely an argument as to why your upload is low risk."
You should just describe what part of the kernel could be impacted.> 
> [ Other Info ]
> 
> The causing commits were first committed to v6.13-rc1 , and have also been
> backported to v6.12.5, v6.6.69, v6.1.120. The fixes are available in
> v6.15-rc6, also backported to v6.14.7, v6.12.29, v6.6.91, v6.1.139,
> therefore this affects following Ubuntu kernels:
> * linux/plucky: all (currently 6.14.0-22.22)
> * linux-oem-6.14/noble: all (currently 6.14.0-1005.5)
> * linux/oracular: 6.11.0-20.20 to latest (currently 6.11.0-28.28)
> * linux-oem-6.11/noble: 6.11.0-1017.17 to latest (currently 6.11.0-1024.24)
> * linux/noble: 6.8.0-58.60 to latest (currently 6.8.0-62.65)
> * linux-oem-6.8/noble: 6.8.0-1026.26 to latest (currently 6.8.0-1029.29)
> 
> Nominate for linux/noble, linux-oem-6.8/noble, linux/oracular,
> linux-oem-6.11/noble, linux/plucky, and linux-oem-6.14/noble.
> 
> Alex Deucher (4):
>    drm/amdgpu/hdp4: use memcfg register to post the write for HDP flush
>    drm/amdgpu/hdp5: use memcfg register to post the write for HDP flush
>    drm/amdgpu/hdp5.2: use memcfg register to post the write for HDP flush
>    drm/amdgpu/hdp6: use memcfg register to post the write for HDP flush
> 
>   drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c |  7 ++++++-
>   drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c |  7 ++++++-
>   drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c | 12 +++++++++++-
>   drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c |  7 ++++++-
>   4 files changed, 29 insertions(+), 4 deletions(-)
>