Backport of the upstream stable fix e773798e14ac ("drm/amdgpu: Fix
init ordering in amdgpu_vram_mgr_init()"). It registers the dmem
region after the VRAM manager lists and buddy allocator are set up, so
a failed registration leaves a manager that can be torn down safely.
It is a clean cherry-pick, the changed lines are the same as upstream,
and it only moves the registration down, nothing else changes.Not run: we have no node with an AMD GPU. Build-tested only, amdgpu_vram_mgr.o compiles with the 211.52.1.22.19 config. As far as I can tell from reading the code, the crash described upstream cannot be triggered in our kernel. amdgpu_ttm_fini() is only reached from the GMC block sw_fini, and amdgpu_device_ip_fini() skips a block whose sw_init failed, so a failed registration just fails the probe. https://virtuozzo.atlassian.net/browse/VSTOR-153702 Thomas Hellström (1): ms/drm/amdgpu: Fix init ordering in amdgpu_vram_mgr_init() drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) -- 2.43.0 _______________________________________________ Devel mailing list [email protected] https://lists.openvz.org/mailman/listinfo/devel
