On 1/26/2024 03:51, Jonathan Gray wrote:
The latest releases of 6.1.y, 6.6.y and 6.7.y introduce a duplicate
commit of 'drm/amd: Enable PCIe PME from D3'.
Good catch. I think this happened because the same commit ended up in
6.7 final as well as 6.8-rc1 with different hashes. This tends to
happen when we have fixes right at the end of the cycle.
In this case it's fortunately harmless, but yes I think one of them
should be dropped from stable trees.
For example on the 6.6.y branch:
commit 847e6947afd3c46623172d2eabcfc2481ee8668e
Author: Mario Limonciello <mario.limonciello@xxxxxxx>
AuthorDate: Fri Nov 24 09:56:32 2023 -0600
Commit: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
CommitDate: Thu Jan 25 15:35:45 2024 -0800
drm/amd: Enable PCIe PME from D3
commit bd1f6a31e7762ebc99b97f3eda5e5ea3708fa792 upstream.
When dGPU is put into BOCO it may be in D3cold but still able send
PME on display hotplug event. For this to work it must be enabled
as wake source from D3.
When runpm is enabled use pci_wake_from_d3() to mark wakeup as
enabled by default.
Cc: stable@xxxxxxxxxxxxxxx # 6.1+
Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxx>
Acked-by: Alex Deucher <alexander.deucher@xxxxxxx>
Signed-off-by: Alex Deucher <alexander.deucher@xxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 2c35036e4ba2..635b58553583 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -2197,6 +2197,8 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
pci_wake_from_d3(pdev, TRUE);
+ pci_wake_from_d3(pdev, TRUE);
+
/*
* For runpm implemented via BACO, PMFW will handle the
* timing for BACO in and out:
commit 49227bea27ebcd260f0c94a3055b14bbd8605c5e
Author: Mario Limonciello <mario.limonciello@xxxxxxx>
AuthorDate: Fri Nov 24 09:56:32 2023 -0600
Commit: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
CommitDate: Fri Dec 8 08:52:17 2023 +0100
drm/amd: Enable PCIe PME from D3
commit 6967741d26c87300a51b5e50d4acd104bc1a9759 upstream.
When dGPU is put into BOCO it may be in D3cold but still able send
PME on display hotplug event. For this to work it must be enabled
as wake source from D3.
When runpm is enabled use pci_wake_from_d3() to mark wakeup as
enabled by default.
Cc: stable@xxxxxxxxxxxxxxx # 6.1+
Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxx>
Acked-by: Alex Deucher <alexander.deucher@xxxxxxx>
Signed-off-by: Alex Deucher <alexander.deucher@xxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 81edf66dbea8..2c35036e4ba2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -2195,6 +2195,8 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
pm_runtime_mark_last_busy(ddev->dev);
pm_runtime_put_autosuspend(ddev->dev);
+ pci_wake_from_d3(pdev, TRUE);
+
/*
* For runpm implemented via BACO, PMFW will handle the
* timing for BACO in and out:
