How about put pci_enable_atomic_ops_to_root ahead of amdgpu_device_ip_early_init, while move pci_atomic_requested from kfd device to kgd device ? In such way, we can avoid duplicate atomic request from both amdgpu and amdkfd.
Regards,
Hawking
-----Original Message-----
From: amd-gfx <amd-gfx-bounces@xxxxxxxxxxxxxxxxxxxxx> On Behalf Of Xiao, Jack
Sent: 2019年5月27日 18:19
To: amd-gfx@xxxxxxxxxxxxxxxxxxxxx
Cc: Xiao, Jack <Jack.Xiao@xxxxxxx>
Subject: [PATCH] drm/amdgpu: enable PCIE atomics ops support
GPU atomics operation depends on PCIE atomics support.
Always enable PCIE atomics ops support in case that it hasn't been enabled.
Signed-off-by: Jack Xiao <Jack.Xiao@xxxxxxx>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index bdd1fe73..a2c6064 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2562,6 +2562,13 @@ int amdgpu_device_init(struct amdgpu_device *adev,
if (adev->rio_mem == NULL)
DRM_INFO("PCI I/O BAR is not found.\n");
+ /* enable PCIE atomic ops */
+ r = pci_enable_atomic_ops_to_root(adev->pdev,
+ PCI_EXP_DEVCAP2_ATOMIC_COMP32 |
+ PCI_EXP_DEVCAP2_ATOMIC_COMP64);
+ if (r)
+ DRM_INFO("PCIE atomic ops is not supported\n");
+
amdgpu_device_get_pcie_info(adev);
/* early init functions */
--
1.9.1
_______________________________________________
amd-gfx mailing list
amd-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
