From 85150626ea0423fd0adb5ac9b5ab4bbaff9aa30b Mon Sep 17 00:00:00 2001
From: Victor Lu <victorchengchi.lu@amd.com>
Date: Tue, 3 Oct 2023 16:15:52 -0400
Subject: drm/amdgpu: Use correct KIQ MEC engine for gfx9.4.3 (v5)
amdgpu_kiq_wreg/rreg is hardcoded to use MEC engine 0.
Add an xcc_id parameter to amdgpu_kiq_wreg/rreg, define W/RREG32_XCC
and amdgpu_device_xcc_wreg/rreg to use the new xcc_id parameter.
Using amdgpu_sriov_runtime to determine whether to access via kiq or
RLC is sufficient for now.
v5: add condition in amdgpu_device_xcc_w/rreg, remove trace func call
v4: avoid using amdgpu_sriov_w/rreg
v3: use W/RREG32_XCC to handle non-kiq case
v2: define amdgpu_device_xcc_wreg/rreg instead of changing parameters
of amdgpu_device_wreg/rreg
Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 89 +++++++++++++++++++++++++++++-
1 file changed, 87 insertions(+), 2 deletions(-)
(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_device.c')
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index fd8cd8e2d3f2..d93240e8cb6a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -73,6 +73,7 @@
#include "amdgpu_pmu.h"
#include "amdgpu_fru_eeprom.h"
#include "amdgpu_reset.h"
+#include "amdgpu_virt.h"
#include <linux/suspend.h>
#include <drm/task_barrier.h>
@@ -472,7 +473,7 @@ uint32_t amdgpu_device_rreg(struct amdgpu_device *adev,
if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
amdgpu_sriov_runtime(adev) &&
down_read_trylock(&adev->reset_domain->sem)) {
- ret = amdgpu_kiq_rreg(adev, reg);
+ ret = amdgpu_kiq_rreg(adev, reg, 0);
up_read(&adev->reset_domain->sem);
} else {
ret = readl(((void __iomem *)adev->rmmio) + (reg * 4));
@@ -509,6 +510,49 @@ uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset)
BUG();
}
+
+/**
+ * amdgpu_device_xcc_rreg - read a memory mapped IO or indirect register with specific XCC
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ * @acc_flags: access flags which require special behavior
+ * @xcc_id: xcc accelerated compute core id
+ *
+ * Returns the 32 bit value from the offset specified.
+ */
+uint32_t amdgpu_device_xcc_rreg(struct amdgpu_device *adev,
+ uint32_t reg, uint32_t acc_flags,
+ uint32_t xcc_id)
+{
+ uint32_t ret, rlcg_flag;
+
+ if (amdgpu_device_skip_hw_access(adev))
+ return 0;
+
+ if ((reg * 4) < adev->rmmio_size) {
+ if (amdgpu_sriov_vf(adev) &&
+ !amdgpu_sriov_runtime(adev) &&
+ adev->gfx.rlc.rlcg_reg_access_supported &&
+ amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags,
+ GC_HWIP, false,
+ &rlcg_flag)) {
+ ret = amdgpu_virt_rlcg_reg_rw(adev, reg, 0, rlcg_flag, xcc_id);
+ } else if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
+ amdgpu_sriov_runtime(adev) &&
+ down_read_trylock(&adev->reset_domain->sem)) {
+ ret = amdgpu_kiq_rreg(adev, reg, xcc_id);
+ up_read(&adev->reset_domain->sem);
+ } else {
+ ret = readl(((void __iomem *)adev->rmmio) + (reg * 4));
+ }
+ } else {
+ ret = adev->pcie_rreg(adev, reg * 4);
+ }
+
+ return ret;
+}
+
/*
* MMIO register write with bytes helper functions
* @offset:bytes offset from MMIO start
@@ -556,7 +600,7 @@ void amdgpu_device_wreg(struct amdgpu_device *adev,
if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
amdgpu_sriov_runtime(adev) &&
down_read_trylock(&adev->reset_domain->sem)) {
- amdgpu_kiq_wreg(adev, reg, v);
+ amdgpu_kiq_wreg(adev, reg, v, 0);
up_read(&adev->reset_domain->sem);
} else {
writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
@@ -597,6 +641,47 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev,
}
}
+/**
+ * amdgpu_device_xcc_wreg - write to a memory mapped IO or indirect register with specific XCC
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: dword aligned register offset
+ * @v: 32 bit value to write to the register
+ * @acc_flags: access flags which require special behavior
+ * @xcc_id: xcc accelerated compute core id
+ *
+ * Writes the value specified to the offset specified.
+ */
+void amdgpu_device_xcc_wreg(struct amdgpu_device *adev,
+ uint32_t reg, uint32_t v,
+ uint32_t acc_flags, uint32_t xcc_id)
+{
+ uint32_t rlcg_flag;
+
+ if (amdgpu_device_skip_hw_access(adev))
+ return;
+
+ if ((reg * 4) < adev->rmmio_size) {
+ if (amdgpu_sriov_vf(adev) &&
+ !amdgpu_sriov_runtime(adev) &&
+ adev->gfx.rlc.rlcg_reg_access_supported &&
+ amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags,
+ GC_HWIP, true,
+ &rlcg_flag)) {
+ amdgpu_virt_rlcg_reg_rw(adev, reg, v, rlcg_flag, xcc_id);
+ } else if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
+ amdgpu_sriov_runtime(adev) &&
+ down_read_trylock(&adev->reset_domain->sem)) {
+ amdgpu_kiq_wreg(adev, reg, v, xcc_id);
+ up_read(&adev->reset_domain->sem);
+ } else {
+ writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
+ }
+ } else {
+ adev->pcie_wreg(adev, reg * 4, v);
+ }
+}
+
/**
* amdgpu_device_indirect_rreg - read an indirect register
*
--
cgit v1.2.3-70-g09d2