Re: [PATCH] drm/amdgpu: refactor code to reuse system information

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




On 3/19/2024 7:43 PM, Lazar, Lijo wrote:

On 3/19/2024 7:27 PM, Khatri, Sunil wrote:
On 3/19/2024 7:19 PM, Lazar, Lijo wrote:
On 3/19/2024 6:02 PM, Sunil Khatri wrote:
Refactor the code so debugfs and devcoredump can reuse
the common information and avoid unnecessary copy of it.

created a new file which would be the right place to
hold functions which will be used between sysfs, debugfs
and devcoredump.

Cc: Christian König <christian.koenig@xxxxxxx>
Cc: Alex Deucher <alexander.deucher@xxxxxxx>
Signed-off-by: Sunil Khatri <sunil.khatri@xxxxxxx>
---
  drivers/gpu/drm/amd/amdgpu/Makefile         |   2 +-
  drivers/gpu/drm/amd/amdgpu/amdgpu.h         |   1 +
  drivers/gpu/drm/amd/amdgpu/amdgpu_devinfo.c | 151 ++++++++++++++++++++
  drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c     | 118 +--------------
  4 files changed, 157 insertions(+), 115 deletions(-)
  create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_devinfo.c

diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile
b/drivers/gpu/drm/amd/amdgpu/Makefile
index 4536c8ad0e11..05d34f4b18f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -80,7 +80,7 @@ amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o
amdgpu_kms.o \
      amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \
      amdgpu_fw_attestation.o amdgpu_securedisplay.o \
      amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \
-    amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o
+    amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o
amdgpu_devinfo.o
    amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o
  diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 9c62552bec34..0267870aa9b1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -1609,4 +1609,5 @@ extern const struct attribute_group
amdgpu_vram_mgr_attr_group;
  extern const struct attribute_group amdgpu_gtt_mgr_attr_group;
  extern const struct attribute_group amdgpu_flash_attr_group;
  +int amdgpu_device_info(struct amdgpu_device *adev, struct
drm_amdgpu_info_device *dev_info);
  #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_devinfo.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_devinfo.c
new file mode 100644
index 000000000000..d2c15a1dcb0d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_devinfo.c
@@ -0,0 +1,151 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright 2024 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person
obtaining a
+ * copy of this software and associated documentation files (the
"Software"),
+ * to deal in the Software without restriction, including without
limitation
+ * the rights to use, copy, modify, merge, publish, distribute,
sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom
the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO
EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM,
DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amd_pcie.h"
+
+#include <drm/amdgpu_drm.h>
+
+int amdgpu_device_info(struct amdgpu_device *adev, struct
drm_amdgpu_info_device *dev_info)
+{
+    int ret;
+    uint64_t vm_size;
+    uint32_t pcie_gen_mask;
+
+    if (dev_info == NULL)
+        return -EINVAL;
+
+    dev_info->device_id = adev->pdev->device;
+    dev_info->chip_rev = adev->rev_id;
+    dev_info->external_rev = adev->external_rev_id;
+    dev_info->pci_rev = adev->pdev->revision;
+    dev_info->family = adev->family;
+    dev_info->num_shader_engines = adev->gfx.config.max_shader_engines;
+    dev_info->num_shader_arrays_per_engine =
adev->gfx.config.max_sh_per_se;
+    /* return all clocks in KHz */
+    dev_info->gpu_counter_freq = amdgpu_asic_get_xclk(adev) * 10;
+    if (adev->pm.dpm_enabled) {
+        dev_info->max_engine_clock = amdgpu_dpm_get_sclk(adev,
false) * 10;
+        dev_info->max_memory_clock = amdgpu_dpm_get_mclk(adev,
false) * 10;
+        dev_info->min_engine_clock = amdgpu_dpm_get_sclk(adev, true)
* 10;
+        dev_info->min_memory_clock = amdgpu_dpm_get_mclk(adev, true)
* 10;
+    } else {
+        dev_info->max_engine_clock =
+            dev_info->min_engine_clock =
+                adev->clock.default_sclk * 10;
+        dev_info->max_memory_clock =
+            dev_info->min_memory_clock =
+                adev->clock.default_mclk * 10;
+        }
+    dev_info->enabled_rb_pipes_mask =
adev->gfx.config.backend_enable_mask;
+    dev_info->num_rb_pipes = adev->gfx.config.max_backends_per_se *
+        adev->gfx.config.max_shader_engines;
+    dev_info->num_hw_gfx_contexts = adev->gfx.config.max_hw_contexts;
+    dev_info->ids_flags = 0;
+    if (adev->flags & AMD_IS_APU)
+        dev_info->ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
+    if (adev->gfx.mcbp)
+        dev_info->ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION;
+    if (amdgpu_is_tmz(adev))
+        dev_info->ids_flags |= AMDGPU_IDS_FLAGS_TMZ;
+    if (adev->gfx.config.ta_cntl2_truncate_coord_mode)
+        dev_info->ids_flags |= AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD;
+
+    vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
+    vm_size -= AMDGPU_VA_RESERVED_TOP;
+
+    /* Older VCE FW versions are buggy and can handle only 40bits */
+    if (adev->vce.fw_version && adev->vce.fw_version <
AMDGPU_VCE_FW_53_45)
+        vm_size = min(vm_size, 1ULL << 40);
+
+    dev_info->virtual_address_offset = AMDGPU_VA_RESERVED_BOTTOM;
+    dev_info->virtual_address_max = min(vm_size,
AMDGPU_GMC_HOLE_START);
+
+    if (vm_size > AMDGPU_GMC_HOLE_START) {
+        dev_info->high_va_offset = AMDGPU_GMC_HOLE_END;
+        dev_info->high_va_max = AMDGPU_GMC_HOLE_END | vm_size;
+    }
+    dev_info->virtual_address_alignment = max_t(u32, PAGE_SIZE,
AMDGPU_GPU_PAGE_SIZE);
+    dev_info->pte_fragment_size = (1 <<
adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;
+    dev_info->gart_page_size = max_t(u32, PAGE_SIZE,
AMDGPU_GPU_PAGE_SIZE);
+    dev_info->cu_active_number = adev->gfx.cu_info.number;
+    dev_info->cu_ao_mask = adev->gfx.cu_info.ao_cu_mask;
+    dev_info->ce_ram_size = adev->gfx.ce_ram_size;
+    memcpy(&dev_info->cu_ao_bitmap[0],
&adev->gfx.cu_info.ao_cu_bitmap[0],
+           sizeof(adev->gfx.cu_info.ao_cu_bitmap));
+    memcpy(&dev_info->cu_bitmap[0], &adev->gfx.cu_info.bitmap[0],
+           sizeof(dev_info->cu_bitmap));
+    dev_info->vram_type = adev->gmc.vram_type;
+    dev_info->vram_bit_width = adev->gmc.vram_width;
+    dev_info->vce_harvest_config = adev->vce.harvest_config;
+    dev_info->gc_double_offchip_lds_buf =
+        adev->gfx.config.double_offchip_lds_buf;
+    dev_info->wave_front_size = adev->gfx.cu_info.wave_front_size;
+    dev_info->num_shader_visible_vgprs = adev->gfx.config.max_gprs;
+    dev_info->num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
+    dev_info->num_tcc_blocks =
adev->gfx.config.max_texture_channel_caches;
+    dev_info->gs_vgt_table_depth = adev->gfx.config.gs_vgt_table_depth;
+    dev_info->gs_prim_buffer_depth =
adev->gfx.config.gs_prim_buffer_depth;
+    dev_info->max_gs_waves_per_vgt = adev->gfx.config.max_gs_threads;
+
+    if (adev->family >= AMDGPU_FAMILY_NV)
+        dev_info->pa_sc_tile_steering_override =
+            adev->gfx.config.pa_sc_tile_steering_override;
+
+    dev_info->tcc_disabled_mask = adev->gfx.config.tcc_disabled_mask;
+
+    /* Combine the chip gen mask with the platform (CPU/mobo) mask. */
+    pcie_gen_mask = adev->pm.pcie_gen_mask & (adev->pm.pcie_gen_mask
16);
+    dev_info->pcie_gen = fls(pcie_gen_mask);
+    dev_info->pcie_num_lanes =
+        adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X32 ?
32 :
+        adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 ?
16 :
+        adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 ?
12 :
+        adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 ? 8 :
+        adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 ? 4 :
+        adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 ? 2
: 1;
+
+    dev_info->tcp_cache_size = adev->gfx.config.gc_tcp_l1_size;
+    dev_info->num_sqc_per_wgp = adev->gfx.config.gc_num_sqc_per_wgp;
+    dev_info->sqc_data_cache_size =
adev->gfx.config.gc_l1_data_cache_size_per_sqc;
+    dev_info->sqc_inst_cache_size =
adev->gfx.config.gc_l1_instruction_cache_size_per_sqc;
+    dev_info->gl1c_cache_size =
adev->gfx.config.gc_gl1c_size_per_instance *
+                    adev->gfx.config.gc_gl1c_per_sa;
+    dev_info->gl2c_cache_size = adev->gfx.config.gc_gl2c_per_gpu;
+    dev_info->mall_size = adev->gmc.mall_size;
+
+
+    if (adev->gfx.funcs->get_gfx_shadow_info) {
+        struct amdgpu_gfx_shadow_info shadow_info;
+
+        ret = amdgpu_gfx_get_gfx_shadow_info(adev, &shadow_info);
This failure is ignored in the present logic and rest of the information
is filled. Probably, need to continue the same way.
Since this is the information that we are sharing with user and on any
failure we just decide not to populate those fields. By default all the
members are initialized to zero and they

are updated when we have valid values. To me seems the right logic
instead of failing the complete call, we provide the information what we
have.

Actually, this is not what is really done. If you fail here, then this
happens down the call. It doesn't copy, instead frees without copying to
userspace.

+        ret = amdgpu_device_info(adev, dev_info);
+        if (!ret) {
Good point. Actually my intent was only to check the valid memory as amdgpu_device_info will be used in other places too but yes amdgpu_gfx_get_gfx_shadow_infocould also send failure.
I propose that to not send the error for amdgpu_gfx_get_gfx_shadow_info. I will update the patch with that information.

Thanks,
Lijo

Regards
Sunil

Thanks,
Lijo

+        if (!ret) {
+            dev_info->shadow_size = shadow_info.shadow_size;
+            dev_info->shadow_alignment = shadow_info.shadow_alignment;
+            dev_info->csa_size = shadow_info.csa_size;
+            dev_info->csa_alignment = shadow_info.csa_alignment;
+        }
+    }
+    return ret;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index a66d47865e3b..24f775c68a51 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -850,125 +850,15 @@ int amdgpu_info_ioctl(struct drm_device *dev,
void *data, struct drm_file *filp)
      }
      case AMDGPU_INFO_DEV_INFO: {
          struct drm_amdgpu_info_device *dev_info;
-        uint64_t vm_size;
-        uint32_t pcie_gen_mask;
            dev_info = kzalloc(sizeof(*dev_info), GFP_KERNEL);
          if (!dev_info)
              return -ENOMEM;
  -        dev_info->device_id = adev->pdev->device;
-        dev_info->chip_rev = adev->rev_id;
-        dev_info->external_rev = adev->external_rev_id;
-        dev_info->pci_rev = adev->pdev->revision;
-        dev_info->family = adev->family;
-        dev_info->num_shader_engines =
adev->gfx.config.max_shader_engines;
-        dev_info->num_shader_arrays_per_engine =
adev->gfx.config.max_sh_per_se;
-        /* return all clocks in KHz */
-        dev_info->gpu_counter_freq = amdgpu_asic_get_xclk(adev) * 10;
-        if (adev->pm.dpm_enabled) {
-            dev_info->max_engine_clock = amdgpu_dpm_get_sclk(adev,
false) * 10;
-            dev_info->max_memory_clock = amdgpu_dpm_get_mclk(adev,
false) * 10;
-            dev_info->min_engine_clock = amdgpu_dpm_get_sclk(adev,
true) * 10;
-            dev_info->min_memory_clock = amdgpu_dpm_get_mclk(adev,
true) * 10;
-        } else {
-            dev_info->max_engine_clock =
-                dev_info->min_engine_clock =
-                    adev->clock.default_sclk * 10;
-            dev_info->max_memory_clock =
-                dev_info->min_memory_clock =
-                    adev->clock.default_mclk * 10;
-        }
-        dev_info->enabled_rb_pipes_mask =
adev->gfx.config.backend_enable_mask;
-        dev_info->num_rb_pipes = adev->gfx.config.max_backends_per_se *
-            adev->gfx.config.max_shader_engines;
-        dev_info->num_hw_gfx_contexts =
adev->gfx.config.max_hw_contexts;
-        dev_info->ids_flags = 0;
-        if (adev->flags & AMD_IS_APU)
-            dev_info->ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
-        if (adev->gfx.mcbp)
-            dev_info->ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION;
-        if (amdgpu_is_tmz(adev))
-            dev_info->ids_flags |= AMDGPU_IDS_FLAGS_TMZ;
-        if (adev->gfx.config.ta_cntl2_truncate_coord_mode)
-            dev_info->ids_flags |=
AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD;
-
-        vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
-        vm_size -= AMDGPU_VA_RESERVED_TOP;
-
-        /* Older VCE FW versions are buggy and can handle only
40bits */
-        if (adev->vce.fw_version &&
-            adev->vce.fw_version < AMDGPU_VCE_FW_53_45)
-            vm_size = min(vm_size, 1ULL << 40);
-
-        dev_info->virtual_address_offset = AMDGPU_VA_RESERVED_BOTTOM;
-        dev_info->virtual_address_max =
-            min(vm_size, AMDGPU_GMC_HOLE_START);
-
-        if (vm_size > AMDGPU_GMC_HOLE_START) {
-            dev_info->high_va_offset = AMDGPU_GMC_HOLE_END;
-            dev_info->high_va_max = AMDGPU_GMC_HOLE_END | vm_size;
-        }
-        dev_info->virtual_address_alignment = max_t(u32, PAGE_SIZE,
AMDGPU_GPU_PAGE_SIZE);
-        dev_info->pte_fragment_size = (1 <<
adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;
-        dev_info->gart_page_size = max_t(u32, PAGE_SIZE,
AMDGPU_GPU_PAGE_SIZE);
-        dev_info->cu_active_number = adev->gfx.cu_info.number;
-        dev_info->cu_ao_mask = adev->gfx.cu_info.ao_cu_mask;
-        dev_info->ce_ram_size = adev->gfx.ce_ram_size;
-        memcpy(&dev_info->cu_ao_bitmap[0],
&adev->gfx.cu_info.ao_cu_bitmap[0],
-               sizeof(adev->gfx.cu_info.ao_cu_bitmap));
-        memcpy(&dev_info->cu_bitmap[0], &adev->gfx.cu_info.bitmap[0],
-               sizeof(dev_info->cu_bitmap));
-        dev_info->vram_type = adev->gmc.vram_type;
-        dev_info->vram_bit_width = adev->gmc.vram_width;
-        dev_info->vce_harvest_config = adev->vce.harvest_config;
-        dev_info->gc_double_offchip_lds_buf =
-            adev->gfx.config.double_offchip_lds_buf;
-        dev_info->wave_front_size = adev->gfx.cu_info.wave_front_size;
-        dev_info->num_shader_visible_vgprs = adev->gfx.config.max_gprs;
-        dev_info->num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
-        dev_info->num_tcc_blocks =
adev->gfx.config.max_texture_channel_caches;
-        dev_info->gs_vgt_table_depth =
adev->gfx.config.gs_vgt_table_depth;
-        dev_info->gs_prim_buffer_depth =
adev->gfx.config.gs_prim_buffer_depth;
-        dev_info->max_gs_waves_per_vgt =
adev->gfx.config.max_gs_threads;
-
-        if (adev->family >= AMDGPU_FAMILY_NV)
-            dev_info->pa_sc_tile_steering_override =
-                adev->gfx.config.pa_sc_tile_steering_override;
-
-        dev_info->tcc_disabled_mask =
adev->gfx.config.tcc_disabled_mask;
-
-        /* Combine the chip gen mask with the platform (CPU/mobo)
mask. */
-        pcie_gen_mask = adev->pm.pcie_gen_mask &
(adev->pm.pcie_gen_mask >> 16);
-        dev_info->pcie_gen = fls(pcie_gen_mask);
-        dev_info->pcie_num_lanes =
-            adev->pm.pcie_mlw_mask &
CAIL_PCIE_LINK_WIDTH_SUPPORT_X32 ? 32 :
-            adev->pm.pcie_mlw_mask &
CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 ? 16 :
-            adev->pm.pcie_mlw_mask &
CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 ? 12 :
-            adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X8
? 8 :
-            adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X4
? 4 :
-            adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X2
? 2 : 1;
-
-        dev_info->tcp_cache_size = adev->gfx.config.gc_tcp_l1_size;
-        dev_info->num_sqc_per_wgp =
adev->gfx.config.gc_num_sqc_per_wgp;
-        dev_info->sqc_data_cache_size =
adev->gfx.config.gc_l1_data_cache_size_per_sqc;
-        dev_info->sqc_inst_cache_size =
adev->gfx.config.gc_l1_instruction_cache_size_per_sqc;
-        dev_info->gl1c_cache_size =
adev->gfx.config.gc_gl1c_size_per_instance *
-                        adev->gfx.config.gc_gl1c_per_sa;
-        dev_info->gl2c_cache_size = adev->gfx.config.gc_gl2c_per_gpu;
-        dev_info->mall_size = adev->gmc.mall_size;
-
-
-        if (adev->gfx.funcs->get_gfx_shadow_info) {
-            struct amdgpu_gfx_shadow_info shadow_info;
-
-            ret = amdgpu_gfx_get_gfx_shadow_info(adev, &shadow_info);
-            if (!ret) {
-                dev_info->shadow_size = shadow_info.shadow_size;
-                dev_info->shadow_alignment =
shadow_info.shadow_alignment;
-                dev_info->csa_size = shadow_info.csa_size;
-                dev_info->csa_alignment = shadow_info.csa_alignment;
-            }
+        ret = amdgpu_device_info(adev, dev_info);
+        if (!ret) {
+            kfree(dev_info);
+            return ret;
          }
            ret = copy_to_user(out, dev_info,

[Index of Archives]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux