[PATCH v6 10/16] drm/amdgpu: Guard against write accesses after device removal

Christian König ckoenig.leichtzumerken at gmail.com
Wed May 12 14:06:17 UTC 2021


Am 12.05.21 um 16:01 schrieb Andrey Grodzovsky:
> Ping - need a confirmation it's ok to keep this as a single patch given
> my explanation bellow.

It was just an suggestion. Key point is the approach sounds sane to me, 
but I can't say much about the psp code for example.

So maximum I can give you is an Acked-by for that.

Christian.

>
> Andrey
>
> On 2021-05-11 1:52 p.m., Andrey Grodzovsky wrote:
>>
>>
>> On 2021-05-11 2:50 a.m., Christian König wrote:
>>> Am 10.05.21 um 18:36 schrieb Andrey Grodzovsky:
>>>> This should prevent writing to memory or IO ranges possibly
>>>> already allocated for other uses after our device is removed.
>>>>
>>>> v5:
>>>> Protect more places wher memcopy_to/form_io takes place
>>>> Protect IB submissions
>>>>
>>>> v6: Switch to !drm_dev_enter instead of scoping entire code
>>>> with brackets.
>>>>
>>>> Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky at amd.com>
>>>> ---
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c    | 11 ++-
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c       |  9 +++
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c        | 17 +++--
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c       | 63 +++++++++++------
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h       |  2 +
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c      | 70 
>>>> +++++++++++++++++++
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h      | 49 ++-----------
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c       | 31 +++++---
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c       | 11 ++-
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c       | 22 ++++--
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c        |  7 +-
>>>>   drivers/gpu/drm/amd/amdgpu/psp_v11_0.c        | 44 ++++++------
>>>>   drivers/gpu/drm/amd/amdgpu/psp_v12_0.c        |  8 +--
>>>>   drivers/gpu/drm/amd/amdgpu/psp_v3_1.c         |  8 +--
>>>>   drivers/gpu/drm/amd/amdgpu/vce_v4_0.c         | 26 ++++---
>>>>   drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c         | 22 +++---
>>>>   .../drm/amd/pm/powerplay/smumgr/smu7_smumgr.c |  2 +
>>>>   17 files changed, 257 insertions(+), 145 deletions(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>>> index a0bff4713672..94c415176cdc 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>>> @@ -71,6 +71,8 @@
>>>>   #include <drm/task_barrier.h>
>>>>   #include <linux/pm_runtime.h>
>>>> +#include <drm/drm_drv.h>
>>>> +
>>>>   MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
>>>>   MODULE_FIRMWARE("amdgpu/vega12_gpu_info.bin");
>>>>   MODULE_FIRMWARE("amdgpu/raven_gpu_info.bin");
>>>> @@ -281,7 +283,10 @@ void amdgpu_device_vram_access(struct 
>>>> amdgpu_device *adev, loff_t pos,
>>>>       unsigned long flags;
>>>>       uint32_t hi = ~0;
>>>>       uint64_t last;
>>>> +    int idx;
>>>> +     if (!drm_dev_enter(&adev->ddev, &idx))
>>>> +         return;
>>>>   #ifdef CONFIG_64BIT
>>>>       last = min(pos + size, adev->gmc.visible_vram_size);
>>>> @@ -299,8 +304,10 @@ void amdgpu_device_vram_access(struct 
>>>> amdgpu_device *adev, loff_t pos,
>>>>               memcpy_fromio(buf, addr, count);
>>>>           }
>>>> -        if (count == size)
>>>> +        if (count == size) {
>>>> +            drm_dev_exit(idx);
>>>>               return;
>>>> +        }
>>>
>>> Maybe use a goto instead, but really just a nit pick.
>>>
>>>
>>>
>>>>           pos += count;
>>>>           buf += count / 4;
>>>> @@ -323,6 +330,8 @@ void amdgpu_device_vram_access(struct 
>>>> amdgpu_device *adev, loff_t pos,
>>>>               *buf++ = RREG32_NO_KIQ(mmMM_DATA);
>>>>       }
>>>>       spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
>>>> +
>>>> +    drm_dev_exit(idx);
>>>>   }
>>>>   /*
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
>>>> index 4d32233cde92..04ba5eef1e88 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
>>>> @@ -31,6 +31,8 @@
>>>>   #include "amdgpu_ras.h"
>>>>   #include "amdgpu_xgmi.h"
>>>> +#include <drm/drm_drv.h>
>>>> +
>>>>   /**
>>>>    * amdgpu_gmc_pdb0_alloc - allocate vram for pdb0
>>>>    *
>>>> @@ -151,6 +153,10 @@ int amdgpu_gmc_set_pte_pde(struct 
>>>> amdgpu_device *adev, void *cpu_pt_addr,
>>>>   {
>>>>       void __iomem *ptr = (void *)cpu_pt_addr;
>>>>       uint64_t value;
>>>> +    int idx;
>>>> +
>>>> +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>> +        return 0;
>>>>       /*
>>>>        * The following is for PTE only. GART does not have PDEs.
>>>> @@ -158,6 +164,9 @@ int amdgpu_gmc_set_pte_pde(struct amdgpu_device 
>>>> *adev, void *cpu_pt_addr,
>>>>       value = addr & 0x0000FFFFFFFFF000ULL;
>>>>       value |= flags;
>>>>       writeq(value, ptr + (gpu_page_idx * 8));
>>>> +
>>>> +    drm_dev_exit(idx);
>>>> +
>>>>       return 0;
>>>>   }
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
>>>> index 148a3b481b12..62fcbd446c71 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
>>>> @@ -30,6 +30,7 @@
>>>>   #include <linux/slab.h>
>>>>   #include <drm/amdgpu_drm.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "atom.h"
>>>> @@ -137,7 +138,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring 
>>>> *ring, unsigned num_ibs,
>>>>       bool secure;
>>>>       unsigned i;
>>>> -    int r = 0;
>>>> +    int idx, r = 0;
>>>>       bool need_pipe_sync = false;
>>>>       if (num_ibs == 0)
>>>> @@ -169,13 +170,16 @@ int amdgpu_ib_schedule(struct amdgpu_ring 
>>>> *ring, unsigned num_ibs,
>>>>           return -EINVAL;
>>>>       }
>>>> +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>> +        return -ENODEV;
>>>> +
>>>>       alloc_size = ring->funcs->emit_frame_size + num_ibs *
>>>>           ring->funcs->emit_ib_size;
>>>>       r = amdgpu_ring_alloc(ring, alloc_size);
>>>>       if (r) {
>>>>           dev_err(adev->dev, "scheduling IB failed (%d).\n", r);
>>>> -        return r;
>>>> +        goto exit;
>>>>       }
>>>>       need_ctx_switch = ring->current_ctx != fence_ctx;
>>>> @@ -205,7 +209,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring 
>>>> *ring, unsigned num_ibs,
>>>>           r = amdgpu_vm_flush(ring, job, need_pipe_sync);
>>>>           if (r) {
>>>>               amdgpu_ring_undo(ring);
>>>> -            return r;
>>>> +            goto exit;
>>>>           }
>>>>       }
>>>> @@ -286,7 +290,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring 
>>>> *ring, unsigned num_ibs,
>>>>           if (job && job->vmid)
>>>>               amdgpu_vmid_reset(adev, ring->funcs->vmhub, job->vmid);
>>>>           amdgpu_ring_undo(ring);
>>>> -        return r;
>>>> +        goto exit;
>>>>       }
>>>>       if (ring->funcs->insert_end)
>>>> @@ -304,7 +308,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring 
>>>> *ring, unsigned num_ibs,
>>>>           ring->funcs->emit_wave_limit(ring, false);
>>>>       amdgpu_ring_commit(ring);
>>>> -    return 0;
>>>> +
>>>> +exit:
>>>> +    drm_dev_exit(idx);
>>>> +    return r;
>>>>   }
>>>>   /**
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
>>>> index 9e769cf6095b..bb6afee61666 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
>>>> @@ -25,6 +25,7 @@
>>>>   #include <linux/firmware.h>
>>>>   #include <linux/dma-mapping.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "amdgpu_psp.h"
>>>> @@ -39,6 +40,8 @@
>>>>   #include "amdgpu_ras.h"
>>>>   #include "amdgpu_securedisplay.h"
>>>> +#include <drm/drm_drv.h>
>>>> +
>>>>   static int psp_sysfs_init(struct amdgpu_device *adev);
>>>>   static void psp_sysfs_fini(struct amdgpu_device *adev);
>>>> @@ -253,7 +256,7 @@ psp_cmd_submit_buf(struct psp_context *psp,
>>>>              struct psp_gfx_cmd_resp *cmd, uint64_t fence_mc_addr)
>>>>   {
>>>>       int ret;
>>>> -    int index;
>>>> +    int index, idx;
>>>>       int timeout = 20000;
>>>>       bool ras_intr = false;
>>>>       bool skip_unsupport = false;
>>>> @@ -261,6 +264,9 @@ psp_cmd_submit_buf(struct psp_context *psp,
>>>>       if (psp->adev->in_pci_err_recovery)
>>>>           return 0;
>>>> +    if (!drm_dev_enter(&psp->adev->ddev, &idx))
>>>> +        return 0;
>>>> +
>>>>       mutex_lock(&psp->mutex);
>>>>       memset(psp->cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE);
>>>> @@ -271,8 +277,7 @@ psp_cmd_submit_buf(struct psp_context *psp,
>>>>       ret = psp_ring_cmd_submit(psp, psp->cmd_buf_mc_addr, 
>>>> fence_mc_addr, index);
>>>>       if (ret) {
>>>>           atomic_dec(&psp->fence_value);
>>>> -        mutex_unlock(&psp->mutex);
>>>> -        return ret;
>>>> +        goto exit;
>>>>       }
>>>>       amdgpu_asic_invalidate_hdp(psp->adev, NULL);
>>>> @@ -312,8 +317,8 @@ psp_cmd_submit_buf(struct psp_context *psp,
>>>>                psp->cmd_buf_mem->cmd_id,
>>>>                psp->cmd_buf_mem->resp.status);
>>>>           if (!timeout) {
>>>> -            mutex_unlock(&psp->mutex);
>>>> -            return -EINVAL;
>>>> +            ret = -EINVAL;
>>>> +            goto exit;
>>>>           }
>>>>       }
>>>> @@ -321,8 +326,10 @@ psp_cmd_submit_buf(struct psp_context *psp,
>>>>           ucode->tmr_mc_addr_lo = psp->cmd_buf_mem->resp.fw_addr_lo;
>>>>           ucode->tmr_mc_addr_hi = psp->cmd_buf_mem->resp.fw_addr_hi;
>>>>       }
>>>> -    mutex_unlock(&psp->mutex);
>>>> +exit:
>>>> +    mutex_unlock(&psp->mutex);
>>>> +    drm_dev_exit(idx);
>>>>       return ret;
>>>>   }
>>>> @@ -359,8 +366,7 @@ static int psp_load_toc(struct psp_context *psp,
>>>>       if (!cmd)
>>>>           return -ENOMEM;
>>>>       /* Copy toc to psp firmware private buffer */
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -    memcpy(psp->fw_pri_buf, psp->toc_start_addr, psp->toc_bin_size);
>>>> +    psp_copy_fw(psp, psp->toc_start_addr, psp->toc_bin_size);
>>>>       psp_prep_load_toc_cmd_buf(cmd, psp->fw_pri_mc_addr, 
>>>> psp->toc_bin_size);
>>>> @@ -625,8 +631,7 @@ static int psp_asd_load(struct psp_context *psp)
>>>>       if (!cmd)
>>>>           return -ENOMEM;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -    memcpy(psp->fw_pri_buf, psp->asd_start_addr, 
>>>> psp->asd_ucode_size);
>>>> +    psp_copy_fw(psp, psp->asd_start_addr, psp->asd_ucode_size);
>>>>       psp_prep_asd_load_cmd_buf(cmd, psp->fw_pri_mc_addr,
>>>>                     psp->asd_ucode_size);
>>>> @@ -781,8 +786,7 @@ static int psp_xgmi_load(struct psp_context *psp)
>>>>       if (!cmd)
>>>>           return -ENOMEM;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -    memcpy(psp->fw_pri_buf, psp->ta_xgmi_start_addr, 
>>>> psp->ta_xgmi_ucode_size);
>>>> +    psp_copy_fw(psp, psp->ta_xgmi_start_addr, 
>>>> psp->ta_xgmi_ucode_size);
>>>>       psp_prep_ta_load_cmd_buf(cmd,
>>>>                    psp->fw_pri_mc_addr,
>>>> @@ -1038,8 +1042,7 @@ static int psp_ras_load(struct psp_context *psp)
>>>>       if (!cmd)
>>>>           return -ENOMEM;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -    memcpy(psp->fw_pri_buf, psp->ta_ras_start_addr, 
>>>> psp->ta_ras_ucode_size);
>>>> +    psp_copy_fw(psp, psp->ta_ras_start_addr, psp->ta_ras_ucode_size);
>>>>       psp_prep_ta_load_cmd_buf(cmd,
>>>>                    psp->fw_pri_mc_addr,
>>>> @@ -1275,8 +1278,7 @@ static int psp_hdcp_load(struct psp_context 
>>>> *psp)
>>>>       if (!cmd)
>>>>           return -ENOMEM;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -    memcpy(psp->fw_pri_buf, psp->ta_hdcp_start_addr,
>>>> +    psp_copy_fw(psp, psp->ta_hdcp_start_addr,
>>>>              psp->ta_hdcp_ucode_size);
>>>>       psp_prep_ta_load_cmd_buf(cmd,
>>>> @@ -1427,8 +1429,7 @@ static int psp_dtm_load(struct psp_context *psp)
>>>>       if (!cmd)
>>>>           return -ENOMEM;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -    memcpy(psp->fw_pri_buf, psp->ta_dtm_start_addr, 
>>>> psp->ta_dtm_ucode_size);
>>>> +    psp_copy_fw(psp, psp->ta_dtm_start_addr, psp->ta_dtm_ucode_size);
>>>>       psp_prep_ta_load_cmd_buf(cmd,
>>>>                    psp->fw_pri_mc_addr,
>>>> @@ -1573,8 +1574,7 @@ static int psp_rap_load(struct psp_context *psp)
>>>>       if (!cmd)
>>>>           return -ENOMEM;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -    memcpy(psp->fw_pri_buf, psp->ta_rap_start_addr, 
>>>> psp->ta_rap_ucode_size);
>>>> +    psp_copy_fw(psp, psp->ta_rap_start_addr, psp->ta_rap_ucode_size);
>>>>       psp_prep_ta_load_cmd_buf(cmd,
>>>>                    psp->fw_pri_mc_addr,
>>>> @@ -3022,7 +3022,7 @@ static ssize_t 
>>>> psp_usbc_pd_fw_sysfs_write(struct device *dev,
>>>>       struct amdgpu_device *adev = drm_to_adev(ddev);
>>>>       void *cpu_addr;
>>>>       dma_addr_t dma_addr;
>>>> -    int ret;
>>>> +    int ret, idx;
>>>>       char fw_name[100];
>>>>       const struct firmware *usbc_pd_fw;
>>>> @@ -3031,6 +3031,9 @@ static ssize_t 
>>>> psp_usbc_pd_fw_sysfs_write(struct device *dev,
>>>>           return -EBUSY;
>>>>       }
>>>> +    if (!drm_dev_enter(ddev, &idx))
>>>> +        return -ENODEV;
>>>> +
>>>>       snprintf(fw_name, sizeof(fw_name), "amdgpu/%s", buf);
>>>>       ret = request_firmware(&usbc_pd_fw, fw_name, adev->dev);
>>>>       if (ret)
>>>> @@ -3062,16 +3065,30 @@ static ssize_t 
>>>> psp_usbc_pd_fw_sysfs_write(struct device *dev,
>>>>   rel_buf:
>>>>       dma_free_coherent(adev->dev, usbc_pd_fw->size, cpu_addr, 
>>>> dma_addr);
>>>>       release_firmware(usbc_pd_fw);
>>>> -
>>>>   fail:
>>>>       if (ret) {
>>>>           DRM_ERROR("Failed to load USBC PD FW, err = %d", ret);
>>>> -        return ret;
>>>> +        count = ret;
>>>>       }
>>>> +    drm_dev_exit(idx);
>>>>       return count;
>>>>   }
>>>> +void psp_copy_fw(struct psp_context *psp, uint8_t *start_addr, 
>>>> uint32_t bin_size)
>>>> +{
>>>> +    int idx;
>>>> +
>>>> +    if (!drm_dev_enter(&psp->adev->ddev, &idx))
>>>> +        return;
>>>> +
>>>> +    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> +    memcpy(psp->fw_pri_buf, start_addr, bin_size);
>>>> +
>>>> +    drm_dev_exit(idx);
>>>> +}
>>>> +
>>>> +
>>>>   static DEVICE_ATTR(usbc_pd_fw, S_IRUGO | S_IWUSR,
>>>>              psp_usbc_pd_fw_sysfs_read,
>>>>              psp_usbc_pd_fw_sysfs_write);
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
>>>> index 46a5328e00e0..2bfdc278817f 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
>>>> @@ -423,4 +423,6 @@ int psp_get_fw_attestation_records_addr(struct 
>>>> psp_context *psp,
>>>>   int psp_load_fw_list(struct psp_context *psp,
>>>>                struct amdgpu_firmware_info **ucode_list, int 
>>>> ucode_count);
>>>> +void psp_copy_fw(struct psp_context *psp, uint8_t *start_addr, 
>>>> uint32_t bin_size);
>>>> +
>>>>   #endif
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
>>>> index 688624ebe421..e1985bc34436 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
>>>> @@ -35,6 +35,8 @@
>>>>   #include "amdgpu.h"
>>>>   #include "atom.h"
>>>> +#include <drm/drm_drv.h>
>>>> +
>>>>   /*
>>>>    * Rings
>>>>    * Most engines on the GPU are fed via ring buffers.  Ring
>>>> @@ -461,3 +463,71 @@ int amdgpu_ring_test_helper(struct amdgpu_ring 
>>>> *ring)
>>>>       ring->sched.ready = !r;
>>>>       return r;
>>>>   }
>>>> +
>>>> +void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
>>>> +{
>>>> +    int idx;
>>>> +    int i = 0;
>>>> +
>>>> +    if (!drm_dev_enter(&ring->adev->ddev, &idx))
>>>> +        return;
>>>> +
>>>> +    while (i <= ring->buf_mask)
>>>> +        ring->ring[i++] = ring->funcs->nop;
>>>> +
>>>> +    drm_dev_exit(idx);
>>>> +
>>>> +}
>>>> +
>>>> +void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
>>>> +{
>>>> +    int idx;
>>>> +
>>>> +    if (!drm_dev_enter(&ring->adev->ddev, &idx))
>>>> +        return;
>>>> +
>>>> +    if (ring->count_dw <= 0)
>>>> +        DRM_ERROR("amdgpu: writing more dwords to the ring than 
>>>> expected!\n");
>>>> +    ring->ring[ring->wptr++ & ring->buf_mask] = v;
>>>> +    ring->wptr &= ring->ptr_mask;
>>>> +    ring->count_dw--;
>>>> +
>>>> +    drm_dev_exit(idx);
>>>> +}
>>>> +
>>>> +void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
>>>> +                          void *src, int count_dw)
>>>> +{
>>>> +    unsigned occupied, chunk1, chunk2;
>>>> +    void *dst;
>>>> +    int idx;
>>>> +
>>>> +    if (!drm_dev_enter(&ring->adev->ddev, &idx))
>>>> +        return;
>>>> +
>>>> +    if (unlikely(ring->count_dw < count_dw))
>>>> +        DRM_ERROR("amdgpu: writing more dwords to the ring than 
>>>> expected!\n");
>>>> +
>>>> +    occupied = ring->wptr & ring->buf_mask;
>>>> +    dst = (void *)&ring->ring[occupied];
>>>> +    chunk1 = ring->buf_mask + 1 - occupied;
>>>> +    chunk1 = (chunk1 >= count_dw) ? count_dw: chunk1;
>>>> +    chunk2 = count_dw - chunk1;
>>>> +    chunk1 <<= 2;
>>>> +    chunk2 <<= 2;
>>>> +
>>>> +    if (chunk1)
>>>> +        memcpy(dst, src, chunk1);
>>>> +
>>>> +    if (chunk2) {
>>>> +        src += chunk1;
>>>> +        dst = (void *)ring->ring;
>>>> +        memcpy(dst, src, chunk2);
>>>> +    }
>>>> +
>>>> +    ring->wptr += count_dw;
>>>> +    ring->wptr &= ring->ptr_mask;
>>>> +    ring->count_dw -= count_dw;
>>>> +
>>>> +    drm_dev_exit(idx);
>>>> +}
>>>
>>> The ring should never we in MMIO memory, so you can completely drop 
>>> that as far as I can see.
>>
>> Yea, it's in all in GART, missed it for some reason...
>>>
>>> Maybe split that patch by use case so that we can more easily 
>>> review/ack it.
>>
>> In fact everything here is the same use case, once I added unmap of
>> all MMIO ranges (both registers ann VRAM) i got a lot of page faults
>> on device remove around any memcpy to from IO. That where I put the
>> drn_dev_enter/exit scope. Also I searched in code and preemeptivly
>> added guards to any other such place. I did drop amdgpu_schedule_ib
>> from this patch both because it had dma_fence_wait inside and so we
>> will take care of this once we decide on how to handle dma_fence waits.
>>
>> Andrey
>>
>>>
>>> Thanks,
>>> Christian.
>>>
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
>>>> index e7d3d0dbdd96..c67bc6d3d039 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
>>>> @@ -299,53 +299,12 @@ static inline void 
>>>> amdgpu_ring_set_preempt_cond_exec(struct amdgpu_ring *ring,
>>>>       *ring->cond_exe_cpu_addr = cond_exec;
>>>>   }
>>>> -static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
>>>> -{
>>>> -    int i = 0;
>>>> -    while (i <= ring->buf_mask)
>>>> -        ring->ring[i++] = ring->funcs->nop;
>>>> -
>>>> -}
>>>> -
>>>> -static inline void amdgpu_ring_write(struct amdgpu_ring *ring, 
>>>> uint32_t v)
>>>> -{
>>>> -    if (ring->count_dw <= 0)
>>>> -        DRM_ERROR("amdgpu: writing more dwords to the ring than 
>>>> expected!\n");
>>>> -    ring->ring[ring->wptr++ & ring->buf_mask] = v;
>>>> -    ring->wptr &= ring->ptr_mask;
>>>> -    ring->count_dw--;
>>>> -}
>>>> +void amdgpu_ring_clear_ring(struct amdgpu_ring *ring);
>>>> -static inline void amdgpu_ring_write_multiple(struct amdgpu_ring 
>>>> *ring,
>>>> -                          void *src, int count_dw)
>>>> -{
>>>> -    unsigned occupied, chunk1, chunk2;
>>>> -    void *dst;
>>>> -
>>>> -    if (unlikely(ring->count_dw < count_dw))
>>>> -        DRM_ERROR("amdgpu: writing more dwords to the ring than 
>>>> expected!\n");
>>>> -
>>>> -    occupied = ring->wptr & ring->buf_mask;
>>>> -    dst = (void *)&ring->ring[occupied];
>>>> -    chunk1 = ring->buf_mask + 1 - occupied;
>>>> -    chunk1 = (chunk1 >= count_dw) ? count_dw: chunk1;
>>>> -    chunk2 = count_dw - chunk1;
>>>> -    chunk1 <<= 2;
>>>> -    chunk2 <<= 2;
>>>> +void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v);
>>>> -    if (chunk1)
>>>> -        memcpy(dst, src, chunk1);
>>>> -
>>>> -    if (chunk2) {
>>>> -        src += chunk1;
>>>> -        dst = (void *)ring->ring;
>>>> -        memcpy(dst, src, chunk2);
>>>> -    }
>>>> -
>>>> -    ring->wptr += count_dw;
>>>> -    ring->wptr &= ring->ptr_mask;
>>>> -    ring->count_dw -= count_dw;
>>>> -}
>>>> +void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
>>>> +                          void *src, int count_dw);
>>>>   int amdgpu_ring_test_helper(struct amdgpu_ring *ring);
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
>>>> index c6dbc0801604..82f0542c7792 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
>>>> @@ -32,6 +32,7 @@
>>>>   #include <linux/module.h>
>>>>   #include <drm/drm.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "amdgpu_pm.h"
>>>> @@ -375,7 +376,7 @@ int amdgpu_uvd_suspend(struct amdgpu_device *adev)
>>>>   {
>>>>       unsigned size;
>>>>       void *ptr;
>>>> -    int i, j;
>>>> +    int i, j, idx;
>>>>       bool in_ras_intr = amdgpu_ras_intr_triggered();
>>>>       cancel_delayed_work_sync(&adev->uvd.idle_work);
>>>> @@ -403,11 +404,15 @@ int amdgpu_uvd_suspend(struct amdgpu_device 
>>>> *adev)
>>>>           if (!adev->uvd.inst[j].saved_bo)
>>>>               return -ENOMEM;
>>>> -        /* re-write 0 since err_event_athub will corrupt VCPU 
>>>> buffer */
>>>> -        if (in_ras_intr)
>>>> -            memset(adev->uvd.inst[j].saved_bo, 0, size);
>>>> -        else
>>>> -            memcpy_fromio(adev->uvd.inst[j].saved_bo, ptr, size);
>>>> +        if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +            /* re-write 0 since err_event_athub will corrupt VCPU 
>>>> buffer */
>>>> +            if (in_ras_intr)
>>>> +                memset(adev->uvd.inst[j].saved_bo, 0, size);
>>>> +            else
>>>> + memcpy_fromio(adev->uvd.inst[j].saved_bo, ptr, size);
>>>> +
>>>> +            drm_dev_exit(idx);
>>>> +        }
>>>>       }
>>>>       if (in_ras_intr)
>>>> @@ -420,7 +425,7 @@ int amdgpu_uvd_resume(struct amdgpu_device *adev)
>>>>   {
>>>>       unsigned size;
>>>>       void *ptr;
>>>> -    int i;
>>>> +    int i, idx;
>>>>       for (i = 0; i < adev->uvd.num_uvd_inst; i++) {
>>>>           if (adev->uvd.harvest_config & (1 << i))
>>>> @@ -432,7 +437,10 @@ int amdgpu_uvd_resume(struct amdgpu_device *adev)
>>>>           ptr = adev->uvd.inst[i].cpu_addr;
>>>>           if (adev->uvd.inst[i].saved_bo != NULL) {
>>>> -            memcpy_toio(ptr, adev->uvd.inst[i].saved_bo, size);
>>>> +            if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +                memcpy_toio(ptr, adev->uvd.inst[i].saved_bo, size);
>>>> +                drm_dev_exit(idx);
>>>> +            }
>>>>               kvfree(adev->uvd.inst[i].saved_bo);
>>>>               adev->uvd.inst[i].saved_bo = NULL;
>>>>           } else {
>>>> @@ -442,8 +450,11 @@ int amdgpu_uvd_resume(struct amdgpu_device *adev)
>>>>               hdr = (const struct common_firmware_header 
>>>> *)adev->uvd.fw->data;
>>>>               if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
>>>>                   offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
>>>> -                memcpy_toio(adev->uvd.inst[i].cpu_addr, 
>>>> adev->uvd.fw->data + offset,
>>>> - le32_to_cpu(hdr->ucode_size_bytes));
>>>> +                if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> + memcpy_toio(adev->uvd.inst[i].cpu_addr, adev->uvd.fw->data + offset,
>>>> + le32_to_cpu(hdr->ucode_size_bytes));
>>>> +                    drm_dev_exit(idx);
>>>> +                }
>>>>                   size -= le32_to_cpu(hdr->ucode_size_bytes);
>>>>                   ptr += le32_to_cpu(hdr->ucode_size_bytes);
>>>>               }
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
>>>> index ea6a62f67e38..833203401ef4 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
>>>> @@ -29,6 +29,7 @@
>>>>   #include <linux/module.h>
>>>>   #include <drm/drm.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "amdgpu_pm.h"
>>>> @@ -293,7 +294,7 @@ int amdgpu_vce_resume(struct amdgpu_device *adev)
>>>>       void *cpu_addr;
>>>>       const struct common_firmware_header *hdr;
>>>>       unsigned offset;
>>>> -    int r;
>>>> +    int r, idx;
>>>>       if (adev->vce.vcpu_bo == NULL)
>>>>           return -EINVAL;
>>>> @@ -313,8 +314,12 @@ int amdgpu_vce_resume(struct amdgpu_device *adev)
>>>>       hdr = (const struct common_firmware_header *)adev->vce.fw->data;
>>>>       offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
>>>> -    memcpy_toio(cpu_addr, adev->vce.fw->data + offset,
>>>> -            adev->vce.fw->size - offset);
>>>> +
>>>> +    if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +        memcpy_toio(cpu_addr, adev->vce.fw->data + offset,
>>>> +                adev->vce.fw->size - offset);
>>>> +        drm_dev_exit(idx);
>>>> +    }
>>>>       amdgpu_bo_kunmap(adev->vce.vcpu_bo);
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
>>>> index 201645963ba5..21f7d3644d70 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
>>>> @@ -27,6 +27,7 @@
>>>>   #include <linux/firmware.h>
>>>>   #include <linux/module.h>
>>>>   #include <linux/pci.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "amdgpu_pm.h"
>>>> @@ -275,7 +276,7 @@ int amdgpu_vcn_suspend(struct amdgpu_device *adev)
>>>>   {
>>>>       unsigned size;
>>>>       void *ptr;
>>>> -    int i;
>>>> +    int i, idx;
>>>>       cancel_delayed_work_sync(&adev->vcn.idle_work);
>>>> @@ -292,7 +293,10 @@ int amdgpu_vcn_suspend(struct amdgpu_device 
>>>> *adev)
>>>>           if (!adev->vcn.inst[i].saved_bo)
>>>>               return -ENOMEM;
>>>> -        memcpy_fromio(adev->vcn.inst[i].saved_bo, ptr, size);
>>>> +        if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +            memcpy_fromio(adev->vcn.inst[i].saved_bo, ptr, size);
>>>> +            drm_dev_exit(idx);
>>>> +        }
>>>>       }
>>>>       return 0;
>>>>   }
>>>> @@ -301,7 +305,7 @@ int amdgpu_vcn_resume(struct amdgpu_device *adev)
>>>>   {
>>>>       unsigned size;
>>>>       void *ptr;
>>>> -    int i;
>>>> +    int i, idx;
>>>>       for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
>>>>           if (adev->vcn.harvest_config & (1 << i))
>>>> @@ -313,7 +317,10 @@ int amdgpu_vcn_resume(struct amdgpu_device *adev)
>>>>           ptr = adev->vcn.inst[i].cpu_addr;
>>>>           if (adev->vcn.inst[i].saved_bo != NULL) {
>>>> -            memcpy_toio(ptr, adev->vcn.inst[i].saved_bo, size);
>>>> +            if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +                memcpy_toio(ptr, adev->vcn.inst[i].saved_bo, size);
>>>> +                drm_dev_exit(idx);
>>>> +            }
>>>>               kvfree(adev->vcn.inst[i].saved_bo);
>>>>               adev->vcn.inst[i].saved_bo = NULL;
>>>>           } else {
>>>> @@ -323,8 +330,11 @@ int amdgpu_vcn_resume(struct amdgpu_device *adev)
>>>>               hdr = (const struct common_firmware_header 
>>>> *)adev->vcn.fw->data;
>>>>               if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
>>>>                   offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
>>>> -                memcpy_toio(adev->vcn.inst[i].cpu_addr, 
>>>> adev->vcn.fw->data + offset,
>>>> - le32_to_cpu(hdr->ucode_size_bytes));
>>>> +                if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> + memcpy_toio(adev->vcn.inst[i].cpu_addr, adev->vcn.fw->data + offset,
>>>> + le32_to_cpu(hdr->ucode_size_bytes));
>>>> +                    drm_dev_exit(idx);
>>>> +                }
>>>>                   size -= le32_to_cpu(hdr->ucode_size_bytes);
>>>>                   ptr += le32_to_cpu(hdr->ucode_size_bytes);
>>>>               }
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>>>> index 9f868cf3b832..7dd5f10ab570 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>>>> @@ -32,6 +32,7 @@
>>>>   #include <linux/dma-buf.h>
>>>>   #include <drm/amdgpu_drm.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "amdgpu_trace.h"
>>>>   #include "amdgpu_amdkfd.h"
>>>> @@ -1606,7 +1607,10 @@ static int 
>>>> amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
>>>>       struct amdgpu_vm_update_params params;
>>>>       enum amdgpu_sync_mode sync_mode;
>>>>       uint64_t pfn;
>>>> -    int r;
>>>> +    int r, idx;
>>>> +
>>>> +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>> +        return -ENODEV;
>>>>       memset(&params, 0, sizeof(params));
>>>>       params.adev = adev;
>>>> @@ -1715,6 +1719,7 @@ static int amdgpu_vm_bo_update_mapping(struct 
>>>> amdgpu_device *adev,
>>>>   error_unlock:
>>>>       amdgpu_vm_eviction_unlock(vm);
>>>> +    drm_dev_exit(idx);
>>>>       return r;
>>>>   }
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
>>>> index 589410c32d09..2cec71e823f5 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
>>>> @@ -23,6 +23,7 @@
>>>>   #include <linux/firmware.h>
>>>>   #include <linux/module.h>
>>>>   #include <linux/vmalloc.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "amdgpu_psp.h"
>>>> @@ -269,10 +270,8 @@ static int 
>>>> psp_v11_0_bootloader_load_kdb(struct psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy PSP KDB binary to memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->kdb_start_addr, psp->kdb_bin_size);
>>>> +    psp_copy_fw(psp, psp->kdb_start_addr, psp->kdb_bin_size);
>>>>       /* Provide the PSP KDB to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> @@ -302,10 +301,8 @@ static int 
>>>> psp_v11_0_bootloader_load_spl(struct psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy PSP SPL binary to memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->spl_start_addr, psp->spl_bin_size);
>>>> +    psp_copy_fw(psp, psp->spl_start_addr, psp->spl_bin_size);
>>>>       /* Provide the PSP SPL to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> @@ -335,10 +332,8 @@ static int 
>>>> psp_v11_0_bootloader_load_sysdrv(struct psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy PSP System Driver binary to memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size);
>>>> +    psp_copy_fw(psp, psp->sys_start_addr, psp->sys_bin_size);
>>>>       /* Provide the sys driver to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> @@ -371,10 +366,8 @@ static int 
>>>> psp_v11_0_bootloader_load_sos(struct psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy Secure OS binary to PSP memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size);
>>>> +    psp_copy_fw(psp, psp->sos_start_addr, psp->sos_bin_size);
>>>>       /* Provide the PSP secure OS to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> @@ -608,7 +601,7 @@ static int psp_v11_0_memory_training(struct 
>>>> psp_context *psp, uint32_t ops)
>>>>       uint32_t p2c_header[4];
>>>>       uint32_t sz;
>>>>       void *buf;
>>>> -    int ret;
>>>> +    int ret, idx;
>>>>       if (ctx->init == PSP_MEM_TRAIN_NOT_SUPPORT) {
>>>>           DRM_DEBUG("Memory training is not supported.\n");
>>>> @@ -681,17 +674,24 @@ static int psp_v11_0_memory_training(struct 
>>>> psp_context *psp, uint32_t ops)
>>>>               return -ENOMEM;
>>>>           }
>>>> -        memcpy_fromio(buf, adev->mman.aper_base_kaddr, sz);
>>>> -        ret = psp_v11_0_memory_training_send_msg(psp, 
>>>> PSP_BL__DRAM_LONG_TRAIN);
>>>> -        if (ret) {
>>>> -            DRM_ERROR("Send long training msg failed.\n");
>>>> +        if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +            memcpy_fromio(buf, adev->mman.aper_base_kaddr, sz);
>>>> +            ret = psp_v11_0_memory_training_send_msg(psp, 
>>>> PSP_BL__DRAM_LONG_TRAIN);
>>>> +            if (ret) {
>>>> +                DRM_ERROR("Send long training msg failed.\n");
>>>> +                vfree(buf);
>>>> +                drm_dev_exit(idx);
>>>> +                return ret;
>>>> +            }
>>>> +
>>>> +            memcpy_toio(adev->mman.aper_base_kaddr, buf, sz);
>>>> +            adev->hdp.funcs->flush_hdp(adev, NULL);
>>>>               vfree(buf);
>>>> -            return ret;
>>>> +            drm_dev_exit(idx);
>>>> +        } else {
>>>> +            vfree(buf);
>>>> +            return -ENODEV;
>>>>           }
>>>> -
>>>> -        memcpy_toio(adev->mman.aper_base_kaddr, buf, sz);
>>>> -        adev->hdp.funcs->flush_hdp(adev, NULL);
>>>> -        vfree(buf);
>>>>       }
>>>>       if (ops & PSP_MEM_TRAIN_SAVE) {
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c
>>>> index c4828bd3264b..618e5b6b85d9 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c
>>>> @@ -138,10 +138,8 @@ static int 
>>>> psp_v12_0_bootloader_load_sysdrv(struct psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy PSP System Driver binary to memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size);
>>>> +    psp_copy_fw(psp, psp->sys_start_addr, psp->sys_bin_size);
>>>>       /* Provide the sys driver to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> @@ -179,10 +177,8 @@ static int 
>>>> psp_v12_0_bootloader_load_sos(struct psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy Secure OS binary to PSP memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size);
>>>> +    psp_copy_fw(psp, psp->sos_start_addr, psp->sos_bin_size);
>>>>       /* Provide the PSP secure OS to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
>>>> index f2e725f72d2f..d0a6cccd0897 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
>>>> @@ -102,10 +102,8 @@ static int 
>>>> psp_v3_1_bootloader_load_sysdrv(struct psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy PSP System Driver binary to memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->sys_start_addr, psp->sys_bin_size);
>>>> +    psp_copy_fw(psp, psp->sys_start_addr, psp->sys_bin_size);
>>>>       /* Provide the sys driver to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> @@ -143,10 +141,8 @@ static int psp_v3_1_bootloader_load_sos(struct 
>>>> psp_context *psp)
>>>>       if (ret)
>>>>           return ret;
>>>> -    memset(psp->fw_pri_buf, 0, PSP_1_MEG);
>>>> -
>>>>       /* Copy Secure OS binary to PSP memory */
>>>> -    memcpy(psp->fw_pri_buf, psp->sos_start_addr, psp->sos_bin_size);
>>>> +    psp_copy_fw(psp, psp->sos_start_addr, psp->sos_bin_size);
>>>>       /* Provide the PSP secure OS to bootloader */
>>>>       WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36,
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
>>>> index 8e238dea7bef..90910d19db12 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
>>>> @@ -25,6 +25,7 @@
>>>>    */
>>>>   #include <linux/firmware.h>
>>>> +#include <drm/drm_drv.h>
>>>>   #include "amdgpu.h"
>>>>   #include "amdgpu_vce.h"
>>>> @@ -555,16 +556,19 @@ static int vce_v4_0_hw_fini(void *handle)
>>>>   static int vce_v4_0_suspend(void *handle)
>>>>   {
>>>>       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
>>>> -    int r;
>>>> +    int r, idx;
>>>>       if (adev->vce.vcpu_bo == NULL)
>>>>           return 0;
>>>> -    if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
>>>> -        unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
>>>> -        void *ptr = adev->vce.cpu_addr;
>>>> +    if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +        if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
>>>> +            unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
>>>> +            void *ptr = adev->vce.cpu_addr;
>>>> -        memcpy_fromio(adev->vce.saved_bo, ptr, size);
>>>> +            memcpy_fromio(adev->vce.saved_bo, ptr, size);
>>>> +        }
>>>> +        drm_dev_exit(idx);
>>>>       }
>>>>       r = vce_v4_0_hw_fini(adev);
>>>> @@ -577,16 +581,20 @@ static int vce_v4_0_suspend(void *handle)
>>>>   static int vce_v4_0_resume(void *handle)
>>>>   {
>>>>       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
>>>> -    int r;
>>>> +    int r, idx;
>>>>       if (adev->vce.vcpu_bo == NULL)
>>>>           return -EINVAL;
>>>>       if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
>>>> -        unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
>>>> -        void *ptr = adev->vce.cpu_addr;
>>>> -        memcpy_toio(ptr, adev->vce.saved_bo, size);
>>>> +        if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +            unsigned size = amdgpu_bo_size(adev->vce.vcpu_bo);
>>>> +            void *ptr = adev->vce.cpu_addr;
>>>> +
>>>> +            memcpy_toio(ptr, adev->vce.saved_bo, size);
>>>> +            drm_dev_exit(idx);
>>>> +        }
>>>>       } else {
>>>>           r = amdgpu_vce_resume(adev);
>>>>           if (r)
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c 
>>>> b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
>>>> index 3f15bf34123a..df34be8ec82d 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
>>>> @@ -34,6 +34,8 @@
>>>>   #include "vcn/vcn_3_0_0_sh_mask.h"
>>>>   #include "ivsrcid/vcn/irqsrcs_vcn_2_0.h"
>>>> +#include <drm/drm_drv.h>
>>>> +
>>>>   #define mmUVD_CONTEXT_ID_INTERNAL_OFFSET            0x27
>>>>   #define mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET 0x0f
>>>>   #define mmUVD_GPCOM_VCPU_DATA0_INTERNAL_OFFSET 0x10
>>>> @@ -268,16 +270,20 @@ static int vcn_v3_0_sw_init(void *handle)
>>>>   static int vcn_v3_0_sw_fini(void *handle)
>>>>   {
>>>>       struct amdgpu_device *adev = (struct amdgpu_device *)handle;
>>>> -    int i, r;
>>>> +    int i, r, idx;
>>>> -    for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
>>>> -        volatile struct amdgpu_fw_shared *fw_shared;
>>>> +    if (drm_dev_enter(&adev->ddev, &idx)) {
>>>> +        for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
>>>> +            volatile struct amdgpu_fw_shared *fw_shared;
>>>> -        if (adev->vcn.harvest_config & (1 << i))
>>>> -            continue;
>>>> -        fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr;
>>>> -        fw_shared->present_flag_0 = 0;
>>>> -        fw_shared->sw_ring.is_enabled = false;
>>>> +            if (adev->vcn.harvest_config & (1 << i))
>>>> +                continue;
>>>> +            fw_shared = adev->vcn.inst[i].fw_shared_cpu_addr;
>>>> +            fw_shared->present_flag_0 = 0;
>>>> +            fw_shared->sw_ring.is_enabled = false;
>>>> +        }
>>>> +
>>>> +        drm_dev_exit(idx);
>>>>       }
>>>>       if (amdgpu_sriov_vf(adev))
>>>> diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu7_smumgr.c 
>>>> b/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu7_smumgr.c
>>>> index aae25243eb10..d628b91846c9 100644
>>>> --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu7_smumgr.c
>>>> +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu7_smumgr.c
>>>> @@ -405,6 +405,8 @@ int smu7_request_smu_load_fw(struct pp_hwmgr 
>>>> *hwmgr)
>>>>                   UCODE_ID_MEC_STORAGE, 
>>>> &toc->entry[toc->num_entries++]),
>>>>                   "Failed to Get Firmware Entry.", r = -EINVAL; 
>>>> goto failed);
>>>>       }
>>>> +
>>>> +    /* AG TODO Can't call drm_dev_enter/exit because access 
>>>> adev->ddev here ... */
>>>>       memcpy_toio(smu_data->header_buffer.kaddr, smu_data->toc,
>>>>               sizeof(struct SMU_DRAMData_TOC));
>>>>       smum_send_msg_to_smc_with_parameter(hwmgr,
>>>



More information about the dri-devel mailing list