Re: [PATCH v4 11/14] drm/amdgpu: Guard against write accesses after device removal

[Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>]

On 1/19/21 2:16 PM, Andrey Grodzovsky wrote:
>
> On 1/19/21 1:59 PM, Christian König wrote:
>> Am 19.01.21 um 19:22 schrieb Andrey Grodzovsky:
>>>
>>> On 1/19/21 1:05 PM, Daniel Vetter wrote:
>>>> On Tue, Jan 19, 2021 at 4:35 PM Andrey Grodzovsky
>>>> <Andrey.Grodzovsky@amd.com> wrote:
>>>>> There is really no other way according to this article
>>>>> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flwn.net%2FArticles%2F767885%2F&amp;data=04%7C01%7CAndrey.Grodzovsky%40amd.com%7Cee61fb937d2d4baedf6f08d8bcac5b02%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637466795752297305%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=a9Y4ZMEVYaMP7IeMVxQgXGpAkDXSkedMAiWkyqwzEe8%3D&amp;reserved=0 
>>>>>
>>>>>
>>>>> "A perfect solution seems nearly impossible though; we cannot acquire a 
>>>>> mutex on
>>>>> the user
>>>>> to prevent them from yanking a device and we cannot check for a presence 
>>>>> change
>>>>> after every
>>>>> device access for performance reasons. "
>>>>>
>>>>> But I assumed srcu_read_lock should be pretty seamless performance wise, no ?
>>>> The read side is supposed to be dirt cheap, the write side is were we
>>>> just stall for all readers to eventually complete on their own.
>>>> Definitely should be much cheaper than mmio read, on the mmio write
>>>> side it might actually hurt a bit. Otoh I think those don't stall the
>>>> cpu by default when they're timing out, so maybe if the overhead is
>>>> too much for those, we could omit them?
>>>>
>>>> Maybe just do a small microbenchmark for these for testing, with a
>>>> register that doesn't change hw state. So with and without
>>>> drm_dev_enter/exit, and also one with the hw plugged out so that we
>>>> have actual timeouts in the transactions.
>>>> -Daniel
>>>
>>>
>>> So say writing in a loop to some harmless scratch register for many times 
>>> both for plugged
>>> and unplugged case and measure total time delta ?
>>
>> I think we should at least measure the following:
>>
>> 1. Writing X times to a scratch reg without your patch.
>> 2. Writing X times to a scratch reg with your patch.
>> 3. Writing X times to a scratch reg with the hardware physically disconnected.


Just realized, I can't test this part since I don't have eGPU to yank out.

Andrey


>>
>> I suggest to repeat that once for Polaris (or older) and once for Vega or Navi.
>>
>> The SRBM on Polaris is meant to introduce some delay in each access, so it 
>> might react differently then the newer hardware.
>>
>> Christian.
>
>
> Will do.
>
> Andrey
>
>
>>
>>>
>>> Andrey
>>>
>>>
>>>>
>>>>> The other solution would be as I suggested to keep all the device IO ranges
>>>>> reserved and system
>>>>> memory pages unfreed until the device is finalized in the driver but 
>>>>> Daniel said
>>>>> this would upset the PCI layer (the MMIO ranges reservation part).
>>>>>
>>>>> Andrey
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On 1/19/21 3:55 AM, Christian König wrote:
>>>>>> Am 18.01.21 um 22:01 schrieb Andrey Grodzovsky:
>>>>>>> This should prevent writing to memory or IO ranges possibly
>>>>>>> already allocated for other uses after our device is removed.
>>>>>> Wow, that adds quite some overhead to every register access. I'm not sure we
>>>>>> can do this.
>>>>>>
>>>>>> Christian.
>>>>>>
>>>>>>> Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
>>>>>>> ---
>>>>>>>    drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 57 ++++++++++++++++++++++++
>>>>>>>    drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c    |  9 ++++
>>>>>>>    drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c    | 53 +++++++++++++---------
>>>>>>>    drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h    |  3 ++
>>>>>>>    drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c   | 70 
>>>>>>> ++++++++++++++++++++++++++++++
>>>>>>>    drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h   | 49 ++-------------------
>>>>>>>    drivers/gpu/drm/amd/amdgpu/psp_v11_0.c     | 16 ++-----
>>>>>>>    drivers/gpu/drm/amd/amdgpu/psp_v12_0.c     |  8 +---
>>>>>>>    drivers/gpu/drm/amd/amdgpu/psp_v3_1.c      |  8 +---
>>>>>>>    9 files changed, 184 insertions(+), 89 deletions(-)
>>>>>>>
>>>>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>>>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>>>>>> index e99f4f1..0a9d73c 100644
>>>>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>>>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>>>>>> @@ -72,6 +72,8 @@
>>>>>>>      #include <linux/iommu.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");
>>>>>>> @@ -404,13 +406,21 @@ uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev,
>>>>>>> uint32_t offset)
>>>>>>>     */
>>>>>>>    void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, 
>>>>>>> uint8_t
>>>>>>> value)
>>>>>>>    {
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>>        if (adev->in_pci_err_recovery)
>>>>>>>            return;
>>>>>>>    +
>>>>>>> +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> +
>>>>>>>        if (offset < adev->rmmio_size)
>>>>>>>            writeb(value, adev->rmmio + offset);
>>>>>>>        else
>>>>>>>            BUG();
>>>>>>> +
>>>>>>> +    drm_dev_exit(idx);
>>>>>>>    }
>>>>>>>      /**
>>>>>>> @@ -427,9 +437,14 @@ void amdgpu_device_wreg(struct amdgpu_device *adev,
>>>>>>>                uint32_t reg, uint32_t v,
>>>>>>>                uint32_t acc_flags)
>>>>>>>    {
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>>        if (adev->in_pci_err_recovery)
>>>>>>>            return;
>>>>>>>    +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> +
>>>>>>>        if ((reg * 4) < adev->rmmio_size) {
>>>>>>>            if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
>>>>>>>                amdgpu_sriov_runtime(adev) &&
>>>>>>> @@ -444,6 +459,8 @@ void amdgpu_device_wreg(struct amdgpu_device *adev,
>>>>>>>        }
>>>>>>> trace_amdgpu_device_wreg(adev->pdev->device, reg, v);
>>>>>>> +
>>>>>>> +    drm_dev_exit(idx);
>>>>>>>    }
>>>>>>>      /*
>>>>>>> @@ -454,9 +471,14 @@ void amdgpu_device_wreg(struct amdgpu_device *adev,
>>>>>>>    void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev,
>>>>>>>                     uint32_t reg, uint32_t v)
>>>>>>>    {
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>>        if (adev->in_pci_err_recovery)
>>>>>>>            return;
>>>>>>>    +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> +
>>>>>>>        if (amdgpu_sriov_fullaccess(adev) &&
>>>>>>>            adev->gfx.rlc.funcs &&
>>>>>>> adev->gfx.rlc.funcs->is_rlcg_access_range) {
>>>>>>> @@ -465,6 +487,8 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device 
>>>>>>> *adev,
>>>>>>>        } else {
>>>>>>>            writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
>>>>>>>        }
>>>>>>> +
>>>>>>> +    drm_dev_exit(idx);
>>>>>>>    }
>>>>>>>      /**
>>>>>>> @@ -499,15 +523,22 @@ u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 
>>>>>>> reg)
>>>>>>>     */
>>>>>>>    void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
>>>>>>>    {
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>>        if (adev->in_pci_err_recovery)
>>>>>>>            return;
>>>>>>>    +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> +
>>>>>>>        if ((reg * 4) < adev->rio_mem_size)
>>>>>>>            iowrite32(v, adev->rio_mem + (reg * 4));
>>>>>>>        else {
>>>>>>>            iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4));
>>>>>>>            iowrite32(v, adev->rio_mem + (mmMM_DATA * 4));
>>>>>>>        }
>>>>>>> +
>>>>>>> +    drm_dev_exit(idx);
>>>>>>>    }
>>>>>>>      /**
>>>>>>> @@ -544,14 +575,21 @@ u32 amdgpu_mm_rdoorbell(struct amdgpu_device 
>>>>>>> *adev, u32
>>>>>>> index)
>>>>>>>     */
>>>>>>>    void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
>>>>>>>    {
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>>        if (adev->in_pci_err_recovery)
>>>>>>>            return;
>>>>>>>    +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> +
>>>>>>>        if (index < adev->doorbell.num_doorbells) {
>>>>>>>            writel(v, adev->doorbell.ptr + index);
>>>>>>>        } else {
>>>>>>>            DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
>>>>>>>        }
>>>>>>> +
>>>>>>> +    drm_dev_exit(idx);
>>>>>>>    }
>>>>>>>      /**
>>>>>>> @@ -588,14 +626,21 @@ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev,
>>>>>>> u32 index)
>>>>>>>     */
>>>>>>>    void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
>>>>>>>    {
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>>        if (adev->in_pci_err_recovery)
>>>>>>>            return;
>>>>>>>    +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> +
>>>>>>>        if (index < adev->doorbell.num_doorbells) {
>>>>>>>            atomic64_set((atomic64_t *)(adev->doorbell.ptr + index), v);
>>>>>>>        } else {
>>>>>>>            DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
>>>>>>>        }
>>>>>>> +
>>>>>>> +    drm_dev_exit(idx);
>>>>>>>    }
>>>>>>>      /**
>>>>>>> @@ -682,6 +727,10 @@ void amdgpu_device_indirect_wreg(struct amdgpu_device
>>>>>>> *adev,
>>>>>>>        unsigned long flags;
>>>>>>>        void __iomem *pcie_index_offset;
>>>>>>>        void __iomem *pcie_data_offset;
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>> +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> spin_lock_irqsave(&adev->pcie_idx_lock, flags);
>>>>>>>        pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
>>>>>>> @@ -692,6 +741,8 @@ void amdgpu_device_indirect_wreg(struct 
>>>>>>> amdgpu_device *adev,
>>>>>>>        writel(reg_data, pcie_data_offset);
>>>>>>>        readl(pcie_data_offset);
>>>>>>> spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
>>>>>>> +
>>>>>>> +    drm_dev_exit(idx);
>>>>>>>    }
>>>>>>>      /**
>>>>>>> @@ -711,6 +762,10 @@ void amdgpu_device_indirect_wreg64(struct 
>>>>>>> amdgpu_device
>>>>>>> *adev,
>>>>>>>        unsigned long flags;
>>>>>>>        void __iomem *pcie_index_offset;
>>>>>>>        void __iomem *pcie_data_offset;
>>>>>>> +    int idx;
>>>>>>> +
>>>>>>> +    if (!drm_dev_enter(&adev->ddev, &idx))
>>>>>>> +        return;
>>>>>>> spin_lock_irqsave(&adev->pcie_idx_lock, flags);
>>>>>>>        pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
>>>>>>> @@ -727,6 +782,8 @@ void amdgpu_device_indirect_wreg64(struct amdgpu_device
>>>>>>> *adev,
>>>>>>>        writel((u32)(reg_data >> 32), pcie_data_offset);
>>>>>>>        readl(pcie_data_offset);
>>>>>>> spin_unlock_irqrestore(&adev->pcie_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 fe1a39f..1beb4e6 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_get_pde_for_bo - get the PDE for a BO
>>>>>>>     *
>>>>>>> @@ -98,6 +100,10 @@ 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.
>>>>>>> @@ -105,6 +111,9 @@ 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_psp.c
>>>>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
>>>>>>> index 523d22d..89e2bfe 100644
>>>>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
>>>>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
>>>>>>> @@ -37,6 +37,8 @@
>>>>>>>      #include "amdgpu_ras.h"
>>>>>>>    +#include <drm/drm_drv.h>
>>>>>>> +
>>>>>>>    static int psp_sysfs_init(struct amdgpu_device *adev);
>>>>>>>    static void psp_sysfs_fini(struct amdgpu_device *adev);
>>>>>>>    @@ -248,7 +250,7 @@ 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 = 2000;
>>>>>>>        bool ras_intr = false;
>>>>>>>        bool skip_unsupport = false;
>>>>>>> @@ -256,6 +258,9 @@ 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);
>>>>>>> @@ -266,8 +271,7 @@ 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);
>>>>>>> @@ -307,8 +311,8 @@ 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;
>>>>>>>            }
>>>>>>>        }
>>>>>>>    @@ -316,8 +320,10 @@ 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;
>>>>>>>    }
>>>>>>>    @@ -354,8 +360,7 @@ 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);
>>>>>>>    @@ -570,8 +575,7 @@ 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);
>>>>>>> @@ -726,8 +730,7 @@ 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,
>>>>>>> @@ -982,8 +985,7 @@ 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,
>>>>>>> @@ -1219,8 +1221,7 @@ 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,
>>>>>>> @@ -1366,8 +1367,7 @@ 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,
>>>>>>> @@ -1507,8 +1507,7 @@ 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,
>>>>>>> @@ -2778,6 +2777,20 @@ static ssize_t psp_usbc_pd_fw_sysfs_write(struct
>>>>>>> device *dev,
>>>>>>>        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 da250bc..ac69314 100644
>>>>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
>>>>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
>>>>>>> @@ -400,4 +400,7 @@ int psp_init_ta_microcode(struct psp_context *psp,
>>>>>>>                  const char *chip_name);
>>>>>>>    int psp_get_fw_attestation_records_addr(struct psp_context *psp,
>>>>>>>                        uint64_t *output_ptr);
>>>>>>> +
>>>>>>> +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 1a612f5..d656494 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
>>>>>>> @@ -463,3 +465,71 @@ 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);
>>>>>>> +}
>>>>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
>>>>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
>>>>>>> index accb243..f90b81f 100644
>>>>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
>>>>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
>>>>>>> @@ -300,53 +300,12 @@ 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/psp_v11_0.c
>>>>>>> b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
>>>>>>> index bd4248c..b3ce5be 100644
>>>>>>> --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
>>>>>>> +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
>>>>>>> @@ -269,10 +269,8 @@ 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 +300,8 @@ 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 +331,8 @@ 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 +365,8 @@ 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,
>>>>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c
>>>>>>> b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c
>>>>>>> index c4828bd..618e5b6 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 f2e725f..d0a6cccd 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,
>>>>
>>>>
>>> _______________________________________________
>>> amd-gfx mailing list
>>> amd-gfx@lists.freedesktop.org
>>> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&amp;data=04%7C01%7CAndrey.Grodzovsky%40amd.com%7Cee61fb937d2d4baedf6f08d8bcac5b02%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637466795752297305%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=a5MkPkwHh7WkR24K9EoCWSKPdCpiXCJH6RwGbGyhHyA%3D&amp;reserved=0 
>>>
>>
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx