Re: [PATCH v5 1/2] drm/amdgpu: add debugfs for reset registers list

["Christian König" <ckoenig.leichtzumerken@gmail.com>]

Am 15.02.22 um 16:36 schrieb Somalapuram Amaranath:
> List of register populated for dump collection during the GPU reset.
>
> Signed-off-by: Somalapuram Amaranath <Amaranath.Somalapuram@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu.h         |  4 +
>   drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c | 95 +++++++++++++++++++++
>   2 files changed, 99 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> index b85b67a88a3d..2e8c2318276d 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
> @@ -1097,6 +1097,10 @@ struct amdgpu_device {
>   
>   	struct amdgpu_reset_control     *reset_cntl;
>   	uint32_t                        ip_versions[HW_ID_MAX][HWIP_MAX_INSTANCE];
> +
> +	/* reset dump register */
> +	uint32_t			*reset_dump_reg_list;
> +	int                             n_regs;
>   };
>   
>   static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev)
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
> index 164d6a9e9fbb..be4336574fec 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
> @@ -1609,6 +1609,99 @@ DEFINE_DEBUGFS_ATTRIBUTE(fops_ib_preempt, NULL,
>   DEFINE_DEBUGFS_ATTRIBUTE(fops_sclk_set, NULL,
>   			amdgpu_debugfs_sclk_set, "%llu\n");
>   
> +static ssize_t amdgpu_reset_dump_register_list_read(struct file *f,
> +				char __user *buf, size_t size, loff_t *pos)
> +{
> +	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
> +	char reg_offset[11];
> +	int i, r, len = 0;
> +
> +	if (adev->n_regs == 0)
> +		return 0;
> +
> +	for (i = 0; i < adev->n_regs; i++) {
> +		sprintf(reg_offset, "0x%x ", adev->reset_dump_reg_list[i]);
> +		r = copy_to_user(buf + len, reg_offset, strlen(reg_offset));
> +		
> +		if (r)
> +			return -EFAULT;
> +
> +		len += strlen(reg_offset);
> +	}
> +
> +	r = copy_to_user(buf + len, "\n", 1);
> +
> +	if (r)
> +		return -EFAULT;
> +
> +	len++;
> +
> +	if (*pos >= len)
> +		return 0;

I think we should probably move that to the beginning of the function 
and just return 0 when *pos is none null.

Alternatively you can implement it much more clever if you actually make 
use of this, e.g. return the register based on the current *pos.

> +
> +	*pos += len - r;

The - r part can be dropped.

> +
> +	return len;
> +}
> +
> +static ssize_t amdgpu_reset_dump_register_list_write(struct file *f,
> +			const char __user *buf, size_t size, loff_t *pos)
> +{
> +	struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(f)->i_private;
> +	char *reg_offset, *reg, reg_temp[11];
> +	int ret, i = 0, len = 0;
> +
> +	reg_offset = reg_temp;
> +	memset(reg_offset,  0, 11);
> +	ret = copy_from_user(reg_offset, buf + len, (11 < (size) ? 11 : (size)));

Please use min(11, size) here.

> +
> +	if (ret)
> +		return -EFAULT;
> +
> +	if (adev->n_regs > 0) {
> +		adev->n_regs = 0;
> +		kfree(adev->reset_dump_reg_list);
> +		adev->reset_dump_reg_list = NULL;
> +	}

If would be better to swap that in after gathering the complete register 
list.

And you need to grab the reset lock while updating it to make sure that 
a concurrent reset doesn't touch freed data.

> +
> +	while (((reg = strsep(&reg_offset, " ")) != NULL) && len < size) {
> +		adev->reset_dump_reg_list =  krealloc_array(
> +						adev->reset_dump_reg_list, 1,
> +						sizeof(uint32_t), GFP_KERNEL);
> +		ret  = kstrtouint(reg, 16, &adev->reset_dump_reg_list[i]);
> +
> +		if (ret) {
> +			kfree(adev->reset_dump_reg_list);
> +			adev->reset_dump_reg_list = NULL;
> +			return -EINVAL;
> +		}
> +
> +		len += strlen(reg) + 1;
> +		reg_offset = reg_temp;
> +		memset(reg_offset,  0, 11);
> +		ret = copy_from_user(reg_offset, buf + len, (11 < (size-len) ? 11 : (size-len)));
> +
> +		if (ret) {
> +			kfree(adev->reset_dump_reg_list);
> +			adev->reset_dump_reg_list = NULL;
> +			return -EFAULT;
> +		}
> +
> +		i++;
> +	}

I think you should completely rework that loop into a do { ... } while 
(size);

Regards,
Christian.

> +
> +	adev->n_regs = i;
> +
> +	return size;
> +}
> +
> +static const struct file_operations amdgpu_reset_dump_register_list = {
> +	.owner = THIS_MODULE,
> +	.read = amdgpu_reset_dump_register_list_read,
> +	.write = amdgpu_reset_dump_register_list_write,
> +	.llseek = default_llseek
> +};
> +
>   int amdgpu_debugfs_init(struct amdgpu_device *adev)
>   {
>   	struct dentry *root = adev_to_drm(adev)->primary->debugfs_root;
> @@ -1672,6 +1765,8 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev)
>   			    &amdgpu_debugfs_test_ib_fops);
>   	debugfs_create_file("amdgpu_vm_info", 0444, root, adev,
>   			    &amdgpu_debugfs_vm_info_fops);
> +	debugfs_create_file("amdgpu_reset_dump_register_list", 0644, root, adev,
> +			    &amdgpu_reset_dump_register_list);
>   
>   	adev->debugfs_vbios_blob.data = adev->bios;
>   	adev->debugfs_vbios_blob.size = adev->bios_size;