All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Christian König" <christian.koenig@amd.com>
To: Nirmoy Das <nirmoy.das@amd.com>, amd-gfx@lists.freedesktop.org
Cc: alexander.deucher@amd.com
Subject: Re: [PATCH 7/7] drm/amdgpu: do not allocate entries separately
Date: Fri, 21 May 2021 15:01:40 +0200	[thread overview]
Message-ID: <04e2e462-1e00-e5da-bc1b-e0255f09ad45@amd.com> (raw)
In-Reply-To: <20210521124533.4380-7-nirmoy.das@amd.com>

Am 21.05.21 um 14:45 schrieb Nirmoy Das:
> Allocate PD/PT entries while allocating VM BOs and use that
> instead of allocating those entries separately.
>
> Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 31 ++++++++++++++------------
>   1 file changed, 17 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index 120e6b7a0286..4717f075a391 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -880,7 +880,12 @@ static int amdgpu_vm_pt_create(struct amdgpu_device *adev,
>   	bp.domain = amdgpu_bo_get_preferred_pin_domain(adev, bp.domain);
>   	bp.flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
>   		AMDGPU_GEM_CREATE_CPU_GTT_USWC;
> -	bp.bo_ptr_size = sizeof(struct amdgpu_bo_vm);
> +	if (level < AMDGPU_VM_PTB)
> +		bp.bo_ptr_size = struct_size((*vmbo), entries,
> +					     amdgpu_vm_num_entries(adev, level));
> +	else
> +		bp.bo_ptr_size = sizeof(struct amdgpu_bo_vm);
> +

Rather do it like this here:

if (level < AMDGPU_VM_PTB)
     num_entries = amdgpu_vm_num_entries(...)
else
     num_entries = 0;

bp.bo_ptr_size = struct_size(....)

If we have that calculation more than once then it might make sense to 
unify it in a function, but I don't think so of hand.

Regards,
Christian.

>   	if (vm->use_cpu_for_update)
>   		bp.flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
>   
> @@ -954,19 +959,14 @@ static int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
>   	struct amdgpu_bo_vm *pt;
>   	int r;
>   
> -	if (cursor->level < AMDGPU_VM_PTB && !entry->entries) {
> -		unsigned num_entries;
> -
> -		num_entries = amdgpu_vm_num_entries(adev, cursor->level);
> -		entry->entries = kvmalloc_array(num_entries,
> -						sizeof(*entry->entries),
> -						GFP_KERNEL | __GFP_ZERO);
> -		if (!entry->entries)
> -			return -ENOMEM;
> -	}
> -
> -	if (entry->base.bo)
> +	if (entry->base.bo) {
> +		if (cursor->level < AMDGPU_VM_PTB)
> +			entry->entries =
> +				to_amdgpu_bo_vm(entry->base.bo)->entries;
> +		else
> +			entry->entries = NULL;
>   		return 0;
> +	}
>   
>   	r = amdgpu_vm_pt_create(adev, vm, cursor->level, immediate, &pt);
>   	if (r)
> @@ -978,6 +978,10 @@ static int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
>   	pt_bo = &pt->bo;
>   	pt_bo->parent = amdgpu_bo_ref(cursor->parent->base.bo);
>   	amdgpu_vm_bo_base_init(&entry->base, vm, pt_bo);
> +	if (cursor->level < AMDGPU_VM_PTB)
> +		entry->entries = pt->entries;
> +	else
> +		entry->entries = NULL;
>   
>   	r = amdgpu_vm_clear_bo(adev, vm, pt_bo, immediate);
>   	if (r)
> @@ -1005,7 +1009,6 @@ static void amdgpu_vm_free_table(struct amdgpu_vm_pt *entry)
>   			amdgpu_bo_unref(&to_amdgpu_bo_vm(entry->base.bo)->shadow);
>   		amdgpu_bo_unref(&entry->base.bo);
>   	}
> -	kvfree(entry->entries);
>   	entry->entries = NULL;
>   }
>   

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

  reply	other threads:[~2021-05-21 13:01 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-21 12:45 [PATCH 1/7] drm/amdgpu: add amdgpu_bo_vm bo type Nirmoy Das
2021-05-21 12:45 ` [PATCH 2/7] drm/amdgpu: add a new identifier for amdgpu_bo Nirmoy Das
2021-05-21 12:58   ` Christian König
2021-05-21 13:54     ` Nirmoy
2021-05-21 12:45 ` [PATCH 3/7] drm/amdgpu: use amdgpu_bo_vm for vm code Nirmoy Das
2021-05-21 12:45 ` [PATCH 4/7] drm/amdgpu: create shadow bo directly Nirmoy Das
2021-05-21 12:45 ` [PATCH 5/7] drm/amdgpu: switch to amdgpu_bo_vm's shadow Nirmoy Das
2021-05-21 12:45 ` [PATCH 6/7] drm/amdgpu: remove unused code Nirmoy Das
2021-05-21 12:45 ` [PATCH 7/7] drm/amdgpu: do not allocate entries separately Nirmoy Das
2021-05-21 13:01   ` Christian König [this message]
2021-05-21 14:04     ` Nirmoy
2021-05-21 14:54 ` [PATCH 1/7] drm/amdgpu: add amdgpu_bo_vm bo type Alex Deucher
2021-05-21 15:19   ` Nirmoy

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=04e2e462-1e00-e5da-bc1b-e0255f09ad45@amd.com \
    --to=christian.koenig@amd.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=nirmoy.das@amd.com \
    --subject='Re: [PATCH 7/7] drm/amdgpu: do not allocate entries separately' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.