From: Huang Rui <ray.huang-5C7GfCeVMHo@public.gmane.org>
To: "Christian K�nig"
<ckoenig.leichtzumerken-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Cc: amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
Subject: Re: [PATCH 5/9] drm/amdgpu: always recreate bo_list
Date: Tue, 31 Jul 2018 13:29:22 +0800 [thread overview]
Message-ID: <20180731051209.GD9056@hr-amur2> (raw)
In-Reply-To: <20180730145159.13212-5-christian.koenig-5C7GfCeVMHo@public.gmane.org>
On Mon, Jul 30, 2018 at 04:51:55PM +0200, Christian König wrote:
> When changing a list always completely recreate it. This avoids locking
> in the hot path because the list always stays like it is until it is
> unreferenced.
The fpriv->bo_list_handles is allocated by OP_CREATE, so here we just
re-create bo list and replace the handles in OP_UPDATE. Then we don't need
locking to protect amdgpu_bo_list because it always be re-created.
Acked-by: Huang Rui <ray.huang@amd.com>
>
> Signed-off-by: Christian König <christian.koenig@amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c | 23 ++++++++++++-----------
> drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h | 1 -
> drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 3 ---
> 3 files changed, 12 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
> index 6728448167ba..556040e45931 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
> @@ -50,7 +50,6 @@ static void amdgpu_bo_list_release_rcu(struct kref *ref)
> for (i = 0; i < list->num_entries; ++i)
> amdgpu_bo_unref(&list->array[i].robj);
>
> - mutex_destroy(&list->lock);
> kvfree(list->array);
> kfree_rcu(list, rhead);
> }
> @@ -70,7 +69,6 @@ int amdgpu_bo_list_create(struct amdgpu_device *adev,
> return -ENOMEM;
>
> /* initialize bo list*/
> - mutex_init(&list->lock);
> kref_init(&list->refcount);
> r = amdgpu_bo_list_set(adev, filp, list, info, num_entries);
> if (r) {
> @@ -188,7 +186,6 @@ int amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id,
>
> if (*result && kref_get_unless_zero(&(*result)->refcount)) {
> rcu_read_unlock();
> - mutex_lock(&(*result)->lock);
> return 0;
> }
>
> @@ -231,7 +228,6 @@ void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
>
> void amdgpu_bo_list_put(struct amdgpu_bo_list *list)
> {
> - mutex_unlock(&list->lock);
> kref_put(&list->refcount, amdgpu_bo_list_release_rcu);
> }
>
> @@ -242,7 +238,6 @@ void amdgpu_bo_list_free(struct amdgpu_bo_list *list)
> for (i = 0; i < list->num_entries; ++i)
> amdgpu_bo_unref(&list->array[i].robj);
>
> - mutex_destroy(&list->lock);
> kvfree(list->array);
> kfree(list);
> }
> @@ -297,7 +292,7 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
> union drm_amdgpu_bo_list *args = data;
> uint32_t handle = args->in.list_handle;
> struct drm_amdgpu_bo_list_entry *info = NULL;
> - struct amdgpu_bo_list *list;
> + struct amdgpu_bo_list *list, *old;
> int r;
>
> r = amdgpu_bo_create_list_entry_array(&args->in, &info);
> @@ -328,16 +323,22 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
> break;
>
> case AMDGPU_BO_LIST_OP_UPDATE:
> - r = amdgpu_bo_list_get(fpriv, handle, &list);
> + r = amdgpu_bo_list_create(adev, filp, info, args->in.bo_number,
> + &list);
> if (r)
> goto error_free;
>
> - r = amdgpu_bo_list_set(adev, filp, list, info,
> - args->in.bo_number);
> - amdgpu_bo_list_put(list);
> - if (r)
> + mutex_lock(&fpriv->bo_list_lock);
> + old = idr_replace(&fpriv->bo_list_handles, list, handle);
> + mutex_unlock(&fpriv->bo_list_lock);
> +
> + if (IS_ERR(old)) {
> + amdgpu_bo_list_put(list);
> + r = PTR_ERR(old);
> goto error_free;
> + }
>
> + amdgpu_bo_list_put(old);
> break;
>
> default:
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
> index 833f846bfdad..89195fdcb1ef 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h
> @@ -41,7 +41,6 @@ struct amdgpu_bo_list_entry {
> };
>
> struct amdgpu_bo_list {
> - struct mutex lock;
> struct rcu_head rhead;
> struct kref refcount;
> struct amdgpu_bo *gds_obj;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> index 0295666968da..f7154f3ed807 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> @@ -580,9 +580,6 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
> &p->bo_list);
> if (r)
> return r;
> -
> - } else if (p->bo_list) {
> - mutex_lock(&p->bo_list->lock);
> }
>
> if (p->bo_list) {
> --
> 2.14.1
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
next prev parent reply other threads:[~2018-07-31 5:29 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-07-30 14:51 [PATCH 1/9] drm/amdgpu: fix total size calculation Christian König
[not found] ` <20180730145159.13212-1-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-30 14:51 ` [PATCH 2/9] drm/amdgpu: return error if both BOs and bo_list handle is given Christian König
[not found] ` <20180730145159.13212-2-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-31 2:44 ` Huang Rui
2018-07-30 14:51 ` [PATCH 3/9] drm/amdgpu: add new amdgpu_vm_bo_trace_cs() function v2 Christian König
2018-07-30 14:51 ` [PATCH 4/9] drm/amdgpu: move bo_list defines to amdgpu_bo_list.h Christian König
[not found] ` <20180730145159.13212-4-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-31 2:52 ` Huang Rui
2018-07-30 14:51 ` [PATCH 5/9] drm/amdgpu: always recreate bo_list Christian König
[not found] ` <20180730145159.13212-5-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-31 5:29 ` Huang Rui [this message]
2018-07-30 14:51 ` [PATCH 6/9] drm/amdgpu: nuke amdgpu_bo_list_free Christian König
[not found] ` <20180730145159.13212-6-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-31 5:31 ` Huang Rui
2018-07-30 14:51 ` [PATCH 7/9] drm/amdgpu: add bo_list iterators Christian König
[not found] ` <20180730145159.13212-7-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-31 6:12 ` Huang Rui
2018-07-30 14:51 ` [PATCH 8/9] drm/amdgpu: allocate the bo_list array after the list Christian König
[not found] ` <20180730145159.13212-8-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-31 7:12 ` Huang Rui
2018-07-31 7:04 ` Christian König
[not found] ` <79923ec6-25d0-f80a-d631-79a1faa91a0e-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2018-07-31 10:23 ` Huang Rui
2018-07-30 14:51 ` [PATCH 9/9] drm/amdgpu: create an empty bo_list if no handle is provided Christian König
[not found] ` <20180730145159.13212-9-christian.koenig-5C7GfCeVMHo@public.gmane.org>
2018-07-31 1:31 ` Zhou, David(ChunMing)
2018-07-31 7:51 ` Huang Rui
2018-07-31 8:52 ` Christian König
[not found] ` <c630d2c3-25c8-b532-1a1c-07626f82ea12-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2018-07-31 9:00 ` Zhang, Jerry (Junwei)
2018-07-31 9:09 ` Huang Rui
2018-07-31 9:00 ` Christian König
[not found] ` <e35b8029-27d0-fc97-64c2-64f4379eeda8-5C7GfCeVMHo@public.gmane.org>
2018-07-31 9:26 ` Huang Rui
2018-07-31 9:41 ` Zhang, Jerry (Junwei)
2018-07-31 2:34 ` [PATCH 1/9] drm/amdgpu: fix total size calculation Huang Rui
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=20180731051209.GD9056@hr-amur2 \
--to=ray.huang-5c7gfcevmho@public.gmane.org \
--cc=amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
--cc=ckoenig.leichtzumerken-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox