Re: reservation questions

[Christian König <ckoenig.leichtzumerken@xxxxxxxxx>]

When adding fences to a reservation object all newly added fences must be created after all existing fences.

In other words when adding a shared fence the caller must guarantee that all existing fences of the same context must be completed before this new one. Because of this we can safely replace the existing fence with the new one.

When adding an exclusive fence the caller must guarantee that all existing fences of that reservation object must complete before this new one. Because of this we can safely throw away all existing fences as well as the old exclusive fence.

Regards,
Christian.

Am 06.03.2018 um 11:10 schrieb Liu, Monk:

Oh, I see

    for (i =

0; i < fobj->shared_count; ++i) {

 

        struct dma_fence *old_fence;

 

 

Since it use I < fobj->shared_count, then the wild pointer access won’t hit, please ignore question 1

 

The only valuable is question 2: how we treat excl fence and shared fence?

e.g. when we add an excl fence to resv, how to deal with shared ? current logic is just put the all

do we need to wait on their signaling before putting them ?

 

thanks

 

/Monk

 

 

From: Liu, Monk
Sent: 2018年3月6日 17:57
To: dri-devel@xxxxxxxxxxxxxxxxxxxxx; Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>; Koenig, Christian <Christian.Koenig@xxxxxxx>
Subject: Re: reservation questions

 

sorry, I have some mistake in previous thread, correct it as followings.

 

1) considering below sequence:

 

call reservation_object_add_shared_fence,

now assume old->shared_count is now 3

call reservation_object_add_shared_fence,

now assume old->shared_count is now 4,

 

call reservation_object_reserve_shared,

now obj->staged is new allocated, and its shared_max = 8, but not

used by far.

 

call reservation_object_add_excl_fence,

it set obj->fence->shared_count to 0, and put all shared fence from obj->fence without waiting signaling.

(this action looks inappropriate, I think at least before put all those shared fences

we should dma_wait_fence() on them to make sure they are signaled)

 

call reservation_object_reserve_shared,

this time obj->staged isn't NULL, and it is freed (nothing bad now

since obj->fence points to other place),

and obj->staged set to NULL,

 

call reservation_object_add_shared_fence,

this time should going through reservation_object_add_shared_inplace,

since old->shared_count is 1 now, during reservation_object_add_shared_inplace()

it would go through the shared list, but the fence in shared list is now wild pointer:

 

    for (i =

0; i < fobj->shared_count; ++i) {

 

        struct dma_fence *old_fence;

 

        old_fence = rcu_dereference_protected(fobj->shared[i],

 

                        reservation_object_held(obj));

 

        if (old_fence->context == fence->context &&

 

            dma_fence_is_later(fence, old_fence)) {

 

            /* memory barrier is added by write_seqcount_begin */

 

            RCU_INIT_POINTER(fobj->shared[i], fence);

 

            write_seqcount_end(&obj->seq);

 

            preempt_enable();

 

            dma_fence_put(old_fence);

 

            return;

 

        }

 

        if (!signaled &&

dma_fence_is_signaled(old_fence)) {

 

            signaled = old_fence;

 

            signaled_idx = i;

 

        }

 

    }

 

 

see that old_fence is get from fobj, and fobj is from reservation_object_get_list(obj)

in outside, which is obj->fence, and in add_excl_fence, all dma_fence in

obj->fence is already put.

 

/Monk

 

 

---

From: Liu, Monk
Sent: Tuesday, March 6, 2018 5:45:19 PM
To: dri-devel@xxxxxxxxxxxxxxxxxxxxx; Chris Wilson; Koenig, Christian
Subject: reservation questions

 

Hi Christian & Chris

 

two question regarding resv:

 

1) considering below sequence:

 

call reservation_object_add_shared_fence,

now assume old->shared_count is now 3

call reservation_object_add_shared_fence,

now assume old->shared_count is now 4,

 

call reservation_object_reserve_shared,

now obj->staged is new allocated, and its shared_max = 8, but not

used by far.

 

call reservation_object_add_excl_fence,

it set obj->fence->shared_count to 0, and put all shared fence from obj->fence without waiting signaling.

(this action looks inappropriate, I think at least before put all those shared fences

we should dma_wait_fence() on them to make sure they are signaled)

 

call reservation_object_reserve_shared,

this time obj->staged isn't NULL, and it is freed (nothing bad now

since obj->fence points to other place),

and obj->staged set to NULL,

 

call reservation_object_add_shared_fence,

this time should going through reservation_object_add_shared_inplace,

But BUG_ON(old->shared_count >= old->shared_max) will hit !

 

This looks a design flaw in reservation object, shouldn't we fix it ?

 

 

2) in add_excl_fence(), it simply set old->shared_count to 0, and put all shared fences of old

is that correct? if excl fence is really exclusively used, why we still consider both shared fence and

excl fence on wait_timeout_rcu() routine, see blew description of this routine

 

 

/**

 

* reservation_object_wait_timeout_rcu - Wait on reservation's objects

 

* shared and/or exclusive fences.

 

* @obj: the reservation object

 

* @wait_all: if true, wait on all fences, else wait on just exclusive fence

 

* @intr: if true, do interruptible wait

 

* @timeout: timeout value in jiffies or zero to return immediately

 

*

 

* RETURNS

 

* Returns -ERESTARTSYS if interrupted, 0 if the wait timed out, or

 

* greater than zer on success.

 

*/

 

long reservation_object_wait_timeout_rcu(struct reservation_object *obj,

 

                     bool wait_all,

bool intr,

 

                     unsigned

long timeout)

 

 

 

thanks

/Monk

 

 

 

 

_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel