Quoting Matthew Auld (2020-10-13 13:11:39)
> See if can make something explode with too large batch (1ULL << 32),
> while also making the batch_len implicit. We should also try each engine
> since batch_len seems to have different interactions related to that.
>
> Signed-off-by: Matthew Auld <matthew.auld@xxxxxxxxx>
> ---
> tests/i915/gem_exec_params.c | 31 ++++++++++++++++++++++++++++++-
> 1 file changed, 30 insertions(+), 1 deletion(-)
>
> diff --git a/tests/i915/gem_exec_params.c b/tests/i915/gem_exec_params.c
> index f8a94074..e00bbd04 100644
> --- a/tests/i915/gem_exec_params.c
> +++ b/tests/i915/gem_exec_params.c
> @@ -273,7 +273,7 @@ static void mmapped(int i915)
> gem_close(i915, buf);
> }
>
> -static uint32_t batch_create_size(int fd, uint32_t size)
> +static uint32_t batch_create_size(int fd, uint64_t size)
> {
> const uint32_t bbe = MI_BATCH_BUFFER_END;
> uint32_t handle;
> @@ -317,6 +317,32 @@ static void test_invalid_batch_start(int fd)
> gem_close(fd, exec.handle);
> }
>
> +static void test_larger_than_life_batch(int fd)
> +{
> + uint64_t size = 1ULL << 32; /* batch_len is __u32 as per the ABI */
> + struct drm_i915_gem_exec_object2 exec = {
> + .handle = batch_create_size(fd, size),
> + };
> + struct drm_i915_gem_execbuffer2 execbuf = {
> + .buffers_ptr = to_user_pointer(&exec),
> + .buffer_count = 1,
> + };
Needs intel_require_memory(2, size, CHECK_RAM)
> +
> + /*
> + * batch_len seems like it can have different interaction depending on
> + * the engine and HW.
> + */
> + for_each_engine(e, fd) {
> + execbuf.flags = eb_ring(e);
> + /* Make the batch_len implicit */
> + __gem_execbuf(fd, &execbuf);
Expect success.
> +
> + gem_sync(fd, exec.handle);
?
If you just want to be sure the system is idle, put it after the loop.
Otherwise it doesn't/shouldn't affect the interpretation of the params
(if paranoid, do both, though to ensure it is active at the time, you
would need a spinner).
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
