On 4/23/2015 12:36 PM, Chris Wilson wrote:
On Thu, Apr 23, 2015 at 12:23:17PM +0100, [email protected] wrote:
From: Tvrtko Ursulin <[email protected]>
Using imported objects should not leak i915 vmas (and vms).
In practice this simulates Xorg importing fbcon and leaking (or not) one vma
per Xorg startup cycle.
v2: use low-level ioctl wrappers and bo offset to check the leak (Chris)
v3: use the flinked bo as batch (Chris)
Signed-off-by: Tvrtko Ursulin <[email protected]>
Signed-off-by: Daniele Ceraolo Spurio <[email protected]> (v2+)
Cc: Chris Wilson <[email protected]>
Cc: Tvrtko Ursulin <[email protected]>
Bikeshedding for fun aside,
Reviewed-by: Chris Wilson <[email protected]>
+static uint64_t exec_and_get_offset(int fd, uint32_t batch)
+{
+ struct drm_i915_gem_execbuffer2 execbuf;
+ struct drm_i915_gem_exec_object2 exec[1];
+ uint32_t batch_data[2] = { MI_NOOP, MI_BATCH_BUFFER_END };
uint32_t buf[2] = { MI_BATCH_BUFFER_END };
+
+ gem_write(fd, batch, 0, batch_data, sizeof(batch_data));
+
+ memset(exec, 0, sizeof(exec));
+ exec[0].handle = batch;
+ exec[0].relocation_count = 0;
+ exec[0].relocs_ptr = 0;
We just memset(0) these two, so we don't need to clear them again.
+ memset(&execbuf, 0, sizeof(execbuf));
+ execbuf.buffers_ptr = (uintptr_t)exec;
+ execbuf.buffer_count = 1;
+ execbuf.batch_len = sizeof(batch_data);
+ execbuf.flags = 0;
These two can also happily disappear.
Ok for the execbuf.flags and for the other fixes, but why remove also
execbuf.batch_len?
Thanks,
Daniele
+ gem_execbuf(fd, &execbuf);
igt_assert_neq(exec[0].offset, -1)
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
