[Intel-gfx] [PATCH v2 2/2] tests/kms_flip: test a fb backed by a bo too big/small for its own good

oscar.mateo at intel.com oscar.mateo at intel.com
Fri May 16 15:07:12 CEST 2014


From: Oscar Mateo <oscar.mateo at intel.com>

This is a "review by igt test" for a bug located in
i915_gem_object_pin_to_display_plane and fixed by:

commit 392013bdd4b6128795e33c84bd6d6d3fd66ff0a3
Author: Oscar Mateo <oscar.mateo at intel.com>
Date:   Fri May 16 11:23:12 2014 +0100

    drm/i915: Gracefully handle obj not bound to GGTT in is_pin_display

    Otherwise, we do a NULL pointer dereference.

    I've seen this happen while handling an error in
    i915_gem_object_pin_to_display_plane():

    If i915_gem_object_set_cache_level() fails, we call is_pin_display()
    to handle the error. At this point, the object is still not pinned
    to GGTT and maybe not even bound, so we have to check before we
    dereference its GGTT vma.

    v2: Chris Wilson says restoring the old value is easier, but that
    is_pin_display is useful as a theory of operation. Take the solomonic
    decision: at least this way is_pin_display is a little more robust
    (until Chris can kill it off).

v2: Avoid code duplication by using igt_create_fb_with_bo_size() as
requested by Ville Syrjälä (original author of the "too big" test idea).

Signed-off-by: Oscar Mateo <oscar.mateo at intel.com>
---
 tests/kms_flip.c | 22 ++++++++++++++++++----
 1 file changed, 18 insertions(+), 4 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index bb4f71d..f2ec9ef 100644
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -74,6 +74,7 @@
 #define TEST_RPM		(1 << 25)
 #define TEST_SUSPEND		(1 << 26)
 #define TEST_TS_CONT		(1 << 27)
+#define TEST_BO_TOOBIG		(1 << 28)
 
 #define EVENT_FLIP		(1 << 0)
 #define EVENT_VBLANK		(1 << 1)
@@ -1213,6 +1214,7 @@ static void run_test_on_crtc_set(struct test_output *o, int *crtc_idxs,
 {
 	char test_name[128];
 	unsigned elapsed;
+	unsigned bo_size = 0;
 	bool tiled;
 	int i;
 
@@ -1249,12 +1251,17 @@ static void run_test_on_crtc_set(struct test_output *o, int *crtc_idxs,
 	if (o->flags & TEST_FENCE_STRESS)
 		tiled = true;
 
+	/* 256 MB is usually the maximum mappable aperture,
+	 * (make it 4x times that to ensure failure) */
+	if (o->flags & TEST_BO_TOOBIG)
+		bo_size = 4*256*1024*1024;
+
 	o->fb_ids[0] = igt_create_fb(drm_fd, o->fb_width, o->fb_height,
 					 igt_bpp_depth_to_drm_format(o->bpp, o->depth),
 					 tiled, &o->fb_info[0]);
-	o->fb_ids[1] = igt_create_fb(drm_fd, o->fb_width, o->fb_height,
+	o->fb_ids[1] = igt_create_fb_with_bo_size(drm_fd, o->fb_width, o->fb_height,
 					 igt_bpp_depth_to_drm_format(o->bpp, o->depth),
-					 tiled, &o->fb_info[1]);
+					 tiled, &o->fb_info[1], bo_size);
 	o->fb_ids[2] = igt_create_fb(drm_fd, o->fb_width, o->fb_height,
 					 igt_bpp_depth_to_drm_format(o->bpp, o->depth),
 					 true, &o->fb_info[2]);
@@ -1264,7 +1271,8 @@ static void run_test_on_crtc_set(struct test_output *o, int *crtc_idxs,
 		igt_require(o->fb_ids[2]);
 
 	paint_flip_mode(&o->fb_info[0], false);
-	paint_flip_mode(&o->fb_info[1], true);
+	if (!(o->flags & TEST_BO_TOOBIG))
+		paint_flip_mode(&o->fb_info[1], true);
 	if (o->fb_ids[2])
 		paint_flip_mode(&o->fb_info[2], true);
 
@@ -1288,10 +1296,15 @@ static void run_test_on_crtc_set(struct test_output *o, int *crtc_idxs,
 	if (o->flags & TEST_CHECK_TS)
 		sleep(1);
 
-	igt_assert(do_page_flip(o, o->fb_ids[1], true) == 0);
+	if (o->flags & TEST_BO_TOOBIG) {
+		igt_assert(do_page_flip(o, o->fb_ids[1], true) == -E2BIG);
+		goto out;
+	} else
+		igt_assert(do_page_flip(o, o->fb_ids[1], true) == 0);
 	wait_for_events(o);
 
 	o->current_fb_id = 1;
+
 	if (o->flags & TEST_FLIP)
 		o->flip_state.seq_step = 1;
 	else
@@ -1543,6 +1556,7 @@ int main(int argc, char **argv)
 		{ 10, TEST_VBLANK | TEST_DPMS | TEST_SUSPEND | TEST_TS_CONT, "dpms-vs-suspend" },
 		{ 0, TEST_VBLANK | TEST_MODESET | TEST_RPM | TEST_TS_CONT, "modeset-vs-rpm" },
 		{ 0, TEST_VBLANK | TEST_MODESET | TEST_SUSPEND | TEST_TS_CONT, "modeset-vs-suspend" },
+		{ 0, TEST_BO_TOOBIG | TEST_NO_2X_OUTPUT, "bo-too-big" },
 	};
 	int i;
 
-- 
1.9.0




More information about the Intel-gfx mailing list