[PATCH 0/1] MR8421: dxgi/tests: Fix out-of-bound in test_cursor_clipping.

[Yuxuan Shui (＠yshui)]

Missing a bound check for "modes".

[PATCH 1/1] dxgi/tests: Fix out-of-bound in test_cursor_clipping.

[Yuxuan Shui]

From: Yuxuan Shui

Missing a bound check for "modes".
---
 dlls/dxgi/tests/dxgi.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/dlls/dxgi/tests/dxgi.c b/dlls/dxgi/tests/dxgi.c
index cb8bec5c666..47905bed66d 100644
--- a/dlls/dxgi/tests/dxgi.c
+++ b/dlls/dxgi/tests/dxgi.c
@@ -6912,8 +6912,10 @@ static void test_cursor_clipping(IUnknown *device, BOOL is_d3d12)
                 if (modes[mode_idx].Width != width && modes[mode_idx].Height != height)
                     break;
             }
-            ok(modes[mode_idx].Width != width && modes[mode_idx].Height != height,
+            ok(mode_idx < mode_count && modes[mode_idx].Width != width && modes[mode_idx].Height != height,
                     "Failed to find a different mode than %ux%u.\n", width, height);
+            if (mode_idx >= mode_count)
+                continue;

ret = ClipCursor(NULL);
             ok(ret, "ClipCursor failed, error %#lx.\n", GetLastError());

[Elizabeth Figura (＠zfigura)]

I believe the intent of that check *was*, simply, "mode_idx < mode_count". We can just change the ok() to that.

We don't need an "if (...)" continue afterward, either. A test failure is a test failure.

[Yuxuan Shui (＠yshui)]

On Wed Jun 25 13:15:16 2025 +0000, Elizabeth Figura wrote:
> I believe the intent of that check *was*, simply, "mode_idx <
> mode_count". We can just change the ok() to that.
> We don't need an "if (...)" continue afterward, either. A test failure
> is a test failure.

the `if()` is needed because there are uses of `modes[mode_idx]` later, which would be out of bound.