Bug 116646

Summary:

[Cairo] Fail to draw box-shadow if origin is very big and scale is not 1.

Product:

WebKit

Reporter:

Yonggeol Jung <yg48.jung>

Component:

Layout and Rendering

Assignee:

Nobody <webkit-unassigned>

Status:

UNCONFIRMED

Severity:

Normal

CC:

bugs-noreply, cdumez, commit-queue, d-r, hyunki.baik, igor.oliveira, laszlo.gombos, mrobinson

Priority:

Version:

528+ (Nightly build)

Hardware:

Other

OS:

Linux

Attachments:

Description	Flags
testcase	none
Patch	none
Patch	beidson: review-
callstack	none

Yonggeol Jung

Reported 2013-05-22 19:37:49 PDT

When try to draw box-shadow to big origin and page scale factor is not 1, it fails to draw.

Attachments
testcase (550 bytes, text/html) 2013-05-22 19:43 PDT, Yonggeol Jung	no flags	Details
Patch (1.96 KB, patch) 2013-05-22 19:45 PDT, Yonggeol Jung	no flags	Details Formatted Diff Diff
Patch (1.96 KB, patch) 2013-05-22 19:47 PDT, Yonggeol Jung	beidson: review-	Details Formatted Diff Diff
callstack (50.23 KB, application/octet-stream) 2013-05-30 19:23 PDT, Yonggeol Jung	no flags	Details
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.

Yonggeol Jung

Comment 1 2013-05-22 19:43:17 PDT

Created attachment 202636 [details] testcase After clicking "Scroll" button, you can see white circle with blue shadow-box. But webkit-efl port doesn't display this.

Yonggeol Jung

Comment 2 2013-05-22 19:45:53 PDT

Created attachment 202637 [details] Patch

Yonggeol Jung

Comment 3 2013-05-22 19:47:04 PDT

Created attachment 202638 [details] Patch

Chris Dumez

Comment 4 2013-05-23 03:59:21 PDT

Comment on attachment 202638 [details] Patch You probably need a test for it (or rebase existing test if already covered).

Yonggeol Jung

Comment 5 2013-05-23 23:16:03 PDT

It's strange. I made some test case for this issue. When running test case using Minibrowser, it's not showing properly but result image file of layout-test shows fine. Maybe the problem is related to repainting. I'll check more.

Martin Robinson

Comment 6 2013-05-28 14:45:17 PDT

Comment on attachment 202638 [details] Patch Can you show the stack trace of the failing call? It probably makes sense to workaround this issue in drawRectShadow, so that it doesn't need to be fixed in every place that calls the method.

Yonggeol Jung

Comment 7 2013-05-30 19:23:57 PDT

Created attachment 203406 [details] callstack This is call stack for the test-case which I uploaded. I caught when "_cairo_matrix_to_pixman_matrix function" returns CAIRO_STATUS_INVALID_MATRIX. // cairo-matrix.c static cairo_status_t_cairo_matrix_to_pixman_matrix () { ... if (_cairo_matrix_has_unity_scale (matrix)) return CAIRO_STATUS_SUCCESS; if (unlikely (fabs (matrix->xx) > PIXMAN_MAX_INT || fabs (matrix->xy) > PIXMAN_MAX_INT || fabs (matrix->x0) > PIXMAN_MAX_INT || fabs (matrix->yx) > PIXMAN_MAX_INT || fabs (matrix->yy) > PIXMAN_MAX_INT || fabs (matrix->y0) > PIXMAN_MAX_INT)) { return _cairo_error (CAIRO_STATUS_INVALID_MATRIX); // => here } .... }

Brady Eidson

Comment 8 2016-05-24 21:38:32 PDT

Comment on attachment 202638 [details] Patch r- to clear stale patches from the review queue

Note You need to log in before you can comment on or make changes to this bug.