Bug 2377 - Polygon stipple broken.
Summary: Polygon stipple broken.
Status: NEW
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/r200 (show other bugs)
Version: 6.4
Hardware: x86 (IA32) Linux (All)
: high normal
Assignee: Default DRI bug account
QA Contact:
Keywords: regression
Depends on:
Reported: 2005-01-25 00:11 UTC by Jacek Rosik
Modified: 2013-01-17 19:14 UTC (History)
0 users

See Also:
i915 platform:
i915 features:

r200 (2.95 KB, image/png)
2013-01-11 15:29 UTC, smoki
swrast (2.95 KB, image/png)
2013-01-11 15:31 UTC, smoki
r200 render (3.04 KB, image/png)
2013-01-11 15:35 UTC, smoki
ums and mesa 7.5.2 r200 render (1.42 KB, image/png)
2013-01-17 19:14 UTC, smoki

Description Jacek Rosik 2005-01-25 00:11:50 UTC
progs/redbook/polys demo from Mesa source tree. And pattern displayed by it is
diferent from pattern sisplayed by indirect rendederer and renderers form
xorg-x11-6.8.1,  mesa-6.2.1  or fglrx-8.8.5 on 9800SE. I have tested it on
FireGL 8800 and 9200 PCI.

Another issue is that the stippled polygon is'nt visible util you cover it
partiali with otherwindow. But this is probalby related to something else.
Comment 1 Laurent carlier 2006-08-09 07:05:14 UTC
Polygon Stipple still broken under ubuntu dapper 6.06 with radeon 9000 (Mesa 
DRI R200 20041207 AGP 1X TCL and 1.3 Mesa 6.4.1). Line Stipple work fine.

Work fine with nvidia closed-source driver.
Comment 2 Alex Deucher 2013-01-11 14:33:55 UTC
Is this still an issues with KMS and mesa from git?
Comment 3 smoki 2013-01-11 15:29:07 UTC
Created attachment 72860 [details]
Comment 4 smoki 2013-01-11 15:31:07 UTC
Created attachment 72861 [details]

 Curent mesa git master - still render diferently.
Comment 5 smoki 2013-01-11 15:34:37 UTC
Comment on attachment 72860 [details]

 oh no wrong image, sorry :)
Comment 6 smoki 2013-01-11 15:35:43 UTC
Created attachment 72862 [details]
r200 render
Comment 7 Roland Scheidegger 2013-01-17 17:44:09 UTC
Looks to me like the offsets for the stipple pattern are wrong.
These are set in r200UpdateViewportOffset - which has no callers. I noted this in bug 57842, and tried to address it there though the patch probably wasn't really right in any case, but you could try that as a starting point.
(I'm not sure though if actually we don't also need to take viewport into account for the offsets? Maybe not though. I'm pretty sure though stippling also additionally would be broken for fbo's (but certainly should go unnoticed by this test), as the code does as the comment says, always invert the stippling pattern, and the y offset is also always calculated taking the window height into account, and I strongly suspect both aren't right for fbos.)
Comment 8 smoki 2013-01-17 18:20:14 UTC
 I was tested your patch from bug 57842 when you posted it there, but as i remember many random mesa demos i tried started to segfault with it.
Comment 9 smoki 2013-01-17 19:14:05 UTC
Created attachment 73194 [details]
ums and mesa 7.5.2 r200 render

 But overal, it is not quite correct but polys renders better now then it was with UMS.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.