| Priority | medium |
|---|---|
| Bug ID | 60236 |
| Assignee | dri-devel@lists.freedesktop.org |
| Summary | corruption of text and bottom of screen in xonotic menus since 325422c49449acdd8df1eb2ca8ed81f7696c38cc |
| Severity | normal |
| Classification | Unclassified |
| OS | Linux (All) |
| Reporter | pheldens@ziggo.nl |
| Hardware | Other |
| Status | NEW |
| Version | git |
| Component | Drivers/Gallium/r600 |
| Product | Mesa |
bisected to this commit 325422c49449acdd8df1eb2ca8ed81f7696c38cc is the first bad commit commit 325422c49449acdd8df1eb2ca8ed81f7696c38cc Author: Jerome Glisse <jglisse@redhat.com> Date: Mon Jan 7 17:45:59 2013 -0500 r600g: add async for staging buffer upload v2 v2: Add virtual address to dma src/dst offset for cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> :040000 040000 4ef6e784f3acb7f21da0c5e1923810c78917d16d 55f0ce7d9793ce04d392fc7943059545053a9d79 M src
| What | Removed | Added |
|---|---|---|
| Attachment #74238 is obsolete | 1 |
(In reply to comment #1) > linux 3.8-rc6 3.7.4 had it also 325422c49449acdd8df1eb2ca8ed81f7696c38cc (r600g: add async for staging buffer upload v2) only takes affect on 3.8 kernels. Previous kernels didn't support the DMA rings. So if you are seeing the same issue on multiple kernels, it's either something else or you are seeing two issues. It may be the same issue as bug 58042.
| What | Removed | Added |
|---|---|---|
| Attachment #74238 mime type | text/plain | image/png |
Tested it again to be sure Linux borgir 3.7.6 does NOT have the problem, pardon the mistake.
(In reply to comment #0= ) > bisected to this commit >=20 > 325422c49449acdd8df1eb2ca8ed81f7696c38cc is the first bad commit > commit 325422c49449acdd8df1eb2ca8ed81f7696c38cc > Author: Jerome Glisse <jg= lisse@redhat.com> > Date: Mon Jan 7 17:45:59 2013 -0500 >=20 > r600g: add async for staging buffer upload v2 >=20=20=20=20=20 > v2: Add virtual address to dma src/dst offset for cayman >=20=20=20=20=20 > Signed-off-by: Jerome Glisse <jglisse@redhat.com> >=20 > :040000 040000 4ef6e784f3acb7f21da0c5e1923810c78917d16d > 55f0ce7d9793ce04d392fc7943059545053a9d79 M src Looks correct to me on rv790. I get nothing like the corruption in the screens, for me it's just the band= at the bottom of the screen and it's not always there (but seems to be reliable first test after boot from power off). I tested a couple of cycles with mesa reset just before this commit and couldn't see the problem. With mesa on this commit I can see it. Kernel is drm-fixes.
Yes for me this effect varies also if I open , say a browser with a complex webpage, and then start xonotic again the text may be garbled the 2nd start its as if video memory or some caches get corrupted
Created attachment 74300 [details] [review] Fix Can you check if that patch fix it for you ?
still has the problem with that patch noticed this in dmesg: radeon 0000:01:00.0: DMA copy src buffer too small (4295180288 18382848) [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream ! radeon 0000:01:00.0: DMA copy src buffer too small (1032192 1024000) [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !
Created attachment 74303 [details] [review] Fix v2 This v2 avoid the kernel message. Weird you still have issue i don't here with this patch.
Try also applying the patch of https://bugs.freedesktop.org/show_bug.cgi?id=60034
fix v2 ( 3.8.0-rc6 x86_64) radeon 0000:01:00.0: DMA copy src buffer too small (7516413696 18382848) [drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream ! still the xonotic garbage, though it does not happen with every try, best test 3x
just applied
> Try also applying the patch of https://bugs.freedesktop.org/show_bug.cgi?id=60034
same problem still
though it started 3x well, then when I open my browser with some pages, and
retry, the corruption in xonotic is back again..
radeon 0000:01:00.0: DMA copy src buffer too small (3221445120 18382848)
[drm:radeon_cs_ib_chunk] *ERROR* Invalid command stream !
Created attachment 74306 [details]
corruption after patches
looks a bit different, also xonotic seems to start more often right (without
corruption) than wrong with the patches applied
(In reply to comment #18) > https://bugs.freedesktop.org/attachment.cgi?id=74306 I've tried quite hard on both 64 bit and 32 bit setups and with both patches applied can no longer get any corruption. One thing I did notice is that there is a slight difference between running the sdl and glx versions, in that the sdl has a few strange chars in console - but it's the same with llvmpipe. Are you running glx or sdl? Can you get any corruption if you run it with the env LIBGL_ALWAYS_SOFTWARE=1
xonotic-linux64-glx LIBGL_ALWAYS_SOFTWARE=1 starts correctly 5x in a row
Paul do you have an up to date kernel, there are have been change to dma portion of kernel in 3.8 serie. Without my patch i got the corruption but with it i can't get any here.
Especially you should not have kernel message with my patch