Announcement

Collapse
No announcement yet.

xf86-video-ati issues with R5xx (M56P, X1600, MacBook Pro)

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • xf86-video-ati issues with R5xx (M56P, X1600, MacBook Pro)

    Hi,

    I've been unable to switch to a VT with the radeon driver ever since it supports the R5xx. No such issue is present with radeonhd.
    Today I discovered that disabling page flipping solves it.

    With PageFlip disabled, I can switch between VTs and X without problems and the previous font corruption problems I had are nearly gone (they do still happen from time to time).

    I'm currently using EXA, with drm also compiled from git, although I'm still on xorg-server 1.3 and mesa 6.5.2. Besides, I'm not using any composite manager although I have it enabled (no noticeable difference whether I enable or disable it).

    TextureVideo seems to work but I get lots of:
    Code:
    (EE) RADEON(0): ADVANCE_RING count != expected (46 vs 44) at radeon_textured_videofuncs.c:475
    reported whenever I play a video over xv, and, from there on, X is much much slower.

    The only options I have enabled on xorg.conf are EnablePageFlip to off, AccelMethod to EXA and SWcursor to on (sometimes my cursor got corrupted and there was no way to fix it besides rebooting).

    Accelerated EXA DFS is also working.
    Font rendering had never been this fast on this computer

    I also get quite a few:
    Code:
    (II) RADEON(0): Delay xxx usec
    with xxx being mostly 1000.

    Furthermore, I got so many DEBUG outputs from CailReadATIRegister & CailWriteATIRegister on X startup that I just commented them out on the source. They made my Xorg.0.log grow by a 10x factor!!
    It just looked like complete garbage since the value were always the same, 10002 or 20002, in a 20%-80% relation (approximately).

    The only other line worth mentioning from the X's logs is:
    Code:
    (WW) RADEON(0): Failed to set up write-combining range (0x80000000,0x10000000)

  • #2
    I've seen that the newest ati.git commit disables EnablePageFlip for all R500+ hardware until there's proper DRM support, so there should be no more problems for other users...

    And I can confirm these debug outputs in the Xorg.0.log

    Comment

    Working...
    X