Somewhere in the diff (800 lines) from xf86-video-ati-6.7.193 to xf86-video-ati-6.7.194 resume from S3 was broken on my "ATI Technologies Inc M24GL [Mobility FireGL V3200]". The system seems to crash on resume (everytime), however with 6.7.193 it works every time.
Can you use git bisect to track down what commit caused the problem?
Ok i found the faulty patch through git bisect: RADEON: fix up LVDS handling for r3xx and newer b27135bce8b41d69290613b440a338b0a7fe0200 reverting that patch fixes the problem.
does reverting just this part fix it? - save->lvds_pll_cntl = info->SavedReg.lvds_pll_cntl; + save->lvds_pll_cntl = (info->SavedReg.lvds_pll_cntl | + RADEON_LVDS_PLL_EN); + + save->lvds_pll_cntl &= ~RADEON_LVDS_PLL_RESET; How about just removing the RADEON_LVDS_PLL_EN or the RADEON_LVDS_PLL_RESET line?
No that seemd not to help. At least this patch (against xf86-video-ati-6.7.194): --- a/src/radeon_output.c +++ b/src/radeon_output.c @@ -812,10 +812,7 @@ static void RADEONInitLVDSRegisters(xf86OutputPtr output, RADEONSavePtr save, ScrnInfoPtr pScrn = output->scrn; RADEONInfoPtr info = RADEONPTR(pScrn); - save->lvds_pll_cntl = (info->SavedReg.lvds_pll_cntl | - RADEON_LVDS_PLL_EN); - - save->lvds_pll_cntl &= ~RADEON_LVDS_PLL_RESET; + save->lvds_pll_cntl = info->SavedReg.lvds_pll_cntl; save->lvds_gen_cntl = info->SavedReg.lvds_gen_cntl; save->lvds_gen_cntl |= RADEON_LVDS_DISPLAY_DIS; did not solve the problem.
how about just reverting this part? - /*OUTREG(RADEON_LVDS_PLL_CNTL, restore->lvds_pll_cntl); - OUTREG(RADEON_BIOS_4_SCRATCH, restore->bios_4_scratch); + OUTREG(RADEON_LVDS_PLL_CNTL, restore->lvds_pll_cntl); + /*OUTREG(RADEON_BIOS_4_SCRATCH, restore->bios_4_scratch); the LVDS PLL must need some special reset sequence.
This did indeed help. However i was not able to test that with todays HEAD. The revert-patch could not be merged with current HEAD. Sorry for the late answer.
Created attachment 12784 [details] [review] Attempt to fix the resume problem for current HEAD. It still does not work with the current master (6f080d00e6f4f84d5e0d6b4eff302bf42c230e81). I traked the problem down to the same codeline introduced by the initial ofending patch. I have no clue what it does ;) I simply know that applying the attached patch to the current HEAD solves the problem.
*** Bug 14070 has been marked as a duplicate of this bug. ***
I can confirm that the patch from comment #7 fixes the problem for me as well (with the current master).
I've committed a workaround for now.
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.