Bug 12596

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.