Bug 15156

Created attachment 24753 [details]
dmesg 2.6.31.9

Marked as a regression.

If you disable the fbdev driver in config, does the machine boot OK?

what exactly i should do?

nomodeset?

ah you mean .config? let me attach it first

Created attachment 24754 [details]
config-2.6.32.6

here some FB_ keys, which one i should to disable?

Well I was trying to work out whether it was an fbdev problem or a DRM problem.

As you have CONFIG_FB=y and CONFIG_DRM=m then I guess it's fbdev.  I'll see if I can wake up an fbdev developer...

Gack, I can't see which driver you're using.  Please attach a dmesg from the 2.6.31 kernel?

Created attachment 24755 [details]
config-2.6.31.9

If you are using KMS for radeon please change the following config:
CONFIG_FB_VGA16=n
CONFIG_FB_VESA=n
CONFIG_FB_EFI=n
CONFIG_FB_RADEON=n
CONFIG_FB_RADEON_I2C=n
CONFIG_FB_RADEON_BACKLIGHT=n

Basicly disable all FB driver, KMS is already providing an FB driver. Do you have a macbook pro ? (i can't watch youtube video here).

i did:

CONFIG_FB_VGA16=n
CONFIG_FB_VESA=n
CONFIG_FB_EFI=n
CONFIG_FB_RADEON=n
CONFIG_FB_RADEON_I2C=n
CONFIG_FB_RADEON_BACKLIGHT=n


the same. yes i have a macbook pro.

small summary:

-2.6.31
  kms-boot - ok
  kms-suspend - ok
  ums-boot - ok
  ums-suspend - fail
  fb-boot  - ok
  fb-suspend - fail

-2.6.32+
  kms-boot - fail
  kms-suspend - unknown
  ums-boot - ok
  ums-suspend - fail
  fb-boot  - ok
  fb-suspend - fail

Created attachment 24885 [details]
config-2.6.32.7

same for Linux 2.6.33-rc7 (hang on boot with kms). suspend/resume on fb-boot ok/fail.

2.6.33-rc8 dont hang anymore.

but 2.6.33-rc8 has other bug, and i really can describe it. youtube video: http://www.youtube.com/watch?v=a-ogf8whmC0

Can you file another bugreport for the second issue with an appropriate subject if this is still present in current kernels? 

('parts of the screen flimmer' or something like that could probably be used to describe the symptoms)

Regards,
Flo