[arch-general] Persistent black screen of death issues with radeon graphics
I am at a loss as to what the hell is going on with my laptop. It's a HP Pavillion with the dual AMD graphics (7520G + 7500M), and I will never again buy a laptop containing components from either manufacturer. More often than not when I boot, when the kernel tries to do modesetting the screen will just turn off. I can SSH in, so it's clearly booting up, but the only hint that anything unusual is happening in the logs is this error message:
[ 28.879834] radeon 0000:01:00.0: No connectors reported connected with modes [ 28.879842] [drm] Cannot find any crtc or sizes - going 1024x768 [ 28.881975] [drm] fb mappable at 0xE0474000 [ 28.881978] [drm] vram apper at 0xE0000000 [ 28.881980] [drm] size 3145728 [ 28.881982] [drm] fb depth is 24 [ 28.881985] [drm] pitch is 4096 [ 28.882332] radeon 0000:01:00.0: fb1: radeondrmfb frame buffer device [ 28.997034] [drm] Initialized radeon 2.40.0 20080528 for 0000:01:00.0 on minor 1
I feel like I've tried almost everything: - Upgrading and downgrading kernels (I'm currently using 3.17.1) - Various combinations of kernel boot parameters (fbcon=map:0, fbcon=map:1, radeon.modeset=1, radeon.modeset=0, etc) - Adding lines to /etc/X11/xorg.conf.d/20-radeon.conf that tell it which PCI device to map to. - Uninstalling and reinstalling the drivers I used to use catalyst because that was the only driver that worked, but as it doesn't support KMS, I couldn't use GNOME anymore, so switching back to that isn't an option. Searching for the problem doesn't seem to find anything relevant; there's a few forum posts by Ubuntu users but mostly tumbleweed. The bug tracker on freedesktop.org is awkward to navigate so I have no idea if what I'm experiencing has occurred to anyone else. Even just knowing where to go to report a bug or a more appropriate mailing list that could point me in the right direction would be helpful at this stage. The most frustrating bit is that this occurs almost at random. I could use the laptop all day at the library, no issues; turn it off (because suspend is broken on radeon), come home and no screen for what feels like a thousand reboots.
On 26-11-14 12:00, Robbie Smith wrote: > I am at a loss as to what the hell is going on with my laptop. It's a HP > Pavillion with the dual AMD graphics (7520G + 7500M), and I will never > again buy a laptop containing components from either manufacturer. > > More often than not when I boot, when the kernel tries to do modesetting > the screen will just turn off. I can SSH in, so it's clearly booting up, > but the only hint that anything unusual is happening in the logs is this > error message: > >> [ 28.879834] radeon 0000:01:00.0: No connectors reported connected with modes >> [ 28.879842] [drm] Cannot find any crtc or sizes - going 1024x768 >> [ 28.881975] [drm] fb mappable at 0xE0474000 >> [ 28.881978] [drm] vram apper at 0xE0000000 >> [ 28.881980] [drm] size 3145728 >> [ 28.881982] [drm] fb depth is 24 >> [ 28.881985] [drm] pitch is 4096 >> [ 28.882332] radeon 0000:01:00.0: fb1: radeondrmfb frame buffer device >> [ 28.997034] [drm] Initialized radeon 2.40.0 20080528 for 0000:01:00.0 on minor 1 > I feel like I've tried almost everything: > - Upgrading and downgrading kernels (I'm currently using 3.17.1) > - Various combinations of kernel boot parameters (fbcon=map:0, > fbcon=map:1, radeon.modeset=1, radeon.modeset=0, etc) > - Adding lines to /etc/X11/xorg.conf.d/20-radeon.conf that tell it which > PCI device to map to. > - Uninstalling and reinstalling the drivers > > I used to use catalyst because that was the only driver that worked, but > as it doesn't support KMS, I couldn't use GNOME anymore, so switching > back to that isn't an option. > > Searching for the problem doesn't seem to find anything relevant; > there's a few forum posts by Ubuntu users but mostly tumbleweed. The bug > tracker on freedesktop.org is awkward to navigate so I have no idea if > what I'm experiencing has occurred to anyone else. Even just knowing > where to go to report a bug or a more appropriate mailing list that > could point me in the right direction would be helpful at this stage. > > The most frustrating bit is that this occurs almost at random. I could > use the laptop all day at the library, no issues; turn it off (because > suspend is broken on radeon), come home and no screen for what feels > like a thousand reboots. Robbie Smith, This sounds a bit like a problem a friend of mine had with a HP Pavillion DV7 . I assume you have an amd APU + discrete videocard combination ? (lspci -k output would help). some questions / comments : are you using latest bios/Uefi firmware ? Are you sure catalyst was removed completely ? the entire dmesg output from a good and bad boot to console would also help. By comparing them we might find clues about the cause of your problems. Lone_Wolf
I had a similar issue on my laptop on very rare occasions. Make sure it's not just a setting of 0 on your /sys/class/backlight/*/brightness... Otherwise you might have to set it on the kernel command line or in /etc/modprobe.*. Cheers! mar77i
On Wed, 2014-11-26 at 15:33 +0100, LoneVVolf wrote: > > On 26-11-14 12:00, Robbie Smith wrote: > > I am at a loss as to what the hell is going on with my laptop. It's > a HP > > Pavillion with the dual AMD graphics (7520G + 7500M), and I will > never > > again buy a laptop containing components from either manufacturer. > > > > More often than not when I boot, when the kernel tries to do > modesetting > > the screen will just turn off. I can SSH in, so it's clearly > booting up, > > but the only hint that anything unusual is happening in the logs is > this > > error message: > > > >> [ 28.879834] radeon 0000:01:00.0: No connectors reported > connected with modes > >> [ 28.879842] [drm] Cannot find any crtc or sizes - going > 1024x768 > >> [ 28.881975] [drm] fb mappable at 0xE0474000 > >> [ 28.881978] [drm] vram apper at 0xE0000000 > >> [ 28.881980] [drm] size 3145728 > >> [ 28.881982] [drm] fb depth is 24 > >> [ 28.881985] [drm] pitch is 4096 > >> [ 28.882332] radeon 0000:01:00.0: fb1: radeondrmfb frame buffer > device > >> [ 28.997034] [drm] Initialized radeon 2.40.0 20080528 for > 0000:01:00.0 on minor 1 > > I feel like I've tried almost everything: > > - Upgrading and downgrading kernels (I'm currently using 3.17.1) > > - Various combinations of kernel boot parameters (fbcon=map:0, > > fbcon=map:1, radeon.modeset=1, radeon.modeset=0, etc) > > - Adding lines to /etc/X11/xorg.conf.d/20-radeon.conf that tell it > which > > PCI device to map to. > > - Uninstalling and reinstalling the drivers > > > > I used to use catalyst because that was the only driver that > worked, but > > as it doesn't support KMS, I couldn't use GNOME anymore, so > switching > > back to that isn't an option. > > > > Searching for the problem doesn't seem to find anything relevant; > > there's a few forum posts by Ubuntu users but mostly tumbleweed. > The bug > > tracker on freedesktop.org is awkward to navigate so I have no idea > if > > what I'm experiencing has occurred to anyone else. Even just > knowing > > where to go to report a bug or a more appropriate mailing list that > > could point me in the right direction would be helpful at this > stage. > > > > The most frustrating bit is that this occurs almost at random. I > could > > use the laptop all day at the library, no issues; turn it off > (because > > suspend is broken on radeon), come home and no screen for what > feels > > like a thousand reboots. > Robbie Smith, > > This sounds a bit like a problem a friend of mine had with a HP > Pavillion DV7 . > I assume you have an amd APU + discrete videocard combination ? > (lspci -k output would help). > > some questions / comments : > > are you using latest bios/Uefi firmware ? > > Are you sure catalyst was removed completely ? > > the entire dmesg output from a good and bad boot to console would > also help. > By comparing them we might find clues about the cause of your > problems. > > Lone_Wolf The laptop has the APU + discrete videocard setup. The laptop came with Windows 7 and the BIOS is some funky not-UEFI thing that has a UEFI application loader. It feels like a hack that HP have come up with to coerce Windows to boot on a UEFI system with MBR disks, or something. The latest update to it came out six months before I bought the laptop, so I'm pretty sure it is the latest firmware. I did a fresh reinstall about a month ago, and went straight to the open-source drivers because I'd finally gotten them working. Catalyst was never installed this time around. The outputs from dmesg and lspci can both be found here: https://gist.github.com/zoqaeski/c2ac50a1fc3c97614e66 The dmesg is from a "good" boot, but it still has that warning about no connectors reported with modes. Robbie
On Wed, 2014-11-26 at 15:33 +0100, LoneVVolf wrote: > On 26-11-14 12:00, Robbie Smith wrote: > > I am at a loss as to what the hell is going on with my laptop. It's a HP > > Pavillion with the dual AMD graphics (7520G + 7500M), and I will never > > again buy a laptop containing components from either manufacturer. > > > > More often than not when I boot, when the kernel tries to do modesetting > > the screen will just turn off. I can SSH in, so it's clearly booting up, > > but the only hint that anything unusual is happening in the logs is this > > error message: > > > >> [ 28.879834] radeon 0000:01:00.0: No connectors reported connected with modes > >> [ 28.879842] [drm] Cannot find any crtc or sizes - going 1024x768 > >> [ 28.881975] [drm] fb mappable at 0xE0474000 > >> [ 28.881978] [drm] vram apper at 0xE0000000 > >> [ 28.881980] [drm] size 3145728 > >> [ 28.881982] [drm] fb depth is 24 > >> [ 28.881985] [drm] pitch is 4096 > >> [ 28.882332] radeon 0000:01:00.0: fb1: radeondrmfb frame buffer device > >> [ 28.997034] [drm] Initialized radeon 2.40.0 20080528 for 0000:01:00.0 on minor 1 > > I feel like I've tried almost everything: > > - Upgrading and downgrading kernels (I'm currently using 3.17.1) > > - Various combinations of kernel boot parameters (fbcon=map:0, > > fbcon=map:1, radeon.modeset=1, radeon.modeset=0, etc) > > - Adding lines to /etc/X11/xorg.conf.d/20-radeon.conf that tell it which > > PCI device to map to. > > - Uninstalling and reinstalling the drivers > > > > I used to use catalyst because that was the only driver that worked, but > > as it doesn't support KMS, I couldn't use GNOME anymore, so switching > > back to that isn't an option. > > > > Searching for the problem doesn't seem to find anything relevant; > > there's a few forum posts by Ubuntu users but mostly tumbleweed. The bug > > tracker on freedesktop.org is awkward to navigate so I have no idea if > > what I'm experiencing has occurred to anyone else. Even just knowing > > where to go to report a bug or a more appropriate mailing list that > > could point me in the right direction would be helpful at this stage. > > > > The most frustrating bit is that this occurs almost at random. I could > > use the laptop all day at the library, no issues; turn it off (because > > suspend is broken on radeon), come home and no screen for what feels > > like a thousand reboots. > Robbie Smith, > > This sounds a bit like a problem a friend of mine had with a HP > Pavillion DV7 . > I assume you have an amd APU + discrete videocard combination ? > (lspci -k output would help). > > some questions / comments : > > are you using latest bios/Uefi firmware ? > > Are you sure catalyst was removed completely ? > > the entire dmesg output from a good and bad boot to console would also help. > By comparing them we might find clues about the cause of your problems. > > Lone_Wolf I've updated the gist to have a bad dmesg; I did a diff of them locally and the only changes are the time stamps, and that things are initialised in a different order (like network cards and etc).
On 26-11-2014 11:00, Robbie Smith wrote:
I am at a loss as to what the hell is going on with my laptop. It's a HP Pavillion with the dual AMD graphics (7520G + 7500M), and I will never again buy a laptop containing components from either manufacturer.
More often than not when I boot, when the kernel tries to do modesetting the screen will just turn off. I can SSH in, so it's clearly booting up, but the only hint that anything unusual is happening in the logs is this error message:
[ 28.879834] radeon 0000:01:00.0: No connectors reported connected with modes [ 28.879842] [drm] Cannot find any crtc or sizes - going 1024x768 [ 28.881975] [drm] fb mappable at 0xE0474000 [ 28.881978] [drm] vram apper at 0xE0000000 [ 28.881980] [drm] size 3145728 [ 28.881982] [drm] fb depth is 24 [ 28.881985] [drm] pitch is 4096 [ 28.882332] radeon 0000:01:00.0: fb1: radeondrmfb frame buffer device [ 28.997034] [drm] Initialized radeon 2.40.0 20080528 for 0000:01:00.0 on minor 1
I feel like I've tried almost everything: - Upgrading and downgrading kernels (I'm currently using 3.17.1) - Various combinations of kernel boot parameters (fbcon=map:0, fbcon=map:1, radeon.modeset=1, radeon.modeset=0, etc) - Adding lines to /etc/X11/xorg.conf.d/20-radeon.conf that tell it which PCI device to map to. - Uninstalling and reinstalling the drivers
I used to use catalyst because that was the only driver that worked, but as it doesn't support KMS, I couldn't use GNOME anymore, so switching back to that isn't an option.
Searching for the problem doesn't seem to find anything relevant; there's a few forum posts by Ubuntu users but mostly tumbleweed. The bug tracker on freedesktop.org is awkward to navigate so I have no idea if what I'm experiencing has occurred to anyone else. Even just knowing where to go to report a bug or a more appropriate mailing list that could point me in the right direction would be helpful at this stage.
The most frustrating bit is that this occurs almost at random. I could use the laptop all day at the library, no issues; turn it off (because suspend is broken on radeon), come home and no screen for what feels like a thousand reboots.
You seem to be doing late KMS, try doing early KMS. I've had trouble before with late KMS and ever since I started using early KMS the problems are gone. https://wiki.archlinux.org/index.php/kernel_mode_setting#Early_KMS_start -- Mauro Santos
participants (4)
-
LoneVVolf
-
Martti Kühne
-
Mauro Santos
-
Robbie Smith