On Fri, Sep 03, 2010 at 04:22:17PM -0400, Stephen Powell wrote: > On Fri, Sep 03, 2010 at 12:58:21AM -1000, Joel Roth wrote: > > Hello all, > > > > Here's what happening. > > > > After boot: normal console font (about 80 characters wide) > > > > After startx: small console font (maybe 132 characters?) > > > > A little later: no display in console at all (Alt-F7 > > still gets me back to X) > > > > This problem appeared soon after converting to > > dependency-based boot process using insserv. > > > > I've been using a self-compiled amd64 kernel with 32-bit userland. > > No such problems before insserv. > > > > Thanks for any possible leads. I have no idea where to > > even start. > > Yeah, that happened to me too, a while ago. It's the X driver. > Specifically, it's the KMS settings that are messing you up. > I hate KMS. It does solve certain problems for developers, > but causes other problems for users. Maybe in a few years all > these KMS-related bugs will get sorted out. But for now, they > are a pain.
Good lead! For example, we have LKML posts such as: Re: Resume problem with radeon+KMS (2.6.34-rc2 and before) http://lkml.indiana.edu/hypermail/linux/kernel/1003.2/02225.html Which hits one of my problems as well. FWIW, here is my dmesg output regarding my ATI Radeon 3100 chip (RS780MC). I wonder about the 'loading firmware' bit, which seems to fail. [ 1478.833236] [drm] Initialized drm 1.1.0 20060810 [ 1478.934354] [drm] radeon kernel modesetting enabled. [ 1478.934483] radeon 0000:01:05.0: setting latency timer to 64 [ 1478.939239] [drm] radeon: Initializing kernel modesetting. [ 1478.939410] [drm] register mmio base: 0x96300000 [ 1478.939415] [drm] register mmio size: 65536 [ 1478.945526] ATOM BIOS: Tos_SCT10AP_MC [ 1478.945576] [drm] Clocks initialized ! [ 1478.945969] [drm] Detected VRAM RAM=256M, BAR=256M [ 1478.945978] [drm] RAM width 32bits DDR [ 1478.946151] [TTM] Zone kernel: Available graphics memory: 899812 kiB. [ 1478.946186] [drm] radeon: 256M of VRAM memory ready [ 1478.946191] [drm] radeon: 512M of GTT memory ready. [ 1478.946257] [drm] radeon: irq initialized. [ 1478.946263] [drm] GART: num cpu pages 131072, num gpu pages 131072 [ 1478.948761] [drm] Loading RS780 Microcode [ 1478.949103] platform radeon_cp.0: firmware: requesting radeon/RS780_pfp.bin [ 1479.022552] platform radeon_cp.0: firmware: requesting radeon/RS780_me.bin [ 1479.047423] platform radeon_cp.0: firmware: requesting radeon/R600_rlc.bin [ 1479.057460] r600_cp: Failed to load firmware "radeon/R600_rlc.bin" [ 1479.057476] [drm:r600_startup] *ERROR* Failed to load firmware! [ 1479.057483] radeon 0000:01:05.0: disabling GPU acceleration [ 1479.057510] radeon 0000:01:05.0: ffff88006df24000 unpin not necessary [ 1479.057517] radeon 0000:01:05.0: ffff88006df24000 unpin not necessary [ 1479.057530] [drm] Enabling audio support [ 1479.057850] [drm] Radeon Display Connectors [ 1479.057855] [drm] Connector 0: .... > How to fix it depends on your chipset. In one case, I had > an nvidia card, and the upgrade switched me from the nv > driver to the nouveau driver. I could get around it by > blacklisting the nouveau driver and creating an explicit > /etc/X11/xorg.conf file that specifies > > driver = "nv" > > in the appropriate place. For other chipsets, such as Intel, > you may be able to set a module parameter like "kms=0", > or some such thing. Search the internet for your card > manufacturer, Linux, and KMS, and maybe you'll find something. > These techniques are only a temporary stopgap measure. > Eventually, there will be no option. You will either have > to use KMS or not use an X server at all. Oh joy. I expect it will be debugged in time. Regards, Joel > > -- > .''`. Stephen Powell > : :' : > `. `'` > `- > > > -- > To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org > with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org > Archive: > http://lists.debian.org/1744286447.595363.1283545337251.javamail.r...@md01.wow.synacor.com > -- Joel Roth -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100903205759.ga4...@sprite