View previous topic :: View next topic |
Author |
Message |
devsk Advocate
Joined: 24 Oct 2003 Posts: 2995 Location: Bay Area, CA
|
Posted: Wed Jan 09, 2008 5:16 pm Post subject: |
|
|
Does anybody know how the initramfs (containing the v86d) embedded inside kernel interact with the initrd specified in grub.conf? Does it override the initrd or are they exclusive? I need the initrd during boot for dmraid. |
|
Back to top |
|
|
Neo2 Apprentice
Joined: 25 Sep 2006 Posts: 224 Location: Italy
|
Posted: Wed Jan 09, 2008 7:12 pm Post subject: |
|
|
Don't worry about that. The file/directory you specify with the "Initramfs source file(s)" option in the kernel is just an "initial image" to which any other file/configuration you specified will be added. When building the initrd an initial archive is created from the files you specified in "Initramfs source file(s)". After that, the initrd gets built as usual, adding to that archive everything you need. Hence dmraid configuration and any other things that might get into the initrd won't stop working _________________ Neo2
Unofficial minimal liveCD for x86/amd64 w/reiser4+truecrypt |
|
Back to top |
|
|
devsk Advocate
Joined: 24 Oct 2003 Posts: 2995 Location: Bay Area, CA
|
Posted: Wed Jan 09, 2008 10:42 pm Post subject: |
|
|
Another thing: the website says uvesafb is in vanilla starting 2.6.24 but its not there in 2.6.24-rc6. Is it planned? or I missed it in the midst of too many config options?
Scratch that: I had to enable connector first to be able to see uvesafb. So, vanilla 2.6.24 has uvesafb. YAY! No need to wait for gentoo-sources? |
|
Back to top |
|
|
Neo2 Apprentice
Joined: 25 Sep 2006 Posts: 224 Location: Italy
|
Posted: Thu Jan 10, 2008 5:34 am Post subject: |
|
|
As of now the current stable release of gentoo-sources is based upon linux-2.6.23.9. Anyway it does include uvesafb (through the patches system if I remember well). Be careful though because reading this thread suggests that sometimes uvesafb doesn't work without any apparent reason (all setup well according to the guides but no result). _________________ Neo2
Unofficial minimal liveCD for x86/amd64 w/reiser4+truecrypt |
|
Back to top |
|
|
pholthau Guru
Joined: 27 Nov 2005 Posts: 361 Location: Bielefeld, Germany
|
Posted: Thu Jan 10, 2008 6:40 am Post subject: |
|
|
devsk wrote: |
Scratch that: I had to enable connector first to be able to see uvesafb. So, vanilla 2.6.24 has uvesafb. YAY! No need to wait for gentoo-sources? |
Yay, that's good news. Running 2.6.24-rc7 with uvesafb now (especially because CONFIG_NO_HZ is available now for amd64). _________________ The message is Feierei, alder! The message is Gude Laune, alder! [Sven] |
|
Back to top |
|
|
hoschi Advocate
Joined: 19 Jul 2003 Posts: 2517 Location: Ulm, Germany, Europe
|
Posted: Sun Jan 13, 2008 8:05 pm Post subject: |
|
|
Hello,
does "Suspend to RAM" work at any configuration with UVESAFB? I'am affraid that ACPI and UVESAFB still not work, with no special bootloader-append the picture on the screen simply doesn't appear, with "acpi_sleep=s3_bios,s3_mode" I got funny blue, green and red blocks after resume
Any idea?
A hint, I have has the same problem with the old vesafb-tng _________________ Just you and me strogg! |
|
Back to top |
|
|
jdmulloy Tux's lil' helper
Joined: 24 Dec 2004 Posts: 139 Location: Massachusetts, USA
|
Posted: Mon Jan 14, 2008 5:24 am Post subject: |
|
|
Has anyone gotten it to work with intel video chips yet. Specifically the 945GM (aka GMA 950).
I'm getting this in dmesg and no framebuffer.
Code: |
uvesafb: Getting VBE info block failed (eax=0x4f00, err=-3)
uvesafb: vbe_init() failed with -22
uvesafb: probe of uvesafb.0 failed with error -22
|
I've tried the suggestions in this thread including patching v86d. I've tried v86d-0.3.1 and v86d-0.1.3-r1 with the x86emu use flag. I've tried lower resolutions and bit depths as well ad getting rid of mtrr and ywrap options. Anyone know how to get this to work. If there is something actually wrong I'll file a bug report. _________________ Joe Mulloy | http://twitter.com/jdmulloy | Ron Paul in 2012! | 5-1-07 | Unban Playfool | Fire your "Too big to fail" bank http://moveyourmoney.info |
|
Back to top |
|
|
swimmer Veteran
Joined: 15 Jul 2002 Posts: 1330 Location: Netherlands
|
Posted: Mon Jan 14, 2008 12:14 pm Post subject: |
|
|
You're using 915resolution don't you?
Greetz
swimmer |
|
Back to top |
|
|
jdmulloy Tux's lil' helper
Joined: 24 Dec 2004 Posts: 139 Location: Massachusetts, USA
|
|
Back to top |
|
|
swimmer Veteran
Joined: 15 Jul 2002 Posts: 1330 Location: Netherlands
|
Posted: Mon Jan 14, 2008 5:24 pm Post subject: |
|
|
Hmm ok - can't say how it was for vesa-tng since I only have Intel chipset on my new notebook for which I exclusively used uvesafb. 915solution (patched though since I'm on a 965GM) works for me so far ...
HTH
swimmer |
|
Back to top |
|
|
jdmulloy Tux's lil' helper
Joined: 24 Dec 2004 Posts: 139 Location: Massachusetts, USA
|
Posted: Mon Jan 14, 2008 6:39 pm Post subject: |
|
|
swimmer wrote: | Hmm ok - can't say how it was for vesa-tng since I only have Intel chipset on my new notebook for which I exclusively used uvesafb. 915solution (patched though since I'm on a 965GM) works for me so far ...
HTH
swimmer |
I'm at work right now, but I'll try it when I get home. I read on the 915resolution website that it would be integrated into Xorg "soon" but they also listed Xorg 6.8 as a recent version so I think that's why I haven't needed it, at least for X. _________________ Joe Mulloy | http://twitter.com/jdmulloy | Ron Paul in 2012! | 5-1-07 | Unban Playfool | Fire your "Too big to fail" bank http://moveyourmoney.info
Last edited by jdmulloy on Tue Jan 15, 2008 2:48 pm; edited 1 time in total |
|
Back to top |
|
|
jdmulloy Tux's lil' helper
Joined: 24 Dec 2004 Posts: 139 Location: Massachusetts, USA
|
Posted: Tue Jan 15, 2008 7:04 am Post subject: |
|
|
I've been able to get it to work as a module when I load it once my initramfs image has handed over the boot to /sbin/init. Still can't get it to work in initramfs but at least I've got something. _________________ Joe Mulloy | http://twitter.com/jdmulloy | Ron Paul in 2012! | 5-1-07 | Unban Playfool | Fire your "Too big to fail" bank http://moveyourmoney.info |
|
Back to top |
|
|
Habbit Apprentice
Joined: 01 Sep 2007 Posts: 237 Location: 3.7137 W, 40.3873 N
|
Posted: Tue Jan 22, 2008 11:25 am Post subject: |
|
|
Even with the newest uvesafb+v86d, my video card (radeon x850xt pcie) refuses to work properly: the screen goes blank and doesn't come up until X starts (and with distorted resolutions, which proves that uvesafb does _something_ to the video card). Furthermore, if I ever exit/switch off X and try to enter a virtual terminal, the computer freezes, as it does if I compile uvesafb as a module and try to load it later instead of having it built into the kernel.
Dmesg trace: Code: | [ 0.000000] Linux version 2.6.24-rc7-habbit (root@accesodirecto) (gcc version 4.2.2 (Gentoo 4.2.2 p1.0)) #1 PREEMPT Fri Jan 11 02:10:34 CET 2008
[ 0.000000] Command line: root=/dev/sdb6 video=uvesafb:1024x768-32@60 quiet
[ 42.810516] Allocate Port Service[0000:00:0e.0:pcie03]
[ 42.811140] uvesafb: (C) 1988-2003, ATI Technologies Inc. R48001.00, R48001.00, 01.00, OEM: ATI ATOMBIOS(C) 1988-2003, ATI Technologies Inc. R48001.00, VBE v3.0
[ 42.816804] uvesafb: VBIOS/hardware supports DDC2 transfers
[ 42.940818] uvesafb: monitor limits: vf = 75 Hz, hf = 83 kHz, clk = 150 MHz
[ 42.940925] uvesafb: scrolling: redraw
[ 42.944267] uvesafb: mode switch failed (eax=0x34f, err=0). Trying again with default timings.
[ 43.345786] uvesafb: mode switch failed (eax=0x34f, err=0). Trying again with default timings.
[ 44.003124] Console: switching to colour frame buffer device 128x48
[ 44.003162] uvesafb: mode switch failed (eax=0x34f, err=0). Trying again with default timings.
[ 44.659924] uvesafb: framebuffer at 0xe0000000, mapped to 0xffffc20010100000, using 16384k, total 16384k
[ 44.659926] fb0: VESA VGA frame buffer device |
As you can see, uvesafb initialization fails but nevertheless Linux switches to to the "framebuffer" provided by uvesafb ("colour FB 128x48"). It is worth mention that this card works with vesafb (though I can't get 1680x1050, which is what I want uvesafb for) and X. This is a full 64-bit system with a 2.6.24-rc7 vanilla kernel on AMD64. Resolution/options specified to uvesafb does not matter, it always does the same. _________________
Code: | ~ $ objdump -d ./habbit_mind
90 xchg %rax, %rax
EB FD jmp $-3 |
|
|
Back to top |
|
|
Neo2 Apprentice
Joined: 25 Sep 2006 Posts: 224 Location: Italy
|
Posted: Tue Jan 22, 2008 6:51 pm Post subject: |
|
|
Okay, I recently bought a new laptop ("new" means that I've changed my 10-years-old laptop with a 4-years-old one ) and like some of you before in this post I've been dealing with correct configuration but no results. And luckily I made it through.
I made diffs and diffs between the working config on the older laptop and the newer one (I recompiled the kernel something like 15 times in a row) and I've spotted out some things that might help. I'm not sure how much and in which way they are related to uvesafb, I just know that it works because of those options (I simply tried not to touch anything not uvesafb-related).
I guess that the core of this comes down to (every comment line "//" refers to the line just above)
Code: |
General setup ->
[*] Initial RAM filesystem and RAM disk (initramfs/initrd) support
(/usr/share/v86d/initramfs) Initramfs source file(s)
(0) User ID to map to 0 (user root)
(0) Group ID to map to 0 (group root)
Device Drivers ->
<*> Connector - unified userspace <-> kernelspace linker --->
//<*> changes in --- when uvesafb is already selected
--- Connector - unified userspace <-> kernelspace linker
[*] Report process events to userspace
Device Drivers -> Character devices -->
[*] Virtual terminal
[*] Support for console on virtual terminal
[*] Support for binding and unbinding console drivers
//actually uvesafb works also without it but personally I'd recommend it enabled
Device Drivers -> Graphics support -->
<*> Backlight & LCD device support
Display device support --->
<*> Display panel/monitor support
<M/*> Lowlevel video output switch controls
<*> Support for frame buffer devices
[*] Enable firmware EDID
--- Enable Video Mode Handling Helpers (//automatically selected by uvesafb)
[ ] Enable Tile Blitting Support (//this is almost-specific for matrox cards)
--- Frame buffer hardware drivers
<*> Userspace VESA VGA Graphics support
Console display driver support --->
[*] VGA text console
[*] Enable scrollback buffer in System RAM
(128) Scrollback Buffer Size (//user's choice)
//I think that everybody of you has already reached this
//point without me pointing at it
//Next goes what I had to include to make it working
[*] Video mode selection support
<M> MDA text console (dual-headed)
//actually needed for the cards specified by the Help of this option, almost everybody of you should say M or N here)
<*> Framebuffer Console support
//I compiled it as a module, I guess what [b]really[/b] was missing
[*] Map the console to the primary display device
[*] Framebuffer console rotation
//though uvesafb works I guess it isn't needed, I didn't have for the 16th recompile :P
[*] Support for Framebuffer Console Decorations (//absolutely needed for those who want splash)
[*] Select compiled-in fonts
//actually all fonts are selected and I don't want to make a neverending list saying that below this point everything is [*] |
About the fonts, try to include them. It will cost you "just" a compile but if it doesn't work at least you'll know it isn't the fonts you haven't selected (I'm waiting for feedback from whoever will try this).
Please, ensure that uvesafb is the only framebuffer selected (I've seen the kernel defaulting to radeonfb even though uvesafb was specified on boot line). If you're worried about being able to boot again if you're getting a blank screen, before starting genkernel just do a
Code: | mv /boot/kernel-genkernel-x86-2.6.23-gentoo-r3 /boot/kernel-genkernel-x86-2.6.23-gentoo-r3_old |
so that you'll be able to edit the boot line with minimal effort (hopefully you're using GRUB, I've never tried lilo).
As of now I haven't included Bootup logos, but I guess they should work (*should* and not *will*). I'm not a fbsplash/gensplash/splash user, but my uvesafb works fine if I decide to turn it on (now I use radeonfb natively, a lot faster )
One final note on kernel compile. I usually configure my kernels manually by importing the last configuration of the previous kernel with "Load an alternate configuration file" and after configure I save them in a path of my choice with "Save an alternate configuration file", ready for restoring backups or diffing. Doing this, I've noticed that
a) when you hit "exit" from "make menuconfig" the configuration is not automatically saved to .config (the old one is left there), but I guess this is normal and how it should be (from the option "save..."). But far more interesting,
b) when I start genkernel with "genkernel --kernel-config=/path/to/myconfig --dmraid all" the config that is actually used for making the kernel is the one present in /usr/src/linux/.config and not the one I've specified by command-line. I'm not sure why this happens nor how this works, but if you want to recompile the kernel with your config and you're using the "save an alternate config file" option, I strongly suggest to manually copy over your new configuration as /usr/src/linux/.config before starting genkernel. (Hope that someone can feedback on this as well). I've noticed this behaviour on all my systems: 2 AMD64 and 3 x86 (2 laptops) and I'm currently investigating (had almost no time 'til now).
Once rebuilt, restart your kernel with the bare minimum on the boot line:
Code: | kernel /boot/kernel-genkernel-x86-2.6.23-gentoo-r3 real_root=<your_root_device> udev video=uvesafb (dmraid users should also append dmraid, of course) |
Once it boots fine (no blank screen, "fine" means a working screen) navigate into /sys/bus/platform/drivers/uvesafb/uvesafb.0 or directly cat "vbe_modes" found in this directory. Just because your mode is listed as working I guess it doesn't mean it *will* work for sure. I suggest you to do a lot of reboots, everytime switching your mode to a higher one. Start with 640x480 and then go on with 800x600, 1024x768 and so on, until you reach the limit where your card stops changing mode.
Same method applies to colour depth, first try with 15, 16 instead of defaulting to 32.
Avoid specifying the refresh rate on the command line, uvesafb *should* automatically choose the best one. If that is not your case, pass a bare "60" in the form <x>x<y>-<depth>-<vrefresh>. I never had to pass manually the vertical refresh frequency, but I'd suggest you to try with and without it before giving up on a particular mode.
About MTRR's settings... well, they shouldn't harm your system, but before enabling them with mtrr:x (where x is 1 <= x <= 4) try to disable them with mtrr:0 or nomtrr. Whether uvesafb is working or not, try to enable and disable mtrrs to see if anything changes (except scrolling speed).
About ywrap and ypan options: they're supported under x86 class only (no AMD64 or other architectures, it simply won't work and will be ignored). ywrap is faster than ypan, and ypan is faster than the default "redraw" scrolling mode. Sometimes these two options make uvesafb default to 640x480. Again, do your testing with and without it and choose which is better for you. Obviously a boot line like "video=uvesafb:1600x1200-32@60,mtrr:4,ywrap" will result in faster scrolling than one that looks like "video=uvesafb:1600x1200-32@60,nomtrr" but the performance gain is not that impressive.
About modularizing uvesafb, I don't see the reason, though it can be done. If I want to setup a framebuffer it should work regardless of module loading, else I'd probably get a 640x480 screen until the module gets loaded anyway (hence no reason, at least to me). All the info I've provided is about configuring uvesafb as builtin, and not as a module. I have no experience with that and I think I'll never have.
All those info, suggestions, parameters and knowledge come from various sources: kernel docs, wikis, forums/blogs, personal experience. Usually when I setup something I try to do it as if was the last time ever I'd have to set it up, so I document myself in every possible way to avoid spending a lot of time fine-tuning. I try to do it ahead of setup. The method I've suggested in the second part with boot parameters it's recursive and very simple: try to start from bare minimum and change just one variable at once until everything is working at the best (the max you can get from your hardware and software).
I can understand that even after all of the tries above uvesafb simply won't work. I guess you'll have to wait until newer uvesafb comes out (better hw support, fixed bugs, etc...).
Hope to have been useful to someone
If I said something wrong, someone correct me please _________________ Neo2
Unofficial minimal liveCD for x86/amd64 w/reiser4+truecrypt |
|
Back to top |
|
|
Wind Of Silence n00b
Joined: 23 Jan 2008 Posts: 1
|
Posted: Wed Jan 23, 2008 10:07 pm Post subject: Worked like a charm! |
|
|
Thanks heaps man! Had a problem where uvesafb loaded just fine (uvesafb records in /var/log/dmesg etc.), without any results ever occuring (resolution/colour etc. stayed the same dull default). It is however all good now with the stuff on the bottom of your list added to my kernel (didnt need the fonts tho). |
|
Back to top |
|
|
kusi n00b
Joined: 11 Jun 2005 Posts: 21
|
Posted: Sun Jan 27, 2008 4:23 pm Post subject: |
|
|
hoschi wrote: | Hello,
does "Suspend to RAM" work at any configuration with UVESAFB?
|
suspending (RAM as well as harddrive) with tuxonice sources works fine as long as you compile the uvesafb as module and not into kernel as mentioned in spocks install instrucions.
kusi |
|
Back to top |
|
|
jdmulloy Tux's lil' helper
Joined: 24 Dec 2004 Posts: 139 Location: Massachusetts, USA
|
Posted: Sun Jan 27, 2008 10:39 pm Post subject: |
|
|
kusi wrote: | hoschi wrote: | Hello,
does "Suspend to RAM" work at any configuration with UVESAFB?
|
suspending (RAM as well as harddrive) with tuxonice sources works fine as long as you compile the uvesafb as module and not into kernel as mentioned in spocks install instrucions.
kusi |
Don't use suspend2/tuxonice, it's crap. It's hackish and not implemented very cleanly, that's why they won't add it to the mainstream kernel. _________________ Joe Mulloy | http://twitter.com/jdmulloy | Ron Paul in 2012! | 5-1-07 | Unban Playfool | Fire your "Too big to fail" bank http://moveyourmoney.info |
|
Back to top |
|
|
NightMonkey Guru
Joined: 21 Mar 2003 Posts: 356 Location: Philadelphia, PA
|
Posted: Mon Jan 28, 2008 1:47 am Post subject: |
|
|
jdmulloy wrote: | kusi wrote: | hoschi wrote: | Hello,
does "Suspend to RAM" work at any configuration with UVESAFB?
|
suspending (RAM as well as harddrive) with tuxonice sources works fine as long as you compile the uvesafb as module and not into kernel as mentioned in spocks install instrucions.
kusi |
Don't use suspend2/tuxonice, it's crap. It's hackish and not implemented very cleanly, that's why they won't add it to the mainstream kernel. |
Uh, its been working fine for me for over 3 years, on multiple laptops (IBM and Fujitsu) and multiple kernel versions, under heavy daily use. Really nice user interfaces, and It Just Works(tm). _________________
|
|
Back to top |
|
|
Duffy n00b
Joined: 02 Dec 2003 Posts: 33
|
Posted: Fri Feb 15, 2008 1:27 am Post subject: |
|
|
i just re-installed gentoo on my (old) laptop and i wanted to give uvesafb a try
done everything according to the uvesafb project site but when i start my laptop it's as normal and i get the message "probe of uvesafb.0 failed with error -5"
my laptop has an ATI Mobility Radeon 7500 .. i hope it's not the graphic card o_O
any ideas? _________________ "This donut has purple in the middle, purple is a fruit. " -- Homer Jay Simpson |
|
Back to top |
|
|
Ramblurr Tux's lil' helper
Joined: 18 Dec 2006 Posts: 103
|
Posted: Sat Feb 16, 2008 1:04 am Post subject: |
|
|
The wiki article regarding bootsplash is in various states of decay. I think it should be updated (or a new article created) with the latest instructions.
So what do you think: attempt to clean up the old one OR create a new article? |
|
Back to top |
|
|
Duffy n00b
Joined: 02 Dec 2003 Posts: 33
|
Posted: Sat Feb 16, 2008 2:25 pm Post subject: |
|
|
ok, i'm one step closer (i think)
now i configured the kernel according to the wiki ... but .... now i get the message that verbose mode is not working because the kernel tree against which dev-libs/klibc was built was not patched with a compatible version of fbcondecor and that splashuitls was compiled without fbcondecor support
i'm running gentoo-sources so i don't think i need to patch the kernel
and i get a couple of errormessages from splashuitils during systemstartup _________________ "This donut has purple in the middle, purple is a fruit. " -- Homer Jay Simpson |
|
Back to top |
|
|
Ramblurr Tux's lil' helper
Joined: 18 Dec 2006 Posts: 103
|
Posted: Sat Feb 16, 2008 3:55 pm Post subject: |
|
|
Duffy wrote: | ok, i'm one step closer (i think)
now i configured the kernel according to the wiki ... but .... now i get the message that verbose mode is not working because the kernel tree against which dev-libs/klibc was built was not patched with a compatible version of fbcondecor and that splashuitls was compiled without fbcondecor support
i'm running gentoo-sources so i don't think i need to patch the kernel |
What version of gentoo-sources are you running? Also which version owns the /usr/src/linux symlink (ls -l /usr/src/ will till you).
What you need to do is make sure the current gentoo-sources (the one in /usr/src/linux) is the one you have been configuring and then you need to recompile dev-libs/klibc |
|
Back to top |
|
|
Duffy n00b
Joined: 02 Dec 2003 Posts: 33
|
Posted: Sat Feb 16, 2008 4:40 pm Post subject: |
|
|
it's a fresh installation - only one kernel on it - gentoo-sources-2.6.23-r8
the error message states that the kernel is either not patched or not patched with a compatible version of fbcondecor
is my conclusion correct that fbcondecor is the "framebuffer console decoration" option in the kernelconfig? thats enabled so this shouldn't be the problem _________________ "This donut has purple in the middle, purple is a fruit. " -- Homer Jay Simpson |
|
Back to top |
|
|
Ramblurr Tux's lil' helper
Joined: 18 Dec 2006 Posts: 103
|
Posted: Sat Feb 16, 2008 5:38 pm Post subject: |
|
|
What is your output of 'ls -l /usr/src/'?
Try remerging klibc |
|
Back to top |
|
|
Duffy n00b
Joined: 02 Dec 2003 Posts: 33
|
Posted: Sat Feb 16, 2008 6:36 pm Post subject: |
|
|
Ramblurr wrote: | What is your output of 'ls -l /usr/src/'?
Try remerging klibc |
linux -> linux-2.6.23-gentoo-r8
linux-2.6.23-gentoo-r8
i re-emerged klibc and splashutils several times and i always get that error message _________________ "This donut has purple in the middle, purple is a fruit. " -- Homer Jay Simpson |
|
Back to top |
|
|
|