UVesaFB - The Successor to VesaFB-TNG (Bootsplash)

Message

swimmer · Post by **swimmer** » Tue Nov 13, 2007 11:29 am

Is /usr/src/linux linked to the correct kernel? Check it with "eselect kernel list" ...

HTH
swimmer

neiljw · Post by **neiljw** » Tue Nov 13, 2007 12:18 pm

Netolish wrote:I still cannot build v86d.
Error: " You need to compile klibc against a kernel tree patched with uvesafb"

See bugzilla:

http://bugs.gentoo.org/show_bug.cgi?id=198937

Netolish · Post by **Netolish** » Wed Nov 14, 2007 8:55 am

[quote="swimmer"]Is /usr/src/linux linked to the correct kernel? Check it with "eselect kernel list" ...

HTH
swimmer[/quote]

Yes due to nvidia-drivers, vmware-modules and so on.
ls -l /usr/src
lrwxrwxrwx 1 root root 22 2007-11-13 10:22 linux -> linux-2.6.23-gentoo-r1

Netolish · Post by **Netolish** » Wed Nov 14, 2007 9:02 am

[quote="neiljw"][quote="Netolish"]I still cannot build v86d.
Error: " You need to compile klibc against a kernel tree patched with uvesafb"[/quote]

See bugzilla:

[url]http://bugs.gentoo.org/show_bug.cgi?id=198937[/url][/quote]

Thank for advice but in gentoo-sources-2.6.23-r1 this patch is allready included :-(

neiljw · Post by **neiljw** » Wed Nov 14, 2007 11:22 am

Netolish wrote:Thank for advice but in gentoo-sources-2.6.23-r1 this patch is allready included

I don't think you read properly.

The fault is in klibc which uses it's own vanilla sources. They are what need patching.

Habbit · Post by **Habbit** » Wed Nov 14, 2007 11:59 am

neiljw wrote:
Netolish wrote:Thank for advice but in gentoo-sources-2.6.23-r1 this patch is allready included
I don't think you read properly. The fault is in klibc which uses it's own vanilla sources. They are what need patching.

Is there any way we could do that, apart from waiting for upstream to fix it or deploying our own portage overlay?

neiljw · Post by **neiljw** » Wed Nov 14, 2007 12:09 pm

Habbit wrote:
neiljw wrote:
Netolish wrote:Thank for advice but in gentoo-sources-2.6.23-r1 this patch is allready included
I don't think you read properly. The fault is in klibc which uses it's own vanilla sources. They are what need patching.
Is there any way we could do that, apart from waiting for upstream to fix it or deploying our own portage overlay?

The easiest way would be to use an earlier version of klibc for the purpose of compiling v86d. Once it's compiled and installed, you can return to klibc-2.5.7-r1 if you wish.

neiljw · Post by **neiljw** » Thu Nov 15, 2007 1:32 am

klibc-2.5.7-r2 is now in portage which fixes the v86d problem.

Habbit · Post by **Habbit** » Thu Nov 15, 2007 10:34 am

Even with the new klibc, /me keeps having problem with uvesafb. This is an excerpt of my dmesg on an amd64 system with a PCIE ATI R480 (radeon x850xt) video card:

Code: Select all

[   45.717114] 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
[   45.722731] uvesafb: VBIOS/hardware supports DDC2 transfers
[   45.849854] uvesafb: monitor limits: vf = 75 Hz, hf = 83 kHz, clk = 150 MHz
[   45.849986] uvesafb: scrolling: redraw
[   45.853308] uvesafb: mode switch failed (eax=0x34f, err=0). Trying again with default timings.
[   46.180019] Switched to high resolution mode on CPU 0
[   46.264014] uvesafb: mode switch failed (eax=0x34f, err=0). Trying again with default timings.
[   46.918671] Console: switching to colour frame buffer device 128x48
[   46.918708] uvesafb: mode switch failed (eax=0x34f, err=0). Trying again with default timings.
[   47.573394] uvesafb: framebuffer at 0xe0000000, mapped to 0xffffc20010100000, using 16384k, total 16384k
[   47.573397] fb0: VESA VGA frame buffer device

After this, screen goes blank and does not come up again until X starts (with the wrong resolution, so uvesafb is actually doing something with the video card). Furthermore, if I ever leave X and return to the command line, be it switching VTs or just shutting down the computer, it blocks on switching modes - this is not due to the infamous fglrx driver as I use the open-source "ati" driver.
I had initially set a 1680x1050-32 mode for uvesafb, but right now I'm testing conservative configs of 1024x768-32 with the same result. This happens with both vanilla-sources-2.6.24-rc2 and gentoo-sources-2.6.23-r1. My initramfs is OK (I think), with the (statically linked) /sbin/v86d binary and the tty1, zero, console and mem device nodes in /dev. My grub config:

Code: Select all

root (hd0,0)
kernel /kernel-gentoo-last root=/dev/sdb6 video=uvesafb:1024x768-32
initrd /gentoo-initramfs.gz

Schizoid · Post by **Schizoid** » Sun Nov 18, 2007 7:47 am

I have an nVidia 8800 GTS and a Samsung 226BW (1680x1050 native).

I am using the kernel line:

kernel /boot/kernel-2.6.23-gentoo-r1-Sussudio root=/dev/sda3 video=uvesafb:1680x1050-32,mtrr:3,ywrap

The odd thing is that this appears to work, I get a nice widescreen display with lots of tiny console text. The silly thing is I still get the classic error:

[ 43.534340] uvesafb: no monitor limits have been set, default refresh rate will be used
[ 48.526639] uvesafb: VBE state buffer size cannot be determined (eax=0x4f04, err=1)

and a five second pause for no reason!

The bizarre thing is if I try to set a refresh rate, I get the same error and horrible large console text!

To recap:
uvesafb:1680x1050 = Nice, small console text, error with 5 second pause
uvesafb:1680x1050-32 = Nice, small console text, error with 5 second pause
uvesafb:1680x1050-32@60 = Bad, large text, error with 5 second pause
uvesafb:1680x1050-32@59 = Nice, small console text, error with 5 second pause
uvesafb:1680x1050p-59 = Bad, large text, error with 5 second pause

Switching 32 for 8 doesn't ever seem to make a difference in how things work out.

cat /sys/bus/platform/drivers/uvesafb/uvesafb.0/vbe_modes | grep "1680"
1680x1050-8, 0x0168
1680x1050-32, 0x0169

cat /sys/class/graphics/fb0/modes | grep "1680"
U:1680x1050p-59
U:1680x1050p-59

So I am not sure what it wants me to do here. It is just hell bent on giving me that error message no matter how specific (or not) I try to be!

justcme · Post by **justcme** » Sun Nov 18, 2007 2:49 pm

Help!

I've gone through everything I could to try and get UVesaFB to work on my laptop - and for the life of me I just can't make it work!!! AGHHHH

I've looked through almost every relevant post here as well as searching on google for answers... to no avail.

So the issue I have is this, happens on boot:

Code: Select all

uvesafb: Getting VBE info block failed (eax=0x4f00, err=1) 
uvesafb: vbe_init() failed with -22 
uvesafb: probe of uvesafb.0 failed with error -22

EDIT: i get err=-3 when i pass in options mtrr=3, scroll=ywrap,mode=1280x800-16@60

I have an Intel GMA 950, and I get the same error on both 2.6.23 and 2.6.24-rc2 kernels (kamikaze sources).

Relevant kernel configs:

Code: Select all

#
# Graphics Support
#
CONFIG_FB=y
CONFIG_FIRMWARE_EDID=y
...
#
# Frame buffere hardware drivers
#
CONFIG_FB_UVESA=m
...
#
# Console display driver support
#
CONFIG_VIDEO_SELECT=y
CONFIG_FRAMEBUFFER_CONSOLE=y
CONFIG_FB_CON_DECOR=y

I have always had these options enabled (except uvesafb) as I had gensplash/fbcondecor working perfectly with vesa-tng (built as modules) prior to this, using initramfs + 915resolution.

atm all i have relating to FB in grub.conf is video=uvesafb (no other options enabled), and the module insertion during initramfs is the same.

Could really use some help. Any ideas?

neiljw · Post by **neiljw** » Sun Nov 18, 2007 9:54 pm

Have you followed Spock's HowTo? Have you installed v86d?

justcme · Post by **justcme** » Mon Nov 19, 2007 1:40 am

neiljw wrote:Have you followed Spock's HowTo? Have you installed v86d?

Yep klibc compiled against patched and compiled kernel, which then allowed me to compiled v86d properly. Also copied the v86d binary into my initramfs. Didn't use the provided initramfs though. I'll try that later today to see if that changes anything.

Apart from that I can't think of anything else I've deviated from.

neiljw · Post by **neiljw** » Mon Nov 19, 2007 1:46 am

justcme wrote:
neiljw wrote:Have you followed Spock's HowTo? Have you installed v86d?
Yep klibc compiled against patched and compiled kernel, which then allowed me to compiled v86d properly. Also copied the v86d binary into my initramfs. Didn't use the provided initramfs though. I'll try that later today to see if that changes anything.

Apart from that I can't think of anything else I've deviated from.

The way I do it is to load v86d in the kernel.

justcme · Post by **justcme** » Mon Nov 19, 2007 2:56 am

neiljw wrote:
justcme wrote:
neiljw wrote:Have you followed Spock's HowTo? Have you installed v86d?
Yep klibc compiled against patched and compiled kernel, which then allowed me to compiled v86d properly. Also copied the v86d binary into my initramfs. Didn't use the provided initramfs though. I'll try that later today to see if that changes anything.

Apart from that I can't think of anything else I've deviated from.
The way I do it is to load v86d in the kernel.

What do you mean by "load v86d in the kernel"? How would I go about doing that?

justcme · Post by **justcme** » Mon Nov 19, 2007 1:15 pm

Wow, it works finally!!
Very strange...

What I did were some different tests...

1. Compiled as modules, custom initramfs = NO FB (original setup)
2. Compiled as modules, packaged initramfs = NO FB
3. Compiled statically, packaged initramfs = FB OK
4. Compiled statically, custom initramfs = FB OK
5. Compiled as modules, custom initramfs = FB OK!!!! (but this is my original setup!!!)

So yeah... seems to be fixed... without actually being fixed...
Got my widescreen + splash decorated framebuffer again! yay

nixnut · Post by **nixnut** » Mon Nov 19, 2007 7:00 pm

merged above six posts here.

neiljw · Post by **neiljw** » Mon Nov 19, 2007 10:25 pm

justcme wrote:What do you mean by "load v86d in the kernel"? How would I go about doing that?

Under General Setup in make menuconfig.

wxqc11h · Post by **wxqc11h** » Wed Nov 21, 2007 2:09 am

i cannot build klibc (x86_64)

defconfig has changed, please remove or edit .config
make: *** [.config] Błąd 1
*
* ERROR: dev-libs/klibc-1.5.7-r2 failed.
* Call stack:
* ebuild.sh, line 1701: Called dyn_compile
* ebuild.sh, line 1039: Called qa_call 'src_compile'
* ebuild.sh, line 44: Called src_compile
* klibc-1.5.7-r2.ebuild, line 143: Called die
* The specific snippet of code:
* emake \
* EXTRA_KLIBCAFLAGS="-Wa,--noexecstack" \
* EXTRA_KLIBCLDFLAGS="-z,noexecstack" \
* HOSTCC="${HOSTCC}" CC="${CC}" \
* INSTALLDIR="/usr/${libdir}/klibc" \
* KLIBCARCH=${KLIBCARCH} \
* KLIBCASMARCH=${KLIBCASMARCH} \
* SHLIBDIR="/${libdir}" \
* libdir="/usr/${libdir}" \
* mandir="/usr/share/man" \
* ${myargs} || die "Compile failed!"
* The die message:
* Compile failed!
*
* If you need support, post the topmost build error, and the call stack if relevant.
* A complete build log is located at '/mnt/zbiornik/.portos/tmp/portage/dev-libs/klibc-1.5.7-r2/temp/build.log'.
*

do you have any ideas ?
defconfig has changed, please remove or edit .config ????

uzz75 · Post by **uzz75** » Sat Nov 24, 2007 11:06 am

I can use good video=uvesafb:1680x1050-32 but i have problem with F2 key.

it don't work, if i want to see the init status i need to press ALT+F1 and after it's impossible to came back to splash.

Some one have the same problem ?

This is my grub line

kernel /vmlinuz-2.6.23-gentoo-r2 ro root=/dev/sda2 video=uvesafb:1680x1050-32,mtrr:3,ywrap splash=silent,fadein CONSOLE=/dev/tty1 quiet

i don't use initrd

Thank's

musv · Post by **musv** » Mon Nov 26, 2007 8:56 am

Code: Select all

CRC-CCITT functions (CRC_CCITT) [N/m/y/?] n
CRC16 functions (CRC16) [N/m/y/?] n
CRC ITU-T V.41 functions (CRC_ITU_T) [N/m/y/?] n
CRC32 functions (CRC32) [Y/?] y
CRC7 functions (CRC7) [N/m/y/?] n
CRC32c (Castagnoli, et al) Cyclic Redundancy-Check (LIBCRC32C) [N/m/y/?] n
#
# configuration written to .config
#
scripts/kconfig/conf -s arch/x86/Kconfig.i386
  CHK     include/linux/version.h
  UPD     include/linux/version.h
  CHK     include/linux/utsrelease.h
  UPD     include/linux/utsrelease.h
  SYMLINK include/asm -> include/asm-x86
  CC      arch/x86/kernel/asm-offsets.s
  GEN     include/asm-x86/asm-offsets.h
  CALL    scripts/checksyscalls.sh
defconfig has changed, please remove or edit .config
make: *** [.config] Fehler 1
 *
 * ERROR: dev-libs/klibc-1.5.7-r2 failed.
 * Call stack:
 *               ebuild.sh, line 1762:  Called dyn_compile
 *               ebuild.sh, line  891:  Called qa_call 'src_compile'
 *               ebuild.sh, line   35:  Called src_compile
 *             environment, line 2144:  Called die
 * The specific snippet of code:
 *       emake EXTRA_KLIBCAFLAGS="-Wa,--noexecstack" EXTRA_KLIBCLDFLAGS="-z,noexecstack" HOSTCC="${HOSTCC}" CC="${CC}" INSTALLDIR="/usr/${libdir}/klibc" KLIBCARCH=${KLIBCARCH} KLIBCASMARCH=${KLIBCASMARCH} SHLIBDIR="/${libdir}" libdir="/usr/${libdir}" mandir="/usr/share/man" ${myargs} || diefunc "$FUNCNAME" "$LINENO" "$?" "Compile failed!";
 *  The die message:
 *   Compile failed!
 *
 * If you need support, post the topmost build error, and the call stack if relevant.
 * A complete build log is located at '/var/tmp/portage/dev-libs/klibc-1.5.7-r2/temp/build.log'.
 *

defconfig has changed, please remove or edit .config

Same here on x86 (PentiumIII and AthlonXP).

JeliJami · Post by **JeliJami** » Mon Nov 26, 2007 4:21 pm

uzz75 wrote:I can use good video=uvesafb:1680x1050-32 but i have problem with F2 key.

it don't work, if i want to see the init status i need to press ALT+F1 and after it's impossible to came back to splash.

Some one have the same problem ?

Some have:
Bootsplash: Press F2 for verbose mode no longer works

jjwc · Post by **jjwc** » Wed Nov 28, 2007 4:54 am

wxqc11h wrote:i cannot build klibc (amd64)

defconfig has changed, please remove or edit .config
do you have any ideas ?
defconfig has changed, please remove or edit .config ????

I get the same error. What exactly causes it? How can it be fixed?

ronmon · Post by **ronmon** » Wed Nov 28, 2007 5:11 pm

What gets me is that after a recent upgrade, I actually have lost the ability to get 1600x1200 splash working. It makes no sense to me. Here's the story in a nutshell.

My old hardware (up to about 15 days ago) was an amd64 3200+ on an ASUS motherboard with VIA chipsets and a Nvidia 5700LE (AGP) running two 20" 1600x1200 Samsung LCD's on the VGA D-SUB inputs. Since I'm running a 64 bit system I used vesafb rather than vesa-tng and I had no problem running 1600x1200 with "vga = 838" in my lilo.conf.

An apparent PSU problem (which turned out to be a power switch) seemed to fry the motherboard, so I gathered some nice used stuff off of the AnandTech forums and ended up with an amd64 x2 4400+, a BIOSTAR nforce based motherboard, and a PCIe 6800GT which runs the monitors via dual-DVI.

Since the changeover was unexpected, I had to boot from a Gentoo install disk, set up a chroot and rebuild my kernel for the new hardware. No problem. All the new harware works fine, network, I/O, X, etc.

What is killing me is that I can't get 1600x1200 splash resolution working, even if I revert to vesafb. I'm stuck with 1024x768 as the best 4:3 availavble according to /sys/class/graphics/fb0/modes.

It seems like my hardware upgrade has actually cost me fbsplash resolution. Bummer.

Code: Select all

ronmon@maxx $ dmesg | grep uvesafb
uvesafb: NVIDIA Corporation, G70 Board - p317h2  , Chip Rev   , OEM: NVIDIA, VBE v3.0
uvesafb: VBIOS/hardware supports DDC2 transfers
uvesafb: monitor limits: vf = 75 Hz, hf = 81 kHz, clk = 170 MHz
uvesafb: scrolling: redraw
uvesafb: framebuffer at 0xd0000000, mapped to 0xffffc20010100000, using 10240k, total 262144k

Edit::
Should have said that I caught the update to uvesafb before the hardware switch and it worked at 1600x1200 under vesa and uvesafb with the old stuff. The only change that stands out is going from D-SUB to DVI. Any ideas?

musv · Post by **musv** » Thu Nov 29, 2007 8:21 am

It's a known bug, that some Nvidia cards don't show up all framebuffer videomodes.

http://gentoo-wiki.com/HOWTO_gensplash# ... safb.29.3F

UVesaFB - The Successor to VesaFB-TNG (Bootsplash)

Re: Problem to build v86d

Re: Problem to build v86d

Re: Problem to build v86d

Re: Problem to build v86d

Re: Problem to build v86d

[Fixed but not Solved] UVesaFB woes :(

Lower resolutions after hardware upgrade