| View previous topic :: View next topic |
| Author |
Message |
socke
n00b
Joined: 03 Nov 2003
Posts: 19
Location: Böblingen (Germany)
|
 Posted: Fri Jul 09, 2004 3:18 pm Post subject: |
 |
|
I've got the same problem using XFree and nvidia-drivers 6061 + Kernel 2.6.7 from kernel.org.
X + nvidia-driver = black screen, only ctrl+alt+del works to reboot - ssh login too.
X + nv = works fine, till now (first start).
It already worked with X + nvidia-driver yesterday, I had about 4k fps in glxgears and everything was ok, but today when I tried to startx it hanged...
My Hardware:
Mobo: Asus a7n8x-x (nforce2)
Video: GeForce 4 Ti 4600 |
|
| Back to top |
|
 |
Benson
Apprentice
Joined: 23 Feb 2004
Posts: 156
Location: Switzerland
|
| Posted: Sat Jul 10, 2004 8:51 am Post subject: |
|
|
Ok, i had the same problem with my new computer (GeForce FX5700, nforce2 Ultra, Kernel 2.6.7-r . After disabling the USB 2.0 support, it works now! Thx socke for finding this solution!!!
But now i have another problem: after booting it doesn't start kdm, xdm is in the rc.conf and starts in default runlevel, also the X Font server starts, but after saying "starting local" the screen flickers once (with nvidia in xfree, changing it to nv it flickers three thimes!) and i remain in console modus. In Console 7 where X should start, is no login, no X at all started... Login in as root and starting then kdm (or X) works without problems. Searched the forums, re-read the docs, have done everything there was mentioned, still no success to get my kdm automatically up... anyone else having this issue??? |
|
| Back to top |
|
|
ewildgoose
Tux's lil' helper
Joined: 02 Mar 2003
Posts: 75
|
| Posted: Sat Jul 10, 2004 9:28 am Post subject: |
|
|
| Benson wrote: | | But now i have another problem: after booting it doesn't start kdm, xdm is in the rc.conf and starts in default runlevel, also the X Font server starts, but after saying "starting local" the screen flickers once (with nvidia in xfree, changing it to nv it flickers three thimes!) and i remain in console modus. In Console 7 where X should start, is no login, no X at all started... Login in as root and starting then kdm (or X) works without problems. Searched the forums, re-read the docs, have done everything there was mentioned, still no success to get my kdm automatically up... anyone else having this issue??? |
Although you didn't post your X log files, I **assume** that you *did* already check in there and it *didn't* explain why it failed to start up...? |
|
| Back to top |
|
|
beacher
n00b
Joined: 11 Mar 2003
Posts: 30
Location: Atlanta
|
| Posted: Sat Jul 10, 2004 7:04 pm Post subject: |
|
|
| Rcomian wrote: | Ok, looks like it works now. I've got nearly all the main USB options and drivers enabled except EHCI (USB2).
It looks like it's USB2 which is breaking things. |
After 3 days of pulling my hair out, you have found my answer 
current setup: 2.6.7-gentoo-r7 dev-sources nvidia-glx-1.0.6106 nvidia-kernel-1.0.6106
I don't know if updating my kernel-headers helped or not (my gaming machine was offline for 6 months until I finished a serious honeydo list), so I was coming from 2.4.20-r6 -> 2.6.7 upgrade
This was driving me #&*)# insane. glxinfo still segfaults but at least it's using the nvidia driver and the /dev/nvidia* devices are showing (they never did show prior to rebuilding the kernel - USB2). I hope I can figure out whats happening from here.....
| Code: | NForce2 chipset mobo..
Bus 0, device 2, function 0:
USB Controller: nVidia Corporation nForce2 USB Controller (rev 163).
IRQ 10.
Master Capable. No bursts. Min Gnt=3.Max Lat=1.
Non-prefetchable 32 bit memory at 0xee003000 [0xee003fff].
Bus 0, device 2, function 1: |
| Code: | Bus 2, device 0, function 0:
VGA compatible controller: nVidia Corporation NV18 [GeForce4 MX - nForce GPU] (rev 163).
IRQ 10. |
Really starting to wonder if it's a USB or NVidia chipset issue at this point.
Shared IRQs.. Nothing surprising just interesting....
Thanks a million!
B |
|
| Back to top |
|
|
ivor_orrible
n00b
Joined: 09 Nov 2003
Posts: 44
Location: 127.0.0.1
|
| Posted: Sat Jul 10, 2004 7:39 pm Post subject: |
|
|
I manage to get 2.6.7 kernel and latest nvidia-kernel working by
selecting devfs and devfs mounted at boot in the kernel config.
And adding nvidia to modules.autoload/kernel2.6
If devfs mounted at boot was not selected the nvidia device would not
appear and it just gave a black screen.
If nvidia was not in autoload gdm would just try to start a few times then
give up and let me log in in text mode.
I never have had to put nvidia in modules.autoload before right up
to kernel 2.6.3 only with 2.6.7.
Ivor Cave |
|
| Back to top |
|
|
Ateo
Advocate
Joined: 02 Jun 2003
Posts: 2022
Location: Vegas Baby!
|
| Posted: Sat Jul 10, 2004 8:05 pm Post subject: |
|
|
| Rcomian wrote: | Ok, looks like it works now. I've got nearly all the main USB options and drivers enabled except EHCI (USB2).
It looks like it's USB2 which is breaking things. |
Yeap. Looks like EHCI is breaking stuff. I disabled EHCI in my kernel and now I am able to boot into the 2.6.7 series using the nvidia driver.
Is this a kernel bug? A driver bug? An IRQ conflict? Luckily, none of the devices on this system uses USB2 so I'm ok for the time being.... At least now we have concluded that it might be EHCI..
[edit]
Using the following:
nvidia-kernel-1.0.6106
nvidia-glx-1.0.6106
ASUS A7N8X-Deluxe
e-GeForce 5200 MX
gentoo-dev-sources-2.6.7-r9
Driver: nvidia (as opposed to "nv") |
|
| Back to top |
|
|
Narusegawa
Apprentice
Joined: 29 Jun 2004
Posts: 210
Location: Bimringham, UK
|
| Posted: Sat Jul 10, 2004 10:56 pm Post subject: |
|
|
I had a problem with this all day. I've just gone into my kernel and changed EHCI to not loaded (it was module before) and now the nvidia works with Xorg.
Latest version of Xorg from emerge (~x86)
NVidia 6601
and 2.6.7-r9 gentoo-dev kernel
Only issue is when killing X it goes into a flickering color screen and I can only ctrl-alt-delete to reboot to get system back.
I'm using an MSI K7N2G-ILSR mobo with NForce2 chipsets and GF4 MX GPU onboard. |
|
| Back to top |
|
|
big_pig
n00b
Joined: 14 Sep 2002
Posts: 18
Location: Renton, WA, USA
|
| Posted: Sat Jul 10, 2004 11:09 pm Post subject: |
|
|
| Ateo wrote: | Yeap. Looks like EHCI is breaking stuff. I disabled EHCI in my kernel and now I am able to boot into the 2.6.7 series using the nvidia driver.
Is this a kernel bug? A driver bug? An IRQ conflict? Luckily, none of the devices on this system uses USB2 so I'm ok for the time being.... At least now we have concluded that it might be EHCI..
|
I don't think all of EHCI is responible for this behavior. Just to put my experience in...
I had the blank/black screen issue with X and 2.6.7. I'm using the gentoo-dev-sources, 2.6.7-r8 to be exact. I was upgrading from 2.6.5. I have nVidia GeForce 2 GTS in my AGP slot. I was using the nVidia drivers (nvidia-kernel: 6106; nvidia-glx: 6106).
When I complied 2.6.7 the first time included all of EHCI. Here is the sinippet of my kernel config.
| Code: |
CONFIG_USB_EHCI_HCD=m
CONFIG_USB_EHCI_SPLIT_ISO=y
CONFIG_USB_EHCI_ROOT_HUB_TT=y
|
Since I was going to be rebuilding the kernel, I decided to try turning off the EHCI options, CONFIG_USB_EHCI_SPLIT_ISO and CONFIG_USB_EHCI_ROOT_HUB_TT. When I launched the new kernel, I didn't have any issues with X and the nVidia driver.
bp |
|
| Back to top |
|
|
Ateo
Advocate
Joined: 02 Jun 2003
Posts: 2022
Location: Vegas Baby!
|
| Posted: Sat Jul 10, 2004 11:31 pm Post subject: |
|
|
The one thing I have to try is building EHCI as a module whereas all of my issues were when it was built into the kernel..
I'll give that a try....
[edit]
With this config:
| Code: |
CONFIG_USB_EHCI_HCD=m
CONFIG_USB_EHCI_SPLIT_ISO=y
CONFIG_USB_EHCI_ROOT_HUB_TT=y
|
I get the black screen again.
For whatever reason, I thought my camera was USB2. I guess my issue is resolved since I do not require EHCI. |
|
| Back to top |
|
|
mcqueen
n00b
Joined: 27 Feb 2004
Posts: 7
Location: BOS
|
| Posted: Sun Jul 11, 2004 2:12 am Post subject: Wait until 2.6.8 |
|
|
For those of you who fixed this problem by disabling EHCI, this problem should be fixed in 2.6.8. Or, at least, I hope it does.
http://bugzilla.kernel.org/show_bug.cgi?id=2846
|
|
| Back to top |
|
|
Locke355
Tux's lil' helper
Joined: 21 May 2002
Posts: 98
|
| Posted: Tue Jul 13, 2004 10:46 am Post subject: |
|
|
| Narusegawa wrote: | I had a problem with this all day. I've just gone into my kernel and changed EHCI to not loaded (it was module before) and now the nvidia works with Xorg.
Latest version of Xorg from emerge (~x86)
NVidia 6601
and 2.6.7-r9 gentoo-dev kernel
Only issue is when killing X it goes into a flickering color screen and I can only ctrl-alt-delete to reboot to get system back.
I'm using an MSI K7N2G-ILSR mobo with NForce2 chipsets and GF4 MX GPU onboard. |
I have this exact same problem (using a p4p800 board, 865 chipset). The screen doesn't flicker tho, monitor loses video sync and i have to reboot. (also, i am running XFree.. not xorg)
any ideas on how to fix? |
|
| Back to top |
|
|
augury
l33t
Joined: 22 May 2004
Posts: 722
Location: philadelphia
|
| Posted: Tue Jul 13, 2004 1:56 pm Post subject: |
|
|
from /usr/src/linux/arch/i386
| Code: |
# 19990713 Artur Skawina <skawina@geocities.com>
# Added '-march' and '-mpreferred-stack-boundary' support
#
LDFLAGS := -m elf_i386
OBJCOPYFLAGS := -O binary -R .note -R .comment -S
LDFLAGS_vmlinux :=
CHECK := $(CHECK) -D__i386__=1
CFLAGS += -pipe -msoft-float
# prevent gcc from keeping the stack 16 byte aligned
CFLAGS += $(call check_gcc,-mpreferred-stack-boundary=2,)
align := $(subst -functions=0,,$(call check_gcc,-falign-functions=0,-malign-functions=0))
cflags-$(CONFIG_M386) += -march=i386
cflags-$(CONFIG_M486) += -march=i486
cflags-$(CONFIG_M586) += -march=i586
cflags-$(CONFIG_M586TSC) += -march=i586
cflags-$(CONFIG_M586MMX) += $(call check_gcc,-march=pentium-mmx,-march=i586)
cflags-$(CONFIG_M686) += -march=i686
cflags-$(CONFIG_MPENTIUMII) += $(call check_gcc,-march=pentium2,-march=i686)
cflags-$(CONFIG_MPENTIUMIII) += $(call check_gcc,-march=pentium3,-march=i686)
cflags-$(CONFIG_MPENTIUMM) += $(call check_gcc,-march=pentium3,-march=i686)
cflags-$(CONFIG_MPENTIUM4) += $(call check_gcc,-march=prescott,-march=i686)
cflags-$(CONFIG_MK6) += -march=k6
# Please note, that patches that add -march=athlon-xp and friends are pointless.
# They make zero difference whatsosever to performance at this time.
cflags-$(CONFIG_MK7) += $(call check_gcc,-march=athlon,-march=i686 $(align)-functions=4)
cflags-$(CONFIG_MK8) += $(call check_gcc,-march=k8,$(call check_gcc,-march=athlon,-march=i686 $(align)-functions=4))
cflags-$(CONFIG_MCRUSOE) += -march=i686 $(align)-functions=0 $(align)-jumps=0 $(align)-loops=0
cflags-$(CONFIG_MWINCHIPC6) += $(call check_gcc,-march=winchip-c6,-march=i586)
cflags-$(CONFIG_MWINCHIP2) += $(call check_gcc,-march=winchip2,-march=i586)
cflags-$(CONFIG_MWINCHIP3D) += $(call check_gcc,-march=winchip2,-march=i586)
cflags-$(CONFIG_MCYRIXIII) += $(call check_gcc,-march=c3,-march=i486) $(align)-functions=0 $(align)-jumps=0 $(align)-loops=0
cflags-$(CONFIG_MVIAC3_2) += $(call check_gcc,-march=c3-2,-march=i686)
|
the only way i know to set 16k stacks is to comment out the boundary line in this file in late 2.6.6-2.6.7 kernels. had to "make V=1" rather than make to find that out. i compiled 2.6.7-gentoo-r10 with:
HOSTCFLAGS = -Wall -Wstrict-prototypes -fomit-frame-pointer -fno-strict-aliasing -fno-common -O3 -fomit-frame-pointer -march=prescott -DNO_DEBUG -fno-exceptions -fmerge-all-constants -pipe --param max-unrolled-insns=16 -funroll-all-loops -finline-limit=2048 -ffast-math -fsingle-precision-constant -ftracer -fprefetch-loop-arrays
and my compilers default stacks for my arch (since i compile all of my software and it would jam up the registers to have different size stacks to have some different). runs much less jerky--5200asus nvidia p4c800 2.8P4e runs better than 5700asus nvidia p4c800 3.2Pc (less mem too)--except for antialiasing. i had gotten kernel panics because i reduce binutils givs. but ever since that was fixed 2.6.7 has run really well with nvidia drivers. they f'd with qt real bad but so did half a dozen other things. mostly the freetype versions are messing me up. could be xfree--its all fonts anyhow. thought mabye id upgrade tetex (which is like taking the brick that holds your door open and getting a bigger brick and saying you upgraded the brick). i cant get it past this cray hooking kpathsea.a piece though. things designed to break w/flags on purpose. i've got to fix the font thing though, its crashing my gedit and porthole when ever i try to cut and paste. i get
bash-2.05b# porthole
| Code: | Fontconfig warning: line 251: invalid edit binding "same"
Fontconfig warning: line 263: invalid edit binding "same"
No fonts found; this probably means that the fontconfig
library is not correctly configured. You may need to
edit the fonts.conf configuration file. More information
about fontconfig can be found in the fontconfig(3) manual
page and on http://fontconfig.org |
i was there but i didnt see much, maybe ill try again.
oh yeah just so get dislocated for diverging, theres config flags for via and flatscreens and what not in os-registry in /usr/src/nv but you have to extract the NVIDIA-Linux-x86-1.0-6106-pkg1 job with -x. after you build it you cant clean it but you can use it again. |
|
| Back to top |
|
|
augury
l33t
Joined: 22 May 2004
Posts: 722
Location: philadelphia
|
| Posted: Tue Jul 13, 2004 2:15 pm Post subject: |
|
|
| Benson wrote: | Ok, i had the same problem with my new computer (GeForce FX5700, nforce2 Ultra, Kernel 2.6.7-r. After disabling the USB 2.0 support, it works now! Thx socke for finding this solution!!!
But now i have another problem: after booting it doesn't start kdm, xdm is in the rc.conf and starts in default runlevel, also the X Font server starts, but after saying "starting local" the screen flickers once (with nvidia in xfree, changing it to nv it flickers three thimes!) and i remain in console modus. In Console 7 where X should start, is no login, no X at all started... Login in as root and starting then kdm (or X) works without problems. Searched the forums, re-read the docs, have done everything there was mentioned, still no success to get my kdm automatically up... anyone else having this issue??? |
if you reinstalled xfree or xorg you have to rebuild qt. kde should be ok. i had to uninstall all nvidia stuff and fontconfig and freetype to get qt to build. |
|
| Back to top |
|
|
CarlUman
Apprentice
Joined: 07 Jul 2004
Posts: 158
Location: SE Iowa
|
| Posted: Fri Jul 23, 2004 3:01 am Post subject: |
|
|
Removing EHCI from the kernel worked for me too 
Finally  |
|
| Back to top |
|
|
isnogood
Tux's lil' helper
Joined: 09 Jul 2003
Posts: 96
|
| Posted: Fri Jul 23, 2004 2:45 pm Post subject: |
|
|
Thanks to everybody - was ssh'ing my ass off until I found this thread.
BTW:Rcomian - do you got one or two drives in your box?Got the same mobo and that thing crashes every five minutes with two hdd's on the same channel. |
|
| Back to top |
|
|
Rcomian
Apprentice
Joined: 10 Jan 2004
Posts: 174
Location: Uk, Northwest
|
| Posted: Sat Jul 24, 2004 9:43 am Post subject: |
|
|
| isnogood wrote: | | BTW:Rcomian - do you got one or two drives in your box?Got the same mobo and that thing crashes every five minutes with two hdd's on the same channel. |
I got one hdd at the moment, so haven't seen the 2 drive problem. Thanks for letting me know, I'll keep it in mind when I upgrade
Does it work ok sharing the channels between the hdd & cdroms? |
|
| Back to top |
|
|
isnogood
Tux's lil' helper
Joined: 09 Jul 2003
Posts: 96
|
| Posted: Sat Jul 24, 2004 10:35 am Post subject: |
|
|
It seems to work with 2.6.7 .So far the box is stable.I'll never switch kernels again.
Dunno about on two different channels - would run into cable problems with the cdrom. |
|
| Back to top |
|
|
PEETIE
Tux's lil' helper
Joined: 28 Mar 2004
Posts: 144
|
| Posted: Tue Jul 27, 2004 10:32 pm Post subject: |
|
|
When unchecking dev/agpgart in the kernel compile menu do you need to uncheck the dri line that appears below it when you uncheck it?
and does it matter where you add the line Option "NvAGP" "1"' in XF86Config? |
|
| Back to top |
|
|
Rcomian
Apprentice
Joined: 10 Jan 2004
Posts: 174
Location: Uk, Northwest
|
| Posted: Wed Jul 28, 2004 10:54 pm Post subject: |
|
|
| PEETIE wrote: | When unchecking dev/agpgart in the kernel compile menu do you need to uncheck the dri line that appears below it when you uncheck it?
and does it matter where you add the line Option "NvAGP" "1"' in XF86Config? |
Not sure about the dri thing. But the "NvAGP" "1" needs to go inside the 'Section "Device"' for the vid card, as in:
| Code: | Section "Device"
### Available Driver options are:-
### Values: <i>: integer, <f>: float, <bool>: "True"/"False",
### <string>: "String", <freq>: "<f> Hz/kHz/MHz"
### [arg]: arg optional
Option "DigitalVibrance" "4" # <i>
#Option "PageFlip" # [<bool>]
#Option "Dac8Bit" # [<bool>]
#Option "NoLogo" # [<bool>]
#Option "Overlay" # [<bool>]
#Option "UBB" # [<bool>]
#Option "Stereo" # <i>
#Option "WindowFlip" # [<bool>]
#Option "SWcursor" "True" # [<bool>]
#Option "HWcursor" "True" # [<bool>]
#Option "VideoKey" # <i>
Option "NvAGP" "1" # <i>
#Option "IgnoreEDID" # [<bool>]
#Option "NoDDC" # [<bool>]
#Option "ConnectedMonitor" # <str>
#Option "ConnectedMonitors" # <str>
#Option "TVStandard" # <str>
#Option "TVOutFormat" # <str>
#Option "RenderAccel" # [<bool>]
#Option "CursorShadow" # [<bool>]
#Option "CursorShadowAlpha" # <i>
#Option "CursorShadowXOffset" # <i>
#Option "CursorShadowYOffset" # <i>
#Option "UseEdidFreqs" "True" # [<bool>]
#Option "FlatPanelProperties" # <str>
#Option "TwinView" # [<bool>]
#Option "TwinViewOrientation" # <str>
#Option "SecondMonitorHorizSync" # <str>
#Option "SecondMonitorVertRefresh" # <str>
#Option "MetaModes" # <str>
#Option "UseInt10Module" # [<bool>]
#Option "SwapReady" # [<bool>]
#Option "NoTwinViewXineramaInfo" # [<bool>]
#Option "NoRenderExtension" # [<bool>]
#Option "UseClipIDs" # [<bool>]
#Option "CIOverlay" # [<bool>]
#Option "TransparentIndex" # <i>
#Option "OverlayDefaultVisual" # [<bool>]
#Option "NvEmulate" # <i>
#Option "NoBandWidthTest" # [<bool>]
#Option "CustomEDID-CRT-0" # <str>
#Option "CustomEDID-CRT-1" # <str>
#Option "CustomEDID-DFP-0" # <str>
#Option "CustomEDID-DFP-1" # <str>
#Option "CustomEDID-TV-0" # <str>
#Option "CustomEDID-TV-1" # <str>
#Option "TVOverScan" # <f>
#Option "IgnoreDisplayDevices" # <str>
#Option "MultisampleCompatibility" # [<bool>]
Identifier "Card_nvidia"
Driver "nvidia"
# Driver "nv" # builtin
# Driver "vesa"
VendorName "nVidia Corporation"
BoardName "NV17 [GeForce4 MX 440]"
BusID "PCI:2:0:0"
EndSection |
|
|
| Back to top |
|
|
JoZEr
n00b
Joined: 23 May 2004
Posts: 3
|
| Posted: Sun Aug 08, 2004 9:47 am Post subject: |
|
|
| I noticed when booting gentoo-dev-sources2.6.7-r12 and getting the black screen that removing ehci worked for me... the interesting thing is that those shared IRQ's were part of the problem. When ehci loaded it was saying "Disabling IRQ #10" I went into the BIOS on my system and disabled a bunch of stuff I don't use (Serial port, Parallel port, Firewire) thereby shifting the IRQ's around. Now echi says "Disabling IRQ #3" instead and since my video card is still using IRQ 10, everything worked fine. Now you can have USB 2.0 and accelerated drivers at the same time. |
|
| Back to top |
|
|
frameRATE
Guru
Joined: 28 Apr 2003
Posts: 386
Location: Orange County California
|
| Posted: Wed Aug 18, 2004 8:41 pm Post subject: |
|
|
Is everyone solving this problem by disabling EHCI? Because that didn't work for me. Still getting the black screen when using the nvidia drivers, and it'll boot fine with "nv" (just no 3d support) =(
_________________
Linux User | 364705
howto install coldfusion mx 7 under gentoo |
|
| Back to top |
|
|
frameRATE
Guru
Joined: 28 Apr 2003
Posts: 386
Location: Orange County California
|
| Posted: Wed Aug 18, 2004 10:45 pm Post subject: |
|
|
ok i got it. the only thing I can think I did that isn't listed in THIS thread is I unplugged my second monitor from the DVI slot so the ONLY one I have hooked up is the VGA. it seemed to work for someone else, but I'm not sure if that fixed it.
All i know is it's working now!
_________________
Linux User | 364705
howto install coldfusion mx 7 under gentoo |
|
| Back to top |
|
|
Moebius
Tux's lil' helper
Joined: 31 Jul 2003
Posts: 76
Location: Austria
|
| Posted: Thu Aug 19, 2004 8:34 am Post subject: |
|
|
| orick wrote: | Hi,
I had similar problems with the latest 2.6.7-gentoo-r5 kernel and nvidia,emerging the latest masked nvidia-glx-1.0.5336-r2 and nvidia-kernel-1.0.5336-r4 solved my problem:
| Code: | | ACCEPT_KEYWORDS="~x86" emerge -Dv nvidia-glx nvidia-kernel |
However, unfortunately updating with emerge -uDV world insists to bring back the older, stable, nvidia versions. So, until these new nvidia versions are marked stable you should reemerge the latest ones after a world update. |
Worked perfectly for me, thanks!
_________________
There are only 10 types of people in the world. Those who unterstand binary and those who don't. |
|
| Back to top |
|
|
frameRATE
Guru
Joined: 28 Apr 2003
Posts: 386
Location: Orange County California
|
| Posted: Fri Aug 20, 2004 7:40 am Post subject: |
|
|
I would just like to confirm that if I plug in the DVI->VGA adapter into my second slot (DVI) on my ti4600 and hook up the monitor, the nvidia drivers just give me a black screen (and the monitor goes into standby)
I know at least one other person has had this problem... anyone fixed it?
_________________
Linux User | 364705
howto install coldfusion mx 7 under gentoo |
|
| Back to top |
|
|
|
Kernel & Hardware
