NVIDIA 177.13 + KERNEL 2.6.27-rc1+

[ellson]

I've applied these patches, but I'm still not able to build the driver. Can anyone help with this new error on Rawhide kernel-2.6.27-0.244.rc2.git1.fc10.x86_64 ?

-> Performing CC sanity check with CC="cc".
-> Performing CC version check with CC="cc".
-> Kernel source path: '/lib/modules/2.6.27-0.244.rc2.git1.fc10.x86_64/source'
-> Kernel output path: '/lib/modules/2.6.27-0.244.rc2.git1.fc10.x86_64/build'
ERROR: If you are using a Linux 2.4 kernel, please make sure
you either have configured kernel sources matching your
kernel or the correct set of kernel headers installed
on your system.

If you are using a Linux 2.6 kernel, please make sure
you have configured kernel sources matching your kernel
installed on your system. If you specified a separate
output directory using either the "KBUILD_OUTPUT" or
the "O" KBUILD parameter, make sure to specify this
directory with the SYSOUT environment variable or with
the equivalent nvidia-installer command line option.

Depending on where and how the kernel sources (or the
kernel headers) were installed, you may need to specify
their location with the SYSSRC environment variable or
the equivalent nvidia-installer command line option.
ERROR: Installation has failed.
....

[Dragoran]

Originally Posted by ellson

I've applied these patches, but I'm still not able to build the driver. Can anyone help with this new error on Rawhide kernel-2.6.27-0.244.rc2.git1.fc10.x86_64 ?

-> Performing CC sanity check with CC="cc".
-> Performing CC version check with CC="cc".
-> Kernel source path: '/lib/modules/2.6.27-0.244.rc2.git1.fc10.x86_64/source'
-> Kernel output path: '/lib/modules/2.6.27-0.244.rc2.git1.fc10.x86_64/build'
ERROR: If you are using a Linux 2.4 kernel, please make sure
you either have configured kernel sources matching your
kernel or the correct set of kernel headers installed
on your system.

If you are using a Linux 2.6 kernel, please make sure
you have configured kernel sources matching your kernel
installed on your system. If you specified a separate
output directory using either the "KBUILD_OUTPUT" or
the "O" KBUILD parameter, make sure to specify this
directory with the SYSOUT environment variable or with
the equivalent nvidia-installer command line option.

Depending on where and how the kernel sources (or the
kernel headers) were installed, you may need to specify
their location with the SYSSRC environment variable or
the equivalent nvidia-installer command line option.
ERROR: Installation has failed.
....

yum install kernel-devel

[vallimar]

Actually, fedora has a bug in that kernel build where they wrongly moved all the include/asm-x86 headers directly into include/asm. Either remove the empty asm-x86 and symlink it to asm (which is backwards from normal) or rename/move the files from asm -> asm-x86, remove asm, symlink asm -> asm-x86.

[CraigG4C]

177.13 is fairly old and is not recommended for use with GPUs other than GT200, but since there's some demand to get it working with recent kernels until updated display drivers become available ...

Zander, I standardized on this and threw everything else away because on my HP laptop (Go 7150M) with 2.6.25+ kernels, 173.x blew away all the consoles on exit. 177 fixed the problem jes' fine. Come out with a new production version SOON.

By the way, thanks everybody for the patches; this is being posted from .27-rc2!

[ellson]

Thanks for the responses to #25, but no luck yet.

Re: #26 I do have kernel-devel installed
$ rpm -q kernel-devel
kernel-devel-2.6.25.13-105.fc9.x86_64
kernel-devel-2.6.27-0.238.rc2.fc10.x86_64
kernel-devel-2.6.27-0.244.rc2.git1.fc10.x86_64

I'm not new to this, I've been able to build the drivers with earlier 2.6.27 rpms,
but they had some kind of problem such that frame rates would grinding to a halt, so
I'm back at 2.6.25 for now. I was trying to retest.

Re: #29: Perhaps I misunderstand what you are suggesting, but I tried:
$ cd /usr/include
$ ln -s asm asm-x86
and it didn't work. Also checked fc8 and fc9 x86_64 boxes and they don't have a /usr/include/asm-x86.

Any other ideas?

[CraigG4C]

ellson -- the link confusion, I think, is in /usr/src/linux/include/<asm whatever>, not /usr/include. Try it out; the NV installer looks in /lib/modules/<version>/build/include to try to find the appropriate headers.

(At least on Gentoo and other Debian-structured releases. Although I was on the beta mailing list for RH2, I haven't used Red Hat since they went all Fedora on me...)

[ellson]

OK, thanks CraigG4C, that worked better:
$ cd /usr/src/kernels/2.6.27-0.244.rc2.git1.fc10.x86_64/include
$ ln -s asm asm-86

I used Zander's diff from #17, and patch-no-xen from #22

The patch-no-xen was necessary, but not sufficient. I also commented out the line
#define CONFIG_XEN 1
from: /usr/src/kernels/2.6.27-0.244.rc2.git1.fc10.x86_64/include/linux/autoconf.h

I'm sure thats not the right fix, but its working for me.

[siddly]

Using NVIDIA_kernel-177.13-2627763.diff.txt and patch-no-xen.patch_against_Zander.txt, Xen not configured. The module loads on 2.6.27-rc2-git4 but does not work, throws me back to the VC. Boot up errors as follows:-
Aug 10 18:26:45 tindog kernel: nvidia: module license 'NVIDIA' taints kernel.
Aug 10 18:26:45 tindog kernel: ACPI: PCI Interrupt Link [APC6] enabled at IRQ 16
Aug 10 18:26:45 tindog kernel: nvidia 0000:07:00.0: PCI INT A -> Link[APC6] -> GSI 16 (level, low) -> IRQ 16
Aug 10 18:26:45 tindog kernel: nvidia 0000:07:00.0: setting latency timer to 64
Aug 10 18:26:45 tindog kernel: NVRM: loading NVIDIA UNIX x86_64 Kernel Module 177.13 Tue Jun 10 16:42:55 PDT 2008
Aug 10 18:26:45 tindog kernel: BUG: unable to handle kernel NULL pointer dereference at 00000000000000d7
Aug 10 18:26:45 tindog kernel: IP: [<ffffffffa0162377>] nvidia_init_module+0x377/0x7e4 [nvidia]
Aug 10 18:26:45 tindog kernel: PGD 10020b067 PUD 100283067 PMD 0
Aug 10 18:26:45 tindog kernel: Oops: 0000 [1] SMP
Aug 10 18:26:45 tindog kernel: CPU 0
Aug 10 18:26:45 tindog kernel: Modules linked in: nvidia(P+) nfsd lockd nfs_acl auth_rpcgss exportfs it87 hwmon_vid snd_pcm_oss binfmt_misc snd_mixer_oss snd_seq sunrpc af_packet iptable_filter ip_tables ip6table_filter ip6_tables x_tables ipv6 cpufreq_userspace cpufreq_powersave powernow_k8 fuse dm_crypt reiserfs loop dm_mod i2c_dev kvm_amd kvm rt2500usb arc4 ecb crypto_blkcipher dvb_pll bnep rt73usb mt352 rt2x00usb rfcomm cx88_dvb rt2x00lib cx88_vp3054_i2c l2cap rfkill videobuf_dvb snd_usb_audio joydev dvb_core firewire_ohci snd_usb_lib led_class cx8802 firewire_core cx8800 usbhid uvcvideo snd_hda_intel mac80211 snd_rawmidi cx88xx hci_usb hid snd_seq_device compat_ioctl32 bluetooth snd_pcm ir_common ff_memless cfg80211 snd_timer videodev v4l1_compat snd_page_alloc i2c_algo_bit ub ppdev v4l2_common tveeprom snd_hwdep snd ohci1394 videobuf_dma_sg parport_pc i2c_nforce2 sr_mod usblp usb_storage forcedeth k8temp parport ieee1394 videobuf_core btcx_risc rtc_cmos soundcore sg i2c_core ehci_hcd ohci_hcd sd_mod usbcor
Aug 10 18:26:45 tindog kernel: e ide_cd_mod ide_core cdrom jfs edd ext3 mbcache jbd fan pata_jmicron ahci sata_nv pata_amd libata scsi_mod dock thermal processor
Aug 10 18:26:45 tindog kernel: Pid: 7996, comm: insmod Tainted: P 2.6.27-rc2-git4-smp #2
Aug 10 18:26:45 tindog kernel: RIP: 0010:[<ffffffffa0162377>] [<ffffffffa0162377>] nvidia_init_module+0x377/0x7e4 [nvidia]
Aug 10 18:26:45 tindog kernel: RSP: 0018:ffff880100257d48 EFLAGS: 00010202
Aug 10 18:26:45 tindog kernel: RAX: 0000000000000001 RBX: 0000000000000000 RCX: 0000000000000034
Aug 10 18:26:45 tindog kernel: RDX: 0000000000000000 RSI: ffffffff806279d4 RDI: 0000000000000040
Aug 10 18:26:45 tindog kernel: RBP: ffff880100257d98 R08: 0000000000000000 R09: 0000000000000000
Aug 10 18:26:45 tindog kernel: R10: ffff880100257c58 R11: 0000000000000000 R12: 0000000000000001
Aug 10 18:26:45 tindog kernel: R13: ffffffffa0df69c0 R14: 0000000000000001 R15: ffff8801002ec000
Aug 10 18:26:45 tindog kernel: FS: 00007fc07aa786f0(0000) GS:ffffffff80627900(0000) knlGS:00000000f7d28af0
Aug 10 18:26:45 tindog kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
Aug 10 18:26:45 tindog kernel: CR2: 00000000000000d7 CR3: 00000001003aa000 CR4: 00000000000006e0
Aug 10 18:26:45 tindog kernel: DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
Aug 10 18:26:45 tindog kernel: DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Aug 10 18:26:45 tindog kernel: Process insmod (pid: 7996, threadinfo ffff880100256000, task ffff8801050dac60)
Aug 10 18:26:45 tindog kernel: Stack: 0000000000000001 ffffffffa0df69c0 0000000000000001 ffff88010fddb6c0
Aug 10 18:26:45 tindog kernel: 0000000000000001 ffff88010fddb6c0 0000000000000001 ffffffffa0df69c0
Aug 10 18:26:45 tindog kernel: 0000000000000001 ffff8801002e3f00 ffff880100257f78 ffffffff8026081f
Aug 10 18:26:45 tindog kernel: Call Trace:
Aug 10 18:26:45 tindog kernel: [<ffffffff8026081f>] sys_init_module+0x1acf/0x1c77
Aug 10 18:26:45 tindog kernel: [<ffffffff8029452a>] ? vma_link+0x9a/0xab
Aug 10 18:26:45 tindog kernel: [<ffffffff803938e4>] ? acpi_bus_register_driver+0x0/0x40
Aug 10 18:26:45 tindog kernel: [<ffffffff8020c49a>] system_call_fastpath+0x16/0x1b
Aug 10 18:26:45 tindog kernel:
Aug 10 18:26:45 tindog kernel:
Aug 10 18:26:45 tindog kernel: Code: b2 a0 bf 04 00 00 00 bb f4 ff ff ff e8 f5 52 94 00 e9 af 03 00 00 48 c7 c6 d4 79 62 80 bf 34 00 00 00 e8 09 01 94 00 85 c0 74 33 <48> 83 3d 58 dd e9 5f 00 75 04 0f 0b eb fe 48 8d 75 c8 bf 80 00
Aug 10 18:26:45 tindog kernel: RIP [<ffffffffa0162377>] nvidia_init_module+0x377/0x7e4 [nvidia]
Aug 10 18:26:45 tindog kernel: RSP <ffff880100257d48>
Aug 10 18:26:45 tindog kernel: CR2: 00000000000000d7
Aug 10 18:26:45 tindog kernel: ---[ end trace 65e4940d722e4c3a ]---
Aug 10 18:26:59 tindog kernel: warning: process `nvidia-installe' used the deprecated sysctl system call with 1.23.

[vallimar]

Originally Posted by ellson

OK, thanks CraigG4C, that worked better:
The patch-no-xen was necessary, but not sufficient. I also commented out the line
#define CONFIG_XEN 1
from: /usr/src/kernels/2.6.27-0.244.rc2.git1.fc10.x86_64/include/linux/autoconf.h

I'm sure thats not the right fix, but its working for me.

You didn't read the part of my note that said you have to pass IGNORE_XEN_PRESENCE=1 to disable xen checking (just like the normal driver).

[ellson]

Thanks vallimar, you're right, I missed it, but now that Xen is included in the kernel by default, why should the user have to do anything extra? Probably what's needed is a run time check to see what mode the kernel is being used in; not a compile time check.

[KimOlsen]

Got it working on Ubuntu 8.04 with 177.13 and 2.6.27-rc2 with the nvidia-no-xen.patch.txt, but I had to insert and #endif in the nv-vm.c file (because the patch failed on that file). Don't remember where, because I didn't expect it to work, but think it was around line 300.

edit: By the way, I have the 9600m gt

[AbsurdMind]

Originally Posted by zander

177.13 is fairly old and is not recommended for use with GPUs other than GT200, but since there's some demand to get it working with recent kernels until updated display drivers become available, and since there seems to be some confusion with respect to how/why the init_mm symbol is picked up, I attached a patch that updates the 177.13 Linux kernel interface code to be closer to what will be included in upcoming driver releases.

thanks a lot