/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h: At top level:
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:151:2: error: #error “struct file_operations compile test likely failed!”
In file included from /tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv.c:13:0:
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:306:2: error: #error “NV_PCI_DMA_MAPPING_ERROR() undefined!”
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:311:4: warning: “NV_ACPI_WALK_NAMESPACE_ARGUMENT_COUNT” is not defined [-Wundef]
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:313:8: warning: “NV_ACPI_WALK_NAMESPACE_ARGUMENT_COUNT” is not defined [-Wundef]
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:319:2: error: #error “NV_ACPI_WALK_NAMESPACE_ARGUMENT_COUNT value unrecognized!”
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:325:4: warning: “NV_ACPI_OS_WAIT_EVENTS_COMPLETE_ARGUMENT_COUNT” is not defined [-Wundef]
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:328:8: warning: “NV_ACPI_OS_WAIT_EVENTS_COMPLETE_ARGUMENT_COUNT” is not defined [-Wundef]
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:834:2: error: #error “NV_KMEM_CACHE_CREATE() undefined (kmem_cache_create() unavailable)!”
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:893:2: error: #error “NV_VMAP() undefined (vmap() unavailable)!”
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:965:2: error: #error “NV_SMP_CALL_FUNCTION() undefined (smp_call_function() unavailable)!”
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:985:2: error: #error “NV_ON_EACH_CPU() undefined (on_each_cpu() unavailable)!”
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h: In function ‘nv_execute_on_all_cpus’:
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:1003:5: error: implicit declaration of function ‘NV_ON_EACH_CPU’ [-Werror=implicit-function-decl
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h: At top level:
/tmp/selfgz1980/NVIDIA-Linux-x86-313.18/kernel/nv-linux.h:1262:13: error: conflicting types for ‘pm_message_t’
In file included from /usr/src/kernels/linux-3.8/arch/x86/include/asm/apic.h:5:0,
You can fetch the patches to fix this problem here.
I’m quite disappointed we have to resort to 3d party hacks to enable compatibility. NVIDIA used to be vigilant in this regard (I guess they are busy with the Tegra 4).
Stable kernel 3.8.1 is out now and we’re up to stable kernel 3.7.10 in the 3.7 tree. I agree it’s bad there still isn’t an official fix for these compatibility issues. End-users, in my opinion, shouldn’t have to rely on hacks & patches for compatibility over that many kernel versions or kernel trees.
You say the installer fix will be available in the next release driver, but what is that supposed to be released? I see people posting about these compatibility issues everywhere but have yet to see any ETA on an official fix. It reminds me of people asking when humans will land on the mars and the answer they’re given is “sometime in the future”.
When compatibility has fallen several stable kernel releases behind, I don’t think it’s unreasonable to expect an ETA for an official fix. Leaving your users hanging out in the wind is not exactly good PR.
I absolutely disagree with your claim. A lot of people do use their distros stock kernel, but a whole hell of a lot of users don’t. They update or compile their own. This is blatantly obvious when you frequent many of the big Linux forums and mailing lists.
If this wasn’t a problem, posts about it wouldn’t be so readily available upon searching.
While I fully agree it’s far from ideal that community patches are required, this talk of distros “suffering” or “being without a driver” is quite silly. Arch provides working nvidia packages for the latest kernels. And if Arch, a niche distro made by volunteers can do it, then the bigger distros like OpenSuse and Fedora should be able to as well.
Gentoo is having problems as well, with ALL their kernels. They have decided to allow users to throw in patches via their portage scheme but not to do any more without nVidia getting a fixed driver set out.
Gotta chime in as a Lubuntu 12.10 user and Linux 3d game developer with a 3.5.0-26 generic kernel.
That’s selective observation. Just because a minority of loud people make a lot of noise in some forums, and those forums are picked up by search engines, doesn’t mean they represent a huge customer base of people with problems in the real world. To some extent kernel compilers and tweakers create their own misery, and they are not necessarily to be pitied.
What I’d like to see from NVIDIA, selfishly enough, is a good ecology of driver support for 3d game customers on common Linux systems. I’m chasing some 50 Hz vs. 60 Hz frontend vs. backend issue and going up a learning curve about it.