Bug#735630: doesn't work with experimental kernels without the patch

[Maurizio Avogadro]

Hi,

Il 26/01/2014 03:17, Andreas Beckmann ha scritto:
> On 2014-01-26 02:38, Adam Borowski wrote:
>> [...]
> 
> Thats weird. Did anything change in 3.13 w.r.t. to that symbol? Links to
> some commits would be nice ...

https://lkml.org/lkml/2013/11/5/515 should be the original patch
removing EXPORT_SYMBOL(acpi_os_wait_events_complete)

> [...]
> 
>> As for the second problem Maurizio reported: I got no EFI system, nor a
>> relevant laptop, so I can't reproduce.  The original patch is enough for my
>> system, as far as I can tell.

Weird. My system is not EFI too, nevertheless it needs the additional patch.

> At least the need for this patch I understand, thanks to the kernel commit.
> 
> I would probably do something like
> 
> #if !defined(OLDNAME) && defined(NEWNAME)
>   #define OLDNAME NEWNAME
> #endif
> 
> to stay backwards-compatible.
> 
> BTW, this does not look correct:
> 
> #ifdef FOO
> #elif BAR  // should be defined(BAR) instead
> #endif

Sorry for the bad quality of my patch, I'm definitely not a C
programmer; I only wanted to share my experience just in case someone
else will have to face that issue.

Fedora guys are facing the same issue and are applying a patch (pls. see
http://cvs.rpmfusion.org/viewvc/rpms/nvidia-kmod/devel/nvidia_3.13_kernel.patch?hideattic=1&revision=1.3&root=nonfree&view=markup).

The 1.2 version of that patch also includes a define for the missing
DEVICE_ACPI_HANDLE macro (removed in v1.3) which perfectly fixes all
issues here.

> Is there any discussion about kernel 3.313 with a response from nvidia
> in the nvidia forum?

Please see
https://devtalk.nvidia.com/default/topic/644906/linux/331-20-on-3-13-rc1-kernel
but no response from nvidia AFAICS.

Thanks


Maurizio