[15:20] <pevster> Hiya - not sure if this is the right channel but I have a question about how ubuntu deals with modules...?
[15:22] <pevster> I'm trying to build a custom version of a module thats normally in the kernel (can.ko) but not sure if there's a way under ubuntu to do it without re-building / re-installing the stock kernel?
[15:22] <pevster> Ideally UI'm looking for a pointer towards an existing example that I can just dissect ...
[15:40] <apw> pevster, you should be able to build against the "headers" in the normal way treating it as an out of tree driver
[15:40] <apw> of course if you have secure-boot in that environment you may have extra issues
[15:40] <pevster> apw: Thanks - nope, no secure boot here...!
[15:41] <pevster> apw: I also need to override a standard header, should that just work if I put my local include path ahead in the search?
[15:42] <apw> pevster, that i have not tried to do; i am rather spoilt as they builder i use for test builds takes 10m so i tend to just make a whole kernel
[15:42] <apw> pevster, though i would prolly just just ram over the one from the headers package as it is only used for these builds
[15:43] <pevster> 10m? My machine takes about 3 (!) :-O
[15:43] <pevster> For me it's more that I want to keep the ubuntu kernel standard and roll out the two modules onto stock test machines...
[15:44] <apw> yep i can see the use case :)
[15:44] <pevster> feels a bit dirty though :-D
[15:44] <pevster> not a fan of changing standard headers either but I can't see an alternate option...
[15:45] <apw> pevster, the headers an out of tree build uses are not /usr/include but only in /usr/src/* and only used for kernel module builds
[15:47] <pevster> Ah, thanks - I hadn't appreciated that! I'm used to normally just cross compiling whole kernels normally so this is a bit new!
[16:11] <jsalisbury> om26er, Upstream is requesting some data and testing of a set of patches.  I posted the requests to the bug report.
[16:12] <om26er> jsalisbury: yes, I replied to the bug. the latest kernel does not change anything and I have attached the logs
[16:12] <jsalisbury> om26er, great, thanks for the update.  I'll see what they want to do next.
[17:13] <pevster> apw: Thanks for your help ; got most of it working, but can't see how to force my clone of the header to take priority over the "proper" one -the "-I" flag doesnt quite cut it....
[17:14] <apw> pevster, i would just shove it into /usr/src over the top if it was me
[17:16] <apw> pevster, or like cp -lr /usr/src/<headers> to BUILD and then replace it in there, and build against that result 
[17:17] <pevster> Hm... At the moment I've bodged it in the source by using a relative local path and defined the same protection var name to prevent the "proper" one getting included. A bit cludgey though
[17:27] <dgoulet> hi! so I'm having this problem since 4.8.y (vanilla) on Ubuntu with gcc 6.2.0-5ubuntu12:
[17:27] <dgoulet> Cannot use CONFIG_CC_STACKPROTECTOR_STRONG: -fstack-protector-strong available but compiler is broken
[17:28] <dgoulet> whatever stackprotector I put, it fails with the same
[17:32] <apw> i think we specity gcc5 for the kernel build
[17:35] <dgoulet> hrm intersting... this is still failing: "make CC=gcc-5 -j4"
[17:35] <dgoulet> but could be the detection script
[17:39] <pevster> apw: Found it ; prepending LINUXINCLUDE
[17:53] <apw> pevster, sounds good
[18:17] <eagleeyes> infinity: what does dkms do?
[18:17] <eagleeyes> after the bcm kernel source package is installed does it need to have firmware cut?