[llvmlinux] Kernel clang patches rebased to master

Stefan Agner stefan at agner.ch
Thu Apr 20 03:46:00 UTC 2017


Hi Jens,

My branch is not really useful for x86_64 since I dropped some patches.
But the Bernhards branch should afaik, and that branch is ontop of
4.11-rc<something>, hence patches should apply fairly well on newer
trees.

--
Stefan


On 2017-04-17 00:56, Jens Staal wrote:
> Is this rebased version still useful for x86_64 and are the patches easy
> to extract and apply to a vanilla master branch?
> 
> Is there any work still ongoing within in the llvmlinux project?
> last code seems to be for kernel 4.2
> 
> On Mon, Apr 17, 2017 at 12:03:40AM -0700, Stefan Agner wrote:
>> Hi Bernhard,
>>
>>
>>
>> Based on your work, I did a rebase ontop of 4.11-rc7 and compiled for ARMv7
>> (NXP i.MX 7 to be specific). I removed some x86/MIPS specific patches and had
>> to add some arm specific changes. It compiles fine now, boot looks good but as
>> soon as user space kicks in the kernel crashes with "Unable to handle kernel
>> paging request" at fairly random places. The same source tree works fine with
>> gcc. I need to further debug it, but in case you have ideas...
>>
>> My current tree is here:
>>
>> http://git.agner.ch/gitweb/?p=linux.git;a=shortlog;h=refs/heads/
>> armv7-master-clang
>>
>>
>>
>> --
>>
>> Stefan
>>
>>
>>
>> On 2017-02-23 11:59, Bernhard Rosenkränzer wrote:
>>
>>     Hi,
>>     I've rebased all the clang patches to the mainline kernel. The new tree
>>     still lives in the
>>     https://android-git.linaro.org/kernel/hikey-clang.git
>>     repository, master-clang branch (but it's somewhat misnamed because this
>>     branch doesn't have the Hikey or Android patches -- the
>>     android-hikey-linaro-4.9-clang branch in the same repository does).
>>
>>     It compiles for both aarch64 and x86_64 with fairly complete configs, but
>>     currently fails to boot on my notebook (hopefully different hardware will
>>     behave better ;) ).
>>
>>     Patch count is down to 30, some of which are simple enough to go upstream.
>>
>>     I've tested with clang 4.0-rc2, results with older versions may vary (I've
>>     intentionally dropped a couple of patches that were workarounds for bugs in
>>     old clang -- they likely wouldn't get accepted upstream anyway, given "fix
>>     the compiler" is actually the right advice there, and that advice has been
>>     followed already ;) ).
>>
>>     ttyl
>>     bero
>>
>>     _______________________________________________
>>     LLVMLinux mailing list
>>     LLVMLinux at lists.linuxfoundation.org
>>     https://lists.linuxfoundation.org/mailman/listinfo/llvmlinux
>>
>>
>>
>>
> 
>> _______________________________________________
>> LLVMLinux mailing list
>> LLVMLinux at lists.linuxfoundation.org
>> https://lists.linuxfoundation.org/mailman/listinfo/llvmlinux


More information about the LLVMLinux mailing list