GSoC 2015 - Clang support and Eclipse plugin work

Joel Sherrill joel.sherrill at oarcorp.com
Tue Mar 24 16:44:02 UTC 2015



On 3/24/2015 10:05 AM, Gedare Bloom wrote:
> On Tue, Mar 24, 2015 at 10:52 AM, Charith Eranga
> <eranga.charith at gmail.com> wrote:
>> Hi All,
>>
>> I'm Charith Eranga from University of Moratuwa, Sri Lanka. I'm
>> currently a second year student specializing in embedded software
>> development.
>>
>> I'm particularly interested in compiling RTEMS with Clang [1] and the
>> improvements to the Eclipse plugin [2]. I've made a start with RTEMS
>> by compiling toolchain/rtems following the GSoC guide [3].
>>
>> For clang support, I've made start by following the steps of [1]. I
>> also have an industry contact (llvm/clang committer) who's willing to
>> help me on that side of things. One question I have about this is, how
> It would be most helpful if your contact would be willing to sign-up
> in Google Melange as a mentor for RTEMS Project. Please have him/her
> contact me directly for any clarifications.
+1
>> do we define a success measure for this project? I mean, RTEMS seem to
>> already compile with Clang with some local patches [1], but how do we
> Somewhat true, but I believe there are problems with using clang that
> have to do with the pervasive use of gcc spec files. A pre-requisite
> task may be to get the spec files completely eliminated, which there
> is an ongoing effort to do with the "waf branch" of RTEMS [1ing tghe].
I am suspicious that most of what is in the specs files is obsolete and
already
taken into account inside gcc by our rtems specific tweaks.  GCC spec way
extension is a way to accomplish things that can be done inside GCC but
without modifying GCC source.  The specs use predates us having many
gcc tweaks.

The short version is that a careful review may significantly reduce them.

Also I have long had the idea that adding an option for an rtems bsp
to gcc and clang could eliminate some needs also. The -B option is
mostly to set an include and lib path to get the BSP.

Random core dump of thoughts of unsure value.
> I think it makes the most sense to work from the waf branch and try to
> help push it forward.
+1
>> evaluate whether the resulting binary is in good shape? should I also
>> be thinking of some test-plan as part of the project? Secondly, will
> RTEMS has an extensive test-suite [2], and you should be prepared to
> run it before/after. We have a tool that helps with automation [3].
>
>> it be OK if I fork off clang in github and maintain the patches there?
>>
> Yes, we prefer for gsoc development to be done on github.
>
>> I haven't yet tried out the RTEMS Eclipse plugin (on my TODO list),
>> but I have an interest in learning Eclipse plugin development. I
>> suppose the objectives of this project would be implement as much as
>> possible in the TODO list of [2]?
>>
> Yes, but some of those may be obsolete. Perhaps users will chime in
> about what they would like to see in it.
Agreed. And suggesting Eclipse capabilities that are useful is welcomed.
>> Many thanks for your comments.
>>
>> - Charith
>>
>> PS: "Hello World" exercise attached herewith.
> Confirmed, thanks. Feel free to add yourself to the "Tracking Table" for 2015.
>
> [1] https://git.rtems.org/amar/waf.git/
> [2] https://git.rtems.org/rtems/tree/testsuites
> [3] https://git.rtems.org/rtems-tools/tree/tester
>
> Gedare
>
>> [1] https://devel.rtems.org/wiki/Projects/CLANG
>> [2] https://devel.rtems.org/wiki/Developer/Eclipse/Information
>> [3] https://devel.rtems.org/wiki/GSoC/GettingStarted
>>
>> _______________________________________________
>> devel mailing list
>> devel at rtems.org
>> http://lists.rtems.org/mailman/listinfo/devel
> _______________________________________________
> devel mailing list
> devel at rtems.org
> http://lists.rtems.org/mailman/listinfo/devel

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel.sherrill at OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985





More information about the devel mailing list