[llvm-dev] [cfe-dev] [GSOC 2018] Information gathering

Paul Semel via llvm-dev llvm-dev at lists.llvm.org
Tue Mar 20 07:06:40 PDT 2018


Hi,

On 03/20/2018 06:05 AM, Eric Christopher wrote:
> 
> 
> On Fri, Mar 16, 2018 at 1:57 AM Paul Semel <semelpaul at gmail.com 
> <mailto:semelpaul at gmail.com>> wrote:
> 
>     Hi Eric,
> 
> 
>     On 03/15/2018 04:33 PM, Eric Christopher wrote:
>>     Hi Paul,
>>
>>
>>         >> I'm also interested in the command line replacements for
>>         GNU Binutils :
>>         >>
>>         >> - What tools would you like to replace in priority ?
>>         >> - Does this subject imply to add options/features to some
>>         of the
>>         >> tools, or is it only about handling command line ?
>>         >
>>
>>
>>     I just replied with this in another thread:
>>
>>     "It's currently still available. The basic idea is that we'd be
>>     working on getting each of the llvm tools or libraries with a
>>     front end that is command line compatible with the GNU binutils
>>     counterpart to serve as a replacement. Whether or not we made them
>>     output compatible is something else, but we'll probably want to
>>     have a couple different modes there from:
>>
>>     a) The compatible tool,
>>     b) The tool we all want.
>>
>>     A and B could be the same, but then again, they might not. The low
>>     bar for the SoC project is going to be A."
>>
>>     And in priority order I'd probably want to finish off objcopy
>>     support (see the recent thread on llvm-dev) and
>>     objdump/readobj/readelf and then go from there.
>>
>>     Thoughts?
>>
>>     -eric
> 
>     I saw the thread you are talking about. So basically, the idea would
>     be to do the correct calls for either COFF subset of functions of
>     ELF ones wether we have a COFF or ELF file as an input.
>     Am I right ?
> 
> 
> Basically what I'm looking for first is a command line equivalent 
> replacement first for gnu objcopy. I'd focus on ELF first, and then move 
> to COFF/PE. I'd start from the work that Jake (cc'd) has already done 
> and work with Zach (cc'd) on the COFF stuff if he's still interested. Of 
> course, I'll be around for the first bit.
> 
> Then follow up with objcopy, etc as there's time.
> 

I think you meant objdump, right ? (you talked about objcopy in your 
previous paragraph).

>     I am really interested in doing a proposal for this subject. What do
>     you expect to be in it ? I was actually thinking of something like
>     exposing the things I've done in LLVM/CLang, the schedule for the 3
>     months (but for this, I need to talk with you about the high
>     priority tools, as I'm not sure it is possible to do all the
>     frontend tools in such amount of time)..
> 
> 
> Showing off your previous work is absolutely great in a proposal. A 
> timeline and some proof that you've at least looked at what's missing 
> and have ideas at how to do the work would be key. And I don't really 
> expect you to finish all of them - at least not without help, but with 
> some luck there might be other contributors to help :)
> 

Alright, that sounds very good ! For the moment, what I've done is that 
I listed the tools that were needed command line replacements (for some 
of those it is really binign).

Do I need to take LLD into account in my timeline ?

Then, I investigated a bit on the different tools command line, and what 
I have learnt so far is that objdump and objcopy are the ones that 
require the biggest amount of work (again, not took LLD into account so 
far).

> Sound good? We can definitely work on the details as you're interested - 
> I'll also be more responsive in the near future as well.
> 
I have shared my draft in the GSOC 2018 Dashboard, but here is a link so 
that you have it right in the email[0]. I would really like to have 
feedback on it, espacially for the timeline I made. (but I'd really 
appreciate for the rest of the draft too 🙂).

I am actually not sure at all about the time it would take for the 
replacement of llvm-objcopy, so maybe Jake and/or Zach would have an 
idea about it, as they already worked on this subject ! 🙂

> Thanks!
> 
> -eric

Thanks,

-- 
Paul Semel

[0] 
https://docs.google.com/document/d/14gEdNv-X6p_a6Hsqvb1PmQcaXHateCct1yEhLEFb2-I


More information about the llvm-dev mailing list