[PATCH] D24706: [ELF] - Partial support of --gdb-index command line option (Part 1).

[Greg Clayton via llvm-commits]

> On Oct 6, 2016, at 10:02 AM, Adrian Prantl <aprantl at apple.com> wrote:
> 
> 
>> On Oct 6, 2016, at 9:58 AM, Eric Christopher <echristo at gmail.com> wrote:
>> 
>> 
>> 
>> On Thu, Oct 6, 2016 at 9:48 AM George Rimar <grimar at accesssoftek.com> wrote:
>> Thanks for explanation, David !
>>   
>> >If other projects are anything to go by, that rarely, if ever, happens. So we tend to like to frontload some of this work to avoid >divergence which makes the cost more expensive over time and less likely to be done.
>> >
>> >(see LLDB's use of a fork of the LLVM DWARF parsing logic (well, I think LLVM maybe forked LLDB's, but either way) as a >cautionary tale)
>>  

To clarify one thing: the LLVM DWARF parser was a copy of the LLDB DWARF parser and the original port to LLVM didn't change LLDB over to use it as it was too complex for the original person to accomplish. The DWARF parsing code originated in LLDB and the fork exists in LLVM.

That being said, I am working on modifying the current LLVM DWARF parser to do everything that the LLDB DWARF parser does and am trying to switch LLDB over to using the newly modified DWARF parser. Any performance issues will be resolved during this port. If anyone has any questions about anything that is needed that isn't currently in the LLVM DWARF parser, please let me know.

Greg

>> I understand your conserns. But looking on full version of functionality (D24267), I see that linker
>> needs just a little amount of parsing code. And I do not expect we will need much more.
>> If we one day we will have fast reusable llvm parsers, I think it be not hard to switch lld to use them (watching for perfomance changes), but for now I think own short implementation that do only things required for generation gdb_index fits better.
>> 
>> 
>> There's an assertion here that the current implementation is not fast or reusable. I'm not sure where that came from or what prompted it.
>> 
>> Could you elaborate more on why you think these things?
> 
> I would also like to add that Greg Clayton is currently working porting LLDB over to use LLVM's dwarf parser (and add any missing functionality to LLVM's implementation).
> 
> -- adrian