[PATCH] Subclass DynamicTable for each targets to factor out handling of target dependent part of a dynamic table

[Shankar Easwaran]

On 1/29/2014 1:31 AM, Simon Atanasyan wrote:
>    I thought about the code duplication problem but could not figure out an elegant solution.
>
>    Sure I can create a new class say `DefaultDynamicTable` subclassed from the `DynamicTable` and override the `createDefaultEntries` method. Then create `DefaultDynamicTable` for every targets except MIPS. But I do not like the word **Default** in the `DefaultDynamicTable` because using `DT_PLTGOT` for `.got.plt` section referencing is not default behavior but is just one of possible cases.
>
>    Another solution is to add a new virtual function `DynamicTable::getPLTGOTTag()`. By default it returns `DT_PLTGOT`, in case of MIPS it returns `DT_MIPS_PLTGOT`.

This is one way that I want to go about :-

DefaultDynamicTable
    - addTag
    - setTag

GnuLDDynamicTable : DefaultDynamicTable
    - add tags that is needed to duplicate default GnuLD behavior for 
all architectures.
    - add mappings that translate from one tag to another (to permit 
architectures to override).

MipsDynamicTable : GnuLDDynamicTable
     - override specific tags

Thanks

Shankar Easwaran

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by the Linux Foundation