I'll think how to rework this.

Concerning formatting - let's wait for another reviewer's opinion then.

Ping

Rebased on top of actual sources.

Updated:

• Remove IndicesVector typedef
• Replace std::copy with insert into vector
• Add more comments to functions

Updated:

• Use std::vector instead of SmallVector
• Add comments to methods
• Rename fillPhdrs -> fillUnusedPhdrs

Ping

Updated:

• move out some functionality to couple of methods
• improve NONE header handling (it now can be specified with other headers and just ignored)

Updated:

• Add method comment
• Simplify custom header layout loop
• Check if no sections assigned to custom headers
• Improve NONE header handling
• Style fixes in the header layout loop

Updated:

• style fixes in linker script's PHDRS parsing
• move custom header processing out to a separate method

In D14140#285339, @ruiu wrote:

One question is the performance with this patch. Does this slow down the linker? If so, how much?

Updated:

• Rename parseSectionDescriptions -> buildSectionMap

Updated:

• Remove OutputSectionDescription structure
• Simplify checks in scan relocs loop

In D14140#282350, @rafael wrote:

It is not clear if this is needed by real program. For now just leave
it out to have a simpler code. Even if it is needed, it would be
better to just organize the code as

• add sections from input files
• sort
• add linker generated sections
• sort the end + merge

no?

Updated:

• Style fixes in linker script
• Remove InputSectionDescription
• Remove hasCustomSections check
• Rename discardInputSection -> isDiscarded
• Make compareSections static

Updated:

• Replace mapBraces method with procedural checks in the linker script parser
• Change peek() -> skip() in the parser
• Remove custom bounding symbols processing for linker script
• Remove checks for regular section ordering for linker script
• Change section comparison lambda with method
• Remove not used fields of input section description struct

I don't think we need that protection. If users try to reorder such
sections, we should do that as instructed, no?

First of all, I reproduced the case when defining 'specific' section in the SECTIONS command affects its ordering in the resulting file without the check, so this check works as I expected.

Update:

• Style fixes
• Use reference in range-for-loop (auto &)

Updated:

• Remove dynamic casting of output section classes
• Remove separate class for linker script support
• Move linker script-related code to Writer and use conditional checks where needed

Will fix style and sort members before commit.

Updated:

• Rename createELFFile -> createELFInputFile
• Rework error handling in createELFInputFile - use early return, split error messages

In D13055#259679, @rafael wrote:

Just one more nit:

in ELFFileBase::getEMachine() I think you have to keep the
llvm_unreachable outside of the switch to avoid a gcc warning.

LGTM with that.

Updated:

• Remove FirstObjName in favor of always reporting incompatibility with the first object file if available.

Updated:

• Remove isCompatibleWith method
• Rename OPT_emulation -> OPT_m
• Rename applyEmulation -> setELFType

Updated:

• simplify initialization of static synthetic symbol
• add test to check that undefined symbol is not in dynamic table

In D13345#257809, @rafael wrote:

Change the last test to create a shared library and check the the 'unknown' symbol shows up only on the regular symbol table, not the dynamic one.

Updated:

• Remove FirstOjbName Config variable, use Symtab.getFirstELF()->getName() instead
• clang-format sources

Sorry for the previous not full comment, just hit Submit button accidentally.

In D13055#256840, @rafael wrote:

You are missing a testcase where -m *is* compatible with the rest of
the elf files :-)

Updated:

• Rename is64Bits() -> is64()
• Add FirstObjFile config parameter
• Introduce ELFNoneKind
• Replace applyEmulation loop with if-else
• Remove configureTarget function
• Drop getElfMachineType function

Updated:

• Make ElfKind optional
• Move Emulations to the place of use
• Rename createTargetELFFile -> createELFFile
• Code style fixes
• Minimize the number of changes by excluding parts that may be removed

Maybe, but that is independent and pretty big, so it should be in another patch.

@rafael, thanks for the patch, but I'd prefer to leave the getElfMachineType method since it allows to simplify the interface of ELFData class and ObjectFile/SharedFile. It removes bunch of methods from the .cpp file as well.

• Fix style
• Fix clang build error
• Provide source of target incompatibility

Updated once again:

• Rename OPT_m -> OPT_emulation
• Fix word spelling in error message

Updated:

• Determine target from the first input file if no emulation given
• Sort emulations alphabetically in the array
• Revert adding -m option to all test cases, add only where needed

I did this so if the host is not x86-64 machine all the tests still pass.

Rafael, thanks for the effort, but I have to abandon this one since the implemented logic is partially wrong. I might re-implement it, but it's more likely to appear in the new linker some time later.

Updated:

• Remove check for duplicate names of output sections.

Rafael, you're right. There may be more than one output section description with the same name. But headers in PHDRS command should have unique names, so I'll leave only check for duplicate header names.

Ping.

LGTM.

Updated:

• return error immediately when encountered
• use SmallVector instead of std::vector for holding matched headers for sections