Superseded by MaskRay's relaxation diffs

• rebase
• rework HiLoAddend as a union

• rework AlignAddend as a union

• Rebase, minor cleanups, minor comment revisions

• revise comments according to review feedback

• fix indentation, as caught by clang-format

• fix indentation, as caught by clang-format

• fix indentation, as caught by clang-format

• Remove glob patterns from ld.lld command lines in lld/test/ELF/riscv-relax-call-undo.s because Windows doesn't support them.

• Cleanup & improve commentary in lld/test/ELF/riscv-relax-syms.s

• add fillAlignGap(), extracted from fillAdjustGaps()
• add empty cases for HI20/LO12 relaxations, to be fleshed-out in a later diff.

Expand comments before relaxOnce() and fillAdjustGaps()

• restructure NOP-rewriter loop in fillAdjustGaps()

• revise according to review feedback

• Fix bug reported by @luismarques regarding relaxation of calls through PLT

• Add capacity to undo earlier relaxations that go out of range after later passes

• revise according to review feedback

LLD no longer crashes when linking Linux.

• open code the trivial getter/setter functions atop R_RISCV_ALIGN's Relocation::addend for align boundary & nop byte count.

• Rework class SymbolAddr. The broken version's comparison function confused llvm::sort by returning true when given the same object for args a and b. This violates the API because comparisions should return a < b, which is false when a and b are the same object.

• Handle case R_RELAX_HINT in InputSectionBase::getRelocTargetVA()

• Revise according to review feedback

In D125036#3513877, @MaskRay wrote:

Clone https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git , apply this patch:

. . .

I see a segfault

• Fix indentation caught by clang-format

Refactor ...

• s/DeleteRanges/AdjustRanges/: these can now handle deletion and insertion, supporting multiple-pass relaxation where alignment padding can expand, and/or relaxations can be undone
• Build framework for adding more relaxation types in later diffs
• Add comments

• Revise according to review feedback
• Rewrite InputSectionBase::deleteRanges() to use a compact single pass that unifies handling of symbol addresses and sizes
• expand lld/test/ELF/riscv-relax-syms.s

Revise according to review feedback:

• place new bool member adjacent to another bool to avoid size change of lld/ELF/InputSection.cpp
• add test case for symbol-size

RIP

In D124794#3486448, @MaskRay wrote:

I agree that people should just symlink ld.lld to ld.bfd to check this, or hack their local lit.local.cfg.
Such changes do not need to be in the upstream.

In D124794#3486452, @jrtc27 wrote:

What happens when we want to use LLD-specific options or are checking LLD-specific messages? Those tests aren't going to pass. Do we then need to annotate LLD tests with REQUIRES: lld-is-actually-lld? Is there any precedent for doing this on other architectures? This doesn't feel right to me.

Add commented-out 32-bit GNU linker alongside 64-bit

In D124794#3486424, @jrtc27 wrote:

Why can't people just symlink GNU ld to ld.lld?

Fix stale comment

Rebase to get 1 year of LLVM changes

Original author @PkmX has been inactive here for a year, and did not respond to email.

• Revise according to review feedback

• Revise according to review feedback.
• Move 2 more hunks of NFC-ish code into this diff to ignore useless symbols and SUBTRACTOR relocs in __TEXT,__eh_frame. (See code comments)

• fix one remaining s/ParsedSection/Section/

• revert Parsed prefix to shorten class names and lower impact of this diff

Revert previous revision

Major FUBAR revision -- ignore this one.

In D113241#3114368, @thevinster wrote:

I'm a big fan of the variable names switching from s/subsections/sections, s/subsectionMap/subsections. It's clearer to me that there are layers whereas the original subsections->subsectionMap layer is not as easily decipherable.

In D113241#3113825, @int3 wrote:

Not a huge fan of the Parsed stuff either, I don't think there's much confusion about when we are working with parsed vs serialized structures...

In D113241#3113825, @int3 wrote:

The Map instead of Vec thing was intentional... a vector of pairs is semantically a map.

I missed a rename ...

• s/Subsection/ParsedSubsection/

LGTM! Thank you for this. It makes __eh_frame handling much nicer too.
Please fix test failure for MachO/compact-unwind-both-local-and-dylib-personality.s.

Nice! Yes, the terse non-descriptive names were getting out of hand.

Bueno!

LGTM

In D105258#2852740, @thakis wrote:

That looks ok, but doesn't llvm-dwarfdump --eh-frame file do this already?

LGTM!

Groovy!

Bueno!

• revise according to review feedback

Bueno!