[llvm-bugs] [Bug 37257] New: Significant binary section size differences between ld.lld and ld.bfd

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=37257

            Bug ID: 37257
           Summary: Significant binary section size differences between
                    ld.lld and ld.bfd
           Product: lld
           Version: unspecified
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: ELF
          Assignee: unassignedbugs at nondot.org
          Reporter: ppadevski at gmail.com
                CC: llvm-bugs at lists.llvm.org

Hi,

I'm not a linker expert but I tried replacing ld.bfd with ld.lld today and
noticed a couple of differences.

For executables there isn't much difference

text       data      bss      dec        hex       filename
25679274   1969696   240249   27889219   1a98e43   exe+ld.bfd
25716658   1963032   238542   27918232   1a9ff98   exe+ld.lld
  +37384     -6664    -1707     +29013

For shared libraries, however, the difference is

text       data      bss     dec        hex       filename
31035525   3179176   46736   34261437   20ac9bd   so+ld.bfd
31419721   3185072   46736   34651529   210bd89   so+ld.lld
 +384196     +5896       0    +390092

If we take a look at both libraries via 'size -A', we'll see two differences.

so+ld.lld
section                  size
.hash                  746520
.rela.plt              427296

so+ld.bfd
section                  size
.hash                  504348

Both libraries and executables are linked with full RELRO (z relro, z now).
If I omit 'z now' when linking with the shared library with ld.bfd I can see

so+ld.bfd
section                  size
.hash                  504348
.rela.plt              427272

I have two questions:
1. Is there a way for the .hash section produced by lld to get smaller or is it
more efficient this way?
2. Is there a way to get rid of the .rela.plt section when linking with lld or
is it really needed?

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20180426/216223ca/attachment.html>