[PATCH] D107782: ReleaseNotes: add lld/ELF notes

James Henderson via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Aug 16 03:55:00 PDT 2021


jhenderson added a comment.

Some grammar suggestions from me. As with @smeenai, I've not been following development closely enough to confirm that these are all the changes that should be documented though.



================
Comment at: lld/docs/ReleaseNotes.rst:44
+  (`D101996 <https://reviews.llvm.org/D101996>`_)
+* ``__rela_iplt_start`` is not defined for -pie/-shared now.
+  This makes GCC/Clang ``-static-pie`` built executable work.
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:47
+  (`rG8cb78e99 <https://reviews.llvm.org/rf8cb78e99aae9aa3f89f7bfe667db2c5b767f21f>`_)
+* IRELATIVE/TLSDESC relocations support ``-z rel`` now.
+  (`D100544 <https://reviews.llvm.org/D100544>`_)
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:53
+  (`D106228 <https://reviews.llvm.org/D106228>`_)
+* Defined symbols are resolved before undefined symbols.
+  (`D95985 <https://reviews.llvm.org/D95985>`_)
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:55
+  (`D95985 <https://reviews.llvm.org/D95985>`_)
+* ``STB_WEAK`` symbols are preferred over COMMON symbols as a fix to a ``--fortran-common`` regression.
+  (`D105945 <https://reviews.llvm.org/D105945>`_)
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:62
+  (`rGe7a7ad13 <https://reviews.llvm.org/rGe7a7ad134fe182aad190cb3ebc441164470e92f5>`_)
+* ``.`` in the output section description of ``.tbss`` is now correct.
+  (`D107288 <https://reviews.llvm.org/D107288>`_)
----------------
I think this is what is meant?

Also should this and the other linker-script related changes have a note saying they are for linker scripts (e.g. in this case "in the linker script output section description ...")?


================
Comment at: lld/docs/ReleaseNotes.rst:81
+  (`D107234 <https://reviews.llvm.org/D107234>`_)
+* If an object file defines both ``foo`` and ``foo at v1``, only ``foo at v1`` will be in the output.
+  (`D107235 <https://reviews.llvm.org/D107235>`_)
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:87
+  (`D96188 <https://reviews.llvm.org/D96188>`_)
+* The AMDGPU port has now supported ``--amdhsa-code-object-version=4`` object files;
+  (`D95811 <https://reviews.llvm.org/D95811>`_)
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:89
+  (`D95811 <https://reviews.llvm.org/D95811>`_)
+* The ARM port has now accounted for PC biases in range extension thunk creation.
+  (`D97550 <https://reviews.llvm.org/D97550>`_)
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:97
+  (`D94625 <https://reviews.llvm.org/D94625>`_)
+* The PowerPC port has fixed a thunk creation bug when TOC/NOTOC calls are mixed.
+  (`D101837 <https://reviews.llvm.org/D101837>`_)
----------------



================
Comment at: lld/docs/ReleaseNotes.rst:99
+  (`D101837 <https://reviews.llvm.org/D101837>`_)
+* The RISC-V port now resolves undefined weak relocations to current location if not using PLT.
+  (`D103001 <https://reviews.llvm.org/D103001>`_)
----------------



Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D107782/new/

https://reviews.llvm.org/D107782



More information about the llvm-commits mailing list