[PATCH] D18960: [lld] Initial implementation of TLSDESC relocation handling

[Rui Ueyama via llvm-commits]

ruiu added a comment.

This seems to conflict with http://reviews.llvm.org/D18711, so you may want to wait for that.


================
Comment at: ELF/InputSection.cpp:301
@@ +300,3 @@
+      Target->relocateOne(BufLoc, BufEnd, Type, AddrLoc,
+                          Body.getGotPltTlsDescVA<ELFT>() + A);
+      continue;
----------------
I'd name this getTlsDescVA rather than getGotPltTlsDescVA if there's no reason to include "GotPlt".

================
Comment at: ELF/OutputSections.h:164
@@ -156,3 +163,3 @@
 private:
-  std::vector<const SymbolBody *> Entries;
+  std::vector<std::vector<const SymbolBody *>> Entries;
 };
----------------
This code is hard to understand. If I understand it correctly, it is a vector of vectors binned according to their DynReloc_* types. Is this correct?

There seems to be no reason to aggregate multiple vectors into one vector (of vectors) here. Why don't you define multiple vectors?


http://reviews.llvm.org/D18960