[PATCH] D61940: [DebugInfo] Don't always extend variable locations when the reg location is unchanging

[Adrian Prantl via Phabricator via llvm-commits]

aprantl added subscribers: jingham, jasonmolenda.
aprantl added a comment.

> My understanding is that we deliberately don't terminate stack-based variable locations during frame destruction, because the debugger should understand that those become invalid as the frame is destroyed.

CC'ing @jasonmolenda and @jingham to double-check that statement.



================
Comment at: lib/CodeGen/AsmPrinter/DbgEntityHistoryCalculator.cpp:265
             // non-CSRs.
-            for (unsigned I : ChangingRegs.set_bits()) {
-              // Don't consider SP to be clobbered by register masks.
-              if (unsigned(I) != SP && TRI->isPhysicalRegister(I) &&
-                  MO.clobbersPhysReg(I)) {
-                clobberRegisterUses(RegVars, I, DbgValues, LiveEntries, MI);
-              }
+            SmallVector<unsigned, 8> RegsToClobber;
+            // Don't consider SP to be clobbered by register masks.
----------------
My intuition says that your typical function call will clobber more than 8 registers, but I don't know what a good number is either.


================
Comment at: lib/CodeGen/AsmPrinter/DbgEntityHistoryCalculator.cpp:282
 
       if (MI.isDebugValue()) {
         assert(MI.getNumOperands() > 1 && "Invalid DBG_VALUE instruction!");
----------------
For readability it might be a good idea to move the short cases to the top, with a `continue`, then we have more horizontal space for the meat of the function.


================
Comment at: test/DebugInfo/AArch64/struct_by_value.ll:3
+; RUN: llc %s -filetype=obj -o - | llvm-dwarfdump - | FileCheck %s
+; REQUIRES: object-emission
 
----------------
Thanks! Feel free to land this ahead of time as an NFC commit.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D61940/new/

https://reviews.llvm.org/D61940