================ @@ -11,7 +11,6 @@ // NOINLINE-NOT: "-fsplit-dwarf-inlining" // SPLIT-NOT: "-dumpdir" // SPLIT: "-debug-info-kind=constructor" -// SPLIT-SAME: "-ggnu-pubnames" ---------------- dwblaikie wrote:
> As discussed in original PR, #82840 , the idea is if -gsplit-dwarf with > -glldb is used the -ggnu-pubnames wouldn't be passed in. Since LLDB doesn't > benefit from it, and it just uses space. Yep, I'm with you on all of that. > A fix from original is if someone does specify -ggnu-pubnames it will be > generated. Sure. > I added a test for it on linux side. Added a test for what on the linux side? And where is that test? > There is also a test on mac side that used -gsplit-dwarf with -ggnu-pubnames. Right, I see that test here in this `split-debug.c` file at the end. Checking for the absence of `-ggnu-pubnames` if `-gsplit-dwarf -g -glldb` is passed, and the presence if `-gsplit-dwarf -g -glldb -ggnu-pubnames` is passed. So I'm with you there. But what i don't understand is that, with the removal of the `// SPLIT-SAME: "-ggnu-pubnames"` line above - what tests that, for instance `-gsplit-dwarf -g -ggdb` does use `-ggnu-pubnames`? (said another way: If we removed the any code that added `-ggnu-pubnames` when `-gsplit-dwarf` is specified, which lit tests would fail?) https://github.com/llvm/llvm-project/pull/83331 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
