[PATCH] D43366: [DWARF v5] Supporting verbose dumping of .dbg_rnglist entries

[Wolfgang Pieb via Phabricator via llvm-commits]

wolfgangp added inline comments.


================
Comment at: include/llvm/DebugInfo/DWARF/DWARFAddressRange.h:51
 
-  void dump(raw_ostream &OS, uint32_t AddressSize) const;
+  void dump(raw_ostream &OS, uint32_t AddressSize, bool Brackets = true) const;
 };
----------------
JDevlieghere wrote:
> Maybe this is an option we can store in the `DIDumpOptions`? 
Decided on an option "DisplayRawContents" with the intent to tell anybody who looks at it to not semantically interpret the content. It's a bit awkward as it's not settable by a user, but it's perhaps closer to what you had in mind.


================
Comment at: test/tools/llvm-dwarfdump/X86/debug_rnglists.s:19-24
+# VERBOSE-NEXT: 0x0000000c: [DW_RLE_start_end]: [0x0000000000000010, 0x0000000000000020)
+# VERBOSE-NEXT: 0x0000001d: [DW_RLE_start_length]: 0x0000000000000025, 0x0000000000000080
+# VERBOSE-SAME: => [0x0000000000000025, 0x00000000000000a5)
+# VERBOSE-NEXT: 0x00000028: [DW_RLE_end_of_list]
+# VERBOSE-NEXT: 0x00000029: [DW_RLE_start_end]: [0x0000000000000100, 0x0000000000000200)
+# VERBOSE-NEXT: 0x0000003a: [DW_RLE_end_of_list]
----------------
dblaikie wrote:
> Probably worth making the prefix indentation consistent (ie: padding the [DW_RLE_...] with spaces on the right, so the DW_RLE_s line up, and the range lines up on the RHS too?
Collecting the length of the maximal DW_RLE* string and using it to align things. The output still doesn't seem to win any aesthetic awards, though.


https://reviews.llvm.org/D43366