https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88644
Bug ID: 88644
Summary: Unexpected pub type info eliminated after r 246973
(causes pubtypes-*.c to regress).
Product: gcc
Version: 9.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: iains at gcc dot gnu.org
Target Milestone: ---
r246973 addresses PR debug/80263 with :
* dwarf2out.c (modified_type_die): Try harder not to emit internal
sizetype type into debug info.
This appears to do the expected thing for debug_info on Darwin (i.e. same as
Linux):
.byte 0x7 # DW_AT_encoding
- .ascii "sizetype\0" # DW_AT_name
- .byte 0x2 # uleb128 0x2; (DIE (0x18c) DW_TAG_base_type)
+ .ascii "long unsigned int\0" # DW_AT_name
+ .byte 0x2 # uleb128 0x2; (DIE (0x1a4) DW_TAG_base_type)
and..
.byte 0x8 # DW_AT_byte_size
- .byte 0x7 # DW_AT_encoding
- .ascii "long unsigned int\0" # DW_AT_name
- .byte 0x5 # uleb128 0x5; (DIE (0x1ae) DW_TAG_pointer_type)
- .byte 0x8 # DW_AT_byte_size
- .byte 0x4 # uleb128 0x4; (DIE (0x1b0) DW_TAG_typedef)
+ .byte 0x4 # uleb128 0x4; (DIE (0x1b3) DW_TAG_typedef)
but it has an unexpected effect on pubtypes
.ascii "sizetype\0" # external name
- .long 0x18c # DIE offset
+ .long 0x1a4 # DIE offset
+ .byte 0x90 # GDB-index flags
.ascii "char\0" # external name
- .long 0x199 # DIE offset
- .ascii "long unsigned int\0" # external name
- .long 0x1b0 # DIE offset
+ .long 0x1b3 # DIE offset
+ .byte 0x90 # GDB-index flags
So .. here the sizetype is still emitted (not intended) and the underlying type
is not.
Linux does the expected thing.
AFAICT, the only difference appears to be that "long unsigned int" appears
*after* "sizetype" on Darwin but *before* it on Linux.
