Bug report on rl78-elf-ld
井原 央翔 <[email protected]>
| Newsgroups | gmane.comp.gnu.utils.bugs |
|---|---|
| Message-ID | <[email protected]> |
Dear binutils development team,
I am very glad to be able to use free tools for RL78, and I must first appreciate the efficiency of rl78-binutils.
There was trouble when I attempted to create ROM image directly from ld, and I later found out that rl78-elf-ld seems to give an incorrect output in some formats.
I tried Intel HEX, Motorola S-Record, and Binary, all of which worked incorrectly in exactly the same way.
The problem is avoidable as long as the elf32 output is consistent with the objs, so this bug might not be crucial.
Sincerely,
Hiroka Ihara
***************
Environment: MinGW on Windows 8 Professional, build 9200 (Japanese OEM Edition)
rl78-elf-ld 2.23.2 built using gcc-4.7.2, no patch applied
Command:
rl78-elf-as -omain.o program.S
rl78-elf-ld -omain.hex --oformat=ihex -Tldscript.x main.o
Problem:
Some part of the HEX file is replaced with invalid bytes.
This part of the HEX file exported via rl78-elf-objcopy after being linked into elf32:
main2.hex
:05000000FC040000D724
has appeared in the HEX file exported from the rl78-elf-ld as:
main.hex
:05000000FC04000000FB
where the former is correct.
As is easily noticeable, the last byte (D7) is cleared.
The same phenomenon is observed also in Binary and S-Record formats.
***************
----------------------------------------------------------------------------------
The University of Tokyo, Faculty of Arts and Science
Hiroka Ihara
[email protected]
----------------------------------------------------------------------------------
files.zip
(application/x-zip-compressed, 1.1 KB) - not displayed