|Re: [hatari-devel] A few more patches for debuginfo of osheader & system variables|
[ Thread Index |
| More lists.tuxfamily.org/hatari-devel Archives
Thanks for the patches!
On 12/6/18 10:43 AM, Thorsten Otto wrote:
attached are 2 patches which should (hopefully) improve debug output. The
first changes the behaviour of printing the osheader. The current solution
does not print anything when sysbase does not point to the ROM, which is
rather contra productive. It does not only happen with MiNT, but also
when using tools that relocate the TOS to RAM like GEMRAM, etc.
Yes, that's definitely an improvement.
Regarding "os_date" output you added, I've never seen it differ from
"os_dos_date" (except one being internally in hex, and another in
GEMDOS date format). In what situations they differ?
(If they don't differ, it's redundant.)
> also with TOS 2.x and 3.x while running programs from auto-folder,
I couldn't reproduce that. Current "info osheader" looks just
fine for me when TOS 2/3 versions run AUTO programs?
The 2nd patch adds a new subcommand to info which prints all the system
I think this will be something clearly appreciated. :-)
(Like your a.out debug symbols support was.)
The output of it is a bit long however, maybe that should be split
into exception vectors, and other system variables?
Definitely. It's so long that I think xconin/out/stat pointers
could also be on their own, or even dropped. What others think?
I would also want some sub-headings for groups of addresses,
- CPU exception vectors
- Auto-vector interrupts
-> drop "autovector" from individual vector names and
list instead what they're for (e.g. hblank & vblank)?
- TRAP exception vectors
- FPU exception vectors
- MMU exception vectors
- MFP port vectors
- TT MFP port vectors
- SCC interrupt vectors
- System vectors
- System variables
For people (like me) who haven't done as much low level
Atari programming, some of the description could be more
extensive (e.g. "Timer C: 200Hz system clock"). Opinions?
After others have given their opinion, please send revised patch.
(Noticed one typo: uninitiaized -> uninitialized)