This is the mail archive of the
mailing list for the GDB project.
Summary of differences between FSF GDB and ST's Micro Connectversion
- From: "Paul Bartlett" <paul dot bartlett at superh dot com>
- To: <gdb at sources dot redhat dot com>
- Date: Wed, 19 Feb 2003 11:35:27 -0000
- Subject: Summary of differences between FSF GDB and ST's Micro Connectversion
A couple of comments from SuperH...
The current version of gdb available from ST is based on 5.2.1.
There are indeed some changes that can only really be classed as hacks. This
was mainly in response to pressure to release something usable to users and
our changes should not be considered to be complete.
We intend to rebase our changes on 5.3 when time permits and also to resolve
in as clean a fashion as possible all 'workarounds' that have been
We also intend to submit the changes for consideration by the maintainers in
This is very much 'work in progress'.
I don't really want to go into the reasons for the introduction of the
'fork' command right now as the explanation would be somewhat lengthy. If
anybody's interested then please contact me directly.
In a naieve attempt to get my Micro Connect ICE to talk to the latest
gdb I thought I'd better have a look at the diff's between the two
sources and inquire to the list if its worth forward porting any of the
changes into the latest gdb source tree. I'm unsure if some of the arch
specific stuff had been covered in later versions.
The version of gdb that I was supplied is quite old 20020211 and has had
patches made to it by both by SuperH, Inc and ST. I've only skimmed
through the diffs (~6000 lines) so the comments are brief.
Please don't forward the diffs to this list. Since the ownership of the code
is unclear, the last thing GDB want is contamination :-(
BFD Changes (bfd/)
bfd/bfd-in2.h - Added addtional SH relocation types
GDB Changes (gdb/)
Various SH specific changes to deal with breakpoints, frame handling etc
Changes to handle SH virtual registers
Changes to dwarf handling
Some code #if 0'd out to either bypass "bugs" or hardwire things for SH
A fair number of explicit variable initialisations and single line
Sounds like someone has been pulling some nasty hacks :-(
GDB CLI (gdb/cli)
New functions added:
sleep_escape (seems to sleep for a bit)
fork_escape (forks external process setting up some pipes) - this is
used to lauch the external gdbserver in my setup
GDB supports the feature:
target remote |program
Command file tracing added (-v).
So I guess my question is will any of these patches of been forward
ported into 5.3 or submitted by SuperH for inclusion? I know the cli
changes have not but I fear these in themselves may not be enough to get
the latest gdb working with the gdb server.
I don't know.
Are there ChangeLog's? Might help with figuring out where the changes came