This is the mail archive of the
archer@sourceware.org
mailing list for the Archer project.
Re: froggy plans
- From: Jan Kratochvil <jan dot kratochvil at redhat dot com>
- To: Chris Moller <cmoller at redhat dot com>
- Cc: Project Archer <archer at sourceware dot org>
- Date: Wed, 25 Feb 2009 23:03:46 +0100
- Subject: Re: froggy plans
- References: <49A59146.20003@redhat.com>
On Wed, 25 Feb 2009 19:43:18 +0100, Chris Moller wrote:
> From comments during the mtg, I think what I need to do now is back off
> from wedging froggy into gdb for a while
add_file_handler() can be used with fd = froggy_fd (froggy),
proc = something-like-froggy.c:resp_listener.
You should remove at least `linux_nat_event_pipe' and `linux_nat_async' to get
errors on any of its use which should be now replaced by froggy and its
froggy_fd (froggy) handling.
my_waitpid() currently calls queued_waitpid() while
something-like-froggy.c:resp_listener should store the received events by
push_waitpid() to make them retrievable by my_waitpid().
If my_waitpid() finds `waitpid_queue' is empty and it would call waitpid(2)
the froggy port in such (only) case should do its blocking wait on a new event
to come from waitpid_queue. (According to the 3rd paragraph of the
linux-nat.c doc part `Waiting for events in async mode'.)
You can fortunately get rid of `linux_nat_event_pipe' as poll() could not wait
on SIGCHLD. IMO FSF GDB could use ppoll() directly catching both file events
and SIGCHLDs (leading to waitpid()). Therefore to have only `waitpid_queue'
without the need of `linux_nat_event_pipe'. But I may miss something.
Sure I miss a lot of stuff but IMO worth a try.
> * detach with optional wait-until-detached (was already in plan, but
> hadn't gotten around to coding it)
> * kill with optional wait-until-dead-or-timeout
Any waiting/blocking call is not usable for the GDB non-stop/async mode (which
is the target for Eclipse/MI operation mode). IMO the libfroggy functions
should be first more driven by the GDB port requirements, shouldn't they?
When using latest:
cvs -d :ext:sources.redhat.com:/cvs/systemtap co froggy
archer-moller-froggy-v2
CFLAGS='-m32' ./configure --with-froggy=/home/jkratoch/redhat/froggy/froggy i386-unknown-linux-gnu
I still get (and cannot find froggy_singlestep_pid() anywhere):
inf-ptrace-froggy.c:270: error: implicit declaration of function ‘froggy_singlestep_pid’
Also could you please provide a x86_64 port? Without -m32 I get:
froggy.c:424: error: invalid application of ‘sizeof’ to incomplete type ‘struct user_fpxregs_struct’
Regards,
Jan