This is the mail archive of the
frysk@sourceware.org
mailing list for the frysk project.
The Froggy project and Frysk-{Prime} and/or GDB
- From: Phil Muldoon <pmuldoon at redhat dot com>
- To: Frysk List <frysk at sourceware dot org>
- Date: Wed, 16 Jul 2008 18:09:13 +0100
- Subject: The Froggy project and Frysk-{Prime} and/or GDB
Over on the utrace-devel list there is a thread about ntrace, and ideas
about how the user-land interface to utrace should look. Thread is here:
https://www.redhat.com/archives/utrace-devel/2008-July/msg00007.html
Chris Moller has started a project called Froggy which realizes these
user-land requirements. The code base and announcements are here:
https://www.redhat.com/archives/utrace-devel/2008-July/msg00039.html
One of this issues we have been looking at as we examine Frysk and its
replacement are the farther reaching aspects of debuggers/tools in the
future. The utrace project forms a part of this, as does Froggy. It is
important at least from our perspective to get our requirements in early.
Anyone who has worked with ptrace for any amount of time normally has
some pet-peeve, or worse, some horror story. Ptrace does some things
right, and some things wrong. I'm not going down the road of why, but I
believe there is a golden opportunity here to present Froggy and utrace
with a listing of living requirements from the debugger community. I'll
volunteer to organize the requirements document if you volunteer your
opinions in the spirit of open discourse. I'm especially interested in
current ptrace use-case limitations, something like:
"Reading large amounts of memory from an inferior via ptrace is slow.
Peeking word after word, especially when dumping corefiles is plain
painful. Lots of people get around this by accessing the inferior's
/proc/self/mem. But this requires the inferior to be ptrace_attach'ed;
so it can be granted access to /proc/self/mem and direct reads. This is
a problem as is if a ptrace'd child gets a signal, it is stopped. In the
real-time world this is a problem, and is undesirable; we simply want
high-bandwidth access to the inferior's memory."
If we can coalesce these use-cases into requirements, it can form high
bandwidth communications with Froggy and utrace.
Comments?
Regards
Phil