Thanks.
Yes, I saw that code, and changed it to "#if 1", and now it
sort of works - stopping in longjmp and then you say next and
it goes on and stops at the right place.
Makes me a little nervous that the values are not "official",
but it seems to pretty much do the right thing.
I'm "pretty sure" this worked before. We were using gdb 4.16 and 4.17
with gcc 2.7.2 on solaris 2.5 and 2.6 for the past two years or so and I
never got any report of this problem (or noticed it myself) until last week
when we switched to solaris 7, gcc 2.95.2 and gdb 4.18.
----- Original Message -----
From: Peter.Schauer <[EMAIL PROTECTED]>
To: Roger S Reynolds <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Monday, March 06, 2000 2:43 AM
Subject: Re: gdb can't step over longjmp call on solaris
> Are you sure that this didn't work on Solaris 1 (SunOS 4) only ?
>
> AFAICT this never worked with any official version of GDB on any version
> of Solaris 2.x.
> See also the following comment from gdb/config/sparc/tm-sun4sol2.h:
>
> #if 0 /* FIXME Setjmp/longjmp are not as well doc'd in SunOS 5.x yet */
>
> > Recently moved from gdb 4.17 on solaris 2.[56]
> > to gdb 4.18 on solaris 7, and have noticed that gdb no longer steps over
> > a longjmp call as it used to do.
> >
> > That is, if you are single stepping (nexting actually), and you come to a
> > longjmp call, when you say "n" the program just continues and runs to
> > completion or the next breakpoint, if one is set.
> > Previously, it would stop correctly at the first statement after in the "if
> > setjmp" clause.
> >
> > Is this a know bug/limitation, or is it supposed to work?
> > Thanks for any tips...
>
> --
> Peter Schauer [EMAIL PROTECTED]
>
