From mboxrd@z Thu Jan 1 00:00:00 1970 From: Michael Snyder To: Daniel Jacobowitz Cc: Andrew Cagney , Michael Snyder , gdb-patches@sources.redhat.com Subject: Re: PATCH: resume + threads + software stepping == boom Date: Sun, 10 Jun 2001 21:40:00 -0000 Message-id: <3B244B4B.4F8A@redhat.com> References: <20010608123432.A2140@nevyn.them.org> <3B215D60.78921819@cygnus.com> <3B215DBE.E9E4463A@cygnus.com> <20010608164327.A22796@nevyn.them.org> <3B225994.9060502@cygnus.com> <20010609160611.A26105@nevyn.them.org> X-SW-Source: 2001-06/msg00205.html Daniel Jacobowitz wrote: [extensive snipping] > + /* If we stopped on a breakpoint and it was not deleted, we want to continue > + only this thread, so we should be stepping. */ I agree with the premise -- whenever we resume at a breakpoint, we should do a single-thread-step before inserting bps, so that other threads do not have a chance to run (and possibly go past other breakpoints) while breakpoints are not inserted. How do others feel about that premise? I think we had reached that consensus at some earlier time, but it has not been implemented (almost, but not quite).