cvs commit: src/usr.bin/truss Makefile amd64-fbsd.c extern.h i386-fbsd.c i386-linux.c ia64-fbsd.c main.c powerpc-fbsd.c setup.c sparc64-fbsd.c syscall.h syscalls.c truss.1 truss.h

[Kip Macy]

On 4/10/07, Ed Maste <ed.maste at gmail.com> wrote:
> On 10/04/07, Xin LI <delphij at freebsd.org> wrote:
> >   Make use of ptrace(2) instead of procfs in truss(1), eliminating
> >   yet another need of an available /proc/ mount.
>
> I've started to do the same thing to gcore(1), in order to make it
> work properly with threaded applications.  I've removed its dependency
> on /proc/pid/mem, /proc/pid/regs, and /proc/pid/fpregs.  However, it
> still relies on /proc/pid/status and /proc/pid/map which are not
> directly available via ptrace(2).
>
> This would make the -s option to gcore redundant (since the process
> will be stopped after attaching anyway).  I don't know how useful a
> core from a non-stopped process is, anyhow.

Fork it to get a COW snapshot and then gcore the stopped forked process.
(Yes this would probably require a new syscall or a new ptrace
interface, but it would be trivial to add).


                    -Kip