[PATCH 2 wayland] tests: detect if debugger is attached

[Marek Chalupa]

Hi,

Here it is used to suppress the timeout, which just kills the client during
debugging if you forget to turn it off.
Since it is just a way how to workaround forgetting of humans (read: me xD)
I think that I don't care about this patch much.
If no-one else want this, I'm OK with rejecting this patch.

Thanks,
Marek


On 28 November 2014 at 12:54, Pekka Paalanen <ppaalanen at gmail.com> wrote:

> On Wed, 19 Nov 2014 12:20:31 +0100
> Marek Chalupa <mchqwerty at gmail.com> wrote:
>
> > Copy function from libinput/test/litest.c is_debugger_detached()
> > and use it in our test-runner. If debugger is attached, turn off
> > leak checks and timeouts automatically.
> >
> > Revision of libinput: 028513a0a723e97941c39c4aeb17433198723913
> >
> > Signed-off-by: Marek Chalupa <mchqwerty at gmail.com>
> > ---
> >  tests/test-runner.c | 43 +++++++++++++++++++++++++++++++++++++++++--
> >  1 file changed, 41 insertions(+), 2 deletions(-)
> >
> > diff --git a/tests/test-runner.c b/tests/test-runner.c
> > index 1373225..63b25ed 100644
> > --- a/tests/test-runner.c
> > +++ b/tests/test-runner.c
> > @@ -33,6 +33,7 @@
> >  #include <dlfcn.h>
> >  #include <errno.h>
> >  #include <limits.h>
> > +#include <sys/ptrace.h>
> >
> >  #include "test-runner.h"
> >
> > @@ -224,6 +225,39 @@ rmdir_xdg_runtime_dir(void)
> >               perror("Cleaning XDG_RUNTIME_DIR");
> >  }
> >
> > +
> > +/* this function is taken from libinput/test/litest.c
> > + * (rev 028513a0a723e97941c39)
> > + */
> > +static int
> > +is_debugger_attached(void)
> > +{
> > +     int status;
> > +     int rc;
> > +     int pid = fork();
> > +
> > +     if (pid == -1)
> > +             return 0;
> > +
> > +     if (pid == 0) {
> > +             int ppid = getppid();
> > +             if (ptrace(PTRACE_ATTACH, ppid, NULL, NULL) == 0) {
> > +                     waitpid(ppid, NULL, 0);
> > +                     ptrace(PTRACE_CONT, NULL, NULL);
> > +                     ptrace(PTRACE_DETACH, ppid, NULL, NULL);
> > +                     rc = 0;
> > +             } else {
> > +                     rc = 1;
> > +             }
> > +             _exit(rc);
> > +     } else {
> > +             waitpid(pid, &status, 0);
> > +             rc = WEXITSTATUS(status);
> > +     }
> > +
> > +     return rc;
> > +}
> > +
> >  int main(int argc, char *argv[])
> >  {
> >       const struct test *t;
> > @@ -237,8 +271,13 @@ int main(int argc, char *argv[])
> >       sys_malloc = dlsym(RTLD_NEXT, "malloc");
> >       sys_free = dlsym(RTLD_NEXT, "free");
> >
> > -     leak_check_enabled = !getenv("WAYLAND_TESTS_NO_LEAK_CHECK");
> > -     timeouts_enabled = !getenv("WAYLAND_TESTS_NO_TIMEOUTS");
> > +     if (is_debugger_attached()) {
> > +             leak_check_enabled = 0;
> > +             timeouts_enabled = 0;
> > +     } else {
> > +             leak_check_enabled =
> !getenv("WAYLAND_TESTS_NO_LEAK_CHECK");
> > +             timeouts_enabled = !getenv("WAYLAND_TESTS_NO_TIMEOUTS");
> > +     }
> >
> >       if (argc == 2 && strcmp(argv[1], "--help") == 0)
> >               usage(argv[0], EXIT_SUCCESS);
>
> Hi,
>
> I'm a little wary of this kind of magic. You can't force-enable these
> settings under a debugger now, but I'm not sure anyone would want to.
>
> Would need a trivial rebase.
>
> Libinput seems use the debugger detection for preventing test forking,
> and we have our own way for that by running a single sub-test.
>
> Do we care enough to have this patch?
>
>
> Thanks,
> pq
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/wayland-devel/attachments/20141128/6f6054aa/attachment.html>