* Eric Blake (ebl...@redhat.com) wrote:
> On 03/30/2017 02:50 AM, Thomas Huth wrote:
> > HMP commands do not get any automatic testing yet, so on certain
> > QEMU machines, some HMP commands were causing crashes in the past.
> > Thus we should test HMP commands in our test suite, too, to avoid
> > that such problems creep in again in the future.
> >
> > Signed-off-by: Thomas Huth <th...@redhat.com>
> > ---
>
> > +static const char *hmp_cmds[] = {
> > + "boot_set ndc",
>
> > +/* Run through the list of pre-defined commands */
> > +static void test_commands(void)
> > +{
> > + char *response;
> > + int i;
> > +
> > + for (i = 0; hmp_cmds[i] != NULL; i++) {
> > + if (verbose) {
> > + fprintf(stderr, "\t%s\n", hmp_cmds[i]);
> > + }
> > + response = hmp(hmp_cmds[i]);
>
> I failed to notice this sooner, but hmp() is passing its first arg as a
> format string through a printf family. If hmp_cmds[i] ever gets
> modified to include something with a %, it will misbehave. Better is to
> use hmp("%s", variable).
>
> I've patched it locally as part of rebasing my work on avoiding dynamic
> JSON format strings, if no one beats me to a fix (my series also adds
> the gcc format attribute tag, so that the compiler catches any further
> mismatches in hmp() format vs. arguments).
Ah yes, good spot. Please include that fix.
Dave
> > + while (*info) {
> > + /* Extract the info command, ignore parameters and description */
> > + g_assert(strncmp(info, "info ", 5) == 0);
> > + endp = strchr(&info[5], ' ');
> > + g_assert(endp != NULL);
> > + *endp = '\0';
> > + /* Now run the info command */
> > + if (verbose) {
> > + fprintf(stderr, "\t%s\n", info);
> > + }
> > + resp = hmp(info);
>
> Another instance.
>
> --
> Eric Blake, Principal Software Engineer
> Red Hat, Inc. +1-919-301-3266
> Virtualization: qemu.org | libvirt.org
>
--
Dr. David Alan Gilbert / dgilb...@redhat.com / Manchester, UK
