Re: [Qemu-devel] [PATCH V4] vl.c: fixed regression in machine error message

[Eduardo Habkost <ehabkost@redhat.com>]

On Thu, Feb 11, 2016 at 08:49:25PM +0200, Marcel Apfelbaum wrote:
> On 02/11/2016 08:31 PM, Eduardo Habkost wrote:
> >On Mon, Feb 08, 2016 at 01:54:29PM +0200, Marcel Apfelbaum wrote:
> >>Commit e1ce0c3cb(vl.c: fix regression when reading machine type from config file)
> >>fixed the error message when the machine type was supplied inside the
> >>config file. However now the option name is not displayed correctly if
> >>the error happens when the machine is specified at command line.
> >>
> >>Running
> >>     ./x86_64-softmmu/qemu-system-x86_64 -M q35-1.5 -redir tcp:8022::22
> >>will result in the error message:
> >>     qemu-system-x86_64: -redir tcp:8022::22: unsupported machine type
> >>     Use -machine help to list supported machines
> >>
> >>Fixed it by restoring the error location and also extracted the code
> >>dealing with machine options into a separate function.
> >>
> >>Reported-by: Michael S. Tsirkin <mst@redhat.com>
> >>Reviewed-by: Laszlo Ersek <lersek@redhat.com>
> >>Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
> >>---
> >>
> >>v3 -> v4:
> >>    - removed double call of loc_push_none (thanks again Laszlo)
> >>
> >>v2 -> v3:
> >>  - fixed commit message and called qemu_get_machine_opts only once. (thanks Laszlo)
> >>
> >>v1 -> v2:
> >>  - Addressed Laszlo Ersek's comments:
> >>    - no need to save the machine options location, is saved in opts
> >>    - rename the extracted method to set_machine_options
> >>    - added the bug reporter to the CC
> >>
> >>  - tested with and without the config file and the error message is now OK:
> >>  config file:
> >>     - qemu-system-x86_64:machine-bug.conf:3: unsupported machine type
> >>  cli:
> >>    - qemu-system-x86_64: -M q35-1.5: unsupported machine type
> >>
> >>Thanks,
> >>Marcel
> >>
> >>  vl.c | 38 +++++++++++++++++++++++++++-----------
> >>  1 file changed, 27 insertions(+), 11 deletions(-)
> >>
> >>diff --git a/vl.c b/vl.c
> >>index 2c03f54..5e22a35 100644
> >>--- a/vl.c
> >>+++ b/vl.c
> >>@@ -2748,6 +2748,31 @@ static const QEMUOption *lookup_opt(int argc, char **argv,
> >>      return popt;
> >>  }
> >>
> >>+static void set_machine_options(MachineClass **machine_class)
> >>+{
> >>+    const char *optarg;
> >>+    QemuOpts *opts;
> >>+    Location loc;
> >>+
> >>+    loc_push_none(&loc);
> >>+
> >>+    opts = qemu_get_machine_opts();
> >>+    qemu_opts_loc_restore(opts);
> >>+
> >>+    optarg = qemu_opt_get(opts, "type");
> >>+    if (optarg) {
> >>+        *machine_class = machine_parse(optarg);
> >>+    }
> >>+
> >>+    if (*machine_class == NULL) {
> >>+        error_report("No machine specified, and there is no default");
> >>+        error_printf("Use -machine help to list supported machines\n");
> >>+        exit(1);
> >>+    }
> >>+
> >>+    loc_pop(&loc);
> >>+}
> >>+
> >>  static int machine_set_property(void *opaque,
> >>                                  const char *name, const char *value,
> >>                                  Error **errp)
> >>@@ -4028,17 +4053,7 @@ int main(int argc, char **argv, char **envp)
> >>
> >>      replay_configure(icount_opts);
> >>
> >>-    opts = qemu_get_machine_opts();
> >>-    optarg = qemu_opt_get(opts, "type");
> >>-    if (optarg) {
> >>-        machine_class = machine_parse(optarg);
> >>-    }
> >>-
> >>-    if (machine_class == NULL) {
> >>-        error_report("No machine specified, and there is no default");
> >>-        error_printf("Use -machine help to list supported machines\n");
> >>-        exit(1);
> >>-    }
> >>+    set_machine_options(&machine_class);
> >>
> >>      set_memory_options(&ram_slots, &maxram_size, machine_class);
> >[Extra context for reference:]
> >>
> >>    loc_set_none();
> >
> >You are fixing the machine error message, but not the root cause:
> 
> Hi,
> 
> I knew that it would not solve all the cases but
> the machine error message is a regression while other
> were there way before (as far as I know)
> 
> I was specifically looking to fix the machine error regression.
> 
> >
> >   $ ./x86_64-softmmu/qemu-system-x86_64 -m size= -vnc :0
> >   qemu-system-x86_64: -vnc :0: missing 'size' option value
> >   $ ./x86_64-softmmu/qemu-system-x86_64 -icount rr=x -vnc :0
> >   qemu-system-x86_64: -vnc :0: Invalid icount rr option: x
> >
> >Moving the loc_set_none() call above replay_configure() should fix the bug in
> >the three cases.
> 
> I am really no expert in loc stuff, but it seamed to me it is placed
> especially after all the options are parsed. Maybe we need another
> call to loc_set_none before replay_configure instead of moving it?

It needs to be reset immediately after the option-parsing 'for'
loop, because the location info becomes obsolete as soon as we
exit the loop. All the remaining code is supposed to push/pop
location info if necessary, so we don't need another call.

> 
> If you think it should be moved, please go ahead :)
> is an one liner and the idea was yours.

I will do that.

> 
> >
> >Setting location in set_machine_options()
> 
> This is done in this patch.

Yes. And if we reorder my patch and yours, the conflicts look
trivial.

> 
> , set_memory_options(), and
> >replay_configure() would be nice, too.
> 
> I can do it on those functions too, if you agree to the way
> is done in set_machine_options

It looks good to me. I would like location setting to be done
automatically somehow, instead of requiring every option parser
to manually set location, but this can be figured out later.

Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>

-- 
Eduardo