Re: [Qemu-devel] [PATCH 4/6] qapi: change the type of TargetInfo.arch from string to enum SysEmuTarget

[Markus Armbruster]

Laszlo Ersek <address@hidden> writes:

> Now that we have @SysEmuTarget, it makes sense to restict
> @address@hidden to valid sysemu targets at the schema level.
>
> Cc: "Daniel P. Berrange" <address@hidden>
> Cc: David Gibson <address@hidden>
> Cc: Eric Blake <address@hidden>
> Cc: Gerd Hoffmann <address@hidden>
> Cc: Kashyap Chamarthy <address@hidden>
> Cc: Markus Armbruster <address@hidden>
> Cc: Paolo Bonzini <address@hidden>
> Cc: Thomas Huth <address@hidden>
> Signed-off-by: Laszlo Ersek <address@hidden>
> ---
>
> Notes:
>     PATCHv1:
>     
>     - qmp_query_target(): pass (-1) as fallback value [Markus]
>     - qmp_query_target(): catch lookup error with error_abort [Markus]
>     
>     RFCv3:
>     
>     - The patch is new in this version. [Markus]
>
>  qapi/misc.json |  6 ++++--
>  arch_init.c    | 10 +++++++++-
>  2 files changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/qapi/misc.json b/qapi/misc.json
> index 104d013adba6..460866cf542f 100644
> --- a/qapi/misc.json
> +++ b/qapi/misc.json
> @@ -1,18 +1,20 @@
>  # -*- Mode: Python -*-
>  #
>  
>  ##
>  # = Miscellanea
>  ##
>  
> +{ 'include': 'common.json' }
> +
>  ##
>  # @qmp_capabilities:
>  #
>  # Enable QMP capabilities.
>  #
>  # Arguments:
>  #
>  # @enable:   An optional list of QMPCapability values to enable.  The
>  #            client must not enable any capability that is not
>  #            mentioned in the QMP greeting message.  If the field is not
>  #            provided, it means no QMP capabilities will be enabled.
> @@ -2441,28 +2443,28 @@
>  #      ]
>  #    }
>  #
>  ##
>  { 'command': 'query-fdsets', 'returns': ['FdsetInfo'] }
>  
>  ##
>  # @TargetInfo:
>  #
>  # Information describing the QEMU target.
>  #
> -# @arch: the target architecture (eg "x86_64", "i386", etc)
> +# @arch: the target architecture
>  #
>  # Since: 1.2.0
>  ##
>  { 'struct': 'TargetInfo',
> -  'data': { 'arch': 'str' } }
> +  'data': { 'arch': 'SysEmuTarget' } }
>  
>  ##
>  # @query-target:
>  #
>  # Return information about the target for this QEMU
>  #
>  # Returns: TargetInfo
>  #
>  # Since: 1.2.0
>  ##
>  { 'command': 'query-target', 'returns': 'TargetInfo' }
> diff --git a/arch_init.c b/arch_init.c
> index 6ee07478bd11..ee3a57019000 100644
> --- a/arch_init.c
> +++ b/arch_init.c
> @@ -21,22 +21,23 @@
>   * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
>   * THE SOFTWARE.
>   */
>  #include "qemu/osdep.h"
>  #include "qemu-common.h"
>  #include "cpu.h"
>  #include "sysemu/sysemu.h"
>  #include "sysemu/arch_init.h"
>  #include "hw/pci/pci.h"
>  #include "hw/audio/soundhw.h"
>  #include "qapi/qapi-commands-misc.h"
> +#include "qapi/error.h"
>  #include "qemu/config-file.h"
>  #include "qemu/error-report.h"
>  #include "hw/acpi/acpi.h"
>  #include "qemu/help_option.h"
>  
>  #ifdef TARGET_SPARC
>  int graphic_width = 1024;
>  int graphic_height = 768;
>  int graphic_depth = 8;
>  #else
>  int graphic_width = 800;
> @@ -104,15 +105,22 @@ int xen_available(void)
>      return 1;
>  #else
>      return 0;
>  #endif
>  }
>  
>  
>  TargetInfo *qmp_query_target(Error **errp)
>  {
>      TargetInfo *info = g_malloc0(sizeof(*info));
>  
> -    info->arch = g_strdup(TARGET_NAME);
> +    /*
> +     * The fallback enum value is irrelevant here (TARGET_NAME is a
> +     * macro and can never be NULL), so simply pass (-1). Also, the
> +     * lookup should never fail -- if it fails, then @SysEmuTarget needs
> +     * extending. Catch that with "error_abort".
> +     */
> +    info->arch = qapi_enum_parse(&SysEmuTarget_lookup, TARGET_NAME, -1,
> +                                 &error_abort);
>  
>      return info;
>  }

Not sure the comment is carrying its weight; for me, the use of -1 and
&error_abort feels obvious enough.  But my feelings are subjective and
could be off.

Reviewed-by: Markus Armbruster <address@hidden>