Re: [Patch v3 3/4] tools/libxl: Fix libxl__device_nic_from_xs_be()

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 26/11/13 15:39, Ian Jackson wrote:
> Andrew Cooper writes ("Re: [Xen-devel] [Patch v3 3/4] tools/libxl: Fix libxl__device_nic_from_xs_be()"):
>> On 26/11/13 15:08, Ian Jackson wrote:
>>> From: Ian Jackson <ian.jackson@eu.citrix.com>
>>> Date: Tue, 26 Nov 2013 12:08:09 +0000
>>> Subject: [PATCH] libxl: Fix error handling in libxl__device_nic_from_xs_be
> ...
>>> Introduce here a READ_BACKEND macro to make the code less repetitive.
> I was thinking about READ_BACKEND and wondered whether we should have
> something like this in libxl_internal.h:
>
>  /*
>   * const char *XSREADF_OR_RCOUT(libxl__gc*, const char *format, ...);
>   *
>   * Reads the xenstore key at sprintf(format, ...).
>   * On success returns the string (from the gc tgc), or NULL for ENOENT.
>   * On other errors, logs, sets rc, and does "goto out".
>   *
>   * Expects in its scope:
>   *   libxl__gc *gc;  // for the sprintf
>   *   int rc;         // trashed
>   *   out:            // used on error only; jumped to with rc set
>   */
>  #define XSREADF_OR_RCOUT(tgc, format, ...) ({                            \
>          const char *xsreadf_tmp;                                         \
>          rc = libxl__xs_read_checked((tgc), XBT_NULL,                     \
>                                      GCSPRINTF((format), __VA_ARGS__),    \
>                                      &xsreadf_tmp);                       \
>          if (rc) goto out;                                                \
>          xsreadf_tmp;                                                     \
>      )}
>
> We don't presently have anywhere in libxl_internal.h that assumes the
> existence of "out" and "rc" in their scope.
>
> Ian.

So what is happening with this?  The fix in the parent email should be
ok, and might be better this late in the development cycle.

~Andrew