Re: [Xen-devel] [PATCH 12/13] libxl: New API for providing OS events to libxl

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Mon, 2011-11-28 at 16:59 +0000, Ian Jackson wrote:
> @@ -89,10 +99,25 @@ int libxl_ctx_alloc(libxl_ctx **pctx, int version,
> 
>  int libxl_ctx_free(libxl_ctx *ctx)
>  {
> +    int i;
> +    GC_INIT(ctx);

You never call GC_FREE in this function. It's a bit of an odd one to
need in the free function I guess but I think you still need it after
the last call to a function which takes a gc argument?

> +
>      if (!ctx) return 0;
> +
> +    for (i = 0; i < ctx->watch_nslots; i++)
> +        assert(!libxl__watch_slot_contents(gc, i));
> +    libxl__ev_fd_deregister(gc, &ctx->watch_efd);
> +
> +    assert(LIBXL_LIST_EMPTY(&ctx->efds));
> +    assert(LIBXL_TAILQ_EMPTY(&ctx->etimes));

Am I right that libxl__ev_fd_deregister is what ensures this is the
case? If not then who is responsible for unregistering akk events before
freeing a context?

I've just seen you've reposted the series so I'll stop here and pickup
in the repost instead.

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel