From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:49008)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1VLNp5-0000nK-7N
	for qemu-devel@nongnu.org; Sun, 15 Sep 2013 21:42:01 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1VLNoz-0008UI-5P
	for qemu-devel@nongnu.org; Sun, 15 Sep 2013 21:41:55 -0400
Received: from mx1.redhat.com ([209.132.183.28]:51084)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1VLNoy-0008UA-UJ
	for qemu-devel@nongnu.org; Sun, 15 Sep 2013 21:41:49 -0400
Date: Mon, 16 Sep 2013 09:41:36 +0800
From: Fam Zheng <famz@redhat.com>
Message-ID: <20130916014136.GC2700@T430s.nay.redhat.com>
References: <1379066356-14986-1-git-send-email-famz@redhat.com>
	<1379066356-14986-6-git-send-email-famz@redhat.com>
	<5233269B.3030803@twiddle.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <5233269B.3030803@twiddle.net>
Subject: Re: [Qemu-devel] [PATCH v9 5/8] module: implement module loading
Reply-To: famz@redhat.com
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Richard Henderson <rth@twiddle.net>
Cc: peter.maydell@linaro.org, mjt@tls.msk.ru, qemu-devel@nongnu.org, alex@alex.org.uk, pbonzini@redhat.com, vilanova@ac.upc.edu

On Fri, 09/13 07:52, Richard Henderson wrote:
> On 09/13/2013 02:59 AM, Fam Zheng wrote:
> > +    const char *module_whitelist[] = {
> > +        CONFIG_MODULE_WHITELIST
> > +    };
> 
>   static const char * const module_whitelist[] = ...
> 

OK, thanks.


> > +    switch (type) {
> > +    case MODULE_LOAD_BLOCK:
> > +        path = CONFIG_MODDIR "/block/";
> > +        break;
> > +    case MODULE_LOAD_UI:
> > +        path = CONFIG_MODDIR "/ui/";
> > +        break;
> > +    case MODULE_LOAD_NET:
> > +        path = CONFIG_MODDIR "/net/";
> > +        break;
> 
> Also, separate the whitelists by type.  I.e.
> 
>   static const char * const modules_block[] = ...
>   static const char * const modules_ui[] = ...
>   static const char * const modules_net[] = ...
> 
>   switch (type) {
>   case MODULE_LOAD_BLOCK:
>     list = modules_block;
>     n = ARRAY_SIZE(modules_block);
>     break;
>   ...
>   }
> 
> No need for null termination of the array, as you're currently using.
> 

It's that I'd like to be consistent with block whitelist code style as in
bdrv_is_whitelisted().

> > +    for (mp = &module_whitelist[0]; *mp; mp++) {
> > +        fname = g_strdup_printf("%s%s" HOST_DSOSUF, path, *mp);
> > +        module_load_file(fname);
> > +        g_free(fname);
> > +    }
> 
> Why this bizzare mix of g_strdup_printf and compile-time string concatenation?
>  Certainly you could have arranged for HOST_DSOSUF to be built into the module
> name as seen in the arrays.
> 
> Then we're back to the subdirectory vs filename prefix and CONFIG_MODDIR vs any
> of several module search path options, the debate of which I don't believe has
> concluded.
> 

Perhaps using module type as prefix and building into whitelist can save us
creating subdirs for each type and also the separation of the list as you
suggested above:

    switch (type) {
    case MODULE_LOAD_BLOCK:
        prefix = "block-";
        break;
        ...
    }

    for (mp = &module_whitelist[0]; *mp; mp++) {
        if (!strncmp(prefix, *mp, sizeof(prefix))) {
            /* load */
        }
    }