[asterisk-dev] [Code Review] 3970: res_phoneprov: Refactor phoneprov to allow pluggable config providers.

rmudgett reviewboard at asterisk.org
Tue Sep 30 13:24:35 CDT 2014 

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/3970/#review13410
-----------------------------------------------------------



branches/12/include/asterisk/chanvars.h
<https://reviewboard.asterisk.org/r/3970/#comment23912>

    guidelines: add space: while (0)



branches/12/include/asterisk/chanvars.h
<https://reviewboard.asterisk.org/r/3970/#comment23913>

    idem



branches/12/main/chanvars.c
<https://reviewboard.asterisk.org/r/3970/#comment23910>

    void is needed between the parens to match the protototype.  This is C not C++



branches/12/main/chanvars.c
<https://reviewboard.asterisk.org/r/3970/#comment23911>

    guidelines: space between while and paren



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23931>

    Please change this to take the name of the hash function instead of doing a token pasting.  If I didn't know about this macro or how it took parameters, it would be difficult to find the function definition given the function name where the continer is allocated.
    
    I actually did have some difficulty finding the function.



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23932>

    idem



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23914>

    This could be done the same way you did others:
    AST_VAR_LIST_INSERT_TAIL(profile->headp, ast_var_assign(args.varname, args.varval));
    
    This would also eliminate the variable variable.



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23919>

    Existing bug:
    The stringfields in exten are not freed.
    ast_string_field_free_memory(exten)
    
    should call: delete_extension(exten)



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23920>

    Remove these two unref_profile lines.  They cause an unbalance in the profile ref count.



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23921>

    OBJ_SEARCH_OBJECT



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23922>

    ast_free is NULL tolerant



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23923>

    idem



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23924>

    The handling of phoneprov_cfg is incorrect.  When ast_config_load() returns you have to check for the special pointer values:
    #define	CONFIG_STATUS_FILEUNCHANGED	(void *)-1
    #define	CONFIG_STATUS_FILEINVALID	(void *)-2
    
    before you can treat phoneprov_cfg as a normal pointer that ast_config_destroy() can handle.
    
    Passing those special pointers to ast_config_destroy() will cause a crash.



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23925>

    idem



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23926>

    idem



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23928>

    guidelines: Function start curly on its own line



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23927>

    idem



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23929>

    idem



branches/12/res/res_phoneprov.c
<https://reviewboard.asterisk.org/r/3970/#comment23930>

    Check for failure


- rmudgett


On Sept. 30, 2014, 11:02 a.m., George Joseph wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviewboard.asterisk.org/r/3970/
> -----------------------------------------------------------
> 
> (Updated Sept. 30, 2014, 11:02 a.m.)
> 
> 
> Review request for Asterisk Developers.
> 
> 
> Repository: Asterisk
> 
> 
> Description
> -------
> 
> The big piece missing for me to finally transition to pjsip was the ability to mirror the auto provisioning features of res_phoneprov.  The first step (this patch) is to make res_phoneprov more modular so other modules (like pjsip) can provide configuration information instead of res_phoneprov relying solely on users.conf and sip.conf.  To accomplish this a new ast_phoneprov public API is now exposed which allows config providers to register themselves, set defaults (server profile, etc) and add user extensions.
> 
> ast_phoneprov_provider_register registers the provider and provides callbacks for loading default settings and loading users.
> ast_phoneprov_provider_unregister clears the defaults and users.
> ast_phoneprov_add_extension should be called once for each user/extension by the provider's load_users callback to add them.
> ast_phoneprov_delete_extension deletes one extension.
> ast_phoneprov_delete_extensions deletes all extensions for the provider.
> 
> res_phoneprov actually registers itself as the provider for sip/users and is always available and is the default.
> 
> Writing a new provider...
> Since res_phoneprov is also it's own provider, examples of what a new provider would have to do are in load_users() in res_phoneprov.c.  Those functions gather the information from users.conf and sip.conf and call the ast_provider_register and ast_phoneprov_add_extension apis.
> 
> So...
> The provider creates a callback function which calls the ast_phoneprov_add_extension api for each user.  
> It then calls ast_phoneprov_provider_register with the callback.
> res_phoneprov then calls the callback to cause the actual load.
> During normal http server ops, all work is done by res_phoneprov and the provider is never called again unless a reload is needed.
> If the provider wants to reload it can simply unregister and reregister or it can call its own load_users callback.
> If res_phoneprov wants to reload, it iterates over its registry and calls the providers callback.
> 
> NOTE:  If res_phoneprov is actually unloaded, it has no way to know what providers were registered (other than itself) so a subsequent load will have nothing but it's own users.  
> 
> Additional changes...
> I added a few convenience functions to chanvars for creating lists and finding and deleting entries.  No existing code was touched.
> 
> Next steps...
> A provider for res_pjsip.
> 
> 
> Diffs
> -----
> 
>   branches/12/res/res_phoneprov.exports.in PRE-CREATION 
>   branches/12/res/res_phoneprov.c 424175 
>   branches/12/main/chanvars.c 424175 
>   branches/12/include/asterisk/phoneprov.h PRE-CREATION 
>   branches/12/include/asterisk/chanvars.h 424175 
>   branches/12/configs/phoneprov.conf.sample 424175 
> 
> Diff: https://reviewboard.asterisk.org/r/3970/diff/
> 
> 
> Testing
> -------
> 
> I ran through several scenarios including the use of PP_EACH_USER and PP_EACH_EXTENSION to make sure that all existing functionality was preserved.  I actually use it with Grandstream phones and everything worked exactly as expected.
> 
> 
> Thanks,
> 
> George Joseph
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20140930/5ebf663d/attachment-0001.html>

More information about the asterisk-dev mailing list