[asterisk-dev] [Code Review] 4554: clang compiler warning: -Wunused-value

rmudgett reviewboard at asterisk.org
Fri Apr 3 17:40:16 CDT 2015


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/4554/#review15053
-----------------------------------------------------------



/branches/13/channels/chan_iax2.c
<https://reviewboard.asterisk.org/r/4554/#comment25728>

    This change causes bugs.  It is supposed to return peer because it increased the ref.



/branches/13/channels/chan_iax2.c
<https://reviewboard.asterisk.org/r/4554/#comment25729>

    This change is another bug.  It is supposed to return the found user.



/branches/13/channels/chan_iax2.c
<https://reviewboard.asterisk.org/r/4554/#comment25730>

    In this case we have a ref before calling AST_SCHED_DEL and the scheduler item that we just deleted also has a ref.  Therefore peer could never be invalid.
    
    The scheduler is very old with a flawed API.  It doesn't know how to handle callback data that must be released such as ao2 object refs or malloced data.



/branches/13/channels/chan_iax2.c
<https://reviewboard.asterisk.org/r/4554/#comment25732>

    This is another case of dealing with the scheduler.  We are releasing the ref we just tried to give the scheduler that failed to be added.



/branches/13/channels/chan_iax2.c
<https://reviewboard.asterisk.org/r/4554/#comment25733>

    This NULL check is not needed.  peer can never be NULL here without an earlier crash.  This function is a scheduler callback.


- rmudgett


On April 3, 2015, 4:24 p.m., Diederik de Groot wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviewboard.asterisk.org/r/4554/
> -----------------------------------------------------------
> 
> (Updated April 3, 2015, 4:24 p.m.)
> 
> 
> Review request for Asterisk Developers.
> 
> 
> Bugs: ASTERISK-24917
>     https://issues.asterisk.org/jira/browse/ASTERISK-24917
> 
> 
> Repository: Asterisk
> 
> 
> Description
> -------
> 
> clang's static analyzer will throw quite a number warnings / errors during compilation, some of which can be very helpfull in finding corner-case bugs. 
> 
> clang compiler warning:-Wunused-value
> 
> Changes:
> apps/app_agent_pool.c
> multiple occasions where return value from ast_channel_ref and ast_channel_unref is not checked not used.
> return value from AST_LIST_REMOVE should be checked.
> Possible issues with leaked logged reference (added remark)
> 
> channels/chan_iax2.c
> return values from ast_callid_ref and ast_callid_unref can/should be used to update the pointer
> return value from AST_SCHED_DEL should be used to update the scheduled variable
> return value from AST_LIST_REMOVE should be checked.
> 
> channels/iax2/parser.c
> return value from AST_LIST_REMOVE should be checked.
> 
> include/asterisk/stringfields.h
> Quick fix to send the returned value from ast_string_field_ptr_set into the void.
> 
> Work in Progress:
> There are too many sources issues with "-Wunused-value" for one person to create all the fixes. If we want to actually use this warning we will need a couple of extra hands-on.
> 
> Should we suppress "-Wunused-value" ? :
> I think some of the issues pointed out by clang could be very helpfull, for example the "unref/ref/AST_LIST_REMOVE" changes that where needed in chan_iax2.c and apps/app_agent_pool.c. Some others a less interesting, for example the SCHED_DEL fixes.
> 
> 
> Diffs
> -----
> 
>   /branches/13/include/asterisk/stringfields.h 433444 
>   /branches/13/channels/iax2/parser.c 433444 
>   /branches/13/channels/chan_iax2.c 433444 
>   /branches/13/apps/app_agent_pool.c 433444 
> 
> Diff: https://reviewboard.asterisk.org/r/4554/diff/
> 
> 
> Testing
> -------
> 
> Just to make it clear, this was just a sample / proposal to work through large set of -Wunused-value warnings return. Some of them could be interesting, others might not be be. 
> 
> Mostly looking for a discussion about which path to follow, and see if we can divvy up the work a little.
> 
> Compiles / Untested / Pretty certain it contains issue because of different way of dealing with return value from ..._unref.
> 
> 
> File Attachments
> ----------------
> 
> ASTCFLAGS="-Wno-error=unused-value" make &| grep "[-W" -B1 -A2
>   https://reviewboard.asterisk.org/media/uploaded/files/2015/03/28/83f2f382-9ef2-41cf-8e7a-c188c014c17e__make.report
> 
> 
> Thanks,
> 
> Diederik de Groot
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20150403/ab4184ce/attachment-0001.html>


More information about the asterisk-dev mailing list