[Top] [All Lists]

Re: [DISCUSS] Planning for new dev cycle (3.17)

To: Dave Chinner <david@xxxxxxxxxxxxx>, xfs@xxxxxxxxxxx
Subject: Re: [DISCUSS] Planning for new dev cycle (3.17)
From: Eric Sandeen <sandeen@xxxxxxxxxxx>
Date: Tue, 10 Jun 2014 16:57:36 -0500
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <53971184.50008@xxxxxxxxxxx>
References: <20140609223320.GE9508@dastard> <20140610060944.GG9508@dastard> <53971184.50008@xxxxxxxxxxx>
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:24.0) Gecko/20100101 Thunderbird/24.5.0
On 6/10/14, 9:09 AM, Eric Sandeen wrote:
> On 6/10/14, 1:09 AM, Dave Chinner wrote:
>> There is a new version of this work (Eric's patches and the base
>> libxfs restructure) here:
>>      git://oss.sgi.com/xfs/xfs.git xfs-libxfs-restructure
>> It's based on 3.15 with the current for-next branch merged into it
>> and then the changes done over the top, so it's about as up-to-date
>> as it can be. Comments welcome...
> Comment 1: doesn't build ;)
> make[1]: *** No rule to make target `fs/xfs/xfs_dir2_readdir.o', needed by 
> `fs/xfs/xfs.o'.  Stop.
> make[1]: *** Waiting for unfinished jobs....
>   CC [M]  fs/xfs/xfs_buf.o
> make: *** [_module_fs/xfs] Error 2
> problems w/ xfs_rtbitmap.o as well, patch follows.
> Comment 2: Coverity thinks this adds about 25 defects, I'll have to go
> look at why...

(whoops, I forgot to send this earlier)

I take that back, somehow it was still counting defects in the files that got 

At most it added one, but really this code is the same as upstream, so
it's probably just a coverity foible; the issue it finds seems real though,
I'll send a patch:

cond_false: Condition "error", taking false branch
if_end: End of if statement
cond_false: Condition "bp", taking false branch
var_compare_op: Comparing "bp" to null implies that "bp" might be null.
cond_true: Condition "error", taking true branch
var_deref_model: Passing null pointer "bp" to function 
"xfs_trans_brelse(xfs_trans_t *, xfs_buf_t *)", which dereferences it.
cond_true: Condition "tp == NULL", taking true branch
deref_parm: Directly dereferencing parameter "bp".


<Prev in Thread] Current Thread [Next in Thread>