linux-kernel - Re: [Cocci] [PATCH v3 0/8] coccicheck: modernize

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 21 Jun 2016 23:02:49 +0200 (CEST)
From:	Julia Lawall <julia.lawall@...6.fr>
To:	"Luis R. Rodriguez" <mcgrof@...nel.org>
cc:	Julia Lawall <julia.lawall@...6.fr>, dmitry.torokhov@...il.com,
	tiwai@...e.de, ming.lei@...onical.com, stephen.boyd@...aro.org,
	deepa.kernel@...il.com, chunkeey@...glemail.com,
	cocci@...teme.lip6.fr, jwboyer@...oraproject.org, jslaby@...e.com,
	zohar@...ux.vnet.ibm.com, dwmw2@...radead.org, hauke@...ke-m.de,
	broonie@...nel.org, akpm@...ux-foundation.org,
	gregkh@...uxfoundation.org, markivx@...eaurora.org,
	linux-kernel@...r.kernel.org, mmarek@...e.com,
	johannes@...solutions.net, torvalds@...ux-foundation.org
Subject: Re: [Cocci] [PATCH v3 0/8] coccicheck: modernize

On Tue, 21 Jun 2016, Luis R. Rodriguez wrote:

> On Tue, Jun 21, 2016 at 10:13:31PM +0200, Julia Lawall wrote:
> > 
> > 
> > On Tue, 21 Jun 2016, Luis R. Rodriguez wrote:
> > 
> > > This v3 series addresses the feedback from the last v2 series
> > > on the coccicheck enhancements [0], namely:
> > > 
> > > o it drops the indexing heuristics in favor for a .cocciconfig use
> > > o drops glimpse support as its simply not well maintained, recommends
> > >   idutils instead.
> > > o adds a Linux .cocciconfig -- the assumption is you'd run spatch when
> > >   you're at the top level of the kernel. This has not only the side effect
> > >   of picking up .cocciconfig, but also that the coccicheck use of the
> > >   make variables passed on are assumed to be correct given the base
> > >   directory as the current directory.
> > 
> > I don't understand this point.  Coccinelle picks up the .cocciconfig, if 
> > any, of the directory on which you want to work, not of the current one.
> 
> The order of precedence for variables for .coccoconfig is as follows:
> 
>  o Your current user's home directory is processed first
>  o Your directory from which spatch is called is processed next
>  o The directory provided with the --dir option is processed last, if used
> 
> Since coccicheck runs through make, it naturally runs from the kernel proper
> dir, as such the second rule above would be implied for picking up a .cocciconfig.
> That's part of the point I'm making.

OK

> Up next let us consider when M= is used or when it is not used, if used
> it populates KBUILD_EXTMOD.
> 
>     if [ "$KBUILD_EXTMOD" = "" ] ; then
>         OPTIONS="--dir $srctree $COCCIINCLUDE"
>     else                                                                        
>         OPTIONS="--dir $KBUILD_EXTMOD $COCCIINCLUDE"
>     fi
> 
> Either way --dir is used, so the third rule applies and so your .cocciconfig
> from there is also read if one is found. My other point was that $COCCIINCLUDE
> has some useful tidbits of includes for coccinelle, and that also assumes
> one is on the top level dir of the kernel.

OK.

> That is sanitized as follows:
> 
> # spatch only allows include directories with the syntax "-I include"           
> # while gcc also allows "-Iinclude" and "-include include"                      
> COCCIINCLUDE=${LINUXINCLUDE//-I/-I }                                            
> COCCIINCLUDE=${COCCIINCLUDE// -include/ --include} 

I don't get the second case.  Is it to replace -include by --include?  
Coccinelle actually supports both, although it doesn't advertise that.  

Also, in LINUXINCLUDE, what is the meaning of -include?  For Coccinelle, 
it is not the same as -I.  It is for files that should be included that 
are not in the set of includes seen by whatever is the specified include 
strategy (--all-includes, etc).  The argument is a specific file name, not 
a directory.  It is a way of eg not bothering with --recursive-includes 
when there is one or a few key header files that each file will need.

> So the point is to annotate that the .cocconfig is picked up first due
> to the fact make is used and its issued from the top level makefile
> and starts from the top level. The fact that --dir is used is important
> but secondary to its introduction as well.

OK, the original text seemed to me to imply that running from the kernel 
directory was essential to getting the kernels .cocciconfig, so I wanted 
to point out that this is not the case.

julia