From mboxrd@z Thu Jan 1 00:00:00 1970 From: tipecaml@gmail.com (Cyril Roelandt) Date: Thu, 22 Nov 2012 00:51:19 +0100 Subject: [Cocci] [Review needed] Ensure that calls to d_find_alias() are followed by dput(). In-Reply-To: References: <509B0B81.50304@gmail.com> <509C118D.3010109@gmail.com> Message-ID: <50AD68F7.7050607@gmail.com> To: cocci@systeme.lip6.fr List-Id: cocci@systeme.lip6.fr On 11/09/2012 10:29 PM, Julia Lawall wrote: > I have attached a new version. I put some more headers at the top of the > file. For the Confidence you should put High, Moderate, or Low. I put > Moderate, but perhaps you will think that something else is better. > Thanks! > Exists causes the rule to match when there is a single control-flow path > through the function that satisfies the specified pattern. Without it, > all paths from the first thing that matches are required to match the > pattern, unless * is used (context mode). > > <... P ...> allows 0 or more occurrences of P within the path matched by > <... ...>. It is just like ... except that it allows the occurrences of > P (that you might want to modify, collect the position of etc.) > > <+... P ...+> is like <... P ...> except that it requires that there is > at least one occurrence of P in the control-flow path. > Ok, thanks for your explanation. > I am not sure that Peter's false positive is really a false positive. > The missing dput is at the beginning of the function, whereas the patch > addressed the end of the function. Perhaps the author of the patch > thought about the function an realized that a dput wasn't needed at the > beginning. But the patch at least doesn't say why. > Indeed, it was not a false positive, the patch got applied in -mm recently. > Semantic patches for integration into Linux should be sent to Michal > Marek , as well as to the people at COCCINELLE/Semantic > Patches (SmPL) in MAINTAINERS. > Ok, I did that, thanks a lot. Cyril Roelandt.