linux-kernel - Script to do smart sparse diffs (was Re: [git pull] vfs.git regression fix Re: Regression related to ipc shmctl compat)

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]

Message-ID: <87mv5f73lt.fsf@concordia.ellerman.id.au>
Date:   Thu, 28 Sep 2017 16:13:18 +1000
From:   Michael Ellerman <mpe@...erman.id.au>
To:     Christoph Hellwig <hch@...radead.org>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Daniel Axtens <dja@...ens.net>
Cc:     Al Viro <viro@...iv.linux.org.uk>, Kyle Huey <me@...ehuey.com>,
        open list <linux-kernel@...r.kernel.org>,
        Robert O'Callahan <robert@...llahan.org>
Subject: Script to do smart sparse diffs (was Re: [git pull] vfs.git regression fix Re: Regression related to ipc shmctl compat)

Christoph Hellwig <hch@...radead.org> writes:

> On Mon, Sep 25, 2017 at 06:37:28PM -0700, Linus Torvalds wrote:
>> Pulled and pushed out, but I'd like to note that sparse would have
>> caught this. Except we are so far away from being sparse-clean that
>> nobody runs it.
>
> I tend to run sparse over the nvme code before sending pull request
> every time.  But it's a fairly new codebase, so it it actually
> is clean.  I wish we'd just default to running sparse at some point
> so people have to clean their shit up, as it catches a lot of
> useful things.  But maybe for the default we want to tune it down
> a bit (e.g. don't warn about missing statics by default, skip
> the lock imbalance checks which while often useful also generate
> tons of false positives).

Daniel (++Cc) wrote a script a while back that can do a "smart" diff of
the sparse output from two builds. Roughly it sorts the output
(important when using make -j) and does some other munging to try and
give you a minimal diff across runs.

That allows you to check if a commit added new sparse warnings without
the build being clean at the beginning.

Anyway it's here if anyone wants to try it:

  https://github.com/daxtens/smart-sparse-diff


cheers