Re: [PATCH v9] selinux: sidtab: reverse lookup hash table

[Paul Moore <paul@paul-moore.com>]

On Thu, Dec 5, 2019 at 6:47 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
> On Thu, Dec 5, 2019 at 12:52 AM Paul Moore <paul@paul-moore.com> wrote:
> > On Wed, Dec 4, 2019 at 4:11 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
> > > On Tue, Dec 3, 2019 at 1:33 AM Paul Moore <paul@paul-moore.com> wrote:
> > > > Thanks Jeff, as well as everyone else who contributed reviews and feedback.
> > > >
> > > > I've pulled this into a working branch and I'll be merging it with the
> > > > other sidtab patches before posting it to a "next-queue" branch for
> > > > review later this week.  When done, I'll send a note to the list, as
> > > > well as the relevant patch authors; your help in reviewing the merge
> > > > would be greatly appreciated.
> > >
> > > I tried doing the merge on my own here [1], you can use it as a sanity
> > > check if we came to the same/similar result. I based it off your
> > > existing next-queue, which contains only Jeff's patch at the time of
> > > writing. I only build-tested it so far.
> >
> > Thanks, that was a good sanity check.  There are some minor diffs from
> > what I ended up with, but nothing substantive that I can see.
> >
> > Although I'll be honest, the merge wasn't as bad as I thought it would
> > be; most of the fuzz was simply due shuffling and renaming of data
> > structures, which generally isn't too bad.  Although I'm still
> > building the kernel to test it, so let's see if that statement still
> > holds (although it looks like it passed Stephen's testing). ;)
> >
> > If you haven't noticed already, the merge currently lives in the
> > selinux/next-queue branch; if you notice anything off, feel free to
> > send a fixup patch.
>
> It looks OK semantically when compared to my merge. I only see
> reordering/comment/whitespace differences.

Thanks for the double check.  Unfortunately my kernel build locks my
test VM in early boot; it appears to be non-SELinux related and since
the test build is based on selinux/next+patches (which is based off
v5.4-rc1) I imagine there might be some unrelated problems in the
build.  I'm going to rebase my test build to Linus' current and try
this again.

> > > Note that there are two whitespace cleanups included in the string
> > > cache commit that I intuitively did while resolving the merge
> > > conflicts. You might want to move those to the first commit or just
> > > ignore them.
> >
> > When looking at the combined diff between the two sidtab patches and
> > comparing it to your merge I did make a few additional small cosmetic
> > tweaks.  Assuming the testing goes well, I'll probably go over
> > everything one more time to make sure the style looks okay, but today
> > I was focusing more on the correctness.
>
> The whitespace misalignment introduced by Jeff's patch is still there
> in your branch. Personally, I'd prefer that we fix them now rather
> than deferring it to a future patch, because it seems that no one ever
> has time to bother sending whitespace fixup patches :) But I'll
> understand it if you prefer not to touch it more than necessary, so I
> won't fight about this further.

Like I said, I only quickly scanned the combined diff for style
problems so it doesn't surprise me that there are still some issues.
I'll give it a closer look once I can get a kernel build passing all
the tests.

As an aside, I keep debating doing a big style-fix patch (likely
automated via astyle or similar, likely with some additional fixes by
hand, and passed through checkpatch.pl) after one of the -rc1 rebases
to clean up all these little things that have crept into the code over
the years.  I dislike the idea of the churn that would likely bring,
but it should make life a little bit better, and help cut down on the
trivial "checkpatch patches" we get from time to time (although that
really hasn't been too bad of an issue).

-- 
paul moore
www.paul-moore.com