linux-kernel - Re: sun50i-a64-pinctrl WARN

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: <20170317163231.psjkwbluqizojwyj@angband.pl>
Date:   Fri, 17 Mar 2017 17:32:31 +0100
From:   Adam Borowski <kilobyte@...band.pl>
To:     Tejun Heo <tj@...nel.org>
Cc:     Icenowy Zheng <icenowy@...c.xyz>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Andre Przywara <andre.przywara@....com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: sun50i-a64-pinctrl WARN_ON drivers/base/dd.c:349

On Fri, Mar 17, 2017 at 10:44:22AM -0400, Tejun Heo wrote:
> On Fri, Mar 17, 2017 at 10:28:34PM +0800, Icenowy Zheng wrote:
> > > It's warning that the device has resources associated with it on
> > > probe. There gotta be something fishy going on with the probing
> > > sequence. How reproducible is the problem?
> > 
> > Do you mean in the first probing trial the driver didn't clean up well?
> 
> Possibly but devres should have released all resources after the
> previous probe failure or driver disassociation, so I have no idea how
> there can be resources left on that list.
> 
> > With the same driver I didn't see this problem in 4.11-rc{1,2}.
> 
> devres hasn't changed, so I have no idea what changed that.  Which
> kernels are affected?

It's a not-yet-mainlined part (that's why I sent the report to Icenowy and
Andre Przywara -- they work on the DT bindings and the driver itself).

The exact kernel is based on git@...hub.com:Icenowy/linux.git
(icenowy/sunxi64-4.11-rc1); I've removed all additional commits that could
possibly be related.  I need at the very least a f2fs fix to boot this
machine (without reformatting).  I did not bother trimming things that are
obviously irrelevant (n_tty, vt, btrfs, sd); lemme retry with 4.11-rc2 that
has just the f2fs fix and "#define DEBUG" in drivers/base/dd.c to be sure.

I've first noticed the warning on pre-4.11 -next; because of very intense
work in these areas Icenowy, Andre and the rest are doing the pine64 parts
are hard to naively rebase -- and thus I haven't run intermediate versions.

> Can you bisect if the problem is easily reproducible?

While it reproduces 100%, the churn in required commits atop mainline would
make bisecting pretty tricky.  Do you have some other ideas over debugging
the hard way?  (I don't know these parts of the kernel myself thus I
reported first.)

In case my exact .config and tree are relevant:
https://angband.pl/tmp/config-kb-ice-4.11-rc1.xz
git@...hub.com:kilobyte/linux.git kb-ice-4.11-rc1
-- 
⢀⣴⠾⠻⢶⣦⠀ Meow!
⣾⠁⢠⠒⠀⣿⡁
⢿⡄⠘⠷⠚⠋⠀ Collisions shmolisions, let's see them find a collision or second
⠈⠳⣄⠀⠀⠀⠀ preimage for double rot13!