linux-kernel - Re: [PATCHv3] w1: omap-hdq: Simplify driver with PM runtime autosuspend

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:   Mon, 20 Apr 2020 23:11:18 +0200
From:   "H. Nikolaus Schaller" <hns@...delico.com>
To:     Tony Lindgren <tony@...mide.com>
Cc:     Andreas Kemnade <andreas@...nade.info>,
        Evgeniy Polyakov <zbr@...emap.net>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-omap <linux-omap@...r.kernel.org>,
        Adam Ford <aford173@...il.com>,
        "Andrew F . Davis" <afd@...com>, Vignesh R <vigneshr@...com>
Subject: Re: [PATCHv3] w1: omap-hdq: Simplify driver with PM runtime autosuspend

Hi Tony,

> Am 20.04.2020 um 17:08 schrieb Tony Lindgren <tony@...mide.com>:
> 
> * H. Nikolaus Schaller <hns@...delico.com> [200417 21:04]:
>> To me it looks as if reading hqd too quickly after omap_hdq_runtime_resume()
>> may be part of the problem, although it is 0.4 seconds between [   18.355163]
>> and [   18.745269]. So I am not sure about my interpretation.
>> 
>> A different attempt for interpretation may be that trying to read the
>> slave triggers omap_hdq_runtime_resume() just before doing the
>> first hdq_read_byte().
> 
> Hmm so I wonder if adding msleep(100) at the end of
> omap_hdq_runtime_resume() might help?

I have tried and initially it did boot and work once.
But after the second boot/reboot the effect was back.

This is something I have observed previously, that the issue
is there in ca. 9 or 10 boot attempts. So I would assume
some race condition with udev reading the uevent file of the
bq27xxx bus client and hence through hdq.

I already had noticed some hqd_read activity right after probing
success.

I had also tried to change pm_runtime_set_autosuspend_delay(, 1000)
with no success. And I tried to call omap_hdq_runtime_resume() at the
end of probe.

The only maybe important observation was when I disabled all
kernel modules except *hdq*.ko and *bq27*.ko. Then I did only
get an emergency shell so that it is quite similar to the
scenario Andreas has tested. With this setup it did work.

I then tried to reenable other kernel modules but the result
wasn't convincing that it gives a reliable result.

So I have still no clear indication when the problem occurs and
when not.

BR and thanks,
Nikolaus