linux-kernel - Re: [PATCH 1/2] clocksource: timer-ti-dm: Do not restore context on every timer enable

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:   Wed, 26 Feb 2020 09:09:39 -0800
From:   Tony Lindgren <tony@...mide.com>
To:     Lokesh Vutla <lokeshvutla@...com>
Cc:     Daniel Lezcano <daniel.lezcano@...aro.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Linux OMAP Mailing List <linux-omap@...r.kernel.org>,
        linux-kernel@...r.kernel.org, narmstrong@...libre.com,
        Sekhar Nori <nsekhar@...com>, Tero Kristo <t-kristo@...com>
Subject: Re: [PATCH 1/2] clocksource: timer-ti-dm: Do not restore context on
 every timer enable

* Lokesh Vutla <lokeshvutla@...com> [200224 05:09]:
> omap_dm_timer_enable() restores the entire context(including counter)
> based on 2 conditions:
> - If get_context_loss_count is populated and context is lost.
> - If get_context_loss_count is not populated update unconditionally.
> 
> Case2 has a side effect of updating the counter register even though
> context is not lost. When timer is configured in pwm mode, this is
> causing undesired behaviour in the pwm period. So restore context only
> if get_context_loss_count is populated and context is actually lost.

Sounds like this will break things. We have get_context_loss_count()
only on omap4 and later, and even that was only partially implemented
from what I recall.

To mee it seems the right thing to do here is to save and restore
context on CPU_CLUSTER_PM_ENTER and CPU_CLUSTER_PM_EXIT. And I'd
just get rid of the partially implemented custom calls to
get_context_loss_count().

See for example a recent patch on removing the legacy SDMA code
for CPU_CLUSTER_PM_ENTER in commit d2f924879e19 ("thermal:
ti-soc-thermal:  Enable addition power management").

Then if we really need to add checks for context lost at some point,
that can be implemented via reset_control_status() returning -ENOLINK
or something similar.

Regards,

Tony