linux-kernel - [RFC PATCH 0/2] Make qcom-ngd-ctrl not wait indefinitely for already booted ADSP

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-next>] [day] [month] [year] [list]

Message-Id: <20220510144219.806391-1-y.oudjana@protonmail.com>
Date:   Tue, 10 May 2022 18:42:17 +0400
From:   Yassine Oudjana <yassine.oudjana@...il.com>
To:     Andy Gross <agross@...nel.org>,
        Bjorn Andersson <bjorn.andersson@...aro.org>,
        Mathieu Poirier <mathieu.poirier@...aro.org>,
        Srinivas Kandagatla <srinivas.kandagatla@...aro.org>
Cc:     Yassine Oudjana <y.oudjana@...tonmail.com>,
        Yassine Oudjana <yassine.oudjana@...il.com>,
        linux-arm-msm@...r.kernel.org, linux-remoteproc@...r.kernel.org,
        linux-kernel@...r.kernel.org, alsa-devel@...a-project.org
Subject: [RFC PATCH 0/2] Make qcom-ngd-ctrl not wait indefinitely for already booted ADSP

Commit a899d324863a3 ("slimbus: qcom-ngd-ctrl: add Sub System Restart support")
made qcom-ngd-ctrl wait for ADSP to become ready before starting to do its work.
Due to how the SSR notifications currently work though, if qcom-ngd-ctrl probes
after ADSP boots and becomes ready, it never receives a QCOM_SSR_AFTER_POWERUP
event notification and keeps waiting indefinitely, making SLIMbus never come up.

This series makes qcom_register_ssr_notifier call the notifier_call of the newly
registered notifier block with the last SSR event received from the remoteproc,
basically reporting the event that qcom-ngd-ctrl missed by registering late,
stopping it from waiting for an event that has already happened.

I'm not sure if this approach would have any unwanted consequences in other
drivers relying on SSR events however, hence I'm sending this as a RFC. This
can also be considered a bug report, so if anyone has a better fix then I'd
appreciate it getting applied instead of this one.

The second patch is a general fix that became necessary after the first patch,
and should likely be applied anyway.

Yassine Oudjana (2):
  remoteproc: qcom: Report last event on SSR notifier registration
  slimbus: qcom-ngd-ctrl: Initialize ngd_up_work before it can be
    scheduled

 drivers/remoteproc/qcom_common.c | 17 +++++++++++++----
 drivers/slimbus/qcom-ngd-ctrl.c  |  3 ++-
 2 files changed, 15 insertions(+), 5 deletions(-)

-- 
2.36.0