Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
To:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        rafael.j.wysocki@intel.com, swboyd@chromium.org,
        linux-kernel@vger.kernel.org
Cc:     linux-pm@vger.kernel.org,
        Abhishek Pandit-Subedi <abhishekpandit@chromium.org>,
        "Rafael J. Wysocki" <rafael@kernel.org>
Subject: [PATCH 0/1] kobject: Don't emit change events if not in sysfs
Date:   Mon, 19 Oct 2020 15:32:56 -0700
Message-Id: <20201019223257.261223-1-abhishekpandit@chromium.org>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk


Hi maintainers,

A little while ago, I got a bug report of a regression caused by a patch
I submitted a45aca510b73b7 (PM: sleep: core: Emit changed uevent
on wakeup_sysfs_add/remove)

https://bugzilla.kernel.org/show_bug.cgi?id=209469

It seems possible for a "change" event to be sent before the device is
added to the sysfs (so when the rule runs, it can't access the device
path that emitted it). The bug report had the following log that made me
identify this is possible:
        > Use global config file: /etc/usb_modeswitch.conf
        > Use top device dir /sys/bus/usb/devices/2-3
        > Check class of first interface ...
        >  No access to first interface. Exit

I've added a patch to fix the former problem here and confirmed via
udevadm monitor that no CHANGE requests are seen for devices before they
emit the ADD event.

Thanks
Abhishek


Abhishek Pandit-Subedi (1):
  kobject: Don't emit change events if not in sysfs

 lib/kobject_uevent.c | 5 +++++
 1 file changed, 5 insertions(+)

-- 
2.29.0.rc1.297.gfa9743e501-goog