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  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]
Date:   Wed, 26 Oct 2022 15:15:29 +0200
From:   Matthias Schiffer <>
To:     Arnd Bergmann <>,
        Greg Kroah-Hartman <>,
        Rob Herring <>,
        Krzysztof Kozlowski <>
Cc:     Marcel Holtmann <>,
        Johan Hedberg <>,
        Luiz Augusto von Dentz <>,
        Amitkumar Karwar <>,
        Ganapathi Bhat <>,
        Sharvari Harisangam <>,
        Xinming Hu <>,
        Kalle Valo <>,
        "David S. Miller" <>,
        Eric Dumazet <>,
        Jakub Kicinski <>,
        Paolo Abeni <>,,,,,,,
        Matthias Schiffer <>
Subject: [RFC 0/5] "notify-device" for cross-driver readiness notification

This patch series is obviously missing documentation, MAINTAINERS
entries, etc., but I'd like to solicit some basic feedback on whether
this approach makes sense at all before I proceed. If it does, the
naming is also very much open for bikeshedding - I'm not too happy with

The basic problem that the notify-device tries to solve is the
synchronization of firmware loading readiness between the Marvell/NXP
WLAN and Bluetooth drivers, but it may also be applicable to other

The WLAN and Bluetooth adapters are handled by separate drivers, and may
be connected to the CPU using different interfaces (for example SDIO for
WLAN and UART for Bluetooth). However, both adapters share a single
firmware that may be uploaded via either interface.

For the SDIO+UART case, uploading the firmware via SDIO is usually
preferable, but even when the interface doesn't matter, it seems like a
good idea to clearly define which driver should handle it. To avoid
making the Bluetooth driver more complicated than necessary in this case,
we'd like to defer the probing of the driver until the firmware is ready.

For this purpose, we are introducing a notify-device, with the following

- The device is created by a driver as soon as some "readiness
  condition" is satisfied
- Creating the device also binds a stub driver, so deferred probes are
- Looking up the notify device is possible via OF node / phandle reference

This approach avoids a hard dependency between the WLAN and Bluetooth
driver, and works regardless of the driver load order.

The first patch implementes the notify-device driver itself, and the
rest shows how the device could be hooked up to the mwifiex and hci_mrvl
drivers. A device tree making use of the notify-device could look like
the following:

    &sdhci1 {
        wifi@1 {
            compatible = "marvell,sd8987";
            reg = <1>;
            wifi_firmware: firmware-notifier {};

    &main_uart3 {
        bluetooth {
            compatible = "marvell,sd8987-bt";
            firmware-ready = <&wifi_firmware>;

Matthias Schiffer (5):
  misc: introduce notify-device driver
  wireless: mwifiex: signal firmware readiness using notify-device
  bluetooth: hci_mrvl: select firmwares to load by match data
  bluetooth: hci_mrvl: add support for SD8987
  bluetooth: hci_mrvl: allow waiting for firmware load using

 drivers/bluetooth/hci_mrvl.c                |  77 ++++++++++++--
 drivers/misc/Kconfig                        |   4 +
 drivers/misc/Makefile                       |   1 +
 drivers/misc/notify-device.c                | 109 ++++++++++++++++++++
 drivers/net/wireless/marvell/mwifiex/main.c |  14 +++
 drivers/net/wireless/marvell/mwifiex/main.h |   1 +
 include/linux/notify-device.h               |  33 ++++++
 7 files changed, 228 insertions(+), 11 deletions(-)
 create mode 100644 drivers/misc/notify-device.c
 create mode 100644 include/linux/notify-device.h


Powered by blists - more mailing lists