GitLab

fixup! Remove netronome firmware

Also remove any WHENCE references to netronome. This avoids errors when
check_whence.py is run, but more importantly, prevents the creation of
broken symlinks in netronome/ for each 'Link:' entry in that section.
These symlinks are created by copy-firmware.sh during 'make install'.

https://phabricator.endlessm.com/T27025
https://phabricator.endlessm.com/T28497

From internal git commit:
b01ce5f49a7ad6683d9b276fc523291b099b0382
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>

The netronome dir includes firmware for high end network devices for
servers, switches and routers, so not useful for devices running EOS.

Removing it frees ~100MB in the ostree:

  linux-firmware$ du --max-depth=0 netronome/
  98360   netronome/

https://phabricator.endlessm.com/T27025

The current Broadcom FullMAC driver embedded in the S905X linux kernel
needs at configuration time the location of the firmware and NVRAM
file. This is a problem when we have two different boards equipped with
two different broadcom chipsets using the same driver + kernel
configuration, like for TCL and ProConcept/AP boards.

In this situation each board (actaully the brcm driver) needs at least a
set of three board-specific files:

1) Firmware data
2) Config file
3) NVRAM file

We can actually force the brcm driver to automatically search for the
correct set (naming) of files.

a) Firmware data: if the driver defines FW_PATH_AUTO_SELECT, the
firmware file name is automatically generated using the initial
information on the transceiver and the file location (directory) can be
specified by CONFIG_BCMDHD_FW_PATH

b) Config file: if the driver defines CONFIG_PATH_AUTO_SELECT again the
config file name is determined by the chip id and the location
(directory) is determined by CONFIG_BCMDHD_NVRAM_PATH

c) NVRAM: in this case explicitly specifying a NVRAM file in the config
file has higher priority over CONFIG_BCMDHD_NVRAM_PATH. This means that
we can specify the path of the NVRAM file using the 'nv_by_chip='
parameter in the config file.

What we are doing with this patch is linking all the board-specific
file into the general directory /lib/firmware/brcm/ and with a kernel
patch we are going to make both CONFIG_BCMDHD_FW_PATH and
CONFIG_BCMDHD_NVRAM_PATH pointing here. The driver modified with the
correct defines will be able to pickup from here the correct files it
needs.
Signed-off-by: Carlo Caione <carlo@endlessm.com>

Import firmware data from the Android image of the ProConcept/AP6255
board.
Signed-off-by: Carlo Caione <carlo@endlessm.com>

This is the WLAN module on the Weibu F3C.

https://phabricator.endlessm.com/T14622

https://phabricator.endlessm.com/T6305

This is the Bluetooth module on the S905X development board.

https://phabricator.endlessm.com/T14153

This is the WLAN module on the S905X development board.

https://phabricator.endlessm.com/T14148

https://phabricator.endlessm.com/T10572

This is the WLAN+BT module on the ATV185 PC-on-a-stick.

https://phabricator.endlessm.com/T6216

Add firmware for EC-100 bluetooth from
Linux_BT_UART_3.0.20150325_8723BS_BTCOEX_20150119-5844.

http://phabricator.endlessm.com/T5517

This will provide the user a place to drop firmware files needed by the
kernel.

In the future (compat level >= 10) we'll want to move
debian/tmpfiles/linux-firmware.conf to debian/linux-firmware.tmpfile, so
it will be automatically installed by dh_installinit, and we can drop
debian/linux-firmware.install.
Signed-off-by: João Paulo Rechi Vita <jprvita@endlessm.com>

https://phabricator.endlessm.com/T12154

Use the update-initramfs trigger instead of calling /usr/sbin/update-initramfs
directly.
Signed-off-by: João Paulo Rechi Vita <jprvita@endlessm.com>

https://phabricator.endlessm.com/T17058

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

The latest upstream rebase adds some raspi4 firmware files which
are also present in linux-firmware-raspi2. Add a breaks to force
linux-firmware-raspi2 to be upgraded first.
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

BugLink: https://bugs.launchpad.net/bugs/1859387

Intel released a new version of sof firmware to Canonical, and in this
new version there are 3 major changes:
 - fix the hang issue in the firmware for the cml and cnl
 - introduce the ldc file which helps to do the firmware debug
 - divide the firmware to cml and cnl specific ones

And Intel suggested that we should integrate the version v1.3-0f73628
and only integrate the firmware for cnl and cml, since like those apl
cht and icl firmwares, they are not fully tested yet.

(backported from beff732e6642bd77dac5dd08514264f055fc4d10
 https://github.com/thesofproject/linux-firmware.git

)
Signed-off-by: Hui Wang <hui.wang@canonical.com>
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

These files are not properly documented in WHENCE.ubuntu, and
upon inpsection they aren't needed as ARCH_IXP4XX is disabled
in the Ubuntu kernel configs.
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Upstream linux-firmware has removed symlinks from the source
tree, instead relying on Link: entries in WHENCE to generate the
symlinks at install time. Our current packaging uses rsync to
copy firmware to the install location, thus we do not get the
symlinks.

Rather than reinventing the logic to create the symlinks, use
the support already present in linux-firmware. Replace the rsync
with "make install", which is going to do a better job anyway as
we still seem to end up with some non-firmware files in
/lib/firmware. Also switch installation of files for nic-firmware
and scsi-firmware to use the install location as the source so
so that, if necessary, symlinks will be correctly installed for
those packages.
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Upstream linux-firmware has removed all symlinks from the git
repo. The links are now created by copy-firmware.sh from Link:
entries in WHENCE. This means we need to use copy-firmware.sh
to install firmware in order to get symlinks.

Update copy-firmware.sh to also check WHENCE.ubuntu for files to
install. This requires using the -h option for grep to suppress
file names in the output.
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Update the File: entries in WHENCE.ubuntu to specify a signle
file per entry so that copy-firmware.sh will be able to extract
firmware files for installation from it.
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

BugLink: https://bugs.launchpad.net/bugs/1850739



This update caused a regression on cometlake platforms. Revert
until fixed firmware is available.
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Add debian/list-missing-firmware, which will take an Ubuntu
kernel fwinfo abi file and list firmware files which are not
present in linux-firmware.
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

BugLink: https://bugs.launchpad.net/bugs/1847250

We have a couple of Dell machines, after installing the system, the
system will hang randomly. After disabling the sof driver, it will not
hang anymore.

Intel help to debug this issue, and finally provide this new firmware
to us, and we tested this firmware can fix the hang issue.

The source code for this firmware is located at:
https://github.com/thesofproject/sof.git



And this firmware is generated based on the sof.git and is
signed with Intel production key.
Signed-off-by: Hui Wang <hui.wang@canonical.com>
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>