The one we switch to has version 5.90.100.43, while original one has
version 5.90.100.41 (can be checked with: `strings bcmdhd_sta.bin |
tail -n1`). The p4wifi tablet is also using the same blob (but
downloaded from elsewhere).
I believe the blob originates from the msm version of galaxy s2, with
codename "skyrocket". It seems to be the newest blob for brcm4330b1
available. The linux-firmware repo only has a blob for brcm4330b2,
which does not work for i9100.
Unfortunately even with latest available firmware blob mainline linux
still prints errors like:
[ 593.513854] ieee80211 phy1: brcmf_proto_bcdc_query_dcmd: brcmf_proto_bcdc_msg failed w/status -110
[ 593.513926] ieee80211 phy1: brcmf_cfg80211_dump_station: BRCMF_C_GET_ASSOCLIST failed, err=-110
a few times per minute.
In the 6.9 branch during development and rebasing this chunk seems to
have disappeared. Since 6.10 msm8974 release is coming probably not
super soon, patch it now so hlte gets its dts back.
Fixes: 434e01295 ("device/google-sargo: Move from testing to community (MR 5492)")
[ci:skip-vercheck]
[ci:skip-build]: already built successfully in CI
The kernel-side parts necessary to get jack detection to work have been
merged as part of the 6.10 upgrade; now, update the UCM config to allow
for automatically switching to headphones and headset mic when a cable
is inserted.
[ci:skip-build]: already built successfully in CI
And add Richard as codeowner.
Also move linux-postmarketos-qcom-sdm670 and alsa-ucm-conf-qcom-sdm670
which otherwise are used by google-bonito, which remains in testing.
For that, enable the kernel config check and update the config
accordingly.
[ci:skip-vercheck]
The offending patch that has caused this dependency has been fixed:
https://git.kernel.org/torvalds/c/7258fdd7d
This essentially reverts commit fb029ff2d9.
[ci:skip-vercheck] # no rebuild needed for this
[ci:skip-build]: already built successfully in CI
The 4k sector size is required for mobile, however we currently don't
build a split image anyway (which is also necessary). Let's just go for
512 bytes so the prebuilt images can be flashed to a USB drive.
[ci:skip-build]
I've had the kernel panic with a NULL pointer dereference in the
addrconf_rt_table function. I have no idea how it could be triggered,
but the fix is straightforward.
Henrik Grimler