this can cause fetch to fail on some systems because the build system
can not detect the checksum type with this trailing space

Josef Schlehofer authored 2 years ago and Michal Vasilek committed 2 years ago

Packages related to old ISDN4Linux was dropped in Linux kernel since
version 5.3, but they were still present in OpenWrt. [1] In recent
commits, where it was updated our Linux 5.15 support the commit from
OpenWrt master branch is included, thus we need to let users know about
this breaking change, because it does not make sense to postpone it or
even have it in distribution anymore.

[1] https://github.com/openwrt/openwrt/commit/db55dea5fc047190af188f07018e99b0c7a4bdde

This is connected to
https://gitlab.nic.cz/turris/os/build/-/raw/b70855d09775ef2c83eb60f8e120097da753d903/patches/openwrt/to-upstream/0102-gcc-ppc-expect-sf-ld-suffix-even-with-spe-enabled.patch
("patches/openwrt: gcc 8.4.0: mpc85xx SPE: expect '-sf' `ld` suffix even
with SPE enabled")

This was allowed because in OpenWrt 21.02 branch, there is now included
this commit [1].

[1]
https://github.com/openwrt/openwrt/commit/cbce6c6d9519ab86e11c9752910a4c72dd52cb58
("kernel: split kernel version to dedicated files")

This enables MLGRU for Linux 5.15

Difference in power consumption between ondemand and performance
governors is negligible (at least on our boards) while ondemand has
noticeable lag when starting transmissions. Performance was also default
governor in previously shipped kernel, so let's stick to it.

We decided that we are not going to release Turris OS 6.0.2 RC2 with
more changes as we would like to release at least something this week to
improve the current behavior of Rainbow, etc.

So, changes added into RC2 are moved to Turris OS 6.0.3

This patch is required only for the latest Linux backports, but we are
not using it, so we can drop it. Reference:
685634ce ("patches/openwrt: 5.15:
mac80211: downgrade")

This reverts commit 838a2844.

Unfortunately, the latest Linux backports does not work for us as there
are some issues in ath10k and mt76, thus Wi-Fi is not working as we
would like.

I need to revert this as in causes one issue in Jenkins. We are building
three targets on one slave and when it is running at the same time, it
usually fails with:

NOTE: Remote head changed. Local mirror removed for re-clone.
Cloning into bare repository '/home/foo/turris-build/git_mirror/openwrt'...
fatal: '/home/foo/turris-build/git_mirror/openwrt' does not appear to be a git repository
fatal: Could not read from remote repository.

It helps to retrigger manually the job, but thats not wanted behaviour right now.

This reverts commit 36b2b5a9.

Fixes:
build/build_dir/target-arm_cortex-a9+vfpv3-d16_musl_eabi/linux-mvebu_cortexa9/mt76-2022-10-01-72b87836/mt7915/init.c:330:37: error: 'IEEE80211_MAX_AMPDU_BUF' undeclared (first use in this function); did you mean 'IEEE80211_MIN_AMPDU_BUF'?
  hw->max_rx_aggregation_subframes = IEEE80211_MAX_AMPDU_BUF;
                                     ^~~~~~~~~~~~~~~~~~~~~~~
                                     IEEE80211_MIN_AMPDU_BUF