changes after https://reviews.freebsd.org/D48974

Just some nits, see also https://reviews.freebsd.org/D48944

In D48903#1115272, @adrian wrote:

good catch! @jsm what do you think?

man page and fw loading from a bin. otherwise no changes since approval. see also https://reviews.freebsd.org/D48817

Perfectly fine

sed correct use of quantifier and use * instead of +

fixes crashes and fw load on RPI4b

6499c6499

< @@ -0,0 +1,4668 @@

@@ -0,0 +1,4672 @@

7118c7118,7122

< + ic->ic_htcaps = IEEE80211_HTC_HT;

+ ic->ic_htcaps = IEEE80211_HTC_HT
+ | IEEE80211_HTC_AMPDU
+ | IEEE80211_HTC_AMSDU
+ | IEEE80211_HTCAP_MAXAMSDU_3839
+ | IEEE80211_HTCAP_SMPS_OFF;

mostly style. removed some comments, and device_printf, tries to reset on detach for better firmware reloading, but not consistently solving the reinitialization problem.

I think I got most of Gavins, comments, also I got monitor mode working, and put that into the man page. I set force short xfer to 0 again in the TX_BE but I am not sure it relates to device timeouts, I got one under non heavy load. Before ht mode it solved consistent device timeouts.

some ht 20 support. fix ratectl and no need for the short transfer usb hack to avoid device timeout.

In D45179#1032340, @kevlo wrote:

I haven't had time to look at the code, but tested on my Xiaomi mini USB WiFi:

mtw0 on uhub1
mtw0: <MediaTek MI WLAN Adapter, class 0/0, rev 2.01/0.00, addr 3> on usbus0
mtw0: version:0x7601
mtw0: loaded firmware ver 30272.256
mtw0: timeout waiting for MCU to initialize
mtw0: /usr/src/sys/dev/usb/wlan/if_mtw.c:3000 USB_ERR_CANCELLED
device_attach: mtw0 attach returned 6

Unfortunately it doesn't work.

My /boot/loader.conf:
mt7601u_fw_load="YES"
if_mtw_load="YES"

Thanks for testing. I did not manage to get hold of any other devices than a noname one where I never got usb err canceled loading firmware.
. . Is it consistent when unplugging/replugging the device. Can you perhaps post a usbconfig dump and a usbdump. Thanks.

Also add mtw to /usr/src/sbin/devd/devd.conf it escaped my diff.. The firmware port is at https://people.freebsd.org/~jsm/mt7601u-firmware-kmod.tar.gz (It should have the license file set just for ref for now)

add lib_depends on libcurl, make fix portlint warnings (whitespace errors)

Forget half of the patch..

localbase:ldflags works for the math plugins skips patches and postpatch

Assume "trash:///" on FreeBSD as well as Linux, sort USES. use %%LOCALBASE%% instead of LOCALBASE as placeholders.

Thanks for looking into this! Looks good but you could perhaps use upstream with PATCH_SITES and PATCHFILE since they (also ) did the changes https://github.com/lxqt/lxqt-config/commit/6add4e4f0040693e7c4242fbae48c9d3200768.patch ? Just an opinion, I use to do that with non released upstream changes, it makes it easier for me to know which patches are belonging to upstream, and expected in next release..

It does not build in poudriere as is, please add CMAKE_OFF= BUILD_TESTS or add qt5-testlib as dependency. I have not tested the later.

Is the 4991811d9212ec1176af6d1cbe88aa37efad4836 commit part of the new 0.12 version? In that case just delete the patchfiles and path_sites instead of commenting them out. Perhaps put a note in summary about it. Also all other consumers of lxqt-build-tools should be updated or tested before this can becomitted

update to 0.5.0