On Wed, Jun 30, 2021 at 12:39 PM David Heidelberg <david@xxxxxxx> wrote: > > Kernel: stable-5.10.y > HW: chipset APQ8064, Nexus 7 2013 (codename flo or deb) > Problem first occured when I was porting Nexus from 4.19 to 5.4 LTS. > > Hello, > > after reverting this particular commit ( > 498b98e939007f8bb65094dfa229e84b6bf30e62 ), I'm able do > > echo 'start' > > /sys/devices/platform/soc/3200800.riva-pil/remoteproc/remoteproc0/state > > [ 64.271634] remoteproc remoteproc0: powering up 3200800.riva-pil > [ 64.276210] remoteproc remoteproc0: Booting fw image wcnss.mdt, size > 6804 > [ 70.087107] remoteproc remoteproc0: remote processor 3200800.riva-pil > is now up > [ 70.100817] qcom_wcnss_ctrl remoteproc0:smd-edge.WCNSS_CTRL.-1.-1: > WCNSS Version 1.4 1.2 > [ 70.124279] wcn36xx: mac address: 18:00:2d:88:9c:a9 > [ 95.357907] wcn36xx: firmware WLAN version 'WCN v2.0 RadioPhy > vIris_TSMC_2.0 with 48MHz XO' and CRM version '201080' > [ 95.357938] wcn36xx: firmware API 1.4.1.2, 41 stations, 2 bssids > > with current stable-5.4.y - stable-5.10.y (cannot test later since it > crashes without saving pstore dmesg): > I believe that the APQ8064 boards fell out of the various automation setups and we lost track of the status. Hopefully we can resurrect e.g. the ifc6410 to catch these kinds of regressions. > [ 50.972560] remoteproc remoteproc0: powering up 3200800.riva-pil > [ 50.977121] remoteproc remoteproc0: Booting fw image wcnss.mdt, size > 6804 > [ 50.977166] remoteproc remoteproc0: Failed to load program segments: > -22 > [ 50.977225] remoteproc remoteproc0: Boot failed: -22 Seems like the reasons for this would be that the ELF header and signature segment are stacked differently from the firmware I looked at when I came up with that patch. Can you please share the output of "readelf -a wcnss.mdt"? Regards, Bjorn
