Posts by lenormf

    Hi kszaq,

    I've been compiling your tree since v7.0.2.009 along with a `hostapd` package I wrote myself, and everything was working fine when setting up a network on `wlan0` (just had to ask `connman` to ignore this network device).

    […]

    It definitely looks like a driver issue (because of the `ioctl` call), but I don't understand why this issue would completey lock the device up in a boot loop afterwards.

    I tried using the upstream `amlogic-3.14` kernel, and the same issue doesn't occur, I reckon something in the patches added to the fork is causing these issues (using `wlan0` as a regular device seems to work).

    Hi kszaq,

    I've been compiling your tree since v7.0.2.009 along with a `hostapd` package I wrote myself, and everything was working fine when setting up a network on `wlan0` (just had to ask `connman` to ignore this network device).

    However since the 8.x upgrade trying to run `hostapd` on `wlan0` fails massively, the program segfaults and then the box (S905x) enters a boot loop. That makes no sense to me, as everything used to run just fine, I've tried downgrading `connman`, upgrading `hostapd` (2.5 → 2.6), using the proper DTB as opposed to the master one compiled… The only thing I haven't tried is downgrading the kernel because it's not as simple as changing the hash in the `options` file, and the `rtl8189es` driver hasn't changed anyway.

    Do you have any hunch as to what could possibly be going on please ?

    `hostapd` log:

    Code
    write(1, "hostapd_setup_bss(hapd=0x1f6170 "..., 50hostapd_setup_bss(hapd=0x1f6170 (wlan0), first=1)
        ) = 50
        write(1, "wlan0: Flushing old station entr"..., 36wlan0: Flushing old station entries
        ) = 36
        ioctl(4, _IOC(0, 0x8b, 0xfc, 0x00) <unfinished ...>
        +++ killed by SIGSEGV +++
        Segmentation fault

    It definitely looks like a driver issue (because of the `ioctl` call), but I don't understand why this issue would completey lock the device up in a boot loop afterwards.

    Thanks.

    Hi @kszaq,

    I've been trying to build an image in a Debian Stretch docker container, but the compilation of `kodi:host` fails with the following error:

    The gcc version in the sysroot basically doesn't support this -mthumb option, which is weird because I could compile images of previous versions just fine.

    What environment do you build your images in, and have you tried doing a clean build before releasing images for the current version?

    Thanks.

    @kszaq

    Just wanted to let you know that the S905 project doesn't compile currently because of the `qca9377-aml` package, the source code changed upstream (minor name changes, but they didn't bother bumping their version up).

    I modified your patch (qca9377-aml-001-Modify-SDIO-and-firmware-handling-to-support-chips-u.patch) to make it compile, here it is: dpaste: 17HDCXJ.