Slow boot

1st Official Post

    Hello,

    I just migrated from OpenELEC 8.0.4 to LibreELEC 8.2.0 on my Wetek Core.

    I love the HDMI-CEC add-on and the fact that the Kodi version is always up-to-date.

    However, I noticed that the boot time is about 20-30 seconds longer than when on OpeneELEC. Is this normal?

    Thank you!

    Quote from lrusak

    You have the "wait for network" option enabled which holds the boot for 10s. Other than that I don't see anything else out of the ordinary.

    Actually, "Wait for network" was disabled (in LibreELEC settings). I enabled it and now the first 2 lines of systemd-analyze blame are these:

    10.152s network-online.service

    10.139s kodi-waitonnetwork.service


    So, there's an extra 10s when "Wait for network" is enabled (kodi-waitonnetwork.service)

    When I disable the option, I get the output which I uploaded to pastebin, with 10s for network-online.service

    Time between LibreELEC boot and Kodi start is about 25 seconds

    That's it then! I have an external USB drive with an EXT4 and a Journaled HFS+ partition.

    So it must be something related to external USB (specifically EXT4 and/or HFS+ partitions) that's causing the delay. I suspect the HFS+ partition.

    But, like I said, boot time was definitely faster on OE 8.0.4 with this same setup.

    Is there a way to dig deeper into this and see what exactly changed that triggers the extra delay?

    Thank you!

    Quote from lrusak

    ah, Do you have some systemd service that mounts something? The examples depend on Requires=network-online.service which will pull in that dependency.

    So that could be it

    I just found this in dmesg output.

    LibreELEC runs from microSD on the Wetek Core. I am constantly having problems with Android on NAND. Could this be connected?

    [ 8.814880@0] EXT4-fs (mmcblk0p2): mounting with "discard" option, but the device does not support discard

    [ 8.814884@0] EXT4-fs (mmcblk0p2): mounted filesystem with ordered data mode. Opts: (null)

    [ 12.176259@0] Connect Timeout

    [ 12.176316@0] Device Not Connected/Responding

    [ 16.796277@1] WARN::hc_xfer_timeout:2723: hc_xfer_timeout: timeout on channel 3

    [ 16.796335@1] WARN::hc_xfer_timeout:2725: start_hcchar_val 0x01188a00

    [ 16.796393@1] WARN::hc_xfer_timeout:2730: chn-3,ep1-IN:type:2,speed:2,len:4096,addr4


    I also see this error:

    [ 20.696343@0] [aml_sdio_timeout] time_start_cnt:999

    [ 20.696343@0] [aml_sdio_print_err] sdio: Timeout error, Cmd52 arg 0xc00, Xfer 0 Bytes, host->xfer_step=12, host->cmd_is_stop=0, pdata->port=0, virqs=0x13, virqc=0x3f001f00, conf->cmd_clk_divide=177, pdata->clkc=177, conf->bus_width=0, pdata->width=0, conf=0x416270b1, clock=397940

    [ 20.696343@0]

    [ 21.696377@0] [aml_sdio_timeout] time_start_cnt:999

    [ 21.696377@0] <4>[ 22.711456@0] sdio: queuing unknown CIS tuple 0x80 (2 bytes)

    [ 22.713945@0] sdio: queuing unknown CIS tuple 0x80 (3 bytes)

    [ 22.721042@0] sdio: queuing unknown CIS tuple 0x80 (3 bytes)

    [ 22.725698@0] sdio: queuing unknown CIS tuple 0x80 (7 bytes)

    [ 22.817400@0] sdio: new high speed SDIO card at address 0001, clock 35416666, 4-bit-bus-width

    [ 22.820725@0] [sdio_reinit] finish

    UPDATE:

    I started clean with a hard reset, set everything as usual, populated my library and startup time is now normal at about 23s

    Weirdly enough, network-online.service doesn't appear anymore under systemd-analyze

    I haven't installed a few add-ons that I had on my other install, so one of them might be the problem:

    - Youtube v5.4.5

    - nfowatchedstateupdater v1.1.12 (axbmcuser mod)

    - I/O benchmark

    - CPU benchmark

    I also noticed there were a lot of add-on script modules in the other install, probably installed by one of these add-ons. Also, the OpenELEC repository was still there after I migrated.

    So, the issue seems to be resolved, it's still weird though that I had normal boot times with this same setup on OpenELEC 8.0.2