samsung-klte mainline cannot find partitions and no usb rndis
Freshly built pmOS klte mainline image does not boot with linux-postmarketos-qcom-5.2.0-r0 armv7
UART log:
Welcome to minicom 2.7
PARAMETERS: I18n
Build date Feb 26 2019, 21:45:57.
Port /dev/ttyUSB0, 01:17:47
Press CTRL-A Z to get help on shortcuts
Android Bootloader - UART_DM Initialized!!!
[fuelgauge] init_fuel_gauge: battery type = 1
[fuelgauge] init_fuel_gauge: fg_reset_soc = 0
[fuelgauge] init_fuel_gauge: vcell=4177, soc=95, ocv=4310
MODEL_NAME:{{SM-G900F}}
eMMC_SERIAL_NUMBER:{{15010041474e44335200b822f679c100}}
BOOTLOADER VERSION : G900FXXS1CQD1
Get DATA success.
Get DATA success.
check_pit_integrity: invalid pit magic code.(0x0)
AST_POWERON
[eMMC] WP status = 0x00, addr = 0x0, (0x01: Temporary WP; 0x02: Power_on WP)
[eMMC] WP status = 0x02, addr = 0x0, (0x01: Temporary WP; 0x02: Power_on WP)
[ 0.992328] spmi spmi-0: pmic_arb_wait_for_done: transaction failed (0x3)
[ 0.993155] spmi spmi-0: pmic_arb_wait_for_done: transaction failed (0x3)
[ 0.998131] qcom-spmi-mpp fc4cf000.spmi:pm8841@4:mpps@a000: read 0x4 failed
[ 1.010225] qcom-spmi-gpio fc4cf000.spmi:pm8941@0:gpios@c000: incorrect block type 0x1 at 0xd600
[ 1.017260] qcom-spmi-regulator fc4cf000.spmi:pm8941@1:regulators: unsupported regulator: name=s4 type=0x00, subtype=0x00, dig major rev=0x00
[ 1.042019] spmi spmi-0: pmic_arb_wait_for_done: transaction failed (0x3)
[ 1.042049] spmi-temp-alarm fc4cf000.spmi:pm8841@4:temp-alarm@2400: could not read type
### postmarketOS initramfs ###
Configuring kernel firmware image search path
modprobe: module ext4 not found in modules.dep
Trying to mount subpartitions for 10 seconds...
ERROR: failed to mount subpartitions!
NOTE: Waiting 10 seconds for the framebuffer /dev/fb0.
If your device does not have a framebuffer, disable this with:
no_framebuffer=true in <https://postmarketos.org/deviceinfo>
ERROR: /dev/fb0 did not appear!
Assertion failed: fd >= 0 (refresher.c: main: 37)
fbsplash: can't open '/dev/fb0': No such file or directory
Setup usb network
/sys/class/android_usb does not exist, skipping android_usb
Setting up an USB gadget through configfs
No USB Device Controller available
Starting udhcpd
Could not find an interface to run a dhcp server on
Interfaces:
1: lo: <LOOPBACK> mtu 65536 qdisc noop state DOWN qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN qlen 1000
link/ether fa:79:32:f9:a6:7d brd ff:ff:ff:ff:ff:ff
ERROR: boot partition not found!
fbsplash: can't open '/dev/fb0': No such file or directoryProbably the problems as with hammerhead? at least for usb?
Activity
@minlexx The kernel is now based on masney repository, which probably doesn't contain the samsung klte patches...
see commits here -> https://gitlab.com/postmarketOS/linux-postmarketos/commits/qcom
By Daniele Debernardi on 2019-09-24T10:03:36
Edited by Administrator-
But the changes should be submitted to actual mainline anyways and not be stuck in our repository on v4.17... So rebasing on top of either the 5.2 tree or on the latest kernel release or on linux-next is the next step for whoever wants to get it working again.
By Luca Weiss on 2019-09-24T10:28:48
-
So I blame @drebrez for not submitting klte patches to mainline
whoever wants to get it working again.
I'd like to have it working again ..
By Alexey Min on 2019-09-24T14:28:12
Edited by Administrator -
@minlexx I don't think those patches are ready to be mainlined, at the time I asked suggestions in msm-something matrix channel but they didn't help me.
The "problem" is that klte seems the only msm8974 that uses different regulator, so the 2° patch was about disabling the unused regulator and add the new regulator nodes, then re-enable the correct regulators in each device dts. I wasn't sure about this approach.
Then there is another "hack" in this patch (https://gitlab.com/postmarketOS/linux-postmarketos/commit/f9d33c737f791878272d9677cc678407d3455b31) where I initially included the unnecessary regulators dtsi files only because there are nodes using it inqcom-msm8974.dtsi, and then removed by overriding an unrelatedpm8941_gpiosnode.
And then my device power IC burnt....By Daniele Debernardi on 2019-09-24T16:40:50
added category::device + 1 deleted label
-
What if we create a SoC-specific "mainline" kernel packages (like already existing linux-postmarketos-qcom-msm8916,
linux-postmarketos-qcom-msm8974, etc), but that don't target custom kernel forks as source repo, instead target current mainline https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/?h=v5.3 (or stable https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/?h=v5.3.1) fromgit.kernel.organd place all patches that are not mainlined separately as files in package? Like now in every downstream kernel package. Phones that share common SoC should be able to share the same kernel, same defconfig and a common set of patches, too? Well, some patches may be device-specific, but still it should be possible for them to live in the same tree...By Alexey Min on 2019-09-24T20:09:53
Edited by Administrator -
I'm all for this, but in the very case of msm8974 there's 8974AA, AB and AC (with the latter two being considered msm8974pro) and with each and every one of them having more than one revision (but that's less of a deal) it'd be hard to support all of them within one kernel package especially with hacks like setting the clock to a constant value - let's say hardcoding mdss-related freqs cause of the lack of msm_bus, some chips can clock higher and some can clock lower, so the general idea there to keep compatibility would be to set it to a maximum that all of them can do but then it wastes potential on the more capable ones so all of that becomes one big mess.
^ This might not be entirely correct. if someone has more knowledge than I do, I will be happy to be corrected on anything I got wrong there.
By Konrad Dybcio on 2019-09-24T20:09:53
