Problems latest firmware

[flotsam]

I can't say that I've followed this whole thread, but any chance that this is related to my problem? Does your MAC address change with reboots?

[cam1965]

I can't say that I've followed this whole thread, but any chance that this is related to my problem? Does your MAC address change with reboots?

In android,  my wifi saved connections are OK.
In debian I've  checked and at  every boot I have to choose my SSID network and  re-enter my password again.
I've checked the mac address in debian and it really changes at every boot.
Will check later in google if there is a solution for this.
Thank you again.

[mithrandir]

As I said , I think Planet has changed , because every another kernel that I flashed didn't boot on gemini . The last was compiled perfectly, but for some reasons it didn't boot. Only the kernel provided inside the debian image can boot gemini. Also I've tested the kernel from the repository ( not the modular ) and it installs perfectly . No need to dd, because it automatically detects the correct partition. Reinstalled again the modular kernel . The partition is correct. ( boot2 ) . I've checked this. But I've also did a test with the manual dd at the correct partition and only the kernel provided from planet boots gemini.

follow the results of /cat/cmdline attached.
Please, contact planet and tell me how are the steps to compile his kernel. Because I've compiled the kernel modules from the source kernel ( they are also with + sign ) , but I can' t load any module as I said before .Also when I do lsmod ( with my modules compiled ) an error appears saying libkmod error , etc ... and could not open /proc/modules.
Thank you again

From your cmdline you appear to have an unmodified bootloader, so there is no way the kernel package could detect the right boot partition. Otherwise there would be a bootpartition entry in the cmdline. Unless the method of partition detection has changed (@Adam: Has it?), manual flashing with dd is necessary.

[cam1965]

As I said , I think Planet has changed , because every another kernel that I flashed didn't boot on gemini . The last was compiled perfectly, but for some reasons it didn't boot. Only the kernel provided inside the debian image can boot gemini. Also I've tested the kernel from the repository ( not the modular ) and it installs perfectly . No need to dd, because it automatically detects the correct partition. Reinstalled again the modular kernel . The partition is correct. ( boot2 ) . I've checked this. But I've also did a test with the manual dd at the correct partition and only the kernel provided from planet boots gemini.

follow the results of /cat/cmdline attached.
Please, contact planet and tell me how are the steps to compile his kernel. Because I've compiled the kernel modules from the source kernel ( they are also with + sign ) , but I can' t load any module as I said before .Also when I do lsmod ( with my modules compiled ) an error appears saying libkmod error , etc ... and could not open /proc/modules.
Thank you again

From your cmdline you appear to have an unmodified bootloader, so there is no way the kernel package could detect the right boot partition. Otherwise there would be a bootpartition entry in the cmdline. Unless the method of partition detection has changed (@Adam: Has it?), manual flashing with dd is necessary.

Thank you .
As I said I am using the default installation from Planet. ( debian and android oreo images ). Fresh install ( flashed ).
I think something has changed. I also did manual  flashing with dd ( kernel ) and only the kernel from debian image boots gemini.
Thank you again.

[mibry]

As I said , I think Planet has changed , because every another kernel that I flashed didn't boot on gemini . The last was compiled perfectly, but for some reasons it didn't boot. Only the kernel provided inside the debian image can boot gemini. Also I've tested the kernel from the repository ( not the modular ) and it installs perfectly . No need to dd, because it automatically detects the correct partition. Reinstalled again the modular kernel . The partition is correct. ( boot2 ) . I've checked this. But I've also did a test with the manual dd at the correct partition and only the kernel provided from planet boots gemini.

follow the results of /cat/cmdline attached.
Please, contact planet and tell me how are the steps to compile his kernel. Because I've compiled the kernel modules from the source kernel ( they are also with + sign ) , but I can' t load any module as I said before .Also when I do lsmod ( with my modules compiled ) an error appears saying libkmod error , etc ... and could not open /proc/modules.
Thank you again

From your cmdline you appear to have an unmodified bootloader, so there is no way the kernel package could detect the right boot partition. Otherwise there would be a bootpartition entry in the cmdline. Unless the method of partition detection has changed (@Adam: Has it?), manual flashing with dd is necessary.

Thank you .
As I said I am using the default installation from Planet. ( debian and android oreo images ). Fresh install ( flashed ).
I think something has changed. I also did manual  flashing with dd ( kernel ) and only the kernel from debian image boots gemini.
Thank you again.

I also reflashed my gemini with the new debian image from planet support website and had the same kernel as cam1965. I did a dd of the linux-boot.img kernel image in /usr/share/kernel and was able to boot into debian. I also notice before I flashed the image I had the same kernel version as cam1965. The kernel version has now changed to 3.18.41 without the + at the end.

[flotsam]

I can't say that I've followed this whole thread, but any chance that this is related to my problem? Does your MAC address change with reboots?

In android,  my wifi saved connections are OK.
In debian I've  checked and at  every boot I have to choose my SSID network and  re-enter my password again.
I've checked the mac address in debian and it really changes at every boot.
Will check later in google if there is a solution for this.
Thank you again.

Sounds like the same problem. Nice to know I'm not the only one suffering from it. So multi-boot is broken, at least for some of us. Anyone from Planet Computers want to chip in?

[cam1965]

As I said , I think Planet has changed , because every another kernel that I flashed didn't boot on gemini . The last was compiled perfectly, but for some reasons it didn't boot. Only the kernel provided inside the debian image can boot gemini. Also I've tested the kernel from the repository ( not the modular ) and it installs perfectly . No need to dd, because it automatically detects the correct partition. Reinstalled again the modular kernel . The partition is correct. ( boot2 ) . I've checked this. But I've also did a test with the manual dd at the correct partition and only the kernel provided from planet boots gemini.

follow the results of /cat/cmdline attached.
Please, contact planet and tell me how are the steps to compile his kernel. Because I've compiled the kernel modules from the source kernel ( they are also with + sign ) , but I can' t load any module as I said before .Also when I do lsmod ( with my modules compiled ) an error appears saying libkmod error , etc ... and could not open /proc/modules.
Thank you again

From your cmdline you appear to have an unmodified bootloader, so there is no way the kernel package could detect the right boot partition. Otherwise there would be a bootpartition entry in the cmdline. Unless the method of partition detection has changed (@Adam: Has it?), manual flashing with dd is necessary.

Thank you .
As I said I am using the default installation from Planet. ( debian and android oreo images ). Fresh install ( flashed ).
I think something has changed. I also did manual  flashing with dd ( kernel ) and only the kernel from debian image boots gemini.
Thank you again.

I also reflashed my gemini with the new debian image from planet support website and had the same kernel as cam1965. I did a dd of the linux-boot.img kernel image in /usr/share/kernel and was able to boot into debian. I also notice before I flashed the image I had the same kernel version as cam1965. The kernel version has now changed to 3.18.41 without the + at the end.

PERFECT.
Problem solved.
Thank you so much.
Now almost everything is fine.
Only the problem of mac address changing after every boot. ( in debian )

Thank you again.

[cam1965]

I can't say that I've followed this whole thread, but any chance that this is related to my problem? Does your MAC address change with reboots?

In android,  my wifi saved connections are OK.
In debian I've  checked and at  every boot I have to choose my SSID network and  re-enter my password again.
I've checked the mac address in debian and it really changes at every boot.
Will check later in google if there is a solution for this.
Thank you again.

Sounds like the same problem. Nice to know I'm not the only one suffering from it. So multi-boot is broken, at least for some of us. Anyone from Planet Computers want to chip in?

In android I've did a reset in networlk settings . Because my mac address was 02:00:00:00:00:00. After this,  it has  changed .
But in debian the problem persists.
I will test with the new kernel ( please see above )  if the problem persists. But I think it is not related to the kernel.
I will search again on the google . If I found a solution I will return here. Now I am at work and I will check this problem  only at night ( about more 8 hours ).

Thank you again.

[mithrandir]

In android I've did a reset in networlk settings . Because my mac address was 02:00:00:00:00:00. After this,  it has  changed .
But in debian the problem persists.
I will test with the new kernel ( please see above )  if the problem persists. But I think it is not related to the kernel.
I will search again on the google . If I found a solution I will return here. Now I am at work and I will check this problem  only at night ( about more 8 hours ).

Thank you again.

Maybe restoring the NVRAM from backup fixes the mac address issue.

[cam1965]

In android I've did a reset in networlk settings . Because my mac address was 02:00:00:00:00:00. After this,  it has  changed .
But in debian the problem persists.
I will test with the new kernel ( please see above )  if the problem persists. But I think it is not related to the kernel.
I will search again on the google . If I found a solution I will return here. Now I am at work and I will check this problem  only at night ( about more 8 hours ).

Thank you again.

Maybe restoring the NVRAM from backup fixes the mac address issue.

Unfortunately I did not make a backup of NVRAM.
But I saw this :

https://wiki.archlinux.org/index.php/MAC_address_spoofing

Later I will did  a  test ( when I arrive at home ).

Only to know what service connman uses : systemd  ?

Thank you again.

[mithrandir]

Unfortunately I did not make a backup of NVRAM.
But I saw this :

https://wiki.archlinux.org/index.php/MAC_address_spoofing

Later I will did  a  test ( when I arrive at home ).

Only to know what service connman uses : systemd  ?

Thank you again.

The Gemini uses systemd. If you provide your mac/imei/etc. Planet might be able generate a NVRAM "backup" for you. At least there have been rumors... Writing a mail is free...

[cam1965]

Unfortunately I did not make a backup of NVRAM.
But I saw this :

https://wiki.archlinux.org/index.php/MAC_address_spoofing

Later I will did  a  test ( when I arrive at home ).

Only to know what service connman uses : systemd  ?

Thank you again.

The Gemini uses systemd. If you provide your mac/imei/etc. Planet might be able generate a NVRAM "backup" for you. At least there have been rumors... Writing a mail is free...

 Ok.
Thank you so much again.

[flotsam]

Maybe restoring the NVRAM from backup fixes the mac address issue.

Unfortunately, I don't think the problem is quite so simple. At least mine isn't.

I did make an NVRAM backup. And while I didn't restore it, I did another backup after the problem appeared, and compared the two. They were identical. So I don't think that restoring it would make a difference.

Also, when I reflash using "firmware upgrade" mode (but no restore of the original NVRAM backup), I do get my original MAC address back. (I didn't really pay attention to the IMEI; don't know about that). So the MAC address does seem to be uncorrupted in the NVRAM. It just isn't getting set on the interface after the first android boot.

But the problem does seem to be related to the early-boot NVRAM settings. I don't really understand the process well enough to say what is going on. Informed speculation is welcome.

[mithrandir]

Maybe restoring the NVRAM from backup fixes the mac address issue.

Unfortunately, I don't think the problem is quite so simple. At least mine isn't.

I did make an NVRAM backup. And while I didn't restore it, I did another backup after the problem appeared, and compared the two. They were identical. So I don't think that restoring it would make a difference.

Also, when I reflash using "firmware upgrade" mode (but no restore of the original NVRAM backup), I do get my original MAC address back. (I didn't really pay attention to the IMEI; don't know about that). So the MAC address does seem to be uncorrupted in the NVRAM. It just isn't getting set on the interface after the first android boot.

But the problem does seem to be related to the early-boot NVRAM settings. I don't really understand the process well enough to say what is going on. Informed speculation is welcome.

The NVRAM backup of course is just the first thing to try. Once had a similar problem with unknown baseband: https://www.oesf.org/forum/index.php?showtopic=35233&hl=
Tried quite some stuff before I got it working again. Probably erasing the contents of /nvdata or /data/nvram has been part of the solution. Afterwards the directory got refilled from NVRAM (on reboot , if I remember correctly), I think. No guarantees on that, since I have tried too many things to recall the complete fix. You might try out if this helps (KEEP A BACKUP) to solve your problem. This is highly speculative, but you have asked for

[Adam Boardman]

I'm still away but the fix Davide gave me for Gemian and Re4son for Kali had different values: Gemian vs Re4son

Its possible that even though the value of 32 was enough with the early Android 8 bootloader we now need a higher value 40 with the release version?

If that fails it would be interesting to know the partition/LVM layout on the planet supplied image for TP3?