Kernel update to 4.4.15-ti-r35 failure

[Steve Osselton]

Hi,

When updating the TI kernel from 4.4.14-ti-r34 to 4.4.15-ti-r35 on Debian 8.5, the kernel failed to

boot cleanly. On connecting a console, the boot process failed in initramfs when trying to mount the

real boot partition (the eMMC). The mount command from initramfs was complaining about invalid

parameters. Any idea about what is causing this? Edited uEnv.txt to refer to the older kernel and

re-booted ok, so was fixable.

Cheers Steve.

[Robert Nelson]

Hi Steve,

Sigh, yeah it's broken on my bbb too...

with r34, i was chasing why the uuid=<> option was failing to find the microSD card on the x15, i fixed that and pushed as r35..

Then with r35, i noticed the eMMC on x15 had regressed (while the microSD was working), which i fixed in r36, by pulling the patch from 4.5.x to hard code microSD=mmc0, eMMC=mmc1...

and now the bbb is broken.. blah...

So, i'll revert back to r34, but leave the 4.5.x hardcode in place for the x15. ;)

Regards,

--

Robert Nelson
https://rcn-ee.com/

[Jason Kridner]

On Tue, Jul 19, 2016 at 9:37 AM Robert Nelson <robert...@gmail.com> wrote:

Hi Steve,

On Tue, Jul 19, 2016 at 5:00 AM, Steve Osselton <steve.o...@prismtech.com> wrote:

Hi,

When updating the TI kernel from 4.4.14-ti-r34 to 4.4.15-ti-r35 on Debian 8.5, the kernel failed to

boot cleanly. On connecting a console, the boot process failed in initramfs when trying to mount the

real boot partition (the eMMC). The mount command from initramfs was complaining about invalid

parameters. Any idea about what is causing this? Edited uEnv.txt to refer to the older kernel and

re-booted ok, so was fixable.

Sigh, yeah it's broken on my bbb too...

with r34, i was chasing why the uuid=<> option was failing to find the microSD card on the x15, i fixed that and pushed as r35..

Mainline u-boot seems to use a 'blkid' command to set the UUID. Does that work for X15?

 

Then with r35, i noticed the eMMC on x15 had regressed (while the microSD was working), which i fixed in r36, by pulling the patch from 4.5.x to hard code microSD=mmc0, eMMC=mmc1...

and now the bbb is broken.. blah...

FYI, it seems to have also broken WL1835 wifi over SDIO (mmc2).

 

So, i'll revert back to r34, but leave the 4.5.x hardcode in place for the x15. ;)

I'll be looking for the build!

 

Regards,

--

Robert Nelson
https://rcn-ee.com/

--
For more options, visit http://beagleboard.org/discuss
---
You received this message because you are subscribed to the Google Groups "BeagleBoard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beagleboard...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/beagleboard/CAOCHtYi2%2B1GL_dBLySwAAgnd4bnOyot8i4njZe5UKmbqpRqMMQ%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Robert Nelson]

On Tue, Jul 19, 2016 at 9:58 AM, Jason Kridner <jason....@hangerhead.com> wrote:

On Tue, Jul 19, 2016 at 9:37 AM Robert Nelson <robert...@gmail.com> wrote:

Hi Steve,

On Tue, Jul 19, 2016 at 5:00 AM, Steve Osselton <steve.o...@prismtech.com> wrote:

Hi,

When updating the TI kernel from 4.4.14-ti-r34 to 4.4.15-ti-r35 on Debian 8.5, the kernel failed to

boot cleanly. On connecting a console, the boot process failed in initramfs when trying to mount the

real boot partition (the eMMC). The mount command from initramfs was complaining about invalid

parameters. Any idea about what is causing this? Edited uEnv.txt to refer to the older kernel and

re-booted ok, so was fixable.

Sigh, yeah it's broken on my bbb too...

with r34, i was chasing why the uuid=<> option was failing to find the microSD card on the x15, i fixed that and pushed as r35..

Mainline u-boot seems to use a 'blkid' command to set the UUID. Does that work for X15?

Debian Jessie era tools (sfdisk (util-linux 2.25.x) and mkfs.ext4 (e2fsprogs 1.42.12)) don't set the PARTUUID bit unless you force GPT which breaks other things..

e2fsprogrs 1.43, can be backported to jessie, but you still need util-linux 2.27 (which needs a few more things backported)  *( i initially went down this path yesterday, before giving up. ;) )

 

 

Then with r35, i noticed the eMMC on x15 had regressed (while the microSD was working), which i fixed in r36, by pulling the patch from 4.5.x to hard code microSD=mmc0, eMMC=mmc1...

and now the bbb is broken.. blah...

FYI, it seems to have also broken WL1835 wifi over SDIO (mmc2).

r37 is being built right now, so everything should be good again..

 

 

So, i'll revert back to r34, but leave the 4.5.x hardcode in place for the x15. ;)

I'll be looking for the build!

Regards,

[Jason Kridner]

On Tue, Jul 19, 2016 at 10:58 AM Jason Kridner <jason....@hangerhead.com> wrote:

On Tue, Jul 19, 2016 at 9:37 AM Robert Nelson <robert...@gmail.com> wrote:

Hi Steve,

On Tue, Jul 19, 2016 at 5:00 AM, Steve Osselton <steve.o...@prismtech.com> wrote:

Hi,

When updating the TI kernel from 4.4.14-ti-r34 to 4.4.15-ti-r35 on Debian 8.5, the kernel failed to

boot cleanly. On connecting a console, the boot process failed in initramfs when trying to mount the

real boot partition (the eMMC). The mount command from initramfs was complaining about invalid

parameters. Any idea about what is causing this? Edited uEnv.txt to refer to the older kernel and

re-booted ok, so was fixable.

Sigh, yeah it's broken on my bbb too...

with r34, i was chasing why the uuid=<> option was failing to find the microSD card on the x15, i fixed that and pushed as r35..

Mainline u-boot seems to use a 'blkid' command to set the UUID. Does that work for X15?

 

Then with r35, i noticed the eMMC on x15 had regressed (while the microSD was working), which i fixed in r36, by pulling the patch from 4.5.x to hard code microSD=mmc0, eMMC=mmc1...

and now the bbb is broken.. blah...

FYI, it seems to have also broken WL1835 wifi over SDIO (mmc2).

OK, that wasn't the issue. ti-rt-r36 still fails with WL1835.

Useful debug output:

Jul 19 15:12:08 beaglebone kernel: omap_hsmmc 47810000.mmc: card claims to support voltages below defined range

.

.

.

Jul 19 15:12:08 beaglebone kernel: mmc2: error -16 whilst initialising SDIO card

[Jason Kridner]

On Tue, Jul 19, 2016 at 11:18 AM Jason Kridner <jason....@hangerhead.com> wrote:

On Tue, Jul 19, 2016 at 10:58 AM Jason Kridner <jason....@hangerhead.com> wrote:

On Tue, Jul 19, 2016 at 9:37 AM Robert Nelson <robert...@gmail.com> wrote:

Hi Steve,

On Tue, Jul 19, 2016 at 5:00 AM, Steve Osselton <steve.o...@prismtech.com> wrote:

Hi,

When updating the TI kernel from 4.4.14-ti-r34 to 4.4.15-ti-r35 on Debian 8.5, the kernel failed to

boot cleanly. On connecting a console, the boot process failed in initramfs when trying to mount the

real boot partition (the eMMC). The mount command from initramfs was complaining about invalid

parameters. Any idea about what is causing this? Edited uEnv.txt to refer to the older kernel and

re-booted ok, so was fixable.

Sigh, yeah it's broken on my bbb too...

with r34, i was chasing why the uuid=<> option was failing to find the microSD card on the x15, i fixed that and pushed as r35..

Mainline u-boot seems to use a 'blkid' command to set the UUID. Does that work for X15?

 

Then with r35, i noticed the eMMC on x15 had regressed (while the microSD was working), which i fixed in r36, by pulling the patch from 4.5.x to hard code microSD=mmc0, eMMC=mmc1...

and now the bbb is broken.. blah...

FYI, it seems to have also broken WL1835 wifi over SDIO (mmc2).

OK, that wasn't the issue. ti-rt-r36 still fails with WL1835.

Useful debug output:

Jul 19 15:12:08 beaglebone kernel: omap_hsmmc 47810000.mmc: card claims to support voltages below defined range

.

.

.

Jul 19 15:12:08 beaglebone kernel: mmc2: error -16 whilst initialising SDIO card

Is -r37 on the way or do I need to look at this as a potential device tree issue related to the new kernel and start hacking the tree again?

[Robert Nelson]

FYI, it seems to have also broken WL1835 wifi over SDIO (mmc2).

OK, that wasn't the issue. ti-rt-r36 still fails with WL1835.

Useful debug output:

Jul 19 15:12:08 beaglebone kernel: omap_hsmmc 47810000.mmc: card claims to support voltages below defined range

.

.

.

Jul 19 15:12:08 beaglebone kernel: mmc2: error -16 whilst initialising SDIO card

Is -r37 on the way or do I need to look at this as a potential device tree issue related to the new kernel and start hacking the tree again?

Most of the r37 builds are done, still waiting on a few "rt" builds, this will work right now in wheezy/jessie/stretch/sid

sudo apt-get update ; sudo apt-get install linux-image-4.4.15-ti-r37

[Jason Kridner]

That did the trick! WL1835 is working out-of-the-box.

 

Regards,

--

Robert Nelson
https://rcn-ee.com/

--

For more options, visit http://beagleboard.org/discuss
---
You received this message because you are subscribed to the Google Groups "BeagleBoard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beagleboard...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/beagleboard/CAOCHtYjCjPbE%2BT6oMd-rsLmzzL5fvuRVeb7M7rG2s-V%3D2KM7%2Bw%40mail.gmail.com.

[Steve Osselton]

Working for me :)

Thanks.