[meta-freescale] Issue with freescale git server at git.freescale.com

[georg]

Hello,

I'm new to meta-freescale and trying to build a system for imx6 from
https://github.com/Freescale/fsl-community-bsp-platform. But the
do_fetch step on the imx kernel fails always with a 'Connection reset by
peer'. It seems that the freescale server is somehow broken. I have
tried a simple git clone on git://git.freescale.com/imx/linux-imx.git
with a similar issue. Most of the time it fails with 'Connection reset
by peer'. Sometime it starts with 'Counting objects' for a very long
time before it gets disconnected.

Has anybody experienced the same issue (I have started trying 3 days
ago)? Is there any mirror which I can use instead?

Thanks in advance,
Georg
--

[Otavio Salvador]

Hello,

I added Jun and Lauren on Cc so they can look at the issue.

Post by georg
Hello,
I'm new to meta-freescale and trying to build a system for imx6 from
https://github.com/Freescale/fsl-community-bsp-platform. But the
do_fetch step on the imx kernel fails always with a 'Connection reset by
peer'. It seems that the freescale server is somehow broken. I have
tried a simple git clone on git://git.freescale.com/imx/linux-imx.git
with a similar issue. Most of the time it fails with 'Connection reset
by peer'. Sometime it starts with 'Counting objects' for a very long
time before it gets disconnected.
Has anybody experienced the same issue (I have started trying 3 days
ago)? Is there any mirror which I can use instead?
Thanks in advance,
Georg
--
_______________________________________________
meta-freescale mailing list
https://lists.yoctoproject.org/listinfo/meta-freescale

[georg]

Hello,

I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.

Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

Regards,
Georg

Post by georg
Hello,
I added Jun and Lauren on Cc so they can look at the issue.

Post by georg
Hello,
I'm new to meta-freescale and trying to build a system for imx6 from
https://github.com/Freescale/fsl-community-bsp-platform. But the
do_fetch step on the imx kernel fails always with a 'Connection reset by
peer'. It seems that the freescale server is somehow broken. I have
tried a simple git clone on git://git.freescale.com/imx/linux-imx.git
with a similar issue. Most of the time it fails with 'Connection reset
by peer'. Sometime it starts with 'Counting objects' for a very long
time before it gets disconnected.
Has anybody experienced the same issue (I have started trying 3 days
ago)? Is there any mirror which I can use instead?
Thanks in advance,
Georg
--
_______________________________________________
meta-freescale mailing list
https://lists.yoctoproject.org/listinfo/meta-freescale

--

[Otavio Salvador]

Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

[georg]

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.
--

[Otavio Salvador]

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

[Bas Mevissen]

Post by Otavio Salvador

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

I can confirm git.freescale.com is slow. This might easily lead to
timeouts.

git clone git://git.freescale.com/imx/linux-imx.git
Cloning into 'linux-imx'...
eatremote: Counting objects: 3791253
remote: Counting objects: 5278893, done.
remote: Compressing objects: 100% (839205/839205), done.
Receiving objects: 4% (261343/5278893), 112.78 MiB | 87.00 KiB/s

Connection speed starts around 400KiB/s and drops sometimes to less than
12KiB/s, even on a well-connected host. Checking out stuff from Github
with this Germany based host flies.

I have had a similar experience in the past, but assumed it to be a
temporarily issue. Apparently, it is more structural...

Cheers,

Bas.


--

[georg]

Post by Otavio Salvador

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

Hello,

I have done my previous tests on master (morty doesn't support imx6ull).

So if have tested rocko and change this file:
meta-freescale/conf/machine/imx6ullevk.conf. I have removed the
following lines:
PREFERRED_PROVIDER_u-boot = "u-boot-imx"
PREFERRED_PROVIDER_virtual/bootloader = "u-boot-imx"
PREFERRED_PROVIDER_virtual/kernel = "linux-imx"

It seems that imx-base.inc let chose bitbake u-boot-fslc and linux-fslc-imx.
I have u-boot-fslc-v2017.09+gitAUTOINC+a6a15fedd1-r0 and
linux-fslc-imx-4.1-2.0.x+gitAUTOINC+6c1ad49339-r0 in my buildstats. So
now it's using u-boot-fslc and linux-fslc-imx.

The download is slow too, but without connection errors. After several
hours, the kernel sources are downloaded.

The build is successful, but boot sucks. U-boot boots correctly, but not
linux. I have a reboot without any kernel trace near the end of the

Post by Otavio Salvador
Configuring network interfaces... fec 20b4000.ethernet eth0: Freescale
FEC PHY driver [Micrel KSZ8081 or KSZ8091]
(mii_bus:phy_addr=20b4000.ethernet:01, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
udhcpc (v1.24.1) started
Sending discover...
Sending discover...
fec 20b4000.ethernet eth0: Link is Up - 100Mbps/Full - flow control >rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Sending discover...
U-Boot 2017.09+fslc+ga6a15fe (Oct 17 2017 - 13:56:43 +0200)

I have no idea why. There is no kernel crash trace so it's perhaps
something like the watchdog.

Regards,
Georg


--

[Otavio Salvador]

Hello Georg,

Post by georg

Post by Otavio Salvador

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

Hello,
I have done my previous tests on master (morty doesn't support imx6ull).
meta-freescale/conf/machine/imx6ullevk.conf. I have removed the
PREFERRED_PROVIDER_u-boot = "u-boot-imx"
PREFERRED_PROVIDER_virtual/bootloader = "u-boot-imx"
PREFERRED_PROVIDER_virtual/kernel = "linux-imx"
It seems that imx-base.inc let chose bitbake u-boot-fslc and linux-fslc-imx.
I have u-boot-fslc-v2017.09+gitAUTOINC+a6a15fedd1-r0 and
linux-fslc-imx-4.1-2.0.x+gitAUTOINC+6c1ad49339-r0 in my buildstats. So
now it's using u-boot-fslc and linux-fslc-imx.
The download is slow too, but without connection errors. After several
hours, the kernel sources are downloaded.
The build is successful, but boot sucks. U-boot boots correctly, but not
linux. I have a reboot without any kernel trace near the end of the

Post by Otavio Salvador
Configuring network interfaces... fec 20b4000.ethernet eth0: Freescale
FEC PHY driver [Micrel KSZ8081 or KSZ8091]
(mii_bus:phy_addr=20b4000.ethernet:01, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
udhcpc (v1.24.1) started
Sending discover...
Sending discover...
fec 20b4000.ethernet eth0: Link is Up - 100Mbps/Full - flow control >rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Sending discover...
U-Boot 2017.09+fslc+ga6a15fe (Oct 17 2017 - 13:56:43 +0200)

I have no idea why. There is no kernel crash trace so it's perhaps
something like the watchdog.

Thanks for running the tests.

I did look about differences and you did test against kernel 4.9 it seems.

I'd like to know if you can run same test but against linux-fslc (it
uses 4.13 mainline). Is it possible?

[georg]

Hello Otavio,

Post by Otavio Salvador
Hello Georg,

Post by georg

Post by Otavio Salvador

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

Hello,
I have done my previous tests on master (morty doesn't support imx6ull).
meta-freescale/conf/machine/imx6ullevk.conf. I have removed the
PREFERRED_PROVIDER_u-boot = "u-boot-imx"
PREFERRED_PROVIDER_virtual/bootloader = "u-boot-imx"
PREFERRED_PROVIDER_virtual/kernel = "linux-imx"
It seems that imx-base.inc let chose bitbake u-boot-fslc and linux-fslc-imx.
I have u-boot-fslc-v2017.09+gitAUTOINC+a6a15fedd1-r0 and
linux-fslc-imx-4.1-2.0.x+gitAUTOINC+6c1ad49339-r0 in my buildstats. So
now it's using u-boot-fslc and linux-fslc-imx.
The download is slow too, but without connection errors. After several
hours, the kernel sources are downloaded.
The build is successful, but boot sucks. U-boot boots correctly, but not
linux. I have a reboot without any kernel trace near the end of the

Post by Otavio Salvador
Configuring network interfaces... fec 20b4000.ethernet eth0: Freescale
FEC PHY driver [Micrel KSZ8081 or KSZ8091]
(mii_bus:phy_addr=20b4000.ethernet:01, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
udhcpc (v1.24.1) started
Sending discover...
Sending discover...
fec 20b4000.ethernet eth0: Link is Up - 100Mbps/Full - flow control >rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Sending discover...
U-Boot 2017.09+fslc+ga6a15fe (Oct 17 2017 - 13:56:43 +0200)

I have no idea why. There is no kernel crash trace so it's perhaps
something like the watchdog.

Thanks for running the tests.
I did look about differences and you did test against kernel 4.9 it seems.
I'd like to know if you can run same test but against linux-fslc (it
uses 4.13 mainline). Is it possible?

I have already seen in the release-notes that there are many different
kernel versions. It's not obvious to me which version to use for my
future project. I think mainline kernel is better for latest bugfixes.
Is my listing of kernels from freescale to mainline at the rate of
changes to mainline correct?:
* linux-imx
* linux-fslc-imx
* linux-fslc
* mainline
Thats just to know why this two version exist (linux-fslc-imx and
linux-fslc).

For the test I have added the following lines to machine/imx6ullevk.conf:

PREFERRED_PROVIDER_virtual/kernel = "linux-fslc"
PREFERRED_PROVIDER_kernel = "linux-fslc"

Not sure if the .._virtual... line is needed but it seems to use the
right kernel version. I get my log in:
build/tmp/work/imx6ullevk-poky-linux-gnueabi/linux-fslc/4.13+gitAUTOINC+e12ac43313-r0/temp/log.do_compile.30043

I get the following compilation error:

| make[3]: *** No rule to make target
'arch/arm/boot/dts/imx6ull-14x14-evk-btwifi.dtb'. Stop.
| arch/arm/Makefile:341: recipe for target
'imx6ull-14x14-evk-btwifi.dtb' failed
| make[2]: *** [imx6ull-14x14-evk-btwifi.dtb] Error 2
| Makefile:145: recipe for target 'sub-make' failed
| make[1]: *** [sub-make] Error 2
| Makefile:24: recipe for target '__sub-make' failed
| make: *** [__sub-make] Error 2
| ERROR: oe_runmake failed

It seems that imx6ull-14x14-evk-btwifi.dtb is missing.
--

[Otavio Salvador]

Hello Georg,

Post by georg
Hello Otavio,

Post by Otavio Salvador
Hello Georg,

Post by georg

Post by Otavio Salvador

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

Hello,
I have done my previous tests on master (morty doesn't support imx6ull).
meta-freescale/conf/machine/imx6ullevk.conf. I have removed the
PREFERRED_PROVIDER_u-boot = "u-boot-imx"
PREFERRED_PROVIDER_virtual/bootloader = "u-boot-imx"
PREFERRED_PROVIDER_virtual/kernel = "linux-imx"
It seems that imx-base.inc let chose bitbake u-boot-fslc and linux-fslc-imx.
I have u-boot-fslc-v2017.09+gitAUTOINC+a6a15fedd1-r0 and
linux-fslc-imx-4.1-2.0.x+gitAUTOINC+6c1ad49339-r0 in my buildstats. So
now it's using u-boot-fslc and linux-fslc-imx.
The download is slow too, but without connection errors. After several
hours, the kernel sources are downloaded.
The build is successful, but boot sucks. U-boot boots correctly, but not
linux. I have a reboot without any kernel trace near the end of the

Post by Otavio Salvador
Configuring network interfaces... fec 20b4000.ethernet eth0: Freescale
FEC PHY driver [Micrel KSZ8081 or KSZ8091]
(mii_bus:phy_addr=20b4000.ethernet:01, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
udhcpc (v1.24.1) started
Sending discover...
Sending discover...
fec 20b4000.ethernet eth0: Link is Up - 100Mbps/Full - flow control >rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Sending discover...
U-Boot 2017.09+fslc+ga6a15fe (Oct 17 2017 - 13:56:43 +0200)

I have no idea why. There is no kernel crash trace so it's perhaps
something like the watchdog.

Thanks for running the tests.
I did look about differences and you did test against kernel 4.9 it seems.
I'd like to know if you can run same test but against linux-fslc (it
uses 4.13 mainline). Is it possible?

I have already seen in the release-notes that there are many different
kernel versions. It's not obvious to me which version to use for my
future project. I think mainline kernel is better for latest bugfixes.
Is my listing of kernels from freescale to mainline at the rate of
* linux-imx
* linux-fslc-imx
* linux-fslc
* mainline
Thats just to know why this two version exist (linux-fslc-imx and
linux-fslc).

Basically:

linux-imx: NXP Linux kernel fork
linux-fslc-imx: NXP Linux kernel fork plus FSLC changes (and stable
releases merged)
linux-fslc: Mainline Linux kernel fork with FSLC chages

Post by georg
PREFERRED_PROVIDER_virtual/kernel = "linux-fslc"
PREFERRED_PROVIDER_kernel = "linux-fslc"
Not sure if the .._virtual... line is needed but it seems to use the
build/tmp/work/imx6ullevk-poky-linux-gnueabi/linux-fslc/4.13+gitAUTOINC+e12ac43313-r0/temp/log.do_compile.30043

Right, please remove the dtb from KERNEL_DEVICETREE list so we can
make sure it all works ;-)

Post by georg
| make[3]: *** No rule to make target
'arch/arm/boot/dts/imx6ull-14x14-evk-btwifi.dtb'. Stop.
| arch/arm/Makefile:341: recipe for target
'imx6ull-14x14-evk-btwifi.dtb' failed
| make[2]: *** [imx6ull-14x14-evk-btwifi.dtb] Error 2
| Makefile:145: recipe for target 'sub-make' failed
| make[1]: *** [sub-make] Error 2
| Makefile:24: recipe for target '__sub-make' failed
| make: *** [__sub-make] Error 2
| ERROR: oe_runmake failed
It seems that imx6ull-14x14-evk-btwifi.dtb is missing.
--
_______________________________________________
meta-freescale mailing list
https://lists.yoctoproject.org/listinfo/meta-freescale

[georg]

Post by Otavio Salvador
Hello Georg,

Post by georg
Hello Otavio,

Post by Otavio Salvador
Hello Georg,

Post by georg

Post by Otavio Salvador

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

Hello,
I have done my previous tests on master (morty doesn't support imx6ull).
meta-freescale/conf/machine/imx6ullevk.conf. I have removed the
PREFERRED_PROVIDER_u-boot = "u-boot-imx"
PREFERRED_PROVIDER_virtual/bootloader = "u-boot-imx"
PREFERRED_PROVIDER_virtual/kernel = "linux-imx"
It seems that imx-base.inc let chose bitbake u-boot-fslc and linux-fslc-imx.
I have u-boot-fslc-v2017.09+gitAUTOINC+a6a15fedd1-r0 and
linux-fslc-imx-4.1-2.0.x+gitAUTOINC+6c1ad49339-r0 in my buildstats. So
now it's using u-boot-fslc and linux-fslc-imx.
The download is slow too, but without connection errors. After several
hours, the kernel sources are downloaded.
The build is successful, but boot sucks. U-boot boots correctly, but not
linux. I have a reboot without any kernel trace near the end of the

Post by Otavio Salvador
Configuring network interfaces... fec 20b4000.ethernet eth0: Freescale
FEC PHY driver [Micrel KSZ8081 or KSZ8091]
(mii_bus:phy_addr=20b4000.ethernet:01, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
udhcpc (v1.24.1) started
Sending discover...
Sending discover...
fec 20b4000.ethernet eth0: Link is Up - 100Mbps/Full - flow control >rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Sending discover...
U-Boot 2017.09+fslc+ga6a15fe (Oct 17 2017 - 13:56:43 +0200)

I have no idea why. There is no kernel crash trace so it's perhaps
something like the watchdog.

Thanks for running the tests.
I did look about differences and you did test against kernel 4.9 it seems.
I'd like to know if you can run same test but against linux-fslc (it
uses 4.13 mainline). Is it possible?

I have already seen in the release-notes that there are many different
kernel versions. It's not obvious to me which version to use for my
future project. I think mainline kernel is better for latest bugfixes.
Is my listing of kernels from freescale to mainline at the rate of
* linux-imx
* linux-fslc-imx
* linux-fslc
* mainline
Thats just to know why this two version exist (linux-fslc-imx and
linux-fslc).

linux-imx: NXP Linux kernel fork
linux-fslc-imx: NXP Linux kernel fork plus FSLC changes (and stable
releases merged)
linux-fslc: Mainline Linux kernel fork with FSLC chages

Post by georg
PREFERRED_PROVIDER_virtual/kernel = "linux-fslc"
PREFERRED_PROVIDER_kernel = "linux-fslc"
Not sure if the .._virtual... line is needed but it seems to use the
build/tmp/work/imx6ullevk-poky-linux-gnueabi/linux-fslc/4.13+gitAUTOINC+e12ac43313-r0/temp/log.do_compile.30043

Right, please remove the dtb from KERNEL_DEVICETREE list so we can
make sure it all works ;-)

I didn't notice that this is coming form the machine configuration file.
I have remove it, the build works now.
But the boot don't, I have the same behavior as with the other kernel.
Near the end of the boot process, it restarts without any kernel trace
indicating a problem.

Post by Otavio Salvador

Post by georg
| make[3]: *** No rule to make target
'arch/arm/boot/dts/imx6ull-14x14-evk-btwifi.dtb'. Stop.
| arch/arm/Makefile:341: recipe for target
'imx6ull-14x14-evk-btwifi.dtb' failed
| make[2]: *** [imx6ull-14x14-evk-btwifi.dtb] Error 2
| Makefile:145: recipe for target 'sub-make' failed
| make[1]: *** [sub-make] Error 2
| Makefile:24: recipe for target '__sub-make' failed
| make: *** [__sub-make] Error 2
| ERROR: oe_runmake failed
It seems that imx6ull-14x14-evk-btwifi.dtb is missing.
--
_______________________________________________
meta-freescale mailing list
https://lists.yoctoproject.org/listinfo/meta-freescale

--

[Otavio Salvador]

Post by georg

Post by Otavio Salvador
Hello Georg,

Post by georg
Hello Otavio,

Post by Otavio Salvador
Hello Georg,

Post by georg

Post by Otavio Salvador

Post by georg

Post by georg
Hello,

Post by georg
I have started a script this weekend which tries to do a git clone with
--mirror in a loop every hour till it succeeds. This was working, I have
now a cloned linux-imx.git repository which I can use in all future builds.
Nonetheless it's very annoying to not be able to do a simple "bitbake
core-image-minimal" with a standard freescale evaluation board.

I am curious to know which machine as we use linux-fslc-imx for most
machines as default.

Ah, that's interesting. I thought it's the same repository for all
freescale processors. It's for imx6ull.

If you can, test rocko and change the machine to use u-boot-fslc and
linux-fslc-imx so we confirm it works there. If it does, we can
change.

Hello,
I have done my previous tests on master (morty doesn't support imx6ull).
meta-freescale/conf/machine/imx6ullevk.conf. I have removed the
PREFERRED_PROVIDER_u-boot = "u-boot-imx"
PREFERRED_PROVIDER_virtual/bootloader = "u-boot-imx"
PREFERRED_PROVIDER_virtual/kernel = "linux-imx"
It seems that imx-base.inc let chose bitbake u-boot-fslc and linux-fslc-imx.
I have u-boot-fslc-v2017.09+gitAUTOINC+a6a15fedd1-r0 and
linux-fslc-imx-4.1-2.0.x+gitAUTOINC+6c1ad49339-r0 in my buildstats. So
now it's using u-boot-fslc and linux-fslc-imx.
The download is slow too, but without connection errors. After several
hours, the kernel sources are downloaded.
The build is successful, but boot sucks. U-boot boots correctly, but not
linux. I have a reboot without any kernel trace near the end of the

Post by Otavio Salvador
Configuring network interfaces... fec 20b4000.ethernet eth0: Freescale
FEC PHY driver [Micrel KSZ8081 or KSZ8091]
(mii_bus:phy_addr=20b4000.ethernet:01, irq=-1)
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
udhcpc (v1.24.1) started
Sending discover...
Sending discover...
fec 20b4000.ethernet eth0: Link is Up - 100Mbps/Full - flow control >rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Sending discover...
U-Boot 2017.09+fslc+ga6a15fe (Oct 17 2017 - 13:56:43 +0200)

I have no idea why. There is no kernel crash trace so it's perhaps
something like the watchdog.

Thanks for running the tests.
I did look about differences and you did test against kernel 4.9 it seems.
I'd like to know if you can run same test but against linux-fslc (it
uses 4.13 mainline). Is it possible?

I have already seen in the release-notes that there are many different
kernel versions. It's not obvious to me which version to use for my
future project. I think mainline kernel is better for latest bugfixes.
Is my listing of kernels from freescale to mainline at the rate of
* linux-imx
* linux-fslc-imx
* linux-fslc
* mainline
Thats just to know why this two version exist (linux-fslc-imx and
linux-fslc).

linux-imx: NXP Linux kernel fork
linux-fslc-imx: NXP Linux kernel fork plus FSLC changes (and stable
releases merged)
linux-fslc: Mainline Linux kernel fork with FSLC chages

Post by georg
PREFERRED_PROVIDER_virtual/kernel = "linux-fslc"
PREFERRED_PROVIDER_kernel = "linux-fslc"
Not sure if the .._virtual... line is needed but it seems to use the
build/tmp/work/imx6ullevk-poky-linux-gnueabi/linux-fslc/4.13+gitAUTOINC+e12ac43313-r0/temp/log.do_compile.30043

Right, please remove the dtb from KERNEL_DEVICETREE list so we can
make sure it all works ;-)

I didn't notice that this is coming form the machine configuration file.
I have remove it, the build works now.
But the boot don't, I have the same behavior as with the other kernel.
Near the end of the boot process, it restarts without any kernel trace
indicating a problem.

Fabio, is it possible for you to test ULL using mainline kernel? Georg
is testing using 4.13 (using 2017.09+fslc U-Boot).

[Fabio Estevam]

On Wed, Oct 18, 2017 at 10:27 AM, Otavio Salvador

Post by Otavio Salvador
Fabio, is it possible for you to test ULL using mainline kernel? Georg
is testing using 4.13 (using 2017.09+fslc U-Boot).

Yes, I can reproduce the reboot problem here.

A quick workaround to avoid the reboot:

--- a/arch/arm/boot/dts/imx6ul-14x14-evk.dts
+++ b/arch/arm/boot/dts/imx6ul-14x14-evk.dts
@@ -262,6 +262,7 @@
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_wdog>;
fsl,ext-reset-output;
+ status = "disabled";
};

&iomuxc {

Currently I don't have time for debugging this issue. I do not see it
with mx6ul, only on mx6ull.
--

[Otavio Salvador]

Post by Fabio Estevam
On Wed, Oct 18, 2017 at 10:27 AM, Otavio Salvador

Post by Otavio Salvador
Fabio, is it possible for you to test ULL using mainline kernel? Georg
is testing using 4.13 (using 2017.09+fslc U-Boot).

Yes, I can reproduce the reboot problem here.
--- a/arch/arm/boot/dts/imx6ul-14x14-evk.dts
+++ b/arch/arm/boot/dts/imx6ul-14x14-evk.dts
@@ -262,6 +262,7 @@
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_wdog>;
fsl,ext-reset-output;
+ status = "disabled";
};
&iomuxc {
Currently I don't have time for debugging this issue. I do not see it
with mx6ul, only on mx6ull.

I did look at 4.14 and there are no changes related to it as well. It
seems this is still happening.

[Fabio Estevam]

On Thu, Oct 19, 2017 at 2:17 PM, Otavio Salvador

Post by Otavio Salvador
I did look at 4.14 and there are no changes related to it as well. It
seems this is still happening.

Yes, I understand the issue now and will send a proper patch later
today to the arm-linux-kernel list.
--

[Fabio Estevam]

Post by Fabio Estevam
Yes, I understand the issue now and will send a proper patch later
today to the arm-linux-kernel list.

I was mistaken and this needs more investigation. I won't be able to
debug this soon.
--

[Fabio Estevam]

Post by Fabio Estevam

Post by Fabio Estevam
Yes, I understand the issue now and will send a proper patch later
today to the arm-linux-kernel list.

I was mistaken and this needs more investigation. I won't be able to
debug this soon.

Here is a temporary hack patch that would allow you to use the board
while a proper fix is not available:

https://pastebin.com/pJWqrSiQ
--

[Otavio Salvador]

Hello Fabio,
Hello Georg,

Fabio, thanks for looking at this.

Post by Fabio Estevam

Post by Fabio Estevam

Post by Fabio Estevam
Yes, I understand the issue now and will send a proper patch later
today to the arm-linux-kernel list.

I was mistaken and this needs more investigation. I won't be able to
debug this soon.

Here is a temporary hack patch that would allow you to use the board
https://pastebin.com/pJWqrSiQ

Georg, is it possible for you test the above patch?

[Otavio Salvador]

Hello Carlos,

On Fri, Oct 20, 2017 at 10:51 AM, Bejar-Colonia, Carlos

Is this a general issue for the git.freescale.com.imx.linux-imx.git and git.freescale.com.imx.uboot-imx.git ?
I am trying to build an image for imx7dsabresd and do_fetch() for both repos fails. I tried Morty and Pyro, same result.

It seems so. I am including Lauren and Jun to escalate it internally.

[georg]

Hello,

Post by Otavio Salvador
Hello Fabio,
Hello Georg,
Fabio, thanks for looking at this.

Post by Fabio Estevam

Post by Fabio Estevam

Post by Fabio Estevam
Yes, I understand the issue now and will send a proper patch later
today to the arm-linux-kernel list.

I was mistaken and this needs more investigation. I won't be able to
debug this soon.

Here is a temporary hack patch that would allow you to use the board
https://pastebin.com/pJWqrSiQ

Georg, is it possible for you test the above patch?

Thanks Otavio for looking on this. I have tried to the patch, but it
doesn't work. I get: "error: arch/arm/boot/dts/imx6ull-14x14-evk.dts:
patch does not apply". I'm working with
git://github.com/Freescale/linux-fslc.git.

I have search for the pinctrl_wdog line and replace the register like this:

diff --git a/arch/arm/boot/dts/imx6ull-14x14-evk.dts
b/arch/arm/boot/dts/imx6ull-14x14-evk.dts
index b0fb913..b016657 100644
--- a/arch/arm/boot/dts/imx6ull-14x14-evk.dts
+++ b/arch/arm/boot/dts/imx6ull-14x14-evk.dts
@@ -552,7 +552,7 @@

pinctrl_wdog: wdoggrp {
fsl,pins = <
- MX6UL_PAD_LCD_RESET__WDOG1_WDOG_ANY
0x30b0
+ MX6UL_PAD_LCD_RESET__GPIO3_IO04 0x30b0

Post by Otavio Salvador
;

};
};

[Fabio Estevam]

Hi Georg,

Post by georg
This is working for me, the evk board doesn't reboot anymore.
I haven't understood why, is it soc or the evk board related?

I finally managed to solve this reboot issue on mx6ull.

It was a U-Boot issue. Here are the two patches:

https://lists.denx.de/pipermail/u-boot/2017-November/312318.html

and

https://lists.denx.de/pipermail/u-boot/2017-November/312319.html

Otavio,

Please apply these two patches into u-boot-fslc.

Thanks,

Fabio Estevam
--

[georg]

Hi Fabio,

Thank you very match. I have just tested the patch on my mx6ull evk
board and it works fine.

Regards,
Georg

Post by Fabio Estevam
Hi Georg,

Post by georg
This is working for me, the evk board doesn't reboot anymore.
I haven't understood why, is it soc or the evk board related?

I finally managed to solve this reboot issue on mx6ull.
https://lists.denx.de/pipermail/u-boot/2017-November/312318.html
and
https://lists.denx.de/pipermail/u-boot/2017-November/312319.html
Otavio,
Please apply these two patches into u-boot-fslc.
Thanks,
Fabio Estevam

--