[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Tom Rini]

Hey all,

It's release day and v2017.09-rc4 is out.  It's a week until release
day.  Please let me know if you know of any regressions, thanks!

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170904/d76dc135/attachment.sig>

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Paul Barker]

On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
> Hey all,
>
> It's release day and v2017.09-rc4 is out.  It's a week until release
> day.  Please let me know if you know of any regressions, thanks!
>
> --
> Tom

I looks like u-boot v2017.09-rc4 is still unable to boot Linux on
Raspberry Pi devices. I've just tested this now on a Raspberry Pi
(original Model B) and confirmed that it gets stuck at "Starting
kernel ...". This was also broken in v2017.07 but it's a regression
against v2017.05 where this was working. For reference see
https://lists.denx.de/pipermail/u-boot/2017-August/300857.html and
replies.

I don't think we have a root cause or a full solution yet, but
re-enabling CONFIG_OF_EMBED in all Raspberry Pi defconfigs should at
least give us a workaround for now that will allow Linux to boot. Is
this something we can get into the v2017.09 release at this stage? If
so I'm happy to send the patch and do the testing.

Cheers,

-- 
Paul Barker
Togán Labs Ltd

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Simon Glass]

Hi Paul,

On 5 September 2017 at 03:55, Paul Barker <pbarker@toganlabs.com> wrote:
> On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
>> Hey all,
>>
>> It's release day and v2017.09-rc4 is out.  It's a week until release
>> day.  Please let me know if you know of any regressions, thanks!
>>
>> --
>> Tom
>
> I looks like u-boot v2017.09-rc4 is still unable to boot Linux on
> Raspberry Pi devices. I've just tested this now on a Raspberry Pi
> (original Model B) and confirmed that it gets stuck at "Starting
> kernel ...". This was also broken in v2017.07 but it's a regression
> against v2017.05 where this was working. For reference see
> https://lists.denx.de/pipermail/u-boot/2017-August/300857.html and
> replies.
>
> I don't think we have a root cause or a full solution yet, but
> re-enabling CONFIG_OF_EMBED in all Raspberry Pi defconfigs should at
> least give us a workaround for now that will allow Linux to boot. Is
> this something we can get into the v2017.09 release at this stage? If
> so I'm happy to send the patch and do the testing.

Yes I think that is better than what we have now. Please go ahead.

I did plan to try to root-cause this, and Stephen Warren gave some
hints, but so far I am in the dark.

One problem I have is that booting raspbian from U-Boot leaves me
without a working keyboard/mouse/ethernet, which seems different to
the problem you have. I am not sure whether this is a device tree
problem or something else.

Regards,
Simon

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Peter Robinson]

On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
> Hey all,
>
> It's release day and v2017.09-rc4 is out.  It's a week until release
> day.  Please let me know if you know of any regressions, thanks!

I'm seeing an issue with a "make defconfig" we use in Fedora for
various utils on x86 which has regressed since RC2 (sorry, currently
traveling so not sure if it was before/after RC3). The error is:

make -f ../scripts/Makefile.build obj=scripts
  GEN     ./Makefile
make -f ../scripts/Makefile.build obj=tools/environ
  gcc -O2 -g -pipe -Wall -Werror=format-security
-Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong
--param=ssp-buffer-size=4 -grecord-gcc-switches
-specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic
-Wp,-MD,scripts/.bin2c.d -Iscripts -Wall -Wstrict-prototypes -O2
-fomit-frame-pointer -o scripts/bin2c ../scripts/bin2c.c
../scripts/Makefile.build:59: ../tools/environ/Makefile: No such file
or directory
make[2]: *** No rule to make target '../tools/environ/Makefile'.  Stop.
make[1]: *** [/builddir/build/BUILD/u-boot-2017.09-rc4/Makefile:1469:
environ] Error 2
make[1]: *** Waiting for unfinished jobs....
if [ -f ../.config -o -d ../include/config ]; then \
echo >&2 "  .. is not clean, please run 'make mrproper'"; \
echo >&2 "  in the '..' directory.";\
/bin/false; \
fi;
make[1]: Leaving directory '/builddir/build/BUILD/u-boot-2017.09-rc4/builds'

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Tom Rini]

On Tue, Sep 05, 2017 at 06:26:59PM +0100, Peter Robinson wrote:
> On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
> > Hey all,
> >
> > It's release day and v2017.09-rc4 is out.  It's a week until release
> > day.  Please let me know if you know of any regressions, thanks!
> 
> I'm seeing an issue with a "make defconfig" we use in Fedora for
> various utils on x86 which has regressed since RC2 (sorry, currently
> traveling so not sure if it was before/after RC3). The error is:
> 
> make -f ../scripts/Makefile.build obj=scripts
>   GEN     ./Makefile
> make -f ../scripts/Makefile.build obj=tools/environ
>   gcc -O2 -g -pipe -Wall -Werror=format-security
> -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong
> --param=ssp-buffer-size=4 -grecord-gcc-switches
> -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic
> -Wp,-MD,scripts/.bin2c.d -Iscripts -Wall -Wstrict-prototypes -O2
> -fomit-frame-pointer -o scripts/bin2c ../scripts/bin2c.c
> ../scripts/Makefile.build:59: ../tools/environ/Makefile: No such file
> or directory
> make[2]: *** No rule to make target '../tools/environ/Makefile'.  Stop.
> make[1]: *** [/builddir/build/BUILD/u-boot-2017.09-rc4/Makefile:1469:
> environ] Error 2
> make[1]: *** Waiting for unfinished jobs....
> if [ -f ../.config -o -d ../include/config ]; then \
> echo >&2 "  .. is not clean, please run 'make mrproper'"; \
> echo >&2 "  in the '..' directory.";\
> /bin/false; \
> fi;
> make[1]: Leaving directory '/builddir/build/BUILD/u-boot-2017.09-rc4/builds'

What's the full make command (or set of comands) ?  It looks like you're
doing mkdir builds ; cd builds ; make -C .. O=`pwd` defconfig and then
something else?  Thanks!

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170905/f41eb90d/attachment.sig>

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Tom Rini]

On Tue, Sep 05, 2017 at 04:55:48PM +0800, Simon Glass wrote:
> Hi Paul,
> 
> On 5 September 2017 at 03:55, Paul Barker <pbarker@toganlabs.com> wrote:
> > On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
> >> Hey all,
> >>
> >> It's release day and v2017.09-rc4 is out.  It's a week until release
> >> day.  Please let me know if you know of any regressions, thanks!
> >>
> >> --
> >> Tom
> >
> > I looks like u-boot v2017.09-rc4 is still unable to boot Linux on
> > Raspberry Pi devices. I've just tested this now on a Raspberry Pi
> > (original Model B) and confirmed that it gets stuck at "Starting
> > kernel ...". This was also broken in v2017.07 but it's a regression
> > against v2017.05 where this was working. For reference see
> > https://lists.denx.de/pipermail/u-boot/2017-August/300857.html and
> > replies.
> >
> > I don't think we have a root cause or a full solution yet, but
> > re-enabling CONFIG_OF_EMBED in all Raspberry Pi defconfigs should at
> > least give us a workaround for now that will allow Linux to boot. Is
> > this something we can get into the v2017.09 release at this stage? If
> > so I'm happy to send the patch and do the testing.
> 
> Yes I think that is better than what we have now. Please go ahead.
> 
> I did plan to try to root-cause this, and Stephen Warren gave some
> hints, but so far I am in the dark.
> 
> One problem I have is that booting raspbian from U-Boot leaves me
> without a working keyboard/mouse/ethernet, which seems different to
> the problem you have. I am not sure whether this is a device tree
> problem or something else.

Yes, please, lets get at least a work-around for now.  Thanks!

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170905/00296bd2/attachment.sig>

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Peter Robinson]

On Tue, Sep 5, 2017 at 6:42 PM, Tom Rini <trini@konsulko.com> wrote:
> On Tue, Sep 05, 2017 at 06:26:59PM +0100, Peter Robinson wrote:
>> On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
>> > Hey all,
>> >
>> > It's release day and v2017.09-rc4 is out.  It's a week until release
>> > day.  Please let me know if you know of any regressions, thanks!
>>
>> I'm seeing an issue with a "make defconfig" we use in Fedora for
>> various utils on x86 which has regressed since RC2 (sorry, currently
>> traveling so not sure if it was before/after RC3). The error is:
>>
>> make -f ../scripts/Makefile.build obj=scripts
>>   GEN     ./Makefile
>> make -f ../scripts/Makefile.build obj=tools/environ
>>   gcc -O2 -g -pipe -Wall -Werror=format-security
>> -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong
>> --param=ssp-buffer-size=4 -grecord-gcc-switches
>> -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic
>> -Wp,-MD,scripts/.bin2c.d -Iscripts -Wall -Wstrict-prototypes -O2
>> -fomit-frame-pointer -o scripts/bin2c ../scripts/bin2c.c
>> ../scripts/Makefile.build:59: ../tools/environ/Makefile: No such file
>> or directory
>> make[2]: *** No rule to make target '../tools/environ/Makefile'.  Stop.
>> make[1]: *** [/builddir/build/BUILD/u-boot-2017.09-rc4/Makefile:1469:
>> environ] Error 2
>> make[1]: *** Waiting for unfinished jobs....
>> if [ -f ../.config -o -d ../include/config ]; then \
>> echo >&2 "  .. is not clean, please run 'make mrproper'"; \
>> echo >&2 "  in the '..' directory.";\
>> /bin/false; \
>> fi;
>> make[1]: Leaving directory '/builddir/build/BUILD/u-boot-2017.09-rc4/builds'
>
> What's the full make command (or set of comands) ?  It looks like you're
> doing mkdir builds ; cd builds ; make -C .. O=`pwd` defconfig and then
> something else?  Thanks!

Full build log on x86_64 is
https://kojipkgs.fedoraproject.org//work/tasks/9100/21659100/build.log

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Tom Rini]

On Tue, Sep 05, 2017 at 06:47:20PM +0100, Peter Robinson wrote:
> On Tue, Sep 5, 2017 at 6:42 PM, Tom Rini <trini@konsulko.com> wrote:
> > On Tue, Sep 05, 2017 at 06:26:59PM +0100, Peter Robinson wrote:
> >> On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
> >> > Hey all,
> >> >
> >> > It's release day and v2017.09-rc4 is out.  It's a week until release
> >> > day.  Please let me know if you know of any regressions, thanks!
> >>
> >> I'm seeing an issue with a "make defconfig" we use in Fedora for
> >> various utils on x86 which has regressed since RC2 (sorry, currently
> >> traveling so not sure if it was before/after RC3). The error is:
> >>
> >> make -f ../scripts/Makefile.build obj=scripts
> >>   GEN     ./Makefile
> >> make -f ../scripts/Makefile.build obj=tools/environ
> >>   gcc -O2 -g -pipe -Wall -Werror=format-security
> >> -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong
> >> --param=ssp-buffer-size=4 -grecord-gcc-switches
> >> -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic
> >> -Wp,-MD,scripts/.bin2c.d -Iscripts -Wall -Wstrict-prototypes -O2
> >> -fomit-frame-pointer -o scripts/bin2c ../scripts/bin2c.c
> >> ../scripts/Makefile.build:59: ../tools/environ/Makefile: No such file
> >> or directory
> >> make[2]: *** No rule to make target '../tools/environ/Makefile'.  Stop.
> >> make[1]: *** [/builddir/build/BUILD/u-boot-2017.09-rc4/Makefile:1469:
> >> environ] Error 2
> >> make[1]: *** Waiting for unfinished jobs....
> >> if [ -f ../.config -o -d ../include/config ]; then \
> >> echo >&2 "  .. is not clean, please run 'make mrproper'"; \
> >> echo >&2 "  in the '..' directory.";\
> >> /bin/false; \
> >> fi;
> >> make[1]: Leaving directory '/builddir/build/BUILD/u-boot-2017.09-rc4/builds'
> >
> > What's the full make command (or set of comands) ?  It looks like you're
> > doing mkdir builds ; cd builds ; make -C .. O=`pwd` defconfig and then
> > something else?  Thanks!
> 
> Full build log on x86_64 is
> https://kojipkgs.fedoraproject.org//work/tasks/9100/21659100/build.log

OK, so tools-only is broken.  I see it now, thanks.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170905/437867f1/attachment.sig>

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Peter Robinson]

On Tue, Sep 5, 2017 at 6:55 PM, Tom Rini <trini@konsulko.com> wrote:
> On Tue, Sep 05, 2017 at 06:47:20PM +0100, Peter Robinson wrote:
>> On Tue, Sep 5, 2017 at 6:42 PM, Tom Rini <trini@konsulko.com> wrote:
>> > On Tue, Sep 05, 2017 at 06:26:59PM +0100, Peter Robinson wrote:
>> >> On Mon, Sep 4, 2017 at 2:05 PM, Tom Rini <trini@konsulko.com> wrote:
>> >> > Hey all,
>> >> >
>> >> > It's release day and v2017.09-rc4 is out.  It's a week until release
>> >> > day.  Please let me know if you know of any regressions, thanks!
>> >>
>> >> I'm seeing an issue with a "make defconfig" we use in Fedora for
>> >> various utils on x86 which has regressed since RC2 (sorry, currently
>> >> traveling so not sure if it was before/after RC3). The error is:
>> >>
>> >> make -f ../scripts/Makefile.build obj=scripts
>> >>   GEN     ./Makefile
>> >> make -f ../scripts/Makefile.build obj=tools/environ
>> >>   gcc -O2 -g -pipe -Wall -Werror=format-security
>> >> -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong
>> >> --param=ssp-buffer-size=4 -grecord-gcc-switches
>> >> -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic
>> >> -Wp,-MD,scripts/.bin2c.d -Iscripts -Wall -Wstrict-prototypes -O2
>> >> -fomit-frame-pointer -o scripts/bin2c ../scripts/bin2c.c
>> >> ../scripts/Makefile.build:59: ../tools/environ/Makefile: No such file
>> >> or directory
>> >> make[2]: *** No rule to make target '../tools/environ/Makefile'.  Stop.
>> >> make[1]: *** [/builddir/build/BUILD/u-boot-2017.09-rc4/Makefile:1469:
>> >> environ] Error 2
>> >> make[1]: *** Waiting for unfinished jobs....
>> >> if [ -f ../.config -o -d ../include/config ]; then \
>> >> echo >&2 "  .. is not clean, please run 'make mrproper'"; \
>> >> echo >&2 "  in the '..' directory.";\
>> >> /bin/false; \
>> >> fi;
>> >> make[1]: Leaving directory '/builddir/build/BUILD/u-boot-2017.09-rc4/builds'
>> >
>> > What's the full make command (or set of comands) ?  It looks like you're
>> > doing mkdir builds ; cd builds ; make -C .. O=`pwd` defconfig and then
>> > something else?  Thanks!
>>
>> Full build log on x86_64 is
>> https://kojipkgs.fedoraproject.org//work/tasks/9100/21659100/build.log
>
> OK, so tools-only is broken.  I see it now, thanks.

Correct, sorry, that's what I meant when I said "various utils" but
clearly my brain is operating differently to my fingers today :-D

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Tuomas Tynkkynen]

Hi,

On 09/04/2017 10:55 PM, Paul Barker wrote:
> 
> I looks like u-boot v2017.09-rc4 is still unable to boot Linux on
> Raspberry Pi devices. I've just tested this now on a Raspberry Pi
> (original Model B) and confirmed that it gets stuck at "Starting
> kernel ...".

FWIW, works for me. I think you need to be more specific since there are 
huge number of variables related to running Linux on the RPi. My 
configuration is:

- Device is Raspberry Pi Model B (512M RAM)
- Firmware is tagged 1.20170811
- U-Boot v2017.09-rc4
- The Foundation/Raspbian kernel tagged 1.20170811, based on 4.9.41. The 
kernel is self-built. Or well, built by the package manager in NixOS but 
I am the maintainer of the RPi kernel package there.
- As for the device tree, I am letting U-Boot load bcm2708-rpi-b.dtb 
built by the Foundation/Raspbian kernel from disk (i.e. firmware is not
involved with the device tree at all)
- The framebuffer, USB, keyboard & Ethernet work in Linux. I have never 
tested the other peripherals (or GPU acceleration).

My setup has been unchanged (except for routine package upgrades of the 
firmware, kernel & U-Boot obviously) since July 2016. That's when the 
Foundation/Raspbian kernel started requiring a device tree.

[U-Boot] [ANN] U-Boot v2017.09-rc4 released

[Paul Barker]

On Tue, Sep 5, 2017 at 11:11 PM, Tuomas Tynkkynen
<tuomas.tynkkynen@iki.fi> wrote:
> Hi,
>
> On 09/04/2017 10:55 PM, Paul Barker wrote:
>>
>>
>> I looks like u-boot v2017.09-rc4 is still unable to boot Linux on
>> Raspberry Pi devices. I've just tested this now on a Raspberry Pi
>> (original Model B) and confirmed that it gets stuck at "Starting
>> kernel ...".
>
>
> FWIW, works for me. I think you need to be more specific since there are
> huge number of variables related to running Linux on the RPi. My
> configuration is:
>
> - Device is Raspberry Pi Model B (512M RAM)
> - Firmware is tagged 1.20170811
> - U-Boot v2017.09-rc4
> - The Foundation/Raspbian kernel tagged 1.20170811, based on 4.9.41. The
> kernel is self-built. Or well, built by the package manager in NixOS but I
> am the maintainer of the RPi kernel package there.
> - As for the device tree, I am letting U-Boot load bcm2708-rpi-b.dtb built
> by the Foundation/Raspbian kernel from disk (i.e. firmware is not
> involved with the device tree at all)

I think this is the difference. We're passing $fdt_addr through to
Linux. This is set to the saved fw_dtb_pointer which is the address
the firmware passed into u-boot at start up by the code in
board/raspberrypi/rpi/.

> - The framebuffer, USB, keyboard & Ethernet work in Linux. I have never
> tested the other peripherals (or GPU acceleration).
>
> My setup has been unchanged (except for routine package upgrades of the
> firmware, kernel & U-Boot obviously) since July 2016. That's when the
> Foundation/Raspbian kernel started requiring a device tree.

Thanks,

-- 
Paul Barker
Togán Labs Ltd