[U-Boot] Cyclone 5 U-boot 2015.10

[U-Boot] Cyclone 5 U-boot 2015.10

[Simon Brown]

Hi,
I've been in discussion with Marek on this issue, and he suggested I post
to the u-boot mailing list.

I've git cloned uboot 2015.10 and checked it out in my own branch.  I
wasn't about to submit anything back, but this way I hopefully wouldn't
break anything if I did!

I've created a new board, "Axim".  I've setup the DTS, Kconfig, defconfig
files etc similar to the Atlas board addition, and things are compiling
ok.  However, when I run U-boot it gets to a relatively early point and
then says "### ERROR ### Please reset the board ###".  I can't see it say
this ,as it no longer talks to the UART, but I can see that it points to
the memory address containing this string in the debugger, so assume that's
what it prints out, before entering an endless loop.

My setup is slightly odd, in that I need to get U-boot running before I can
configure an SD card.  This means I'm programming my FPGA via JTAG, then
loading the preloader with the ARM DS-5 debugger via JTAG and running it.
I get output on the UART (via FPGA pins), to my PC, to suggest that all is
running well.  I can then upload my u-boot.img file to the preloader using
Ymodem (preloader configured to accept this), and I can run U-boot.

Before I was using the preconfigured 2013.01.01 U-boot as provided on
Rocketboards.org, but since we are using the KSZ9031 chip for our PHY felt
I needed a newer version to support this.  But 2013.01.01 runs ok, and I
can get to the U-boot prompt on the UART, whereas 2015 hangs.

Any clues gratefully received.

Simon

[U-Boot] Cyclone 5 U-boot 2015.10

[Marek Vasut]

On Tuesday, November 17, 2015 at 02:08:21 PM, Simon Brown wrote:
> Hi,

Hi,

> I've been in discussion with Marek on this issue, and he suggested I post
> to the u-boot mailing list.
> 
> I've git cloned uboot 2015.10 and checked it out in my own branch.  I
> wasn't about to submit anything back, but this way I hopefully wouldn't
> break anything if I did!
> 
> I've created a new board, "Axim".  I've setup the DTS, Kconfig, defconfig
> files etc similar to the Atlas board addition, and things are compiling
> ok.  However, when I run U-boot it gets to a relatively early point and
> then says "### ERROR ### Please reset the board ###".  I can't see it say
> this ,as it no longer talks to the UART, but I can see that it points to
> the memory address containing this string in the debugger, so assume that's
> what it prints out, before entering an endless loop.
> 
> My setup is slightly odd, in that I need to get U-boot running before I can
> configure an SD card.

Why ?

> This means I'm programming my FPGA via JTAG, then
> loading the preloader with the ARM DS-5 debugger via JTAG and running it.

OK, so what I suggest you do (to make things easy) is:

1) edit arch/arm/mach-socfpga/spl.c
2) locate board_init_f()
3) Add the following code all the way to the end of board_init_f() (which is,
   incidentally, at the end of the file):
   asm volatile("x: b x");

This will make sure that the SPL starts spinning in an endless loop at this 
point. Once you load this modified SPL with your debugger, it will init the
hardware, but it won't do anything funny anymore.

So once your hardware is running, load U-Boot (u-boot-dtb.bin) to 0x01000040
and execute it from that address. This should get you the prompt.

> I get output on the UART (via FPGA pins), to my PC, to suggest that all is
> running well.  I can then upload my u-boot.img file to the preloader using
> Ymodem (preloader configured to accept this), and I can run U-boot.
> 
> Before I was using the preconfigured 2013.01.01 U-boot as provided on
> Rocketboards.org, but since we are using the KSZ9031 chip for our PHY felt
> I needed a newer version to support this.  But 2013.01.01 runs ok, and I
> can get to the U-boot prompt on the UART, whereas 2015 hangs.
> 
> Any clues gratefully received.

Try the above method, hope that helps.

Best regards,
Marek Vasut

[U-Boot] Cyclone 5 U-boot 2015.10

[Simon Brown]

Ok, that seems to have helped - I think one problem I had was using
u-boot.img, not u-boot-dtb.img - what is the difference?

It now prints the U-boot startup prompt, followed by some information about
setup, up to DRAM: 1 GiB
then it says "Error binding driver 'gpio-dwapb'" three times, followed by
some other error messages:
>Some drivers failed to bind
>Error binding driver 'generic_simple_bus'
>Some drivers failed to bind
finishing with "initcall sequence 3ffb1ce4 failed at call 010031bf(err=-1)".

As regards not configuring SD card until after we boot - we made the
decision to use an EMMC chip rather than an SD card, but no means of
programming it without Linux!

Anyway, will try and workout what the above errors signify.
Kind regards,

Simon



On 17 November 2015 at 14:18, Marek Vasut <marex@denx.de> wrote:

> On Tuesday, November 17, 2015 at 02:08:21 PM, Simon Brown wrote:
> > Hi,
>
> Hi,
>
> > I've been in discussion with Marek on this issue, and he suggested I post
> > to the u-boot mailing list.
> >
> > I've git cloned uboot 2015.10 and checked it out in my own branch.  I
> > wasn't about to submit anything back, but this way I hopefully wouldn't
> > break anything if I did!
> >
> > I've created a new board, "Axim".  I've setup the DTS, Kconfig, defconfig
> > files etc similar to the Atlas board addition, and things are compiling
> > ok.  However, when I run U-boot it gets to a relatively early point and
> > then says "### ERROR ### Please reset the board ###".  I can't see it say
> > this ,as it no longer talks to the UART, but I can see that it points to
> > the memory address containing this string in the debugger, so assume
> that's
> > what it prints out, before entering an endless loop.
> >
> > My setup is slightly odd, in that I need to get U-boot running before I
> can
> > configure an SD card.
>
> Why ?
>
> > This means I'm programming my FPGA via JTAG, then
> > loading the preloader with the ARM DS-5 debugger via JTAG and running it.
>
> OK, so what I suggest you do (to make things easy) is:
>
> 1) edit arch/arm/mach-socfpga/spl.c
> 2) locate board_init_f()
> 3) Add the following code all the way to the end of board_init_f() (which
> is,
>    incidentally, at the end of the file):
>    asm volatile("x: b x");
>
> This will make sure that the SPL starts spinning in an endless loop at this
> point. Once you load this modified SPL with your debugger, it will init the
> hardware, but it won't do anything funny anymore.
>
> So once your hardware is running, load U-Boot (u-boot-dtb.bin) to
> 0x01000040
> and execute it from that address. This should get you the prompt.
>
> > I get output on the UART (via FPGA pins), to my PC, to suggest that all
> is
> > running well.  I can then upload my u-boot.img file to the preloader
> using
> > Ymodem (preloader configured to accept this), and I can run U-boot.
> >
> > Before I was using the preconfigured 2013.01.01 U-boot as provided on
> > Rocketboards.org, but since we are using the KSZ9031 chip for our PHY
> felt
> > I needed a newer version to support this.  But 2013.01.01 runs ok, and I
> > can get to the U-boot prompt on the UART, whereas 2015 hangs.
> >
> > Any clues gratefully received.
>
> Try the above method, hope that helps.
>
> Best regards,
> Marek Vasut
>

[U-Boot] Cyclone 5 U-boot 2015.10

[Marek Vasut]

On Tuesday, November 17, 2015 at 04:03:03 PM, Simon Brown wrote:
> Ok, that seems to have helped - I think one problem I had was using
> u-boot.img, not u-boot-dtb.img - what is the difference?

The later has the device tree attached to the end of the binary,
the former does not.

> It now prints the U-boot startup prompt, followed by some information about
> setup, up to DRAM: 1 GiB
> then it says "Error binding driver 'gpio-dwapb'" three times, followed by
> 
> some other error messages:
> >Some drivers failed to bind
> >Error binding driver 'generic_simple_bus'
> >Some drivers failed to bind
> 
> finishing with "initcall sequence 3ffb1ce4 failed at call
> 010031bf(err=-1)".

Is this stock 2015.10 or u-boot/master ?

> As regards not configuring SD card until after we boot - we made the
> decision to use an EMMC chip rather than an SD card, but no means of
> programming it without Linux!

Oh, we do the same on MCVEVK. At one point, I just cooked stripped down
version of U-Boot with USB mass storage support and placed that into
RAM cells in the FPGA. Then I made the SoCFPGA boot from FPGA.

The MCVEVK behaves as a USB stick and the eMMC is the backing store.

> Anyway, will try and workout what the above errors signify.
> Kind regards,
> 
> Simon
> 
> On 17 November 2015 at 14:18, Marek Vasut <marex@denx.de> wrote:
> > On Tuesday, November 17, 2015 at 02:08:21 PM, Simon Brown wrote:
> > > Hi,
> > 
> > Hi,
> > 
> > > I've been in discussion with Marek on this issue, and he suggested I
> > > post to the u-boot mailing list.
> > > 
> > > I've git cloned uboot 2015.10 and checked it out in my own branch.  I
> > > wasn't about to submit anything back, but this way I hopefully wouldn't
> > > break anything if I did!
> > > 
> > > I've created a new board, "Axim".  I've setup the DTS, Kconfig,
> > > defconfig files etc similar to the Atlas board addition, and things
> > > are compiling ok.  However, when I run U-boot it gets to a relatively
> > > early point and then says "### ERROR ### Please reset the board ###". 
> > > I can't see it say this ,as it no longer talks to the UART, but I can
> > > see that it points to the memory address containing this string in the
> > > debugger, so assume
> > 
> > that's
> > 
> > > what it prints out, before entering an endless loop.
> > > 
> > > My setup is slightly odd, in that I need to get U-boot running before I
> > 
> > can
> > 
> > > configure an SD card.
> > 
> > Why ?
> > 
> > > This means I'm programming my FPGA via JTAG, then
> > > loading the preloader with the ARM DS-5 debugger via JTAG and running
> > > it.
> > 
> > OK, so what I suggest you do (to make things easy) is:
> > 
> > 1) edit arch/arm/mach-socfpga/spl.c
> > 2) locate board_init_f()
> > 3) Add the following code all the way to the end of board_init_f() (which
> > is,
> > 
> >    incidentally, at the end of the file):
> >    asm volatile("x: b x");
> > 
> > This will make sure that the SPL starts spinning in an endless loop at
> > this point. Once you load this modified SPL with your debugger, it will
> > init the hardware, but it won't do anything funny anymore.
> > 
> > So once your hardware is running, load U-Boot (u-boot-dtb.bin) to
> > 0x01000040
> > and execute it from that address. This should get you the prompt.
> > 
> > > I get output on the UART (via FPGA pins), to my PC, to suggest that all
> > 
> > is
> > 
> > > running well.  I can then upload my u-boot.img file to the preloader
> > 
> > using
> > 
> > > Ymodem (preloader configured to accept this), and I can run U-boot.
> > > 
> > > Before I was using the preconfigured 2013.01.01 U-boot as provided on
> > > Rocketboards.org, but since we are using the KSZ9031 chip for our PHY
> > 
> > felt
> > 
> > > I needed a newer version to support this.  But 2013.01.01 runs ok, and
> > > I can get to the U-boot prompt on the UART, whereas 2015 hangs.
> > > 
> > > Any clues gratefully received.
> > 
> > Try the above method, hope that helps.
> > 
> > Best regards,
> > Marek Vasut

[U-Boot] Cyclone 5 U-boot 2015.10

[Simon Brown]

I thought I checked out "v2015.10", rather than a rc.  I assumed that was
the proper version.

The startup message says:
U-Boot 2015.10-dirty (Nov 17 2015 - 14:31:48 +0000)

I don't have USB either - only input means before ethernet is serial port.
Not very clever, but at least we have ONE means of getting comms going.

Regards,
Simon

On 17 November 2015 at 15:15, Marek Vasut <marex@denx.de> wrote:

> On Tuesday, November 17, 2015 at 04:03:03 PM, Simon Brown wrote:
> > Ok, that seems to have helped - I think one problem I had was using
> > u-boot.img, not u-boot-dtb.img - what is the difference?
>
> The later has the device tree attached to the end of the binary,
> the former does not.
>
> > It now prints the U-boot startup prompt, followed by some information
> about
> > setup, up to DRAM: 1 GiB
> > then it says "Error binding driver 'gpio-dwapb'" three times, followed by
> >
> > some other error messages:
> > >Some drivers failed to bind
> > >Error binding driver 'generic_simple_bus'
> > >Some drivers failed to bind
> >
> > finishing with "initcall sequence 3ffb1ce4 failed at call
> > 010031bf(err=-1)".
>
> Is this stock 2015.10 or u-boot/master ?
>
> > As regards not configuring SD card until after we boot - we made the
> > decision to use an EMMC chip rather than an SD card, but no means of
> > programming it without Linux!
>
> Oh, we do the same on MCVEVK. At one point, I just cooked stripped down
> version of U-Boot with USB mass storage support and placed that into
> RAM cells in the FPGA. Then I made the SoCFPGA boot from FPGA.
>
> The MCVEVK behaves as a USB stick and the eMMC is the backing store.
>
> > Anyway, will try and workout what the above errors signify.
> > Kind regards,
> >
> > Simon
> >
> > On 17 November 2015 at 14:18, Marek Vasut <marex@denx.de> wrote:
> > > On Tuesday, November 17, 2015 at 02:08:21 PM, Simon Brown wrote:
> > > > Hi,
> > >
> > > Hi,
> > >
> > > > I've been in discussion with Marek on this issue, and he suggested I
> > > > post to the u-boot mailing list.
> > > >
> > > > I've git cloned uboot 2015.10 and checked it out in my own branch.  I
> > > > wasn't about to submit anything back, but this way I hopefully
> wouldn't
> > > > break anything if I did!
> > > >
> > > > I've created a new board, "Axim".  I've setup the DTS, Kconfig,
> > > > defconfig files etc similar to the Atlas board addition, and things
> > > > are compiling ok.  However, when I run U-boot it gets to a relatively
> > > > early point and then says "### ERROR ### Please reset the board ###".
> > > > I can't see it say this ,as it no longer talks to the UART, but I can
> > > > see that it points to the memory address containing this string in
> the
> > > > debugger, so assume
> > >
> > > that's
> > >
> > > > what it prints out, before entering an endless loop.
> > > >
> > > > My setup is slightly odd, in that I need to get U-boot running
> before I
> > >
> > > can
> > >
> > > > configure an SD card.
> > >
> > > Why ?
> > >
> > > > This means I'm programming my FPGA via JTAG, then
> > > > loading the preloader with the ARM DS-5 debugger via JTAG and running
> > > > it.
> > >
> > > OK, so what I suggest you do (to make things easy) is:
> > >
> > > 1) edit arch/arm/mach-socfpga/spl.c
> > > 2) locate board_init_f()
> > > 3) Add the following code all the way to the end of board_init_f()
> (which
> > > is,
> > >
> > >    incidentally, at the end of the file):
> > >    asm volatile("x: b x");
> > >
> > > This will make sure that the SPL starts spinning in an endless loop at
> > > this point. Once you load this modified SPL with your debugger, it will
> > > init the hardware, but it won't do anything funny anymore.
> > >
> > > So once your hardware is running, load U-Boot (u-boot-dtb.bin) to
> > > 0x01000040
> > > and execute it from that address. This should get you the prompt.
> > >
> > > > I get output on the UART (via FPGA pins), to my PC, to suggest that
> all
> > >
> > > is
> > >
> > > > running well.  I can then upload my u-boot.img file to the preloader
> > >
> > > using
> > >
> > > > Ymodem (preloader configured to accept this), and I can run U-boot.
> > > >
> > > > Before I was using the preconfigured 2013.01.01 U-boot as provided on
> > > > Rocketboards.org, but since we are using the KSZ9031 chip for our PHY
> > >
> > > felt
> > >
> > > > I needed a newer version to support this.  But 2013.01.01 runs ok,
> and
> > > > I can get to the U-boot prompt on the UART, whereas 2015 hangs.
> > > >
> > > > Any clues gratefully received.
> > >
> > > Try the above method, hope that helps.
> > >
> > > Best regards,
> > > Marek Vasut
>

[U-Boot] Cyclone 5 U-boot 2015.10

[Marek Vasut]

On Tuesday, November 17, 2015 at 04:20:16 PM, Simon Brown wrote:

Hi,

> I thought I checked out "v2015.10", rather than a rc.  I assumed that was
> the proper version.
> 
> The startup message says:
> U-Boot 2015.10-dirty (Nov 17 2015 - 14:31:48 +0000)

Try git describe HEAD ;-)

> I don't have USB either - only input means before ethernet is serial port.
> Not very clever, but at least we have ONE means of getting comms going.

What about ethernet ? Hell, you have an FPGA, you can route the eMMC pins on
some edge connector and plug that thing into a card reader :b

> Regards,
> Simon
> 
> On 17 November 2015 at 15:15, Marek Vasut <marex@denx.de> wrote:
> > On Tuesday, November 17, 2015 at 04:03:03 PM, Simon Brown wrote:
> > > Ok, that seems to have helped - I think one problem I had was using
> > > u-boot.img, not u-boot-dtb.img - what is the difference?
> > 
> > The later has the device tree attached to the end of the binary,
> > the former does not.
> > 
> > > It now prints the U-boot startup prompt, followed by some information
> > 
> > about
> > 
> > > setup, up to DRAM: 1 GiB
> > > then it says "Error binding driver 'gpio-dwapb'" three times, followed
> > > by
> > > 
> > > some other error messages:
> > > >Some drivers failed to bind
> > > >Error binding driver 'generic_simple_bus'
> > > >Some drivers failed to bind
> > > 
> > > finishing with "initcall sequence 3ffb1ce4 failed at call
> > > 010031bf(err=-1)".
> > 
> > Is this stock 2015.10 or u-boot/master ?
> > 
> > > As regards not configuring SD card until after we boot - we made the
> > > decision to use an EMMC chip rather than an SD card, but no means of
> > > programming it without Linux!
> > 
> > Oh, we do the same on MCVEVK. At one point, I just cooked stripped down
> > version of U-Boot with USB mass storage support and placed that into
> > RAM cells in the FPGA. Then I made the SoCFPGA boot from FPGA.
> > 
> > The MCVEVK behaves as a USB stick and the eMMC is the backing store.
> > 
> > > Anyway, will try and workout what the above errors signify.
> > > Kind regards,
> > > 
> > > Simon
> > > 
> > > On 17 November 2015 at 14:18, Marek Vasut <marex@denx.de> wrote:
> > > > On Tuesday, November 17, 2015 at 02:08:21 PM, Simon Brown wrote:
> > > > > Hi,
> > > > 
> > > > Hi,
> > > > 
> > > > > I've been in discussion with Marek on this issue, and he suggested
> > > > > I post to the u-boot mailing list.
> > > > > 
> > > > > I've git cloned uboot 2015.10 and checked it out in my own branch. 
> > > > > I wasn't about to submit anything back, but this way I hopefully
> > 
> > wouldn't
> > 
> > > > > break anything if I did!
> > > > > 
> > > > > I've created a new board, "Axim".  I've setup the DTS, Kconfig,
> > > > > defconfig files etc similar to the Atlas board addition, and things
> > > > > are compiling ok.  However, when I run U-boot it gets to a
> > > > > relatively early point and then says "### ERROR ### Please reset
> > > > > the board ###". I can't see it say this ,as it no longer talks to
> > > > > the UART, but I can see that it points to the memory address
> > > > > containing this string in
> > 
> > the
> > 
> > > > > debugger, so assume
> > > > 
> > > > that's
> > > > 
> > > > > what it prints out, before entering an endless loop.
> > > > > 
> > > > > My setup is slightly odd, in that I need to get U-boot running
> > 
> > before I
> > 
> > > > can
> > > > 
> > > > > configure an SD card.
> > > > 
> > > > Why ?
> > > > 
> > > > > This means I'm programming my FPGA via JTAG, then
> > > > > loading the preloader with the ARM DS-5 debugger via JTAG and
> > > > > running it.
> > > > 
> > > > OK, so what I suggest you do (to make things easy) is:
> > > > 
> > > > 1) edit arch/arm/mach-socfpga/spl.c
> > > > 2) locate board_init_f()
> > > > 3) Add the following code all the way to the end of board_init_f()
> > 
> > (which
> > 
> > > > is,
> > > > 
> > > >    incidentally, at the end of the file):
> > > >    asm volatile("x: b x");
> > > > 
> > > > This will make sure that the SPL starts spinning in an endless loop
> > > > at this point. Once you load this modified SPL with your debugger,
> > > > it will init the hardware, but it won't do anything funny anymore.
> > > > 
> > > > So once your hardware is running, load U-Boot (u-boot-dtb.bin) to
> > > > 0x01000040
> > > > and execute it from that address. This should get you the prompt.
> > > > 
> > > > > I get output on the UART (via FPGA pins), to my PC, to suggest that
> > 
> > all
> > 
> > > > is
> > > > 
> > > > > running well.  I can then upload my u-boot.img file to the
> > > > > preloader
> > > > 
> > > > using
> > > > 
> > > > > Ymodem (preloader configured to accept this), and I can run U-boot.
> > > > > 
> > > > > Before I was using the preconfigured 2013.01.01 U-boot as provided
> > > > > on Rocketboards.org, but since we are using the KSZ9031 chip for
> > > > > our PHY
> > > > 
> > > > felt
> > > > 
> > > > > I needed a newer version to support this.  But 2013.01.01 runs ok,
> > 
> > and
> > 
> > > > > I can get to the U-boot prompt on the UART, whereas 2015 hangs.
> > > > > 
> > > > > Any clues gratefully received.
> > > > 
> > > > Try the above method, hope that helps.
> > > > 
> > > > Best regards,
> > > > Marek Vasut