[PATCH v3 00/20] Refactor the architecture parts of mt7628

[Mauro Condarelli <mc5686@mclink.it>]

Sorry Daniel,
I seem unable to pass the message through.

I have, in front of me two, very similar directories:

u-boot.master:
built from master + Weijie.v3 patches + my vocore2 patch.

u-boot-mips:
built from u-boot-mips/testing + my vocore2 patch.

I was very careful not to change anything after patching.

In both directories I do:

make ARCH=mips CROSS_COMPILE=mipsel-linux- distclean
make ARCH=mips CROSS_COMPILE=mipsel-linux- vocore2_defconfig
make ARCH=mips CROSS_COMPILE=mipsel-linux-

Both produce similarly-sized u-boot-mtmips.bin (Flash version)
and u-boot.bin (RAM version).

On my target I do:

setenv autoload no; dhcp; tftpboot 80200000 192.168.7.101:u-boot.bin; go
${fileaddr}

In the case of the "master dir I get:

=> setenv autoload no; dhcp; tftpboot 80200000 192.168.7.101:u-boot.bin;
go ${fileaddr}
BOOTP broadcast 1
DHCP client bound to address 192.168.7.135 (115 ms)
Using eth at 10110000 device
TFTP from server 192.168.7.101; our IP address is 192.168.7.135
Filename 'u-boot.bin'.
Load address: 0x80200000
Loading: ####################################
     1.1 MiB/s
done
Bytes transferred = 527090 (80af2 hex)
## Starting application at 0x80200000 ...


U-Boot 2020.01-00648-gc46834052a-dirty (Feb 11 2020 - 09:30:52 +0100)

CPU:   MediaTek MT7628A ver:1 eco:2
Boot:  DDR2, SPI-NOR 3-Byte Addr, CPU clock from XTAL
Clock: CPU: 580MHz, Bus: 193MHz, XTAL: 40MHz
Model: VoCore2
DRAM:  128 MiB
WDT:   Started with servicing (60s timeout)
MMC:   mmc at 10130000: 0
Loading Environment from SPI Flash... SF: Detected w25q128 with page
size 256 Bytes, erase size 4 KiB, total 16 MiB
OK
Model: VoCore2
Net:  
Warning: eth at 10110000 (eth0) using random MAC address - 26:9f:0f:24:ae:39
eth0: eth at 10110000
Hit any key to stop autoboot:  0
=>

If I do the same with product of "mips" dir I get:

=> setenv autoload no; dhcp; tftpboot 80200000 192.168.7.101:u-boot.bin;
go ${fileaddr}
BOOTP broadcast 1
BOOTP broadcast 2
BOOTP broadcast 3
BOOTP broadcast 4
DHCP client bound to address 192.168.7.112 (3136 ms)
Using eth at 10110000 device
TFTP from server 192.168.7.101; our IP address is 192.168.7.112
Filename 'u-boot.bin'.
Load address: 0x80200000
Loading: ####################################
     1 MiB/s
done
Bytes transferred = 527390 (80c1e hex)
## Starting application at 0x80200000 ...
<*DEAD*>

I am trying to understand where difference lies... and failing, to date :(

I already sent "my vocore2 patch" (Note: that is a "git rebase -i"
squashing the three commits in one, no change) earlier today;
if You think it might be useful I can send also the complete
"git format-patch" from a known u-boot master commit from
where I branched off
(currently 8b9cc858e0239823b051a9324431d511baf2b998)

My "git log --oneline" on "master" is:
c46834052a (HEAD -> weijie.v3.vocore.net, syno0/weijie.v3.vocore.net)
Move back environment to SPI NOR. Changes to environment default settings.
7dbd7fc8bc Added network setup.
375435c35e (syno0/weijie.v3.vocore, weijie.v3.vocore) Add support for
SoM "VoCore2".
f4fa7f499e (syno0/weijie.v3, weijie.v3) mips: mtmips: add support for
mt7628-rfb
724463aa9e mips: mtmips: enable SPL for all boards
1fe4eda6c3 mips: mtmips: add SPL support
5ea8181206 spl: nor: add lzma decompression support for legacy image
173aa9cb6c tools: binman: add etype file for u-boot-lzma-img
789a9085d3 Makefile: add support to generate LZMA compressed u-boot image
e4721a8b98 lib: enable lzma decompression support for SPL build
b769d64488 mips: add an option to enable u_boot_list section for SPL
loaders in u-boot-spl.lds
c704313c7e mips: enable support for appending dtb to spl binary
8801d2b226 dts: mtmips: add alternative pinmux node for uart2
4cda40c3ea mips: mtmips: rewrite lowlevel codes of mt7628
13299d7651 mips: add a option to support not reserving malloc space on
initial stack
eb8a3689b0 mips: add a mtmips-specific field to architecture-specific
global data
f5fda08bed configs: enable CONFIG_RESTORE_EXCEPTION_VECTOR_BASE for all
mtmips boards
e95f8beabe mips: mtmips: make use of sysreset-resetctrl for mt7628 soc
58246e0175 sysreset: add reset controller based reboot driver
b6a20ebced mips: start.S: avoid overwriting outside gd when clearing
global data in stack
c2359e8947 mips: add an option to support initialize SRAM for initial stack
a6eecdd28e mips: mtmips: add predefined i-cache/d-cache size and linesize
b096b2e05f mips: add support to restore exception vector base before
booting linux
2e5b1c00fd MIPS: allow override of get_tbclk()
8b9cc858e0 mx7ulp: Move SoC base address to a common file
...

while in the "mips" directory I have:
cd23ee2179 (HEAD -> VoCore2) Move back environment to SPI NOR. Changes
to environment default settings.
ddc07fb968 Added network setup.
9eabc0c0bb Add support for SoM "VoCore2".
b3d1d5d05b (origin/testing, testing, master) mips: mtmips: add support
for mt7628-rfb

As said: there must be some difference, somewhere,
but it's not evident (to me) where it is and how to find it.

Thanks in Advance for any advice
Best Regards
Mauro

On 2/11/20 2:54 PM, Daniel Schwierzeck wrote:
> On Tue, Feb 11, 2020 at 11:58 AM Mauro Condarelli <mc5686@mclink.it> wrote:
>> Thanks Daniel.
>>
>> On 2/10/20 10:28 PM, Daniel Schwierzeck wrote:
>>> Hi Mauro,
>>>
>>> Am 10.02.20 um 21:20 schrieb Mauro Condarelli:
>>>> FYI
>>>> I've been using this patchset for over a week without any adverse effect.
>>>> It allowed me to port to VoCore2 board.
>>>> Should I add a "Tested-by" flag?
>>>> If so: how should I do it?
>>>>
>>>> Regards
>>>> Mauro Codarelli
>>>>
>>> sorry that I could respond to your questions earlier. I've pushed the
>>> complete patch set from Weijie to:
>>>
>>> https://gitlab.denx.de/u-boot/custodians/u-boot-mips/commits/testing
>> I tried to use this repo/branch, but something is wrong (or I goofed badly).
>>> Maybe this helps you with development. If you have a bootable patch set
>>> (you can do MMC later) for your VoCore2 board, please submit a regular
>>> patch series based on that branch so that we can review again.
>> I *do* have a bootable patchset built on top of master + Nemirovsky
>> "reconfigurable cpu clocks" + Weijie v3.
>> Result is fully working, including net and mmc/SD.
>>
>> This patchset applies cleanly to uboot-mips/testing and compile
>> apparently without errors, but resulting u-boot.bin does not work.
>> By "does not work" I mean it does not utter a single char on debug
>> serial.
>>
>> I tried to do a complete diff between my working tree and this
>> non-working one and there are tons of differences, but I couldn't
>> spot anything that might be relevant.
>>
>> I am unsure on how to proceed; please advise.
> but do you reach the U-Boot prompt? If I apply your patch and build
> it, I need to be able to program the resulting U-Boot image to SPI
> flash and boot to U-Boot prompt. Also note that the final U-Boot image
> with SPL is now called u-boot-mtmips.bin. Don't try to use u-boot.bin.
>
> Regarding debug UART you also need to configure the serial driver with
> "make menuconfig" (register base, clocks, baud rate etc.). There is no
> driver model or device-tree involved at that point. I think Stefan
> already told you the correct settings and you need this
> board_early_init_f magic for the pinmux (don't forget to enable
> CONFIG_BOARD_EARLY_INIT_F). But for the production mode debug UART
> should be disabled anyway. SPL UART is different than debug UART.
> AFAIU the regular driver model and device-tree stuff should work
> there. For the pinmux you'll need the new CONFIG_SPL_UART2_SPIS_PINMUX
> option. According to the code, you'll also need to configure
> CONFIG_CONS_INDEX = 3.
>