[U-Boot] qemu and u-boot-console and other ports?

[Stephen Warren]

On 12/02/2016 12:18 PM, Tom Rini wrote:
> On Fri, Dec 02, 2016 at 11:03:00AM -0700, Stephen Warren wrote:
>> On 12/02/2016 07:40 AM, Tom Rini wrote:
>>> Hey,
>>>
>>> I'm trying to debug adding sh4 and r2dplus support to test.py.  Until my
>>> current round of testing is applied you'll need
>>> http://patchwork.ozlabs.org/bundle/trini/fix-sh4-support-v2/ in order
>>> for us to get a functional U-Boot.  After that, I have a "normal" conf
>>> file for the board for QEMU that looks like:
>>> console_impl=qemu
>>> qemu_machine="r2d"
>>> qemu_binary="qemu-system-sh4"
>>> qemu_extra_args="-nographic -serial null -serial mon:stdio"
>>> qemu_kernel_args="-kernel ${U_BOOT_BUILD_DIR}/u-boot.bin"
>>> reset_impl=none
>>> flash_impl=none
>>>
>>> And here's where things get funny.  When I kick off test.py with --bd
>>> r2dplus --id qemu --build -s -k sleep it will build and launch the board
>>> and since I have -s I can see stdio and I see U-Boot come up and give me
>>> the prompt, and then test.py says it times out waiting for the prompt.
>>> Any ideas where to poke next?  I want to say something is funny with
>>> respect to what we see and where we see it (in terms of pipes) due to
>>> having to say -serial null -serial mon:stdio so that we see port #2 on
>>> the "board" rather than port #1 as this is the port that U-Boot and
>>> Linux use by default.  Thanks!
>>
>> Does "-serial null -serial mon:stdio" cause qemu to re-open file
>> descriptors rather than just using stdout directly, or something
>> like that? If so, perhaps its output is getting into the overall
>> script's stdout (e.g. your terminal/console?) rather than the pipe
>> that test/py is reading. If so, you'd see the qemu output but
>> test/py wouldn't.
>>
>> I think you can test this assumption by removing the -s option. If
>> you still see qemu output, then qemu is sending it directly to
>> wherever the overall stdout is going. If you no longer see qemu's
>> output, then test/py must be reading it and only echo'ing it due to
>> the -s option, and so you'd have to look somewhere else.
>
> When I drop out -s instead I get:
> ----------------------------------- Captured stdout ------------------------------------
> +u-boot-test-reset r2dplus qemu
> long write to SH7750_WCR1_A7 (0x000000001f800008) ignored
> long write to SH7750_WCR2_A7 (0x000000001f80000c) ignored
> long write to SH7750_WCR3_A7 (0x000000001f800010) ignored
> long write to SH7750_MCR_A7 (0x000000001f800014) ignored
> word write to SH7750_RTCNT_A7 (0x000000001f800020) ignored
> word write to SH7750_RTCOR_A7 (0x000000001f800024) ignored
> Write access to refresh count register
> word write to SH7750_RTCSR_A7 (0x000000001f80001c) ignored
> Read access to refresh count register, incrementing
> long write to SH7750_MCR_A7 (0x000000001f800014) ignored
>
>
> U-Boot 2016.11-00492-ged6a1e9bbf0c (Dec 02 2016 - 14:14:17 -0500)
>
> CPU: SH4
> BOARD: Renesas Solutions R2D Plus
> DRAM:  64 MiB
> Flash: ERROR: too many flash sectors
> 8 MiB
> *** Warning - bad CRC, using default environment
>
> PCI: SH7751 PCI host bridge found.
> long read to SH7750_WCR1_A7 (0x000000001f800008) ignored
> long read to SH7750_WCR2_A7 (0x000000001f80000c) ignored
> long read to SH7750_WCR3_A7 (0x000000001f800010) ignored
> long read to SH7750_MCR_A7 (0x000000001f800014) ignored
> PCI:   Bus Dev VenId DevId Class Int
> PCI:
>   00:00.0     - 1054:350e - Bridge device
>   00:02.0     - 10ec:8139 - Network controller
> In:    serial
> Out:   serial
> Err:   serial
> Net:   RTL8139#0
> Error: RTL8139#0 address not set.
>
> IDE:   Bus 0: not available
> => => s
> =========================== 92 tests deselected by '-ksleep' ===========================
> ======================= 1 failed, 92 deselected in 31.20 seconds =======================
>
> So it's seeing output, but for some reason not matching on it?

If you drop -s, you shouldn't see any of U-Boot's output. It looks like 
qemu is writing its serial port output to a file descriptor other than 
the stdout that's fed to its parent.

Oh, you can also confirm this by looking in the test-log.html generated 
by test/py; that will show you whether test/py sees qemu's output at 
all, since all of qemu's output is logged to that file. I guess this 
assumes you're running locally, since Travis doesn't save that file for 
you to look at. Perhaps you could cat it after running test/py though.