[U-Boot] help on using buildman

[York Sun <yorksun@freescale.com>]

On 11/20/2014 11:13 AM, Simon Glass wrote:
> Hi York,
> 
> On 20 November 2014 18:47, York Sun <yorksun@freescale.com> wrote:
>> Simon,
>>
>> Can you shed some light on my trouble? I am trying to use buildman for
>> regression tests. I got inconsistency and am trying to find out why. Take one
>> example MPC832XEMDS_HOST_66 in this case (I randomly pick a failed board)
>>
>> On one computer
>>
>> $ ./tools/buildman/buildman -b working_upstream_qoriq --force-build
>> --force-reconfig MPC832XEMDS_HOST_66 -ve
>> No section: 'make-flags'
>> winegcc: gcc-4.5 failed
>> boards.cfg is up to date. Nothing to do.
>> Building 15 commits for 1 boards (1 thread, 4 jobs per thread)
>> 01: ARM: atmel: add sama5d4 xplained ultra board support
>> 02: qe: add u-qe support to arm board
>> 03: qe: add qe support to ls1021aqds
>> 04: qe: add qe support for ls1021a-twr board
>> 05: arm: ls102xa: Fixed a register definition error
>> 06: u_qe: add u_qe_upload_firmware for u_qe
>> 07: powerpc: Move fsl_errata.h out of arch/powerpc
>> 08: driver/usb/fsl: Make FSL USB driver common
>> 09: drivers: usb: fsl: Define USB configs for LS102XA
>> 10: ls1: config: Enable USB EHCI Host on LS1021AQDS
>> 11: arm: ls102xa: Add snoop disable for slave port 0, 1 and 2
>> 12: arm: ls102xa: Remove bit reversing for SCFG registers
>> 13: ls102xa: ifc: nor: fix the write issue when bytes unaligned
>> 14: arm: ls102xa: Add SystemID EEPROM support for LS1021ATWR board
>> 15: arm: ls102xa: Select ge2_clk125 for eTSEC clock muxing
>>    15    0    0 /15     MPC832XEMDS_HOST_66
>>
>> No error.
>>
>> On another computer
>>
>> $ ./tools/buildman/buildman -b working_upstream_qoriq --force-build
>> --force-reconfig MPC832XEMDS_HOST_66 -se
> 
> -s means it won't actually build, just report a previous build. But I
> suppose you did a previous build.

Right. I posted the command to show summary by mistake. Here is the one to build it.

$ ./tools/buildman/buildman -b working_upstream_qoriq --force-build
--force-reconfig MPC832XEMDS_HOST_66 -ve
No section: 'make-flags'
boards.cfg is up to date. Nothing to do.
Building 15 commits for 1 boards (1 thread, 24 jobs per thread)
Cloning repo for thread 0
01: ARM: atmel: add sama5d4 xplained ultra board support
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
02: qe: add u-qe support to arm board
03: qe: add qe support to ls1021aqds
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
04: qe: add qe support for ls1021a-twr board
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
05: arm: ls102xa: Fixed a register definition error
06: u_qe: add u_qe_upload_firmware for u_qe
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
07: powerpc: Move fsl_errata.h out of arch/powerpc
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
08: driver/usb/fsl: Make FSL USB driver common
09: drivers: usb: fsl: Define USB configs for LS102XA
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
10: ls1: config: Enable USB EHCI Host on LS1021AQDS
11: arm: ls102xa: Add snoop disable for slave port 0, 1 and 2
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
12: arm: ls102xa: Remove bit reversing for SCFG registers
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
13: ls102xa: ifc: nor: fix the write issue when bytes unaligned
14: arm: ls102xa: Add SystemID EEPROM support for LS1021ATWR board
   powerpc: +   MPC832XEMDS_HOST_66
+make[1]: *** No rule to make target `../arch//cpu/u-boot.lds', needed by
`u-boot.lds'.  Stop.
+make: *** [sub-make] Error 2
15: arm: ls102xa: Select ge2_clk125 for eTSEC clock muxing
    6    0    9 /15     MPC832XEMDS_HOST_66


> 
> It seems to have failed to configure things - there is no $ARCH perhaps?
> 

When I built for more targets, for example 'arm & freescale' powerpc aarch64, I
see this kind of errors, but not with all boards. The ARCH is not defined in
command line or environment when calling buildman. Beside, this error comes and
goes for the same board for different commit. It doesn't make much sense.

>>
>> You can see one error comes on and off. How can I debug it? I don't use python
>> enough to diagnose the code. Is there a way to dump the make command to check?
>>
>> York
> 
> Any change you are missing a toolchain or it is broken?
> 
> See ../working_upstream_qoriq/01*/MPC832XEMDS_HOST_66/toolchain for
> possible clues.
> 

The toolchain is OK. I use the same setup to build other targets. Not all have
this error. Using the same toolchain, I have no error calling MAKEALL.

> You can only see the make output if you comment out this line in
> builderthread.py:
> 
>                 #args.append('-s')
> 
> Then it will be in a file called 'log'. It would be nice to have an
> option for that...
> 

Let me try that.

York