[yocto] wic appending ".1" to a file name on multiple images build

[Alexander Kanavin]

It probably goes to some log file?

Alex

> On 1 Jun 2018, at 14.56, Alan Martinovic <alan.martinovic at senic.com> wrote:
> 
> Yeah, a deep dive it is then. :)
> 
> Am just trying to print from the code using the Python debugger (as
> used in the source file).
>    self._log.debug("Print this")
> 
> but it's not printing anything even when running bitbake with -DDDvvv
> 
> On Fri, Jun 1, 2018 at 12:52 PM, Alexander Kanavin
> <alex.kanavin at gmail.com> wrote:
>> 2018-06-01 13:38 GMT+03:00 Alan Martinovic <alan.martinovic at senic.com>:
>>> Building them at the same time causes an issue:
>>> 
>>> Traceback (most recent call lastd):
>>>  File "/home/alan/workspace/senic-os-dev/oe/scripts/lib/wic/filemap.py",
>>> line 120, in _open_image_file
>>>     self._f_image = open(self._image_path, 'rb')
>>>  FileNotFoundError: [Errno 2] No such file or directory:
>>> '/home/alan/workspace/senic-os-dev/build/tmp-glibc/deploy/images/senic-hub/u-boot-sunxi-with-spl.bin.1'
>>> 
>>> 
>>> The cause of the issue seems to be that `wic create`,
>>> even though it gets the correct file path through the
>>> wks file ("u-boot-sunxi-with-spl.bin")
>>> somewhere in the execution of wic create that
>>> fiile gets renamed (to "u-boot-sunxi-with-spl.bin.1")
>>> and results in an above issue.
>>> 
>>> This only happens when the images are built in parallel,
>>> running each separately gives no errors.
>>> 
>>> Am open for suggestions on how to approach this. :)
>> 
>> By applying code archaeology skills obviously, which are a must-have
>> for anyone involved in Yocto :)
>> 
>> Specifically here: where and how is self._image_path set? What are the
>> places in code that call into the path setter? Where do they get the
>> image name from? Does any of them manipulate the name and/or rename
>> the file and why?
>> Should the name include the image name to avoid such clashes and
>> races? And so on. :)
>> 
>> Alex