all bset email

[Chris Johns]

On 9/11/19 10:00 am, Joel Sherrill wrote:
> 
> 
> On Fri, Nov 8, 2019 at 4:24 PM Chris Johns <chrisj at rtems.org
> <mailto:chrisj at rtems.org>> wrote:
> 
>     On 9/11/19 3:04 am, Joel Sherrill wrote:
>     > When building the rtems-all.bset, the build log is quite large, especially
>     when
>     > there are failures.
> 
>     I have increased the limit on that list.
> 
>     > Would it make more sense for automated testing to build each target
>     > independently? This would result in more, smaller emails.
> 
>     It is easier to track failures.
>      
>     > I am thinking it might be easier to figure out the specific failures as
>     well as
>     > avoid having messages blocked to build@ for size.
> 
>     Yes.
> 
>     > I started this thought from having a msys2 build that took almost 3 days to
>     > complete have multiple failures and get blocked for size at build@
> 
>     There is no pending email to that list from you for a Windows build. There were
>     3 emails of 1.1M pending, two from Sebastian and one from you. I have accepted
>     all 3.
> 
> 
> Grrr... is anything wrong with this command line?
> 
>  ../source-builder/sb-set-builder --prefix=/home/jrs007/mingw-tools-5
> --keep-going --mail --mail-to=build at rtems.org <mailto:build at rtems.org>
> --mail-from=joel at rtems.org <mailto:joel at rtems.org> --log=l-all.txt 5/rtems-all

It depends on your mail set up and other settings such as the environment. I
would have thought --smtp-host would have been supplied on the command line. I
have no idea how your mingw shell would find your SMTP host.

> I wonder why no email got out. Any ideas? Might there be a hint in the log?

I suggest you check your logs and if they have a valid connection to the
rtems.org I can take a look.

> FWIW only 12 gcc's got installed. 
> 
> $ ls ~/mingw-tools-5/bin/*gcc.exe
> /home/jrs007/mingw-tools-5/bin/aarch64-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/bfin-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/i386-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/m68k-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/microblaze-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/mips-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/or1k-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/powerpc-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/sh-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/sparc64-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/v850-rtems5-gcc.exe
> /home/jrs007/mingw-tools-5/bin/x86_64-rtems5-gcc.exe
> 
> I don't see arm, epiphany, lm32, moxie, nios2, riscv, or sparc.
>
> ARM failed for this:
> 
> In file included from
> ../../../../../../../../../../gnu-mirror-gcc-fb371a33fa6/newlib/libm/machine/arm/s_floor.c:39:0:
> ../../../../../../../../../../gnu-mirror-gcc-fb371a33fa6/newlib/libm/machine/arm/../../math/s_floor.c:65:10:
> fatal error: fdlibm.h: No such file or directory
>  #include "fdlibm.h"
>           ^~~~~~~~~~
> compilation terminated.
> 
> That looks quite odd. Is there still some issue with path length on the build
> system that I should have done something special for?
> 
> Any other ideas?

I think sparc is an sis build failure. It is on my list.

You need to check the details in the log for each arch. It is time consuming.

Chris