Re: [Qemu-devel] [PATCH v4 06/10] tests: Add NetBSD image

[Fam Zheng]

On Tue, 08/29 18:47, Philippe Mathieu-Daudé wrote:
> Hi Fam,
> 
> On 08/29/2017 09:09 AM, Philippe Mathieu-Daudé wrote:
> [...]>> +if __name__ == "__main__":
> > > +    sys.exit(basevm.main(NetBSDVM))
> > 
> > This one is failing:
> > 
> > DEBUG:root:ssh_cmd: ssh -q -o StrictHostKeyChecking=no -o
> > UserKnownHostsFile=/dev/null -o ConnectTimeout=1 -p 34091 -i
> > /tmp/qemu-vm-59XYOj/id_rsa address@hidden
> >          set -e;
> >          cd $(mktemp -d /var/tmp/qemu-test.XXXXXX);
> >          tar -xf /dev/rld1a;
> >          ./configure --python=python2.7 ;
> >          gmake -j4;
> >          gmake check;
> > 
> > ...
> >    CC      bt-host.o
> >    CC      bt-vhci.o
> >    CC      dma-helpers.o
> >    CC      vl.o
> >    CC      tpm.o
> > In file included from vl.c:72:0:
> > /var/tmp/qemu-test.ht0XHU/include/hw/loader.h:4:29: fatal error:
> > hw/nvram/fw_cfg.h: No such file or directory
> >   #include "hw/nvram/fw_cfg.h"
> >                               ^
> > compilation terminated.
> > /var/tmp/qemu-test.ht0XHU/rules.mak:66: recipe for target 'vl.o' failed
> > gmake: *** [vl.o] Error 1
> > gmake: *** Waiting for unfinished jobs....
> > tests/vm/Makefile.include:32: recipe for target 'vm-build-netbsd' failed
> > make: *** [vm-build-netbsd] Error 3
> 
> Probably false alarm, this seems to be an ENOMEM host error.
> 
> I later got:
> 
>       CHK version_gen.h
> Makefile:342: recipe for target 'subdir-dtc' failed
> gmake[1]: *** No rule to make target 'dtc/libfdt/fdt.h', needed by
> 'libfdt/fdt.o'.  Stop.
> gmake: *** [subdir-dtc] Error 2
> gmake: *** Waiting for unfinished jobs....
> 
> which I solved running "gmake -C dtc" after ./configure but I'm not sure it
> is necessary or can come from my tree, this testing is veeeeeeery slow and
> mostly kill my laptop, I ended wondering on what kind of hardware you
> developed this series without going crazy nut :S

Not too beefy, just a Lenovo W541 laptop with i7-4810. To save you from being
killed again, maybe we should reduce the default -smp to nr_cores / 2?

Fam