phosphor-pid-control build failure with yocto head

[Patrick Venture]

On Mon, Sep 9, 2019 at 12:26 PM Patrick Venture <venture at google.com> wrote:
>
> On Mon, Sep 9, 2019 at 10:25 AM Patrick Venture <venture at google.com> wrote:
> >
> > On Tue, Sep 3, 2019 at 9:40 PM Benjamin Fair <benjaminfair at google.com> wrote:
> > >
> > > On Tue, Sep 3, 2019 at 5:28 PM James Feist <james.feist at linux.intel.com> wrote:
> > > >
> > > > On 9/2/2019 1:15 AM, Brad Bishop wrote:
> > > > > Hi James, Patrick
> > > > >
> > > > > Updating our poky subtree to master HEAD exposes a build failure in
> > > > > phosphor-pid-control.
> > > >
> > > > Here's the bug, I've seen this before, it seems random when it happens
> > > > and changing a few lines can make it go away. Something to do with lto...
> > > >
> > > > main.o swampd-util.o  ./.libs/libswampd.a -lstdc++fs -lphosphor_logging
> > > > -lsdbusplus -lsystemd -lphosphor_dbus
> > > > 00:59:59 | lto1: internal compiler error: in add_symbol_to_partition_1,
> > > > at lto/lto-partition.c:154
> > > > 00:59:59 | Please submit a full bug report,
> > > > 00:59:59 | with preprocessed source if appropriate.
> > > > 00:59:59 | See <https://gcc.gnu.org/bugs/> for instructions.
> > > > 00:59:59 | lto-wrapper: fatal error: arm-openbmc-linux-gnueabi-g++
> > > > returned 1 exit status
> > > > 00:59:59 | compilation terminated.
> > > > 00:59:59 |
> > > > /tmp/openbmc/work/armv7a-openbmc-linux-gnueabi/phosphor-pid-control/0.1+gitAUTOINC+35906cc3d0-r1/recipe-sysroot-native/usr/bin/arm-openbmc-linux-gnueabi/../../libexec/arm-openbmc-linux-gnueabi/gcc/arm-openbmc-linux-gnueabi/9.2.0/ld:
> > > > error: lto-wrapper failed
> > > > 00:59:59 | collect2: error: ld returned 1 exit status
> > > > 00:59:59 | Makefile:919: recipe for target 'swampd' failed
> > > > 00:59:59 | make[2]: *** [swampd] Error 1
> > > >
> > > > What platform is gsj? We're currently behind tip, and it'll probably
> > > > take me some time to catch up to be able to try anything this new.
> > >
> > > GSJ is a Quanta board with a Nuvoton NPCM7xx BMC.
> > >
> > > This could be showing up there since the CPU cores are ARMv7 unlike in Aspeed.
> >
> > Will this build if we drop the lto stuff?  I have seen this issue
> > before, but I cannot recall the solution... but it did go-away at some
> > point.  So, I'm confident it can be addressed.  I'm going to try
> > building without lto and see if it's happy.
>
> I tested the patch without the flto option enabled and it started
> building again.

https://gerrit.openbmc-project.xyz/c/openbmc/phosphor-pid-control/+/25048
>
> >
> > >
> > > >
> > > > >
> > > > > GCC is asking for a bug to be opened.  Could find someone to have a look
> > > > > at this so we can continue to pick up poky changes?
> > > > >
> > > > > If you would like an openbmc tree to pull give this a try:
> > > > >
> > > > > https://gerrit.openbmc-project.xyz/c/openbmc/openbmc/+/24783
> > > > >
> > > > > thx - brad