Re: rdma compile error

[Olga Kornievskaia <aglo@umich.edu>]

On Thu, Dec 19, 2019 at 10:57 AM Chuck Lever <chuck.lever@oracle.com> wrote:
>
>
>
> > On Dec 4, 2019, at 2:09 PM, Olga Kornievskaia <aglo@umich.edu> wrote:
> >
> > On Wed, Dec 4, 2019 at 1:25 PM Chuck Lever <chuck.lever@oracle.com> wrote:
> >>
> >>
> >>
> >>> On Dec 4, 2019, at 1:12 PM, Olga Kornievskaia <aglo@umich.edu> wrote:
> >>>
> >>> On Wed, Dec 4, 2019 at 1:02 PM Chuck Lever <chuck.lever@oracle.com> wrote:
> >>>>
> >>>> Hi Olga-
> >>>>
> >>>>> On Dec 4, 2019, at 11:15 AM, Olga Kornievskaia <aglo@umich.edu> wrote:
> >>>>>
> >>>>> Hi Chuck,
> >>>>>
> >>>>> I git cloned your origin/cel-testing, it's on the following commit.
> >>>>> commit 37e235c0128566e9d97741ad1e546b44f324f108
> >>>>> Author: Chuck Lever <chuck.lever@oracle.com>
> >>>>> Date:   Fri Nov 29 12:06:00 2019 -0500
> >>>>>
> >>>>>  xprtrdma: Invoke rpcrdma_ep_create() in the connect worker
> >>>>>
> >>>>> And I'm getting the following compile error.
> >>>>>
> >>>>> CC [M]  drivers/infiniband/core/cma_trace.o
> >>>>> In file included from drivers/infiniband/core/cma_trace.h:302:0,
> >>>>>               from drivers/infiniband/core/cma_trace.c:16:
> >>>>> ./include/trace/define_trace.h:95:43: fatal error: ./cma_trace.h: No
> >>>>> such file or directory
> >>>>> #include TRACE_INCLUDE(TRACE_INCLUDE_FILE)
> >>>>>                                         ^
> >>>>> Is this known?
> >>>>
> >>>> I haven't had any complaints from lkp.
> >>>>
> >>>> f73179592745 ("RDMA/cma: Add trace points in RDMA Connection Manager")
> >>>>
> >>>> should have added drivers/infiniband/core/cma_trace.h .
> >>>>
> >>>
> >>> The file "cma_trace.h" is there in the "core" directory. But for some
> >>> reason my compile expects it to be in include/trace directory (if I
> >>> were to copy it there I can compile).
> >>
> >> The end of cma_trace.h should have:
> >>
> >> #undef TRACE_INCLUDE_PATH
> >> #define TRACE_INCLUDE_PATH .
> >> #define TRACE_INCLUDE_FILE cma_trace
> >
> > It does have it.
> >
> >> That is supposed to steer the compiler to the cma_trace.h in core/ .
> >>
> >> Does a "make mrproper; git clean -d -f -x" help? Feels like there's
> >> a stale generated file somewhere that's breaking things.
> >
> > I probably do have something uncleaned. I have tried what you
> > suggested but it's not helping. This build is a tar of a git clone
> > tree then copied into an internal lab (with rdma hardware).
>
> I found a very similar compile issue yesterday. The fix is in the
> current cel-testing topic branch, if you are interested.

Thanks. I'll give it a try!

>
>
> --
> Chuck Lever
>
>
>