Re: [Starpu-devel] segfault with dynamic partitioning and STARPU_PROFILING=1

[Samuel Thibault <samuel.thibault@inria.fr>]

Alfredo Buttari, on Wed 31 Aug 2016 11:24:57 +0200, wrote:
>   call fstarpu_data_unpartition_submit(hdla, 0, shdlsa, 0)

I dont have this in my fortran bindings in the trunk, so couldn't build
the testcase

> and here is what valgrind --tool=memcheck says about its execution:
> 
> 
> ==2241== Invalid read of size 8
> ==2241==    at 0x7EBA547: _starpu_fetch_nowhere_task_input 
> (coherency.c:1169)
> ==2241==    by 0x7E9BD74: _starpu_repush_task (sched_policy.c:463)
> ==2241==    by 0x7E75AC4: _starpu_enforce_deps_and_schedule (jobs.c:644)
> ==2241==    by 0x7E76D58: _starpu_submit_job (task.c:372)
> ==2241==    by 0x7E79418: starpu_task_submit (task.c:682)
> ==2241==    by 0x7EDBAD0: _starpu_task_insert_v (starpu_task_insert.c:137)
> ==2241==    by 0x7EDBFF8: starpu_task_insert (starpu_task_insert.c:159)
> ==2241==    by 0x7EC28CA: starpu_data_partition_submit (filters.c:604)
> ==2241==    by 0x404E7E: __dqrm_dsmat_mod_MOD_dqrm_block_partition1
> (dqrm_dsmat_mod.F90:1294)
> ==2241==    by 0x40B695: __dqrm_dsmat_mod_MOD_dqrm_dsmat_qr_facto_async
> (dqrm_dsmat_mod.F90:505)
> ==2241==    by 0x40B7E8: __dqrm_dsmat_mod_MOD_dqrm_dsmat_qr_facto
> (dqrm_dsmat_mod.F90:384)
> ==2241==    by 0x403A3B: MAIN__ (dqrm_test_dns.F90:94)

Could you not strip the output? :) It should be notably explaining
before and after that snippet why it thinks that the read is invalid
(which I really don't understand since the very same function does the
same accesses at its beginning...)

Samuel