From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: by sourceware.org (Postfix, from userid 48) id 86C243858C33; Sat, 20 Jan 2024 14:38:06 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 86C243858C33 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org; s=default; t=1705761486; bh=AATxN+WZSMUwKUZizpP10xQATm3ChC/aA+GLLKnloso=; h=From:To:Subject:Date:In-Reply-To:References:From; b=iTNEhZAdQNuqHahqYzc7Czcxsz7g8o/X5VDrMXUWksI8Lr+Pryr1gp4RjbfxRrlA1 vHHpY3YFR6Wng0ByJjQcaM9pYammSyG4lpioxGvJ3DMX9NLWniyuHfv6+RRzNUataC BUFI5Z5/wKAI3A9RhjOZ8COimrZ6w3SL3lBOEhZY= From: "burnus at gcc dot gnu.org" To: gcc-bugs@gcc.gnu.org Subject: [Bug libgomp/113513] [OpenMP] libgomp: cuCtxGetDevice error with OMP_DISPLAY_ENV=true OMP_TARGET_OFFLOAD="mandatory" for libgomp.c/target-52.c Date: Sat, 20 Jan 2024 14:38:05 +0000 X-Bugzilla-Reason: CC X-Bugzilla-Type: changed X-Bugzilla-Watch-Reason: None X-Bugzilla-Product: gcc X-Bugzilla-Component: libgomp X-Bugzilla-Version: 14.0 X-Bugzilla-Keywords: openmp, wrong-code X-Bugzilla-Severity: normal X-Bugzilla-Who: burnus at gcc dot gnu.org X-Bugzilla-Status: UNCONFIRMED X-Bugzilla-Resolution: X-Bugzilla-Priority: P3 X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org X-Bugzilla-Target-Milestone: --- X-Bugzilla-Flags: X-Bugzilla-Changed-Fields: Message-ID: In-Reply-To: References: Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/ Auto-Submitted: auto-generated MIME-Version: 1.0 List-Id: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D113513 --- Comment #1 from Tobias Burnus --- Looking at the called GOMP_OFFLOAD_* function, in the failing case, there i= s: ... DEBUG GOMP_OFFLOAD_run DEBUG GOMP_OFFLOAD_dev2host DEBUG GOMP_OFFLOAD_free DEBUG: nvptx_attach_host_thread_to_device - 0 and in the successful case: DEBUG GOMP_OFFLOAD_fini_device 0 <<< called before unregister DEBUG: GOMP_offload_unregister_ver dev=3D0; state=3DGOMP_DEVICE_FINALIZED DEBUG: GOMP_offload_unregister_ver dev=3D0; state=3DGOMP_DEVICE_FINALIZED and then - in the failing case: DEBUG: GOMP_offload_unregister_ver dev=3D0; state=3DGOMP_DEVICE_INITIALIZED DEBUG: GOMP_offload_unregister_ver dev=3D0; state=3DGOMP_DEVICE_INITIALIZED DEBUG: gomp_unload_image_from_device DEBUG GOMP_OFFLOAD_unload_image, 0, 196609 DEBUG: gomp_target_fini; dev=3D0, state=3DGOMP_DEVICE_INITIALIZED DEBUG GOMP_OFFLOAD_fini_device 0 DEBUG: nvptx_attach_host_thread_to_device - 0 libgomp: cuCtxGetDevice error: unknown cuda error Thus, for some reason, GOMP_OFFLOAD_fini_device then GOMP_offload_unregister_ver is swapped when OMP_DISPLAY_ENV=3Dtrue and OMP_TARGET_OFFLOAD=3D"mandatory" are set - but not otherwise. The call to omp_target_fini comes from: if (atexit (gomp_target_fini) !=3D 0) gomp_fatal ("atexit failed"); While the call to GOMP_offload_unregister_ver comes from mkoffload: fprintf (out, "static __attribute__((destructor)) void fini (void)\n" "{\n" " GOMP_offload_unregister_ver (%#x, __OFFLOAD_TABLE__," " %d/*NVIDIA_PTX*/, &nvptx_data);\n" "};\n", * * *=20 Actually, the same problem occurs when compiled with: -foffload=3Ddisable With that flag + no 'mandatory': DEBUG GOMP_OFFLOAD_version DEBUG GOMP_OFFLOAD_get_caps DEBUG GOMP_OFFLOAD_get_num_devices 0 DEBUG GOMP_OFFLOAD_get_name DEBUG GOMP_OFFLOAD_get_type DEBUG GOMP_OFFLOAD_init_device 0 DEBUG: nvptx_open_device - 0 DEBUG: gomp_target_fini; dev=3D0, state=3DGOMP_DEVICE_INITIALIZED DEBUG GOMP_OFFLOAD_fini_device 0 DEBUG: nvptx_attach_host_thread_to_device - 0 And with 'mandatory' + OMP_DISPLAY_ENV=3Dverbose: DEBUG GOMP_OFFLOAD_version DEBUG GOMP_OFFLOAD_get_caps DEBUG GOMP_OFFLOAD_get_num_devices 0 DEBUG GOMP_OFFLOAD_get_name DEBUG GOMP_OFFLOAD_get_type < omp_display_env output> DEBUG GOMP_OFFLOAD_init_device 0 DEBUG: nvptx_open_device - 0 libgomp: OMP_TARGET_OFFLOAD is set to MANDATORY, but device cannot be used = for offloading DEBUG: gomp_target_fini; dev=3D0, state=3DGOMP_DEVICE_INITIALIZED DEBUG GOMP_OFFLOAD_fini_device 0 DEBUG: nvptx_attach_host_thread_to_device -=20 libgomp: cuCtxGetDevice error: unknown cuda error libgomp: device finalization failed Thus, the error message is the same =E2=80=93 but here no offloading code e= xists and just gomp_target_fini is called. - However, there is a prior call to=20 'gomp_fatal' which probably messes things up for the plugin handling - whi= le in the original code, we have a valid code. * * * If there is no offloading code but OMP_DISPLAY_ENV=3Dverbose OMP_TARGET_OFFLOAD=3D"mandatory" is used, it works: DEBUG GOMP_OFFLOAD_version DEBUG GOMP_OFFLOAD_get_caps DEBUG GOMP_OFFLOAD_get_num_devices 0 DEBUG GOMP_OFFLOAD_get_name DEBUG GOMP_OFFLOAD_get_type OPENMP DISPLAY ENVIRONMENT BEGIN ... OPENMP DISPLAY ENVIRONMENT END DEBUG: gomp_target_fini; dev=3D0, state=3D0 * * * If there is only one or none of the two env vars, there is no need to search for devices - and, hence, the nvptx plugin is not called at all and it, obviously, works as well.=