I think the part_persist.h issue just came up in a GitHub issue: https://github.com/open-mpi/ompi/issues/13847
> On May 6, 2026, at 8:09 AM, [email protected] wrote: > > Dear Jeff, > > You can forget about my difficulties. > > Yesterday I started again from scratch and, this time, the entire compilation > is > (almost) normal. The almost is because I had to edit the file: > > emacs ${SOURCE}/ompi/mca/part/persist/part_persist.h, > > where I had to change the definition of mca_part_persist_start (around line > 484) and > comment out the __always_inline__ directive, due to a persistent error > message: > > inlined from ‘mca_part_persist_start’ at > ../../../../../ompi/mca/part/persist/part_persist.h:500:23: > ../../../../../ompi/mca/part/persist/part_persist.h:484:1: error: inlining > failed in call to ‘always_inline’ ‘mca_part_persist_start’: recursive inlining > 484 | mca_part_persist_start(size_t count, ompi_request_t** requests) > > Previously, I tried increasing the value of the max-inline-insns-single > parameter, > but the error persisted. > > I've attached two files (config.log and ompi_info_param.log). > > Many thanks for your time, and my apologies for bothering you. > > Regards, > Jorge D'Elia. > > Jeff Squyres <[email protected] <mailto:[email protected]>> escribió: > >> The two files are constructed exactly the same; the only difference is in >> how they are compressed in just about the last step (during the >> Automake-generated “make dist” step). >> >> >> I am unable to replicate your problem. Could you send all the logs of how >> you download and expand the .gz tarball, and all the logs described here: >> https://docs.open-mpi.org/en/v5.0.x/getting-help.html#for-problems-building-or-installing-open-mpi >> >> Specifically, when I download, examine, and build the 5.0.10 gz and bz2 >> tarballs, they function identically for me. For example, I wrote this short >> comparison script: >> >> ——— >> $ cat test.sh >> >> #!/bin/sh >> >> >> TARBALL=openmpi-5.0.10.tar >> >> URL_STEM=https://download.open-mpi.org/release/open-mpi/v5.0 >> >> >> # Hand-copied from https://www.open-mpi.org/software/ompi/v5.0/ >> >> UPSTREAM_SHA256_BZ2=0acecc4fc218e5debdbcb8a41d182c6b0f1d29393015ed763b2a91d5d7374cc6 >> >> UPSTREAM_SHA256_GZ=5692cc80554a7117c99eaa725d35100edd8bbf73423a5e265ff867979192df7d >> >> >> echo "Downloading bz2" >> >> rm -f $TARBALL.bz2 >> >> wget -q $URL_STEM/$TARBALL.bz2 >> >> echo "Downloading gz" >> >> rm -f $TARBALL.gz >> >> wget -q $URL_STEM/$TARBALL.gz >> >> >> echo "Computing sha256 bz2" >> >> DL_SHA256_BZ2=$(sha256sum $TARBALL.bz2 | awk '{print $1}') >> >> echo "Computing sha256 gz" >> >> DL_SHA256_GZ=$(sha256sum $TARBALL.gz | awk '{print $1}') >> >> >> if [[ $UPSTREAM_SHA256_BZ2 != $DL_SHA256_BZ2 ]]; then >> >> echo "bz2 SHA256's don't match!" >> >> echo "- upstream: $UPSTREAM_SHA256_BZ2" >> >> echo "- downloaded: $DL_SHA256_BZ2" >> >> else >> >> echo "bz2: good download" >> >> fi >> >> >> if [[ $UPSTREAM_SHA256_GZ != $DL_SHA256_GZ ]]; then >> >> echo "gz SHA256's don't match!" >> >> echo "- upstream: $UPSTREAM_SHA256_GZ" >> >> echo "- downloaded: $DL_SHA256_GZ" >> >> else >> >> echo "gz: good download" >> >> fi >> >> >> DIR=3rd-party/openpmix/ >> >> echo "Number of files in $DIR in bz2" >> >> tar tf $TARBALL.bz2 | grep $DIR | wc -l >> >> echo "Number of files in $DIR in gz" >> >> tar tf $TARBALL.gz | grep $DIR | wc -l >> >> ——— >> >> And when I run it: >> >> $ ./test.sh >> >> Downloading bz2 >> >> Downloading gz >> >> Computing sha256 bz2 >> >> Computing sha256 gz >> >> bz2: good download >> >> gz: good download >> >> Number of files in 3rd-party/openpmix/ in bz2 >> >> 1291 >> >> Number of files in 3rd-party/openpmix/ in gz >> >> 1291 >> >> >> Same number of files in 3rd-party/openpmix in both tarballs. And both >> tarballs expand and build properly for me. >> >> >> >>> On May 3, 2026, at 10:05 AM, [email protected] wrote: >>> >>> > Dear Jeff. > > Many thanks for your suggestions. In short: > > 1) The problem was (and still is present) with the source code of alternative > 1 > (https://download.open-mpi.org/release/open-mpi/v5.0/openmpi-5.0.10.tar.gz), > because that file does not contain all source codes in /3rd-party. > > 2) But with the source code of alternative 2 > (https://download.open-mpi.org/release/open-mpi/v5.0/openmpi-5.0.10.tar.bz2), > everything returned to the normal way (sources, configuration, building, and > installation). > > Perhaps, the construction of the *.tar.gz file available in the repository > should be reviewed. > > Regards. > Jorge D'Elia > > Jeff Squyres <[email protected] <mailto:[email protected]>> escribió: > >> Just like libevent, OpenPMIx is also in that one tarball. Most people can >> do the “quick start” in our docs: >> https://docs.open-mpi.org/en/v5.0.x/installing-open-mpi/quickstart.html#building-from-source >> >> >> I.e., download, untar, run ./configure, run make, run make install. >> >> If you’re doing something else, the instructions may be a bit different. If >> you’re building from a git clone, for example, you’ll want to see the >> Developer’s guide: https://docs.open-mpi.org/en/v5.0.x/developers/index.html >> >> >> >>> On Apr 29, 2026, at 1:42 PM, [email protected] wrote: >>> >>> Dear Jeff Squyres, >>> >>> Thank you so much for your suggestions. They helped me make some progress, >>> but I've gotten stuck again, this time with: >>> >>> *** Configuring PMIx >>> configure: ===== configuring 3rd-party/openpmix ===== >>> configure: WARNING: no configuration information is in 3rd-party/openpmix >>> configure: ===== done with 3rd-party/openpmix configure ===== >>> configure: error: Could not find viable pmix build. >>> >>> Please, where can I find a reliable PMIX source? Thanks in advance. >>> >>> Regards. >>> Jorge D'Elia. >>> >>> -- >>> >>> Jeff Squyres <[email protected]> escribió: >>> >>>> The libevent file is there in the 5.0.10 tarball: >>>> >>>> $ wget >>>> https://download.open-mpi.org/release/open-mpi/v5.0/openmpi-5.0.10.tar.gz >>>> …snipped… >>>> $ tar tf openmpi-5.0.10.tar.gz| grep libevent-2.1.12-stable-ompi.tar.gz >>>> openmpi-5.0.10/3rd-party/libevent-2.1.12-stable-ompi.tar.gz >>>> $ >>>> >>>> I think you might want to check how the original tarball was unzipped, >>>> and/or make sure that the backing store didn’t somehow exclude saving that >>>> file. >>>> >>>> >>>> >>>>> On Apr 28, 2026, at 7:09 PM, [email protected] wrote: >>>>> >>>>> Hello Howard, >>>>> >>>>> Thank you so much for your time and help! >>>>> >>>>> "'Pritchard Jr., Howard' via Open MPI users" <[email protected]> >>>>> escribió: >>>>> >>>>> >>>>>> Hello Jorge, >>>>>> >>>>>> Could you check if /home/bigpack/openmpi-paq/openmpi-5.0.10/3rd-party >>>>>> >>>>> >>>>> I followed the usual procedure, i.e. I downloaded the file >>>>> https://download.open-mpi.org/release/open-mpi/v5.0/openmpi-5.0.10.tar.gz >>>>> and configure it. >>>>> >>>>>> has libevent-2.1.12-stable-ompi.tar.gz present? >>>>>> >>>>> >>>>> However, there is no such file, i.e. >>>>> >>>>> configure: Expanding ./3rd-party/libevent-2.1.12-stable-ompi.tar.gz in >>>>> 3rd-party >>>>> ./configure: line 60632: >>>>> /home/bigpack/openmpi-paq/openmpi-5.0.10/3rd-party/libevent-2.1.12-stable-ompi.tar.gz: >>>>> No such file or directory >>>>> tar: This does not look like a tar archive >>>>> tar: Exiting with failure status due to previous errors >>>>> configure: error: failed to extract >>>>> 3rd-party/libevent-2.1.12-stable-ompi.tar.gz >>>>> configure: ===== configuring 3rd-party/libevent-2.1.12-stable-ompi ===== >>>>> configure: WARNING: no configuration information is in >>>>> 3rd-party/libevent-2.1.12-stable-ompi >>>>> configure: ===== done with 3rd-party/libevent-2.1.12-stable-ompi >>>>> configure ===== >>>>> ./configure: line 60909: test: -eq: unary operator expected >>>>> configure: error: Could not find viable libevent build. >>>>> >>>>>> It should be there. It may be possible something is wrong with the >>>>>> g(un)zip utilities on your system? >>>>>> >>>>> >>>>> Ok, I'll check it just in case. >>>>> >>>>>> If the tarball is present in 3rd-party why don’t you try running the >>>>>> following in the following in the 3rd-party directory: >>>>>> >>>>>> gunzip libevent-2.1.12-stable-ompi.tar.gz | tar xf – >>>>>> >>>>> >>>>> Ok but, again, there is not any file. >>>>> >>>>>> and see what’s happens. >>>>>> >>>>>> You may also want to check in the config.log in the top level directory >>>>>> of your unpacked Open MPI for “gzip” and see where the configury thinks >>>>>> your gzip command is located. >>>>>> >>>>> >>>>> Ok. >>>>> >>>>> Thanks again! >>>>> >>>>> Regards. >>>>> Jorge. >>>>> >>>>> >>>>>> Howard >>>>>> >>>>>> >>>>>> >>>>>> From: <[email protected]> on behalf of "[email protected]" >>>>>> <[email protected]> >>>>>> Reply-To: "[email protected]" <[email protected]> >>>>>> Date: Tuesday, April 28, 2026 at 1:23 PM >>>>>> To: Open MPI Users <[email protected]> >>>>>> Subject: [EXTERNAL] [OMPI users] Where is the >>>>>> libevent-2.1.12-stable-ompi.tar.gz file ? >>>>>> >>>>>> >>>>>> Hi all, >>>>>> >>>>>> I downloaded the >>>>>> openmpi-5.0.10.tar.gz<https://urldefense.com/v3/__https:/download.open-mpi.org/release/open-mpi/v5.0/openmpi-5.0.10.tar.gz__;!!Bt8fGhp8LhKGRg!AYxajar_k4gcP69SeOSe8tLYLNhyv3fxY4cqZldI_S9QOUvlEbVTQz9foo2jgbv5QqJH4opKZaTfoDJ7jggZd44$> >>>>>> package and I'm trying to build it using the >>>>>> script given given in "Sec. 2.2. For problems building or installing >>>>>> Open MPI" of >>>>>> https://docs.open-mpi.org/en/main/getting-help.html<https://urldefense.com/v3/__https:/docs.open-mpi.org/en/main/getting-help.html__;!!Bt8fGhp8LhKGRg!AYxajar_k4gcP69SeOSe8tLYLNhyv3fxY4cqZldI_S9QOUvlEbVTQz9foo2jgbv5QqJH4opKZaTfoDJ7_MkDPGU$>. >>>>>> >>>>>> But it's failing me with the missing file >>>>>> /3rd-party/libevent-2.1.12-stable-ompi.tar.gz, >>>>>> as follows: >>>>>> >>>>>> *** Configuring Libevent >>>>>> checking for libevent pkg-config name... libevent_core >>>>>> checking if libevent pkg-config module exists... no >>>>>> configure: Searching for libevent in default search paths >>>>>> checking for libevent cppflags... >>>>>> checking for libevent ldflags... >>>>>> checking for libevent libs... -levent_core >>>>>> checking for libevent static libs... -levent_core >>>>>> checking for event2/event.h... no >>>>>> configure: Expanding ./3rd-party/libevent-2.1.12-stable-ompi.tar.gz in >>>>>> 3rd-party >>>>>> ./configure: line 60632: >>>>>> /home/bigpack/openmpi-paq/openmpi-5.0.10/3rd-party/libevent-2.1.12-stable-ompi.tar.gz: >>>>>> No such file or directory >>>>>> >>>>>> Then, please, where can I download it? >>>>>> >>>>>> Thanks in advance. >>>>>> >>>>>> Regards. >>>>>> Jorge D'Elia. >>>>>> -- >>>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>>> an email to >>>>>> [email protected]<mailto:[email protected]>. >>>>>> >>>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>>> an email to [email protected]. >>>>>> >>>>> >>>>> >>>>> >>>>> To unsubscribe from this group and stop receiving emails from it, send an >>>>> email to [email protected]. >>>>> >>>> >>>> >>>> -- >>>> {+} Jeff Squyres >>>> >>>> To unsubscribe from this group and stop receiving emails from it, send an >>>> email to [email protected]. >>> >>> >>> >>> To unsubscribe from this group and stop receiving emails from it, send an >>> email to [email protected]. >>> >> >> >> -- >> {+} Jeff Squyres > > > > > > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected] > <mailto:[email protected]>. > > > > -- > {+} Jeff Squyres > > > > > > > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected] > <mailto:[email protected]>. > <config.log.gz><ompi_info_param.log.gz> -- {+} Jeff Squyres To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
