Update TBB to version v2021.1.1

[fwyzard]

See the release notes for more information.

Issues in the v2021.1.1 release (alrady resolved in the master branch):

• patch tbb's CMakeFile.txt to find the hwloc library and enable support for NUMA-aware task_arena.
• disable the deprecated copy warning (-Wdeprecated-copy) from GCC 9.

[cmsbuild]

A new Pull Request was created by @fwyzard (Andrea Bocci) for branch IB/CMSSW_11_3_X/master.

@cmsbuild, @smuzaffar, @mrodozov can you please review it and eventually sign? Thanks.
cms-bot commands are listed here

[fwyzard]

please test

[fwyzard]

@smuzaffar @mrodozov just to check if the v2021.1.1 release is more friendly, or fails like the master branch.

[cmsbuild]

-1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-bd605d/12235/summary.html
COMMIT: 76f6257
CMSSW: CMSSW_11_3_X_2021-01-11-2300/slc7_amd64_gcc900

External Build

I found compilation warning when building: See details on the summary page.

[fwyzard]

mhm... ERROR: unsuccessful bootstrap is hopefully unrelated ?

[fwyzard]

please test

[cmsbuild]

-1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-bd605d/12236/summary.html
COMMIT: 76f6257
CMSSW: CMSSW_11_3_X_2021-01-11-2300/slc7_amd64_gcc900

External Build

I found compilation error when building:

then mv -f ".deps/DefTable.Tpo" ".deps/DefTable.Po"; else rm -f ".deps/DefTable.Tpo"; exit 1; fi
if g++ -DHAVE_CONFIG_H -I. -I. -I../../include/HepPDT -I../../include -I../../include  -I/data/cmsbld/jenkins/workspace/ib-run-pr-tests/testBuildDir/slc7_amd64_gcc900/external/tbb/v2021.1.1-6f94df567e79f364c99ad4d828ddb6b1/include -O -ansi -pedantic -Wall -D_GNU_SOURCE -O2 -std=c++14 -MT HeavyIonUnknownID.o -MD -MP -MF ".deps/HeavyIonUnknownID.Tpo" -c -o HeavyIonUnknownID.o HeavyIonUnknownID.cc; \
then mv -f ".deps/HeavyIonUnknownID.Tpo" ".deps/HeavyIonUnknownID.Po"; else rm -f ".deps/HeavyIonUnknownID.Tpo"; exit 1; fi
In file included from HeavyIonUnknownID.cc:8:
../../include/HepPDT/ParticleDataTable.hh: In member function 'size_t HepPDT::ParticleDataTable::PIDhash::operator()(const HepPDT::ParticleID&) const':
../../include/HepPDT/ParticleDataTable.hh:52:19: error: 'tbb_hasher' is not a member of 'tbb'
52 |       return tbb::tbb_hasher(p.pid());
|                   ^~~~~~~~~~
make[1]: *** [HeavyIonUnknownID.o] Error 1
make[1]: Leaving directory `/data/cmsbld/jenkins/workspace/ib-run-pr-tests/testBuildDir/BUILD/slc7_amd64_gcc900/external/heppdt/3.03.00-bbd2cf88c6760dea82f170b4c3941f8a/heppdt-3.03.00/src/HepPDT'
make: *** [all-recursive] Error 1


[fwyzard]

OK, looks like v2021.1.1 and the master branch have the same issue for our code base.

[mrodozov]

ROOT already have an issue about this incompatibility.
For reference
root-project/root#6933

[fwyzard]

OK, so we can make the switch only after ROOT 6.24.00 ?

[smuzaffar]

@mrodozov , are there any other externals ( except heppdt and root) which fail due to newer TBB?

[mrodozov]

yes, at least mkfit which I'm reading what's wrong with (I think it was the hasher)
edit: its the tbb/task_scheduler_init.h as in ROOT.

[mrodozov]

https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-bd605d/12236/external/mkfit/2.0.1-2f404f8f229ee770f473d6fb2e1cbde5/log

[fwyzard]

Yes, looks like the same issue:

make: *** No rule to make target 'tbb/task_scheduler_init.h', needed by 'main.o'. Stop.

I guess this we can send back to the mkFit team to fix ?

[mrodozov]

In this document:
https://software.intel.com/sites/default/files/managed/b2/d2/TBBRevamp.pdf
Intel described substitutions to be used. I can see the tbb::hash -> std::hash and what to use for tbb::task_scheduler_init ,
but not how 😅

[fwyzard]

Yes... well, I meant that since mkFit is a(n external) package developed by CMS people, they could do the work of updating it to the current version of TBB.

[mrodozov]

I've submitted a request (issue 21 on their github)

[mrodozov]

Dan Riley already fixed that.
trackreco/mkFit#286
he just backported it trackreco/mkFit#291

[mrodozov]

mkfit builds using 27f790c3edc96630d7c658f7d4d4b86389986488 from devel.
new official tag is coming. when they tag it we can get it. I asked ROOT what they plan to do

[mrodozov]

ROOT are backporting the fix for from master to 6.22

[mrodozov]

ROOT just opened the 6.22 backport
root-project/root#7065

[smuzaffar]

@mrodozov , can we open a PR for cms-sw/root cms/6.22 branch to already start testing it?

[smuzaffar]

please test with root-project/root#7065
let's see if bot can directly test it :-)

[smuzaffar]

looks like not, bot only recognize repos from our organizations i.e. cms-sw, cms-data and cms-external

[mrodozov]

looks like not, bot only recognize repos from our organizations i.e. cms-sw, cms-data and cms-external

I'm opening a normal 6.22 update + PR on cms-sw/root for that

[smuzaffar]

closing in favor of #6792