poudriere reported host amd64 vs x86_64 #645
Comments
|
can I ping this issue: |
|
|
bdrewery
added
the
Ports Issue
I tried this but that did not work. The $(ARCH) variable is referenced many different times throughout the build. Even setting ARCH from within the ports Makefile this was not enough. If I set ARCH from the makefile the first check would pass but then it would fail when trying to match |
|
This issue keeps cropping up and I am hoping that you guys can offer a fix. Here's another example: math/scalapack has a SLmake.in.example file; contents ############################################################################
#
# Program: ScaLAPACK
#
# Module: SLmake.inc
#
# Purpose: Top-level Definitions
#
# Creation date: February 15, 2000
#
# Modified: October 13, 2011
#
# Send bug reports, comments or suggestions to scalapack@cs.utk.edu
#
############################################################################
#
# C preprocessor definitions: set CDEFS to one of the following:
#
# -DNoChange (fortran subprogram names are lower case without any suffix)
# -DUpCase (fortran subprogram names are upper case without any suffix)
# -DAdd_ (fortran subprogram names are lower case with "_" appended)
CDEFS = -DAdd_
#
# The fortran and C compilers, loaders, and their flags
#
FC = mpif90
CC = mpicc
NOOPT = -O0
FCFLAGS = -O3
CCFLAGS = -O3
FCLOADER = $(FC)
CCLOADER = $(CC)
FCLOADFLAGS = $(FCFLAGS)
CCLOADFLAGS = $(CCFLAGS)
#
# The archiver and the flag(s) to use when building archive (library)
# Also the ranlib routine. If your system has no ranlib, set RANLIB = echo
#
ARCH = ar
ARCHFLAGS = cr
RANLIB = ranlib
#
# The name of the ScaLAPACK library to be created
#
SCALAPACKLIB = libscalapack.a
#
# BLAS, LAPACK (and possibly other) libraries needed for linking test programs
#
BLASLIB = -lblas
LAPACKLIB = -llapack
LIBS = $(LAPACKLIB) $(BLASLIB)You can see ARCH defined there running poudriere testport the jail starts up, build is going just fine and then failure mpif90 -c -O3 pzunmr2.f
mpif90 -c -O3 pzunmrq.f
mpif90 -c -O3 pzunmtr.f
mpif90 -c -O3 pzlaevswp.f
mpif90 -c -O3 pzlarzb.f
mpif90 -c -O3 pzlarzt.f
mpif90 -c -O3 pzlarz.f
mpif90 -c -O3 pzlarzc.f
mpif90 -c -O3 pzlatrz.f
mpif90 -c -O3 pztzrzf.f
mpif90 -c -O3 pzunmr3.f
mpif90 -c -O3 pzunmrz.f
mpif90 -c -O3 pzmax1.f
mpif90 -c -O3 pdzsum1.f
mpif90 -c -O3 pzlamr1d.f
mpif90 -c -O3 zdbtf2.f
mpif90 -c -O3 zdbtrf.f
mpif90 -c -O3 zdttrf.f
mpif90 -c -O3 zdttrsv.f
mpif90 -c -O3 zpttrsv.f
mpif90 -c -O3 zsteqr2.f
mpif90 -c -O3 ztrmvt.f
mpif90 -c -O3 zlamsh.f
mpif90 -c -O3 zlaref.f
mpif90 -c -O3 zlanv2.f
mpif90 -c -O3 zlahqr2.f
mpif90 -c -O3 pzheevr.f
amd64 cr ../libscalapack.a psdbsv.o psdbtrf.o psdbtrs.o psdbtrsv.o psdtsv.o psdttrf.o psdttrs.o psdttrsv.o psgbsv.o psgbtrf.o psgbtrs.o psgebd2.o psgebrd.o psgecon.o psgeequ.o psgehd2.o psgehrd.o psgelq2.o psgelqf.o psgels.o psgeql2.o psgeqlf.o psgeqpf.o psgeqr2.o psgeqrf.o psgerfs.o psgerq2.o psgerqf.o psgesv.o psgesvd.o psgesvx.o psgetf2.o psgetrf.o psgetri.o psgetrs.o psggqrf.o psggrqf.o pslabrd.o pslacon.o pslacp2.o pslacpy.o pslahrd.o pslange.o pslanhs.o pslansy.o pslantr.o pslapiv.o pslapv2.o pslaqge.o pslaqsy.o pslarf.o pslarfb.o pslarfg.o pslarft.o pslase2.o pslaset.o pslascl.o pslassq.o pslaswp.o pslatra.o pslatrd.o pslatrs.o pslauu2.o pslauum.o psorg2l.o psorg2r.o psorgl2.o psorglq.o psorgql.o psorgqr.o psorgr2.o psorgrq.o psorm2l.o psorm2r.o psormbr.o psormhr.o psorml2.o psormlq.o psormql.o psormqr.o psormr2.o psormrq.o psormtr.o pspocon.o pspbsv.o pspbtrf.o pspbtrs.o pspbtrsv.o psptsv.o pspttrf.o pspttrs.o pspttrsv.o pspoequ.o psporfs.o psposv.o psposvx.o pspotf2.o pspotrf.o pspotri.o pspotrs.o psrscl.o psstein.o pssyev.o pssyevd.o pssyevx.o pssygs2.o pssygst.o pssygvx.o pssyngst.o pssyntrd.o pssyttrd.o pssytd2.o pssytrd.o pstrti2.o pstrtri.o pstrtrs.o pslaevswp.o pslarzb.o pslarzt.o pslarz.o pslatrz.o pstzrzf.o psormr3.o psormrz.o pslahqr.o pslaconsb.o pslacp3.o pslawil.o pslasmsub.o pslared2d.o pslamr1d.o slaref.o slamsh.o slasorte.o ssteqr2.o sdbtf2.o sdbtrf.o sdttrf.o sdttrsv.o spttrsv.o strmvt.o pssyevr.o bslaapp.o bslaexc.o bstrexc.o pstrord.o pstrsen.o psgebal.o pshseqr.o pslamve.o pslaqr0.o pslaqr1.o pslaqr2.o pslaqr3.o pslaqr4.o pslaqr5.o psrot.o slaqr6.o pslabad.o pslaed0.o pslaed1.o pslaed2.o pslaed3.o pslaedz.o pslaiect.o pslamch.o pslared1d.o pslasrt.o psstebz.o psstedc.o slapst.o slasrt2.o sstein2.o slar1va.o slarrb2.o slarrd2.o slarre2.o slarre2a.o slarrf2.o slarrv2.o sstegr2.o sstegr2a.o sstegr2b.o slamov.o clamov.o \
pjlaenv.o pbchkvect.o getpbbuf.o pilaenvx.o piparmq.o pilaver.o pmpim2.o pmpcol.o
gmake[2]: amd64: Command not found
gmake[2]: *** [Makefile:196: single] Error 127
gmake[2]: *** Waiting for unfinished jobs....
gmake[2]: Leaving directory '/wrkdirs/usr/ports/math/scalapack/work/scalapack-2.0.2/SRC'
gmake[1]: *** [Makefile:68: scalapacklib] Error 2
gmake[1]: Leaving directory '/wrkdirs/usr/ports/math/scalapack/work/scalapack-2.0.2'
===> Compilation failed unexpectedly.
Try to set MAKE_JOBS_UNSAFE=yes and rebuild before reporting the failure to
the maintainer.
*** Error code 1You can clearly see that the ${AR} command is replaced by ARCH defined in /usr/ports/Mk/bsd.port.mk I could apply NO_ARCH everywhere but that just feels like pointing a rifle at my toes. Do you have any suggested fix for this? |
|
This seems like a bug in /usr/ports/Mk/bsd.port.mk # Get the architecture
.if !defined(ARCH)
ARCH!= ${UNAME} -p
.endif
HOSTARCH:= ${ARCH}
.if defined(CROSS_TOOLCHAIN)
ARCH= ${CROSS_TOOLCHAIN:C,-.*$,,}
.endif
_EXPORTED_VARS+= ARCHThey use ARCH as a temp variable to get the hostarch, cross tools and added it to exported vars. Maybe a fix is to unset ARCH after all this is completed? |
|
Like @bdrewery said, simply patch the software's Makefile to use another variable instead of |
This problem only happens in poudriere on the first attempt at poudriere testport; If you go interactive after the failure and run the exact same commands; the bug does not appear unless you go and set the env variable again. Also the $(ARCH) variable is used in almost every makefile that i've been working with recently, there's nothing that you guys are willing to do about this? |
|
Well, we are not willing to rename a variable that is used by base, ports and docs, so, you can patch the files, that's about it. |
It's not about patching out ARCH. It seems like when poudriere starts testport it doesn't go into an active shell because even after ${SETENV} ${MAKE_ENV} in any of the port stages I cannot get them to stick; Makefiles seems to get the wrong values but CMake does not seem to exhibit similar behavior. I'll put together a sample dummy port that exhibits the behavior and upload it here. |
This has bitten me a few times and I finally have a project that I can reliably reproduce it in.
I created an issue upstream but also creating one here to see if this is something that could get fixed here.
OpenMathLib/OpenBLAS#1796 (comment)
I can provide a port makefile that should reproduce the issue.
In this particular case if I start the poudriere jail, login and run gmake in the src directory, The env is properly detected and there's no problems.
If I do poudriere test port, it will fail. Using interactive mode after the failure and going to the source directory and running gmake the env is detected properly and the project gets built.
The problem only occurs when doing poudriere testport.
steps to test: git clone the openblas repo,
go to the src directory and run gmake; no problems.
host system: FreeBSD blubee 12.0-ALPHA5 FreeBSD 12.0-ALPHA5 #0 r338520: Fri Sep 7 23:31:46 CST 2018 root@:/usr/obj/usr/src/amd64.amd64/sys/GENERIC amd64
The text was updated successfully, but these errors were encountered: