JuliaLang / julia

The Julia Programming Language
https://julialang.org/
MIT License
45.73k stars 5.48k forks source link

Fail to compile mbedtls for julia 0.5.0-rc1 #17920

Closed ronisbr closed 8 years ago

ronisbr commented 8 years ago

Hi guys!

I'm trying to pack julia 0.5.0-rc1, but the mbedtls is failing with this message:

[   51s] CMake Error at library/cmake_install.cmake:144 (file):
[   51s]   file RPATH_CHANGE could not write new RPATH:
[   51s] 
[   51s]     $ORIGIN
[   51s] 
[   51s]   to the file:
[   51s] 
[   51s]     /home/abuild/rpmbuild/BUILD/julia-0.5.0-rc1/usr-staging/mbedtls-2.3.0/home/abuild/rpmbuild/BUILD/julia-0.5.0-rc1/usr/lib/libmbedcrypto.so.2.3.0
[   51s] 
[   51s]   No valid ELF RPATH or RUNPATH entry exists in the file;
[   51s] Call Stack (most recent call first):
[   51s]   cmake_install.cmake:37 (include)
[   51s] 
[   51s] 
[   51s] make[2]: *** [Makefile:130: install] Error 1

Can anyone help me?

Keno commented 8 years ago

Seems like CMake is having trouble rewriting the RPATH? For some libraries we do that manually using patchelf. I guess we could do the same for MbedTLS, though it is surprising that no rpath is being set in the first place. Can you look in the full compile commands if that's just not being passed or if the compiler is just not adding it for some reason?

ronisbr commented 8 years ago

Where can I find this?

ronisbr commented 8 years ago

Maybe here:

/usr/bin/gcc -m64 -fPIC  -Wall -Wextra -W -Wdeclaration-after-statement -Wwrite-strings -Wlogical-op -Wshadow -Wmissing-declarations -Wmissing-prototypes -O2  -shared -Wl,-soname,libmbedcrypto.so.0 -o libmbedcrypto.so.2.3.0 CMakeFiles/mbedcrypto.dir/aes.c.o CMakeFiles/mbedcrypto.dir/aesni.c.o CMakeFiles/mbedcrypto.dir/arc4.c.o CMakeFiles/mbedcrypto.dir/asn1parse.c.o CMakeFiles/mbedcrypto.dir/asn1write.c.o CMakeFiles/mbedcrypto.dir/base64.c.o CMakeFiles/mbedcrypto.dir/bignum.c.o CMakeFiles/mbedcrypto.dir/blowfish.c.o CMakeFiles/mbedcrypto.dir/camellia.c.o CMakeFiles/mbedcrypto.dir/ccm.c.o CMakeFiles/mbedcrypto.dir/cipher.c.o CMakeFiles/mbedcrypto.dir/cipher_wrap.c.o CMakeFiles/mbedcrypto.dir/ctr_drbg.c.o CMakeFiles/mbedcrypto.dir/des.c.o CMakeFiles/mbedcrypto.dir/dhm.c.o CMakeFiles/mbedcrypto.dir/ecdh.c.o CMakeFiles/mbedcrypto.dir/ecdsa.c.o CMakeFiles/mbedcrypto.dir/ecjpake.c.o CMakeFiles/mbedcrypto.dir/ecp.c.o CMakeFiles/mbedcrypto.dir/ecp_curves.c.o CMakeFiles/mbedcrypto.dir/entropy.c.o CMakeFiles/mbedcrypto.dir/entropy_poll.c.o CMakeFiles/mbedcrypto.dir/error.c.o CMakeFiles/mbedcrypto.dir/gcm.c.o CMakeFiles/mbedcrypto.dir/havege.c.o CMakeFiles/mbedcrypto.dir/hmac_drbg.c.o CMakeFiles/mbedcrypto.dir/md.c.o CMakeFiles/mbedcrypto.dir/md2.c.o CMakeFiles/mbedcrypto.dir/md4.c.o CMakeFiles/mbedcrypto.dir/md5.c.o CMakeFiles/mbedcrypto.dir/md_wrap.c.o CMakeFiles/mbedcrypto.dir/memory_buffer_alloc.c.o CMakeFiles/mbedcrypto.dir/oid.c.o CMakeFiles/mbedcrypto.dir/padlock.c.o CMakeFiles/mbedcrypto.dir/pem.c.o CMakeFiles/mbedcrypto.dir/pk.c.o CMakeFiles/mbedcrypto.dir/pk_wrap.c.o CMakeFiles/mbedcrypto.dir/pkcs12.c.o CMakeFiles/mbedcrypto.dir/pkcs5.c.o CMakeFiles/mbedcrypto.dir/pkparse.c.o CMakeFiles/mbedcrypto.dir/pkwrite.c.o CMakeFiles/mbedcrypto.dir/platform.c.o CMakeFiles/mbedcrypto.dir/ripemd160.c.o CMakeFiles/mbedcrypto.dir/rsa.c.o CMakeFiles/mbedcrypto.dir/sha1.c.o CMakeFiles/mbedcrypto.dir/sha256.c.o CMakeFiles/mbedcrypto.dir/sha512.c.o CMakeFiles/mbedcrypto.dir/threading.c.o CMakeFiles/mbedcrypto.dir/timing.c.o CMakeFiles/mbedcrypto.dir/version.c.o CMakeFiles/mbedcrypto.dir/version_features.c.o CMakeFiles/mbedcrypto.dir/xtea.c.o -lz -Wl,-rpath,::::::: 
Keno commented 8 years ago

Yep -Wl,-rpath,:::::::, so it looks like it somehow messes up computing what the RPATH should be.

ronisbr commented 8 years ago

This patch allowed me to compile mbedtls:

Index: julia-0.5.0-rc1/deps/mbedtls.mk
===================================================================
--- julia-0.5.0-rc1.orig/deps/mbedtls.mk
+++ julia-0.5.0-rc1/deps/mbedtls.mk
@@ -23,7 +23,7 @@ MBEDTLS_OPTS += -DENABLE_ZLIB_SUPPORT=ON
 endif

 ifeq ($(OS),Linux)
-MBEDTLS_OPTS += -DCMAKE_INSTALL_RPATH="\$$ORIGIN"
+MBEDTLS_OPTS += -DCMAKE_SKIP_BUILD_RPATH=FALSE -DCMAKE_BUILD_WITH_INSTALL_RPATH=FALSE -DCMAKE_INSTALL_RPATH="\$$ORIGIN"
 endif

 $(SRCDIR)/srccache/$(MBEDTLS_SRC).tgz: | $(SRCDIR)/srccache

However, I have no idea if this is the right approach.

Keno commented 8 years ago

Does the resulting binary have an RPATH set (~/julia/usr/tools/patchelf --print-rpath or objdump -x to check)

ronisbr commented 8 years ago

I need to check this in which file?

ronisbr commented 8 years ago

If you mean julia executable, then:

patchelf --print-rpath ./julia-0.5-0.5.0.rc1-0.x86_64/usr/bin/julia                                                                                                                                  17:29:12 
$ORIGIN/../lib64/julia:$ORIGIN/../lib64
tkelman commented 8 years ago

I think Keno meant the mbedcrypto library. What distro and cmake version is this? If you're building a package against system copies of libgit2, you might not need mbedtls at all if the libgit2 you're building against supports https and proxies.

ronisbr commented 8 years ago

@tkelman, I'm pretty sure that libgit2 in openSUSE has support to https and proxies. Hence, what should I do? Set USE_SYSTEM_MBEDTLS=1 and not install mbedtls?

tkelman commented 8 years ago

I don't know whether USE_SYSTEM_MBEDTLS=1 by itself will work right now. USE_SYSTEM_LIBGIT2=1 if you have a new enough version should be able to avoid building any of mbedtls, libssh2, or libcurl, but the makefile logic may need adjusting to make that actually happen correctly.

ronisbr commented 8 years ago

Without USE_SYSTEM_MBEDTLS=1, the build process tries to download the tarball.

tkelman commented 8 years ago

ref https://github.com/JuliaLang/julia/issues/17884

ronisbr commented 8 years ago

Good! With USE_SYSTEM_MBEDTLS=1 julia builds fine. I'll send to build farm, and when it is completed, I will run the tests. Thanks for the help @tkelman and @Keno

ronisbr commented 8 years ago

The runtest.jl fails:

# julia runtests.jl 
    From worker 7:       * linalg/special        in   4.15 seconds, maxrss  215.82 MB
    From worker 6:       * linalg/schur          in  15.78 seconds, maxrss  259.24 MB
    From worker 7:       * linalg/svd            in  32.70 seconds, maxrss  295.14 MB
    From worker 6:       * linalg/lapack         in  26.38 seconds, maxrss  298.28 MB
    From worker 9:       * linalg/bunchkaufman   in  42.31 seconds, maxrss  260.75 MB
    From worker 4:       * linalg/dense          in  47.39 seconds, maxrss  316.79 MB
    From worker 8:       * linalg/eigen          in  50.50 seconds, maxrss  298.29 MB
    From worker 4:       * linalg/pinv           in   4.94 seconds, maxrss  405.73 MB
    From worker 8:       * linalg/givens         in   3.63 seconds, maxrss  302.54 MB
    From worker 7:       * linalg/tridiag        in  21.90 seconds, maxrss  321.18 MB
    From worker 6:       * linalg/bidiag         in  25.34 seconds, maxrss  355.48 MB
    From worker 3:       * linalg/qr             in  71.71 seconds, maxrss  339.54 MB
    From worker 7:       * linalg/symmetric      in  13.57 seconds, maxrss  343.98 MB
    From worker 6:       * linalg/generic        in   5.46 seconds, maxrss  359.21 MB
    From worker 6:       * linalg/hessenberg     in   0.64 seconds, maxrss  361.27 MB
    From worker 3:       * linalg/uniformscaling in   4.21 seconds, maxrss  347.07 MB
    From worker 9:       * linalg/diagonal       in  37.84 seconds, maxrss  328.90 MB
WARNING: Method definition f(Tuple{Vararg{Int64, #N<:Any}}, AbstractArray{#T<:Any, #N<:Any}) in module Main at /usr/share/julia/test/core.jl:706 overwritten at /usr/share/julia/test/core.jl:712.
    From worker 4:       * linalg/cholesky       in  31.72 seconds, maxrss  417.71 MB
    From worker 4:       * inference             in   0.45 seconds, maxrss  417.81 MB
    From worker 4:       * keywordargs           in   0.98 seconds, maxrss  417.97 MB
WARNING: Method definition f() in module JLCall14301 at /usr/share/julia/test/core.jl:3529 overwritten at /usr/share/julia/test/core.jl:3539.
    From worker 8:       * linalg/lu             in  36.62 seconds, maxrss  359.25 MB
    From worker 6:       * linalg/arnoldi        in  22.24 seconds, maxrss  424.84 MB
    From worker 6:       * char                  in   0.55 seconds, maxrss  426.89 MB
    From worker 9:       * core                  in  19.39 seconds, maxrss  350.79 MB
    From worker 9:       * triplequote           in   0.01 seconds, maxrss  350.79 MB
    From worker 8:       * printf                in  11.04 seconds, maxrss  365.68 MB
    From worker 9:       * unicode               in   3.38 seconds, maxrss  366.82 MB
    From worker 6:       * string                in   7.74 seconds, maxrss  432.71 MB
    From worker 6:       * hashing               in   3.71 seconds, maxrss  435.04 MB
    From worker 5:       * linalg/matmul         in 108.17 seconds, maxrss  310.10 MB
    From worker 6:       * iobuffer              in   0.53 seconds, maxrss  435.07 MB
    From worker 5:       * staged                in   0.62 seconds, maxrss  310.67 MB
    From worker 9:       * dict                  in   7.28 seconds, maxrss  377.04 MB
    From worker 9:       * tuple                 in   1.18 seconds, maxrss  380.23 MB
WARNING: could not attach metadata for @simd loop.
    From worker 9:       * reduce                in   3.01 seconds, maxrss  402.70 MB
    From worker 9:       * reducedim             in  10.57 seconds, maxrss  419.09 MB
    From worker 8:       * dates                 in  26.04 seconds, maxrss  431.94 MB
    From worker 4:       * numbers               in  43.74 seconds, maxrss  516.32 MB
    From worker 4:       * intfuncs              in   0.43 seconds, maxrss  516.32 MB
    From worker 4:       * simdloop              in   0.91 seconds, maxrss  516.32 MB
    From worker 4:       * vecelement            in   1.29 seconds, maxrss  516.32 MB
    From worker 9:       * random                in   8.34 seconds, maxrss  452.50 MB
    From worker 7:       * linalg/lq             in  62.31 seconds, maxrss  398.43 MB
    From worker 6:       * offsetarray           in  31.02 seconds, maxrss  490.89 MB
    From worker 6:       * copy                  in   1.00 seconds, maxrss  491.06 MB
    From worker 5:       * arrayops              in  34.68 seconds, maxrss  473.50 MB
    From worker 5:       * fastmath              in   3.31 seconds, maxrss  473.50 MB
    From worker 4:       * blas                  in  23.90 seconds, maxrss  546.77 MB
    From worker 4:       * operators             in   0.41 seconds, maxrss  546.86 MB
    From worker 4:       * path                  in   3.07 seconds, maxrss  551.50 MB
    From worker 4:       * ccall                 in   3.40 seconds, maxrss  553.09 MB
    From worker 4:       * parse                 in   1.39 seconds, maxrss  553.40 MB
    From worker 4:       * loading               in   0.05 seconds, maxrss  553.48 MB
    From worker 4:       * bigint                in   1.01 seconds, maxrss  553.80 MB
    From worker 4:       * bigfloat              in   0.03 seconds, maxrss  553.81 MB
    From worker 5:       * functional            in  20.05 seconds, maxrss  473.50 MB
    From worker 6:       * math                  in  32.17 seconds, maxrss  528.79 MB
    From worker 6:       * spawn                       [stdio passthrough ok]
    From worker 5:       * statistics            in   6.70 seconds, maxrss  473.50 MB
    From worker 5:       * backtrace                From worker 8:       * abstractarray         in  47.32 seconds, maxrss  489.87 MB
    From worker 8:       * priorityqueue         in   1.02 seconds, maxrss  505.51 MB
    From worker 6:   in  15.94 seconds, maxrss  534.40 MB
WARNING: file permissions tests skipped due to running tests as root (not recommended)
WARNING: file permissions tests skipped due to running tests as root (not recommended)
    From worker 10:      * read                  in  12.55 seconds, maxrss  293.01 MB
    From worker 10:      * version               in   0.74 seconds, maxrss  294.03 MB
    From worker 10:      * resolve               in   2.58 seconds, maxrss  305.36 MB
    From worker 4:       * sorting               in  30.02 seconds, maxrss  613.45 MB
    From worker 10:      * pollfd                in   3.41 seconds, maxrss  307.42 MB
    From worker 6:       * mmap                  in   7.84 seconds, maxrss  534.42 MB
    From worker 4:       * mpfr                  in   1.46 seconds, maxrss  613.73 MB
    From worker 4:       * socket                in   2.40 seconds, maxrss  614.71 MB
    From worker 4:       * floatapprox           in   0.19 seconds, maxrss  615.35 MB
    From worker 6:       * complex               in   4.82 seconds, maxrss  534.42 MB
    From worker 4:       * datafmt               in   4.94 seconds, maxrss  625.24 MB
    From worker 4:       * regex                 in   0.31 seconds, maxrss  626.02 MB
    From worker 6:       * reflection            in   3.51 seconds, maxrss  534.42 MB
    From worker 6:       * combinatorics         in   0.28 seconds, maxrss  534.42 MB
    From worker 4:       * float16               in   0.63 seconds, maxrss  627.14 MB
    From worker 6:       * sysinfo               in   0.57 seconds, maxrss  534.42 MB
    From worker 4:       * rounding              in   0.46 seconds, maxrss  628.78 MB
    From worker 4:       * mod2pi                in   0.11 seconds, maxrss  628.88 MB
    From worker 4:       * euler                 in   0.54 seconds, maxrss  638.28 MB
    From worker 8:       * file                  in  36.00 seconds, maxrss  516.92 MB
    From worker 8:       * lineedit              in   2.09 seconds, maxrss  516.92 MB
    From worker 10:      * broadcast             in  19.63 seconds, maxrss  325.60 MB
    From worker 8:       * replcompletions       in   2.59 seconds, maxrss  516.92 MB
    From worker 4:       * show                  in  12.42 seconds, maxrss  669.59 MB
    From worker 10:      * repl                  in   3.33 seconds, maxrss  330.26 MB
    From worker 8:       * replutil             Test Failed
    From worker 8:    Expression: contains(err_str,"Cannot raise an integer x to a negative power -n")
    From worker 6:       * ranges                in  14.79 seconds, maxrss  536.28 MB
    From worker 6:       * goto                  in   0.03 seconds, maxrss  536.28 MB
    From worker 6:       * llvmcall              in   0.09 seconds, maxrss  536.40 MB
    From worker 4:       * sets                  in   2.06 seconds, maxrss  671.77 MB
    From worker 6:       * grisu                 in   1.65 seconds, maxrss  540.54 MB
    From worker 6:       * meta                  in   0.21 seconds, maxrss  542.66 MB
    From worker 11:      * stacktraces           in   1.06 seconds, maxrss  215.05 MB
    From worker 10:      * test                  in   5.37 seconds, maxrss  340.32 MB
    From worker 6:       * profile               in   2.69 seconds, maxrss  550.80 MB
    From worker 2:       * linalg/triangular     in 225.88 seconds, maxrss  681.95 MB
    From worker 2:       * base64                in   0.20 seconds, maxrss  682.11 MB
    From worker 10:      * docs                  in   2.27 seconds, maxrss  354.54 MB
    From worker 4:       * nullable              in   7.07 seconds, maxrss  676.41 MB
    From worker 6:       * markdown              in   2.84 seconds, maxrss  551.36 MB
Warning: threaded loop executed in order
    From worker 6:       * enums                 in   1.36 seconds, maxrss  554.39 MB
    From worker 2:       * serialize             in   3.96 seconds, maxrss  686.52 MB
    From worker 2:       * i18n                  in   0.01 seconds, maxrss  686.52 MB
    From worker 4:       * threads               in   2.53 seconds, maxrss  685.72 MB
    From worker 4:       * libdl                 in   0.38 seconds, maxrss  685.72 MB
    From worker 2:       * workspace             in   0.53 seconds, maxrss  686.52 MB
    From worker 11:      * libgit2               in   8.88 seconds, maxrss  247.88 MB
    From worker 4:       * int                   in   1.86 seconds, maxrss  685.72 MB
    From worker 11:      * intset                in   0.41 seconds, maxrss  247.98 MB
    From worker 10:      * misc                  in   7.13 seconds, maxrss  379.36 MB
    From worker 10:      * inline                in   0.13 seconds, maxrss  379.37 MB
    From worker 4:       * floatfuncs            in   1.83 seconds, maxrss  685.72 MB
    From worker 4:       * error                 in   0.64 seconds, maxrss  685.72 MB
    From worker 10:      * boundscheck           in   1.66 seconds, maxrss  379.39 MB
    From worker 10:      * cartesian             in   0.01 seconds, maxrss  379.39 MB
WARNING: Method definition ambig(Any, Integer) in module Main at /usr/share/julia/test/ambiguous.jl:7 overwritten at /usr/share/julia/test/ambiguous.jl:79.
    From worker 4:       * ambiguous            Skipping Base.<|
    From worker 4:  Skipping Base.>:
    From worker 4:  Skipping Base.active_repl
    From worker 4:  Skipping Base.active_repl_backend
    From worker 4:   in   0.91 seconds, maxrss  685.72 MB
    From worker 2:       * checked               in   7.81 seconds, maxrss  702.62 MB
    From worker 7:       * bitarray              in 104.68 seconds, maxrss  444.86 MB
    From worker 4:       * dsp                   in   7.33 seconds, maxrss  689.88 MB
    From worker 10:      * fft                   in  13.05 seconds, maxrss  403.82 MB
    From worker 11:      * parallel              in  50.31 seconds, maxrss  248.05 MB
    From worker 9:       * sparse                in 176.08 seconds, maxrss  715.23 MB
    From worker 3:       * subarray              in 247.49 seconds, maxrss  814.52 MB
    From worker 6:       * cmdlineargs          Test Failed
    From worker 6:    Expression: contains(bt,"in include_from_node1")
Exception running test backtrace :
On worker 5:
LoadError: BoundsError: attempt to access svec()
  at index [1]
 in macro expansion; at /usr/share/julia/test/backtrace.jl:7 [inlined]
 in anonymous at ./<missing>:?
 in macro expansion at ./util.jl:226 [inlined]
 in runtests at /usr/share/julia/test/testdefs.jl:7
 in #16 at /usr/share/julia/test/runtests.jl:44
 in #503 at ./multi.jl:1410
 in run_work_thunk at ./multi.jl:996
 in macro expansion at ./multi.jl:1410 [inlined]
 in #502 at ./event.jl:46
while loading /usr/share/julia/test/backtrace.jl, in expression starting on line 5
Exception running test replutil :
On worker 8:
LoadError: There was an error during testing
 in record at ./test.jl:397
 in do_test at ./test.jl:281
 in macro expansion at ./util.jl:226 [inlined]
 in runtests at /usr/share/julia/test/testdefs.jl:7
 in #16 at /usr/share/julia/test/runtests.jl:44
 in #503 at ./multi.jl:1410
 in run_work_thunk at ./multi.jl:996
 in macro expansion at ./multi.jl:1410 [inlined]
 in #502 at ./event.jl:46
while loading /usr/share/julia/test/replutil.jl, in expression starting on line 261
Exception running test cmdlineargs :
On worker 6:
LoadError: There was an error during testing
 in record at ./test.jl:397
 in do_test at ./test.jl:281
 in macro expansion; at /usr/share/julia/test/cmdlineargs.jl:299 [inlined]
 in anonymous at ./<missing>:?
 in macro expansion at ./util.jl:226 [inlined]
 in runtests at /usr/share/julia/test/testdefs.jl:7
 in #16 at /usr/share/julia/test/runtests.jl:44
 in #503 at ./multi.jl:1410
 in run_work_thunk at ./multi.jl:996
 in macro expansion at ./multi.jl:1410 [inlined]
 in #502 at ./event.jl:46
while loading /usr/share/julia/test/cmdlineargs.jl, in expression starting on line 296
ERROR: LoadError: Some tests exited with errors.
 in (::##11#19)() at /usr/share/julia/test/runtests.jl:72
 in cd(::##11#19, ::String) at ./file.jl:59
while loading /usr/share/julia/test/runtests.jl, in expression starting on line 26

But I think it is not related to this problem, right?

ronisbr commented 8 years ago

I opened a new bug #17941 I think we can close this issue since I can now build julia, right?

tkelman commented 8 years ago

What distro and cmake version is this?

And what was rpath on the mbedcrypto library? We would like to resolve the issue you were having with source builds if it's reproducible with a standard-configuration make in your environment.

ronisbr commented 8 years ago

@tkelman it turns out that mbedtls was not built. The source was not even unpacked. I have no idea what happened, but I could not make it build even if the patch I proposed. I think that the build process (openSUSE) added the libgit2 and julia skipped mbedtls. As of now, I cannot build mbedtls together with julia.

tkelman commented 8 years ago

Are you setting anything in Make.user or on the make command line, or do you even have this problem in a plain git clone https://github.com/JuliaLang/julia && make -C julia ?

ronisbr commented 8 years ago

@tkelman actually I'm trying to make a package for openSUSE. Here is what I'm using for the make command line:

NO_GIT=1 \
USE_SYSTEM_LLVM=0 \
USE_SYSTEM_LIBUNWIND=1 \
USE_SYSTEM_READLINE=1 \
USE_SYSTEM_PCRE=1 \
USE_SYSTEM_LIBM=0 \
USE_SYSTEM_OPENLIBM=1 \
USE_SYSTEM_OPENSPECFUN=1 \
USE_SYSTEM_DSFMT=1 \
USE_SYSTEM_BLAS=1 \
LIBBLASNAME=libblas.so.3 \
USE_SYSTEM_LAPACK=1 \
LIBBLASNAME=liblapack.so.3 \
USE_SYSTEM_FFTW=1 \
LIBFFTWNAME=libfftw3_threads.so.3 \
LIBFFTWFNAME=libfftw3f_threads.so.3 \
USE_SYSTEM_GMP=1 \
USE_SYSTEM_MPFR=1 \
USE_SYSTEM_ARPACK=1 \
USE_SYSTEM_SUITESPARSE=1 \
USE_SYSTEM_ZLIB=1 \
USE_SYSTEM_GRISU=1 \
USE_SYSTEM_RMATH=0 \
USE_SYSTEM_LIBUV=0 \
USE_SYSTEM_UTF8PROC=0 \
USE_SYSTEM_PATCHELF=1 \
USE_SYSTEM_LIBGIT2=1 \
USE_SYSTEM_MBEDTLS=1 \
USE_SYSTEM_LIBSSH2=1 \
USE_MKL=0 \
USE_BLAS64=0 \
USE_LLVM_SHLIB=0 \
VERBOSE=1 \
USECLANG=0
ronisbr commented 8 years ago

For the records, if USE_SYSTEM_MBEDTLS=1, then I can build julia even not installing mbedtls at all.

tkelman commented 8 years ago

In that case I'll close this as a duplicate of https://github.com/JuliaLang/julia/issues/17884. That kind of information would be best to include with the initial bug report next time.

ronisbr commented 8 years ago

@tkelman sorry, I didn't understand. I just discovered that I can build julia without mbedtls after I opened this bug report.

tkelman commented 8 years ago

Here is what I'm using for the make command line:

That list would have been valuable to include when you initially opened the issue is all, so we can have as much information as possible to understand what's going on and correctly diagnose the problem.

ronisbr commented 8 years ago

Ah! Ok :) sorry