Attempt to update ispc

[Luya Tshimbalanga]

Hello team,

Upstream ispc updated to 1.10.0.  Unfortunately, the build failed with
broken clang++ compiler according to koji  
(https://koji.fedoraproject.org/koji/taskinfo?taskID=32150411)

Can someone investigate the issue? This package is needed by embree, a
dependency for both Blender, luxcorerender and Yafaray.

Thanks

Luya

[Serge Guelton]

On Sun, Jan 20, 2019 at 11:36:22AM -0800, Luya Tshimbalanga wrote:
> Hello team,
>
> Upstream ispc updated to 1.10.0.  Unfortunately, the build failed with
> broken clang++ compiler according to koji  
> (https://koji.fedoraproject.org/koji/taskinfo?taskID=32150411)
>
> Can someone investigate the issue? This package is needed by embree, a
> dependency for both Blender, luxcorerender and Yafaray.
>
> Thanks
>
> Luya

Hi Luya,

Clang does not support the -fstack-clash-protection flag. We used to silently ignore that flag but it's no longer the case.
Why are you using clang to compile the package? The safe step is to use gcc, ot if clang is needed, to strip -fstack-clash-protection
from the flags passed to clang, but that's not future-proof (clang may end up supporting that flag).

Hope it helps,

Serge

[Luya Tshimbalanga]

On 2019-01-22 8:17 a.m., Serge Guelton wrote:
> Hi Luya,
>
> Clang does not support the -fstack-clash-protection flag. We used to silently ignore that flag but it's no longer the case.
> Why are you using clang to compile the package? The safe step is to use gcc, ot if clang is needed, to strip -fstack-clash-protection
> from the flags passed to clang, but that's not future-proof (clang may end up supporting that flag).
>
> Hope it helps,
>
> Serge

Upstream reason. Trying to build with gcc will fail. See the attacked
CMakeLists.txt from ispc-1.10.0

Luya

[Jonathan Wakely]

On 25/01/19 08:36 -0800, Luya Tshimbalanga wrote:
> On 2019-01-22 8:17 a.m., Serge Guelton wrote:
> > Hi Luya,
> >
> > Clang does not support the -fstack-clash-protection flag. We used to silently ignore that flag but it's no longer the case.
> > Why are you using clang to compile the package? The safe step is to use gcc, ot if clang is needed, to strip -fstack-clash-protection
> > from the flags passed to clang, but that's not future-proof (clang may end up supporting that flag).
> >
> > Hope it helps,
> >
> > Serge
>
> Upstream reason. Trying to build with gcc will fail. See the attacked
> CMakeLists.txt from ispc-1.10.0

That says it wants to build with clang, but doesn't say why building
with GCC will fail. Is it due to a GCC bug? Is it something we can
fix?

[Luya Tshimbalanga]

On 2019-01-25 9:02 a.m., Jonathan Wakely wrote:
> > Upstream reason. Trying to build with gcc will fail. See the attacked
> > CMakeLists.txt from ispc-1.10.0
>
> That says it wants to build with clang, but doesn't say why building
> with GCC will fail. Is it due to a GCC bug? Is it something we can
> fix?

My bad as I quickly wrote before leaving to work. Here is the updated
spec file and the result scratch build:

https://koji.fedoraproject.org/koji/taskinfo?taskID=32250196

Luya

[Dominik 'Rathann' Mierzejewski]

On Saturday, 26 January 2019 at 02:20, Luya Tshimbalanga wrote:
> On 2019-01-25 9:02 a.m., Jonathan Wakely wrote:
> > > Upstream reason. Trying to build with gcc will fail. See the attacked
> > > CMakeLists.txt from ispc-1.10.0
> >
> > That says it wants to build with clang, but doesn't say why building
> > with GCC will fail. Is it due to a GCC bug? Is it something we can
> > fix?
>
> My bad as I quickly wrote before leaving to work. Here is the updated
> spec file and the result scratch build:
>
> https://koji.fedoraproject.org/koji/taskinfo?taskID=32250196

Well, obviously existing CMakeLists file enforces the use of clang.
The developers seem to have a strong preference for clang. The issue
tracker has a couple of closed issues around this topic:

https://github.com/ispc/ispc/issues/1355
https://github.com/ispc/ispc/issues/1352

I would try patching out the clang requirement from CMakeLists
and building with system gcc.

Regards,
Dominik

[Luya Tshimbalanga]

On 2019-01-26 12:24 a.m., Dominik 'Rathann' Mierzejewski wrote:
> I would try patching out the clang requirement from CMakeLists
> and building with system gcc.
>
> Regards,
> Dominik

Done. Unfortunately, the build failed as the compiler still remains
clang and finding the cause is beyond my knowledge. Here is the resulted
scratch build:

https://koji.fedoraproject.org/koji/taskinfo?taskID=32277096

Luya

[Florian Weimer]

* Luya Tshimbalanga:
> On 2019-01-26 12:24 a.m., Dominik 'Rathann' Mierzejewski wrote:
> > I would try patching out the clang requirement from CMakeLists
> > and building with system gcc.
> >
> > Regards,
> > Dominik
>
> Done. Unfortunately, the build failed as the compiler still remains
> clang and finding the cause is beyond my knowledge. Here is the resulted
> scratch build:
>
> https://koji.fedoraproject.org/koji/taskinfo?taskID=32277096

Which sources are you building?  Nothing like this is on the master
branch on src.fedoraproject.org.

Thanks,
Florian

[Luya Tshimbalanga]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 2019-01-27 2:00 p.m., Florian Weimer wrote:
> Luya Tshimbalanga:
> > On 2019-01-26 12:24 a.m., Dominik 'Rathann' Mierzejewski wrote:
> > > I would try patching out the clang requirement from CMakeLists
> > > and building with system gcc.
> > >
> > > Regards,
> > > Dominik
> >
> > Done. Unfortunately, the build failed as the compiler still remains
> > clang and finding the cause is beyond my knowledge. Here is the resulted
> > scratch build:
> >
> > https://koji.fedoraproject.org/koji/taskinfo?taskID=32277096
>
> Which sources are you building? Nothing like this is on the master
> branch on src.fedoraproject.org.
>
> Thanks,
> Florian

I avoided committing the update as I preferred to do a scratch-build
hence the link from koji.
Realizing the change can be seen on the master branch on
src.fedoraproject.org, I applied the update for view:
https://src.fedoraproject.org/rpms/ispc/blob/master/f/ispc.spec

I figured that gcc parameter was missing on %make_build and the failure
occurred on python line despite the available dependency:
https://koji.fedoraproject.org/koji/taskinfo?taskID=32307602

Luya
> -----BEGIN PGP SIGNATURE-----
>
> iQEzBAEBCAAdFiEEWyB+BQtYiFz4GUNDXlKBdNiiYJoFAlxO1GoACgkQXlKBdNii
> YJpvLQgAoqqXkwQwMPwRjO6hQfETtcdU0R/OKJyaHXniTbFPf+stsNPR0YqKMWwD
> FuvGNq7S+KscQG9EgFFMnVc19LSWDPoA+9jQF3KkZJFyaI8THx/4ekgVKHz/DkFt
> lxL7UR3F++HWNwVfaBnJDOYm9+i8ZYRGMC8qw47i55NsfnuIMj5mNe30o0uvuIqP
> xRT077YDisgTE/7dq9dkvWDqLaosqA3hfrtZSmS4EtmCcBvJa++Af72HKkbvC15p
> SitFlMLxV46PwFDinxaSO0zyZQy/Su1nYRV6kGlA56ewEyGR6DJvgSd2IehFyo/F
> JNP1gM0ssc3EiKGzNd40ugHx7ZlnIw==
> =OZJV
> -----END PGP SIGNATURE-----

[Florian Weimer]

* Luya Tshimbalanga:
> On 2019-01-27 2:00 p.m., Florian Weimer wrote:
> > Luya Tshimbalanga:
> > > On 2019-01-26 12:24 a.m., Dominik 'Rathann' Mierzejewski wrote:
> > > > I would try patching out the clang requirement from CMakeLists
> > > > and building with system gcc.
> > > >
> > > > Regards,
> > > > Dominik
> > >
> > > Done. Unfortunately, the build failed as the compiler still remains
> > > clang and finding the cause is beyond my knowledge. Here is the resulted
> > > scratch build:
> > >
> > > https://koji.fedoraproject.org/koji/taskinfo?taskID=32277096
> >
> > Which sources are you building? Nothing like this is on the master
> > branch on src.fedoraproject.org.
> >
> > Thanks,
> > Florian
>
> I avoided committing the update as I preferred to do a scratch-build
> hence the link from koji.
> Realizing the change can be seen on the master branch on
> src.fedoraproject.org, I applied the update for view:
> https://src.fedoraproject.org/rpms/ispc/blob/master/f/ispc.spec

Sorry, the rawhide build now fails for me with:

m4 -Ibuiltins/ -DLLVM_VERSION=LLVM_7_0 -DBUILD_OS=UNIX builtins/dispatch.ll | python bitcode2cpp.py builtins/dispatch.ll > objs/builtins-dispatch.cpp
BUILDSTDERR: /bin/sh: python: command not found
BUILDSTDERR: make: *** [Makefile:330: objs/builtins-dispatch.cpp] Error
127

This looks completely unrelated to clang.

Thanks,
Florian

[Luya Tshimbalanga]

Please if possible. I attempt to use gcc to build it but some
dependecies are linked to clang which is beyond my skill. This compiler
blocks both embree and luxcorerender. See
https://src.fedoraproject.org/rpms/ispc

Luya

On 2019-01-25 9:02 a.m., Jonathan Wakely wrote:
> On 25/01/19 08:36 -0800, Luya Tshimbalanga wrote: >> On 2019-01-22 8:17 a.m., Serge Guelton wrote: >>> >>> Hi Luya, >>>
> > > Clang does not support the -fstack-clash-protection flag. We used to

silently ignore that flag but it's no longer the case. >>> Why are you
using clang to compile the package? The safe step is to use gcc, ot if
clang is needed, to strip -fstack-clash-protection >>> from the flags
passed to clang, but that's not future-proof (clang may end up
supporting that flag). >>> >>> Hope it helps, >>> >>> Serge >>> >> >>
Upstream reason. Trying to build with gcc will fail. See the attacked >>
CMakeLists.txt from ispc-1.10.0 > > That says it wants to build with
clang, but doesn't say why building > with GCC will fail. Is it due to a
GCC bug? Is it something we can > fix? >
_______________________________________________ > devel mailing list --
[e-mail] > To unsubscribe send an email to
[e-mail] > Fedora Code of Conduct:
https://getfedora.org/code-of-conduct.html > List Guidelines:
https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives:
https://lists.fedoraproject.org/archives/list/[e-mail]

[Luya Tshimbalanga]

I managed to resolve some issue but now the problem is related to the
hardened part:

https://koji.fedoraproject.org/koji/taskinfo?taskID=32324093

Here is the following lines for the failure:

BUILDSTDERR: /usr/bin/ld: kernels_ispc_sse2.o: relocation R_X86_64_32
against symbol
`RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_sse2'
can not be used when making a PIE object; recompile with -fPIC
BUILDSTDERR: /usr/bin/ld: kernels_ispc_sse4.o: relocation R_X86_64_32S
against `.rodata.cst16' can not be used when making a PIE object;
recompile with -fPIC
BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx.o: relocation R_X86_64_32
against symbol
`RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx'
can not be used when making a PIE object; recompile with -fPIC
BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx2.o: relocation R_X86_64_32S
against `.rodata.cst32' can not be used when making a PIE object;
recompile with -fPIC
BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx512knl.o: relocation
R_X86_64_32 against symbol
`RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx512knl'
can not be used when making a PIE object; recompile with -fPIC
BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx512skx.o: relocation
R_X86_64_32 against symbol
`RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx512skx'
can not be used when making a PIE object; recompile with -fPIC
BUILDSTDERR: /usr/bin/ld: final link failed: nonrepresentable section on
output
BUILDSTDERR: collect2: error: ld returned 1 exit status
BUILDSTDERR: make[2]: ***
[examples/deferred/CMakeFiles/deferred_shading.dir/build.make:204:
bin/deferred_shading] Error 1
BUILDSTDERR: make[1]: *** [CMakeFiles/Makefile2:370:
examples/deferred/CMakeFiles/deferred_shading.dir/all] Error 2
BUILDSTDERR: make: *** [Makefile:133: all] Error 2

Luya

[Fabio Valentini]

On Tue, Jan 29, 2019, 05:21 Luya Tshimbalanga <[e-mail] wrote:
> I managed to resolve some issue but now the problem is related to the
> hardened part:
>
> https://koji.fedoraproject.org/koji/taskinfo?taskID=32324093
>
> Here is the following lines for the failure:
>
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_sse2.o: relocation R_X86_64_32
> against symbol
>
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_sse2'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_sse4.o: relocation R_X86_64_32S
> against `.rodata.cst16' can not be used when making a PIE object;
> recompile with -fPIC

I've seen this errror before. It's uusally caused by upstream accidentally
overriding CFLAGS instead appending to already set CFLAGS, which breaks
fedora builds. Look for something like "set *CFLAGS* (foo bar)" in
CMakeLists.txt files, and change it to *append* to the already existing
value (coming from fedora settings), instead of overriding it.

Fabio

BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx.o: relocation R_X86_64_32
> against symbol
>
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx2.o: relocation R_X86_64_32S
> against `.rodata.cst32' can not be used when making a PIE object;
> recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx512knl.o: relocation
> R_X86_64_32 against symbol
>
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx512knl'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx512skx.o: relocation
> R_X86_64_32 against symbol
>
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx512skx'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: final link failed: nonrepresentable section on
> output
> BUILDSTDERR: collect2: error: ld returned 1 exit status
> BUILDSTDERR: make[2]: ***
> [examples/deferred/CMakeFiles/deferred_shading.dir/build.make:204:
> bin/deferred_shading] Error 1
> BUILDSTDERR: make[1]: *** [CMakeFiles/Makefile2:370:
> examples/deferred/CMakeFiles/deferred_shading.dir/all] Error 2
> BUILDSTDERR: make: *** [Makefile:133: all] Error 2
>
> Luya
> _______________________________________________
> devel mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives:
> https://lists.fedoraproject.org/archives/list/[e-mail]

[Dridi Boukelmoune]

> I've seen this errror before. It's uusally caused by upstream accidentally overriding CFLAGS instead appending to already set CFLAGS, which breaks fedora builds. Look for something like "set *CFLAGS* (foo bar)" in CMakeLists.txt files, and change it to *append* to the already existing value (coming from fedora settings), instead of overriding it.

Actually, upstream should prepend its own CFLAGS et al to give
precedence to user-defined flags.

Dridi

[Florian Weimer]

* Dridi Boukelmoune:
> > I've seen this errror before. It's uusally caused by upstream accidentally overriding CFLAGS instead appending to already set CFLAGS, which breaks fedora builds. Look for something like "set *CFLAGS* (foo bar)" in CMakeLists.txt files, and change it to *append* to the already existing value (coming from fedora settings), instead of overriding it.
>
> Actually, upstream should prepend its own CFLAGS et al to give
> precedence to user-defined flags.

The ispc compiler unfortunately has a completely different command line
syntax from what we use in CFLAGS, so this is not going to work anyway.

Thanks,
Florian

[Florian Weimer]

* Luya Tshimbalanga:
> I managed to resolve some issue but now the problem is related to the
> hardened part:
>
> https://koji.fedoraproject.org/koji/taskinfo?taskID=32324093
>
> Here is the following lines for the failure:
>
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_sse2.o: relocation R_X86_64_32
> against symbol
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_sse2'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_sse4.o: relocation R_X86_64_32S
> against `.rodata.cst16' can not be used when making a PIE object;
> recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx.o: relocation R_X86_64_32
> against symbol
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx2.o: relocation R_X86_64_32S
> against `.rodata.cst32' can not be used when making a PIE object;
> recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx512knl.o: relocation
> R_X86_64_32 against symbol
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx512knl'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: kernels_ispc_avx512skx.o: relocation
> R_X86_64_32 against symbol
> `RenderTile___uniuniREFs_5B_unInputHeader_5D_REFs_5B_unInputDataArrays_5D_uniun_3C_unT_3E_un_3C_unT_3E_un_3C_unT_3E_avx512skx'
> can not be used when making a PIE object; recompile with -fPIC
> BUILDSTDERR: /usr/bin/ld: final link failed: nonrepresentable section on
> output
> BUILDSTDERR: collect2: error: ld returned 1 exit status
> BUILDSTDERR: make[2]: ***
> [examples/deferred/CMakeFiles/deferred_shading.dir/build.make:204:
> bin/deferred_shading] Error 1
> BUILDSTDERR: make[1]: *** [CMakeFiles/Makefile2:370:
> examples/deferred/CMakeFiles/deferred_shading.dir/all] Error 2
> BUILDSTDERR: make: *** [Makefile:133: all] Error 2

I tried to build with --pic, using the attached patch, but got this:

[ 68%] Generating perfbench_ispc.h, perfbench_ispc.o, perfbench_ispc_sse2.h, perfbench_ispc_sse2.o, perfbench_ispc_sse4.h, perfbench_ispc_sse4.o, perfbench_ispc_avx.h, perfbench_ispc_avx.o, perfbench_ispc_avx2.h, perfbench_ispc_avx2.o, perfbench_ispc_avx512knl.h, perfbench_ispc_avx512knl.o, perfbench_ispc_avx512skx.h, perfbench_ispc_avx512skx.o
cd /builddir/build/BUILD/ispc-1.10.0/examples/perfbench && ../../bin/ispc /builddir/build/BUILD/ispc-1.10.0/examples/perfbench/perfbench.ispc --pic --target=sse2-i32x4,sse4-i32x4,avx1-i32x8,avx2-i32x8,avx512knl-i32x16,avx512skx-i32x16 --arch=x86-64 -h /builddir/build/BUILD/ispc-1.10.0/examples/perfbench/perfbench_ispc.h -o /builddir/build/BUILD/ispc-1.10.0/examples/perfbench/perfbench_ispc.o
make[2]: Leaving directory '/builddir/build/BUILD/ispc-1.10.0'
BUILDSTDERR: /builddir/build/BUILD/ispc-1.10.0/src/main.cpp(353): FATAL ERROR: Unhandled signal sent to process; terminating.
BUILDSTDERR: ***
BUILDSTDERR: *** Please file a bug report at https://github.com/ispc/ispc/issues
BUILDSTDERR: *** (Including as much information as you can about how to reproduce this error).
BUILDSTDERR: *** You have apparently encountered a bug in the compiler that we'd like to fix!
BUILDSTDERR: ***
BUILDSTDERR: make[2]: *** [examples/perfbench/CMakeFiles/perfbench.dir/build.make:65: examples/perfbench/perfbench_ispc.h] Aborted (core dumped)

This looks like something upstream needs to investigate. 8-(

Thanks,
Florian

[Luya Tshimbalanga]

-----BEGIN PGP MESSAGE-----
Charset: UTF-8
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=a/zz
-----END PGP MESSAGE-----

[Fabio Valentini]

On Tue, Jan 29, 2019 at 6:03 PM Luya Tshimbalanga
 wrote:
> -----BEGIN PGP MESSAGE-----
> Charset: UTF-8

I guess the encrypted garble garble below happened by accident - you
might want to send your message again without encrypting it.

Fabio
> hQEMAyZZ5X4GTmHqAQgA3r4Bs+wCUHFsVqpIPN5v0tFnaioogOKYTrDgA9MTKzzV
> GuEEdfFctzFOZw9eAiu7lIYZn41Ajpx7pGDjgxXJYsnhBKmL/n5BLrQ+hlG4FYcM
> LcT/O5DQjwgndjwsx9QlsHoCrk12lMCcGZLPX+w0Gjc1d8uMofQ3F9iZpr71zq99
> 6FF/hmpEghytBTnqL/0BPMu2tJOpdcVxeO8hYzFZoM/906GjLJMqOFXSa9Szlxib
> weoQg667E7i0J3p/GS0r5hU8g35M0pTjvHHlwh30lhQad/BBH6ry7Y/8J37Sh6XL
> whdSLi7xZMEOnV7RuBbiFjwF4FruHYvNi8Ph2JvLtNLqASL7zJZbPjgZh1VeUxbL
> lnQoJbcyiFM+hBxLCqJvnlzatMiwya63GqbEQwTOyU+kHFWoNSgQTtz3kwcSnahq
> q+A0ILXxLt5Sz0tKHRw7C7DwNFqrAj9Pr5MSbllm1AkLu+dAa5AW3CCdohz20iEH
> g+TaVMJjiPhyPaaWjrSkPl1Qbkd/Amy9jisHSBSQVWigviQjphkuFGDzLJByiWQe
> HCS5y4tx3q9ZTmT3RugU/Y4cm9IHHmODfI+B4fHMWubFVlos8i7Eu0lC4tWAo9nL
> MCrJkhJidFZguutfYwGS+/oznV8CkB3MiRkooH77FfLpLMwCtEiMoSSTsHDMVrFZ
> FqqsZXCDa8bZB+A5kb2eQLHDjAv6Y0Uv4l+oxxuz7AaAzgoCKHDe/2zFY/ZPguoV
> aM5CW+W4bKlDLKK3XzmtYqQYxcQcFWwHWDbrbunWFuS9/RlCcW6W4x9ARXkGY+26
> X4483yHxNA9kG+DyDqA3vvNUEgDcTPHe3rOr/ViUvFZZWnsfd7y+fO/wpVJNFywn
> xrzwmpjSI75z/n5kp22ckokm0sJz2NNX+N8ZRRWmsDyve8OgmS2wd61CGPoJ0cY3
> zRh4MJL+2Ly/u7BBocOxBdrlf0gSgBxBGmOQHAcbSgAYultgw4BWgs5UXr1OaEvx
> dFXWbNyfrWpwubl8f96fujQKgV64xi73Y9obSq6AjmWajKWnZFRSph+rV6CkoZqO
> FvUOPlFAbZmuaRb+9mU/8b8u3AhY1Xh8e50g5IW+Hqt8gw/Dz4vWPhaddJlc5o8A
> LR5O3v+CcJq1y+BzJ1GBymrUkuPJCxjRO/RctQbtTopUqMdiGx8jTy7y0mESFvAD
> Ojjc2W0gN9cfP6itGeB/27aeynRP/xSXw8xvg1bN+JW0fqnFidGdsrHahX5nPYRw
> MmVJsUyaIZbQSyD7TUbqzl30eIjPM92XNCXA7Jy+1Zi0lvdPDRvgQW3L0DijQVqu
> fhjvKPELPD+/L4kKDoJhSlaFSAyrd/pWfr2oKXJ8tiRqFShalFv/i2+fmxf18hqb
> AZMHqb3zDFFUQLopk4wf/668HHZVRxzyq4uUshEDAp36Eubtdrc8hMzmv2XfoxuM
> bs41OChDVTzdiekq+vj0hjftUNI/zMdFLYV3LilfI4e0XVF87uBzu6zFNZtjhxQN
> PBf05oA6dd3IaEgiomZOs3XLpZ8L1VjZhPo8EuJBvfgBz6jkGJKIt3vr9Ju2+t3a
> SJs/zIUC1tBcDtU5ZVBV9N2cfmpat6ISqaifB67afJN3APq+h3ZHgzaGc0QrGLW1
> QQY3t4+vdjt+a+Tx3HW+RYwa2IHclZk/L0yA4oa0yj3DRc6s53E3ZSQ4ZxD4jjkI
> KumR7OQ2NsV6zp4iCPgDT9dbO2z5hCkU1B9rAohTc0pqvFeXMlP3O8LOHbv/0j+Z
> rvvPvzuJlXlW0wfF+QEiD0SgNhCdcoeuiC74V4LIKsm1CF6+QRI/QcVH+xFZPJi0
> VorJ+I7BtsklvqmXS6Yc05ujJLvYyaWZhof3fwPf+40IV7BqKdZ9j9o4fa0LOgQD
> wzWw3FgMAL/WDFuPyzu7RZMxwQwqpxDffo/rCUfB61wZ+CAUCipX4kgugl6ZPRIs
> oVs4O4XW9eDaboBuQe/XfWZ3dEFVjIYQ17LMgu2tsMAShAMhtgZD+Giq4dUmKsHv
> D+pM2OPAeON+HF6sOAseocgShprB0t2NQzdnNT+FtW3+H/sw1VY7NoyW1pheC1KS
> aPbygQTcIkj0zDX9pZoCbGa3XkPqflqnViWSMZY8Ap7v4QcZnBWzAkQ8yFdovUtE
> XQ3CRrPX4aMvFL4GHg2CQRpCLCuz7Qab0WUP4v6v0/FvOP+poDKxZdahVreed1DZ
> LSiih/piAz1GG3imiIr20uxRgyBWo7Kt0jL6JthYszWGerkgwQeH8wAlfG5fQWZp
> gUZzaywXQ7EWD9HpA69IeNhblBxrxpTJEya15WORLZtk/1qk47c0cQfgkS+1Rd1e
> wSxs/IZyV+voAkTbHIWG7QCSHIP/ifg/Q/vHe6LBJN/ATqkvryqQDS/Lr32jbRP4
> 3sg2UroeWoYFjsb+2XjxyTWNXiFJ1RE2uQNnO7el5K58w+ROhJd120XQV6kAULaV
> Dswb7M6l/9qbcOQyC1kGz4axc3QeNVdlFSUcyMTmNN8tMXHMNwgpeWeRqiNNc7/j
> zVELKN42l3VP0bDFVEwK0elbU2IFqE8EMuOmVRlldDB8RmO0oqrtTwZVZb7CCLE8
> 7SHqLkIzAnXGASl9
> =a/zz
> -----END PGP MESSAGE-----
> _______________________________________________
> devel mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/[e-mail]

[Luya Tshimbalanga]

Here is. Filed the bug to upstream:

https://github.com/ispc/ispc/issues/1413

Luya

[Luya Tshimbalanga]

Contacting upstream about the issue, a suggestion is use the built-in
script alloy.py to manually build llvm instead of using the stock model
i.e. llvm-devel.

https://github.com/ispc/ispc/issues/1413#issuecomment-458654316

Any feedback?

Luya

[Luya Tshimbalanga]

On 2019-01-29 12:26 p.m., Luya Tshimbalanga wrote:
> Contacting upstream about the issue, a suggestion is use the built-in
> script alloy.py to manually build llvm instead of using the stock model
> i.e. llvm-devel.
>
> https://github.com/ispc/ispc/issues/1413#issuecomment-458654316
>
> Any feedback?
>
> Luya
>
> devel mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/[e-mail]

Added spec file with suggested alloy.py script.

Luya