Fail on mpir-3.0 build

41 views
Skip to first unread message

degski

unread,
May 2, 2019, 1:17:10 AM5/2/19
to mpir-...@googlegroups.com
Good day!

I've been trying to build mpir from repo: https://github.com/BrianGladman/mpir .

I've cloned the repo.
I've run 'mpir_config.py 17', chose 16 (broadwell_avx). [python 3.7.2-x64]
Load solution. [vs2017-15.9.11]
Change to x64.
Build.

This gives the following output:

1>------ Rebuild All started: Project: lib_mpir_broadwell_avx, Configuration: Debug x64 ------
1>building MPIR for broadwell\avx (x64) from directory mpn\x86_64w\broadwell\avx
1>creating mpir.h for x64
1>creating config.h from vs17\cdata\mpn\x86_64w\broadwell\avx\
1>Assembling add_err1_n.asm ==> x64\Debug\add_err1_n.obj
1>Y:\REPOS\mpir\msvc\vsyasm.targets(81,5): error MSB3721: The command ""Y:\COMMANDS\"vsyasm.exe -Xvc -f x64 -g cv8 -i "..\..\..\mpn\x86_64w\\" -o "x64\Debug\mpn\add_err1_n.obj" -rnasm -pnasm   ..\..\..\mpn\x86_64w\add_err1_n.asm" exited with code 1.
1>Done building project "lib_mpir_broadwell_avx.vcxproj" -- FAILED.
========== Rebuild All: 0 succeeded, 1 failed, 0 skipped ==========

vsyasm-1.3.0 exits with -1 (not very helpful [of yasm])
There's 2 things that strike me as slightly odd on that error message.
the quotes around  "Y:\COMMANDS\"and then a bit further, the double back-slash "..\..\..\mpn\x86_64w\\". The latter is possibly the culprit.

I'll obviously keep looking for the solution, but iff some-one could help, that's most welcome.

Have a nice day,

degski
--
Microsoft, please kill Paint3D

Cactus

unread,
May 2, 2019, 4:30:14 AM5/2/19
to mpir-devel


On Thursday, 2 May 2019 06:17:10 UTC+1, degski wrote:
Good day!

I've been trying to build mpir from repo: https://github.com/BrianGladman/mpir .
 
Hi Degski,

Unfortunately I am away at the moment so I am a bit limited in what I can do to help until Tuesday next week.  I have tried the broadwell_avx build and it does produce errors for me as well, but oddly, not the error you are seeing.

If you are using an old version of vsyasm it would be worth updating this.  

The other thing to try is to unload the project in Visual Studio and manually edit the line that refers to add_err1_n.asm to remove the extra double backslash that you are suspicious about to see if this helps.

The error I am seeing is not a build issue but a test failure in the mpn.logic test.  I think I understand this but I need to talk to Jens Nurmann (the author of the assembler source code involved) before I can fix this.

   Brian


degski

unread,
May 2, 2019, 5:37:52 AM5/2/19
to mpir-...@googlegroups.com
Hi Brian,

In the meanwhile, I've tracked down the problem [literally seconds ago]. The problem was/is in the "Yasm Object Filename Setting", which is '$(IntDir)mpn\%(FileName).obj'.

Problem is my paths are overruled by a global .props file (which I have, because I don't like the standard layout). Since you have the paths on default, the overrule kicks in. If the paths would be hard-coded in the Properties -> General -> [Output Directory/Intermediate Directory], they would not be overruled and they would obey whatever you put. So this is probably a good idea, it will make the build more robust.

I'll carry on and if I run into trouble I'll post and be patient till you have more time [or whatever is inhibiting you].

Thank you Brian, take care!

degski

--
You received this message because you are subscribed to the Google Groups "mpir-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mpir-devel+...@googlegroups.com.
To post to this group, send email to mpir-...@googlegroups.com.
Visit this group at https://groups.google.com/group/mpir-devel.
For more options, visit https://groups.google.com/d/optout.

degski

unread,
May 2, 2019, 5:45:06 AM5/2/19
to mpir-...@googlegroups.com
On Thu, 2 May 2019 at 08:30, Cactus <riem...@gmail.com> wrote:
Unfortunately I am away at the moment so I am a bit limited in what I can do to help until Tuesday next week.  I have tried the broadwell_avx build and it does produce errors for me as well, but oddly, not the error you are seeing.

I was tracking my problem in/with the basic broadwell build, that one works. The broadwell_avx gives (a.o.):

1>  Assembling and_n.asm ==> x64\Debug\and_n.obj
1>  ..\..\..\mpn\x86_64w\broadwell\avx\and_n.asm(117): error : undefined symbol `SizeB' (first use)
1>  ..\..\..\mpn\x86_64w\broadwell\avx\and_n.asm(117): error : (Each undefined symbol is reported only once.)
1>  ..\..\..\mpn\x86_64w\broadwell\avx\and_n.asm(152): error : undefined symbol `CountB' (first use)
1>  Y:\REPOS\mpir\msvc\vsyasm.targets(81,5): error MSB3721: The command ""C:\Program Files\yasm\"vsyasm.exe -Xvc -f x64 -g cv8 -i "..\..\..\mpn\x86_64w\\" -o "..\..\..\mpn\and_n.obj" -rnasm -pnasm   ..\..\..\mpn\x86_64w\broadwell\avx\and_n.asm" exited with code 1.

Which seem more like coding issues.

I guess that's what you were looking at. Next week [or whenever you get to it] no problem.

Thanks,

Brian Gladman

unread,
May 2, 2019, 5:24:34 PM5/2/19
to mpir-...@googlegroups.com
Yes, there were errors in three of the Windows x64 assembler code files
for broadwell_avx.

I have just corrected these in the repository and the build should now
succeed and pass the tests.

Brian

degski

unread,
May 3, 2019, 12:12:28 AM5/3/19
to mpir-...@googlegroups.com
On Thu, 2 May 2019 at 21:24, Brian Gladman <b...@gladman.plus.com> wrote:
I have just corrected these in the repository and the build should now
succeed and pass the tests.

Brian, thanks, it works like a dream ...

degski
Reply all
Reply to author
Forward
0 new messages