I recently tried to use Armadillo on iOS to do some matrix computing. The App worked on my development iPhone, but Apple gave me the error message when trying to publish it in the Appstore. It seems that Armadillo calls some BLAS functions which are internal. I searched the web with the message, but had not found anything useful. I also found calling BLAS functions with "cblas_" prefix, e.g. cblas_dgemv, directly from my code would not cause the error. However, that made the use of armadillo meaningless.
I wonder if anyone has encountered the same problem, and what the solution is. I’m suspecting it’s something related to some macro in config.hpp. Thank you very much for your kindly help.
Error message:
Non-public API usage:
• The app references non-public symbols in ***: _sgemm_, _sgemv_, _ssyrk_
The Accelerate BLAS implementation supports a bunch of redundant symbols to facilitate divergent function naming schemes of various fortran compilers. Strictly speaking these are intended to be used (by your fortran compiler) so you probably have some arguing ground that they are not private interfaces. If the AppStore is still giving you trouble, file a bug against Apple and ask them to fix the bookkeeping on the interfaces so they can be used.
It would be simpler to just start using the cblas_ interfaces in the headers though.
Related
Out of educational purposes I'm trying to use Eigen in my pet project. In order to speed the math I included macro EIGEN_USE_BLAS which activates use of blas library.
But I encountered an issue when I tried to upload my project to Testflight. The Apple response to that was:
ITMS-90338: Non-public API usage - The app references non-public
symbols in My-Project: _saxpy_, _sgemm_, _sgemv_, _strmm_,
_strmv_. If method names in your source code match the private Apple
APIs listed above, altering your method names will help prevent this
app from being flagged in future submissions. In addition, note that
one or more of the above APIs may be located in a static library that
was included with your app. If so, they must be removed. For further
information, visit the Technical Support Information at
http://developer.apple.com/support/technical/
I know that Apple has blas library as a part of Accelerate.framework and it should be use strictly through Accelerate API. But the thing is that Eigen also has their own blas included in their source and the library actually has no intention of using Accelerate.framework private guts. So that is why Apple asks me to rename those functions or remove them completely in order to get rid of that inconvenience.
But I'm not even sure that it is possible to do - to tune Eigen and its blas to use alternative names.
Is there maybe a way to solve that issue in some efficient manner? Or maybe I don't know something about Eigen usage in iOS environment?
I was lucky to figure out the problem with few tips from my fellow colleagues and a strong will to get the job done, and now I want to share with you my conclusions.
The problem: Eigen doesn't have BLAS binary provided, it only has it's headers. And when one uses EIGEN_USE_BLAS macro and it actually works, that means that Apple is linking Eigen BLAS functions to their binary - BLAS binary which is a part of Accelerate.framework. The trickiest part is that your project doesn't necessarily should have Accelerate.framework inside. XCode will add necessary files automatically and of course it won't bother to tell you that.
What's weird for me the most here is that using "naked" BLAS provided by Apple's Accelerate.framework is a clear violation because it's private API. So... from my perspective it's kind of XCode violates this rule without my intention and tells me nothing about it. Strange...
So, in order to fix it, you need to build BLAS or OpenBLAS binary (I have built OpenBLAS), add it to the project and link through Other linker flags in Build settings. After that the error is gone and you're able to upload your build to Testflight.
In order to save you some time in building OpenBLAS for iOS, I will leave you few links. I think those would be enough:
http://www.programmersought.com/article/2638161057/ - very clear tutorial on how to build OpenBLAS
https://github.com/xianyi/OpenBLAS/tree/release-0.2.21 - source code of the library
https://github.com/xianyi/OpenBLAS/issues/1531 - discussion where you can find solutions to problems you might encounter
What I found the most confusing part of this whole story is the message from Apple. Here's another copy of that:
ITMS-90338: Non-public API usage - The app references non-public
symbols in My-Project: saxpy, sgemm, sgemv, strmm, strmv. If
method names in your source code match the private Apple APIs listed
above, altering your method names will help prevent this app from
being flagged in future submissions. In addition, note that one or
more of the above APIs may be located in a static library that was
included with your app. If so, they must be removed. For further
information, visit the Technical Support Information at
http://developer.apple.com/support/technical/
I'm not a native English speaker but I feel pretty confident to say that this text is a non-sense and has absolutely nothing to do with the solving of the actual problem.
It says absolutely nothing about the fact that private API was linked to my code and that is the problem. And that I need to provide my own backend for those functions in order to make error disappear. It says about names matching and that altering my names will help to solve the problem, when it clearly will not in my case.
I hope Apple will bring more clarity into this problem and that errors descriptions will have more sense in the future.
I saw them in libc-8XX/pthreads on opensource.apple.com, but since libc-9xx the pthreads folder are gone, so did these two functions. I saw these two functions were used in _Unwind_SjLj_Register in exception handling on 32-bit IOS platform, but where are they now? or am I just misunderstanding something about them?
I have just finished porting a decent amount of c-sources to the iOS platform and packaged them as a universal static framework. I, then, added the framework (not the project) to a sample iOS app in order to test linkage and proper function. That's when I ran into a humbling problem.
In my attempt to solve the problem described here, I also came across some symbols that are composed through the heavy use of macros (i HATE those). Some of those macros use function attributes that are really extensions of gcc rather than of standard C.
Of course I can always add -std=gnu89, but even then, I am not sure it will resolve the original problem of undefined symbols in the static library.
Not only that, I am now worried that my port to iOS of those sources may not be an accurate port and may result in the type of bugs/issues that maybe related to compiler's codeine and/or optimization policies.
If you can share some of your experience/advice in how best to go about that port, I would really appreciate it.
Thanks!
From manual testing with clang 8.0, it seems that both __attribute__((constructor)) and __attribute__((__constructor__)) work for your purpose.
I'm trying to add the Dropbox Sync API (v1.1.2) to an iOS app built with Marmalade (v6.3). I'm getting the following link error:
Undefined symbols for architecture armv7:
"___udivmodsi4", referenced from:
_sqlite3BitvecSet in libDropbox.a(sqlite3.o)
_sqlite3BitvecClear in libDropbox.a(sqlite3.o)
_sqlite3BitvecTest in libDropbox.a(sqlite3.o)
ld: symbol(s) not found for architecture armv7
Googling for pertinent parts of that error message finds a number of users of a certain SQLCipher library experiencing the same issue, and suggestions that the problem is caused by an inconsistency of the compilers used to build that library and the various projects using it.
As the build system for our project is set up by the Marmalade toolset, changing the compiler (currently a version of GCC 4.4 supplied by Marmalade, I believe) is not an option, I think.
Can anyone tell me more precisely what is going wrong? Are there any other workarounds to this problem?
On processors like ARM, with relatively simple instruction sets, some more complex operations are mapped on to function calls rather sequences of instructions. When you link using the "correct" compiler, the implementations of these are pulled in and it works! You on't normally see it.
The obvious solution here would be to use marmalade to compile the dropbox library, then it will use a compatible compiler.
The question then, I guess, is whether there is a reason you are not doing this to start with? Current Marmalade compilers don't support ARC. I guess that would be a reason not to compile under ARC.
The other answer is correct. However, for this specific problem (if these are the only linker errors you are getting) I see two workarounds:
Grab the source from sqlite3 that includes sqlite3BitvecSet and compile those functions in your own project to override the library. They will pick up whatever divmod support is offered by your own compiler.
Implement your own udivmodsi4. You don't have to implement bitwise division (although you can go get that basic C implementation from the GCC source). You just have to implement it in native operations and let your compiler call whatever internal support it needs.
This is untested, but should give you the basic idea. You may need more underscores on the name to match the behavior/naming of the other build environment:
unsigned long
udivmodsi4(unsigned long num, unsigned long den, int modwanted)
{
if (modwanted)
return num % den;
else
return num / den;
}
I have been having a tough time setting up an experiment where I allocate memory with CUDA on the device, take that pointer to memory on the device, use it in OpenCL, and return the results. I want to see if this is possible. I had a tough time getting a CUDA project to work so I just used Nvidia's template project in their SDK. In the makefile I added -lOpenCL to the libs section of the common.mk. Everything is fine when I do that, but when I add #include <CL/cl.h> to template.cu so I can start making OpenCL calls, I get over a 100 errors. They all look similar to this, but with different function names at the end:
/usr/lib/gcc/x86_64-linux-gnu/4.4.1/include/xmmintrin.h(334): error:
identifier "__builtin_ia32_cmpeqps" is undefined
I am having a hard time figuring out why. Please help if you can. Also, if there is an easier way to set up a project that'll be able to call the CUDA and OpenCL APIs let me know.
I haven't really worked with cuda, so I don't know how helpful my answer is.
From what I understand you are trying to use opencl directly from your cuda hostcode, which is if I remember correctly compiled using some compiler from nvidia instead the standard gcc. So the problem is probably that this compiler doesn't implement the necessary builtins to work with the mentioned headers.
Look here for a similar problem and it's solution:
http://forums.nvidia.com/lofiversion/index.php?t88573.html
It seems you have to put everything which needs the opencl api into a different (non cuda) compilation unit so that it will be compiled by the non nvidia compiler.
However I wouldn't count on this working (since opencl buffers aren't just pointers to the memory but should contain some metainformations to), simply because there is no real reason it should work and if it does there is no guarantee that it continues to do so.
What you could try if you really want to is using opengl for the interop, since both opencl and cuda have extensions to allow creating buffers from opengl buffers.
However why do you need to do this? Whats keeping you from using Apple's implementation shortterm, since IIRC it's open source and most of it (the opencl parts) should be platform independent anyways.