internal error: assertion failed: lower_expr: bad kind (shared/edgcpfe/lower_il.c, line 14342)

[Rhys_Ulerich]

Any hints for what would cause this internal error (below) to be emitted from icpc (ICC) 10.1 20081024?
It's in the midst of a mess of stuff. I'm happy to work towards simple recreate, but knowing what's failing will help me do so.

The only other reference in the forums to internal errors in that particular file is in this thread: http://software.intel.com/en-us/forums/showthread.php?t=46593

- Rhys


../../sz/suzerain/blas_et_al.hpp(658): internal error: assertion failed: lower_expr: bad kind (shared/edgcpfe/lower_il.c, line 14342)

return suzerain_blas_zscal(boost::numeric_cast(n),
^

compilation aborted for ../../sz/tests/test_state.cpp (code 4)

[Rhys_Ulerich]

I'm moving again. The problem was casting a (const std::complex *) using reinterpret_cast and incorrectly passing the result to a method expecting a double (*)[2] without the const qualifier.

Give me a shout if anyone's interested in that recreate.

- Rhys

[Rhys_Ulerich]

Nah, I lied. That didn't fix it.

[Rhys_Ulerich]

Moving again on a second try. Issue was somehow related to a reinterpret_cast(std::complex

Once again, let me know if you'd like me to spend time on a recreate.

[Judith_W_Intel]

Yes, we'd be interested in a reproducer to check to see if the problem still exists in our supported compilers. You can just attach a preprocessed file and any special command line options necessary to showthe problemif you'd like (On Linux you should use the -E or -P options to produce a preprocessed file).

thanks,
Judy

[Kittur_G_Intel]

Hi Rhys,
If you could attach a reproducer like Judy mentioned in the post before, I can file an appropriate issue with our developers and keep track of the same, thanks much.
-regards,
Kittur

[Kittur_G_Intel]

Hi Rhys,
Just a friendly reminder - can you please attach thereproducer if possible? You can also create a private thread and do so if need be? Thanks much.
-Kittur

[Rhys_Ulerich]

Hi Kittur,

Sorry-- didn't notice the thread. I think the forum updates hit my spam filter. It may be a little while, but I'll get you a recreate.

- Rhys

[Kittur_G_Intel]

Sure Rhys, thanks much.
-regards,
Kittur

[Rhys_Ulerich]

Hi Kittur,

After about an hour of trying to whittle down the code, I'm afraid I'm unable to get a clean recreate for you. I'm sorry, and I'll keep an eye on getting one should I run into this problem again.

- Rhys

[Kittur_G_Intel]

Thanks Rhys for getting back on our request. Sure, I'll close this thread for now, but please reopen if you see this problem again? Appreciate your prompt response and help.
_Cheers,
Kittur

[Kittur_G_Intel]

Hi Rhys,
Were you able to get some time on this?! Appreciate a reproducer on this when you get a chance
-regards,
Kittur

[amaigil]

Hello,

I have a similar internal error with icc (ICC) 12.0.0 20101116 on linux:
internal error: assertion failed at: "shared/edgcpfe/lower_il.c", line 14558
which is 100% reproducible.
However I don't want to post the source code here. Where can I can conveniently report details?

Regards
Alexander

[TimP]

You could attach to a private reply here, or preferably, submit the case on your support account at premier.intel.com. Note that you are on update 0 of a new compiler, and update 3 is due out shortly.