Error messages

[Doug]

I can't find a list of error messages generated by M2.

I just ran into

                            mmap(PROT_NONE) failed

while testing an automated version of ny new desingularization of function fields code.

This doesn't seem to be related to time or space problems (10 minutes 1 gig), so I'm at a loss to figure out what to do

to get my seemingly functioning code to finish the problem it was working on.

Doug Leonard

[Daniel R. Grayson]

I've never seen that error before.  The message comes from the libgc library that we depend on for memory allocation and garbage collection.  One possible cause might be that you are out of memory, but without running the program in a debugger, it will be impossible to know the precise cause.  The manual page for "mmap" lists 10 separate possible causes for failure, and, unfortunately, libgc neglects to tell us which of the 10 applies, even though it could!  If you could tell us how to reproduce the error, either on your machine or on one of ours, we could try to debug it, but it looks rather mysterious, so trying to find a way to work around the problem might be more fruitful.

See http://man7.org/linux/man-pages/man2/mmap.2.html

[mike stillman]

If it is really not out of memory, this seems like it would be a bug in M2, but it is hard to know that from what is written (i.e. we need more detail as Dan suggests).  In this case, for you, a workaround might be good, but it would be good for us to fix the bug too, if it is one!

[Justin Chen]

I just wanted to add that I have seen this same error multiple times before. Unfortunately I don't remember specific instances which gave rise to it, but my general impression is that it popped up in instances where I already suspected that M2 would use too much memory. (Also, my general use case for M2 is in the Windows Subsystem for Linux.)

Justin

[Daniel R. Grayson]

I've run your code on my machine and killed it when M2-binary got up to 4GB of ram.  So I think you're running out of memory.

[abrahammcs]

Hi, I recently had the exact same error in a server running M2-binary and it definitively didn't consume all the memory (have more than 600GB of ram) nor space.   I used to have an error with the garbage collector when running and old version of my code, but I am not sure if this error is related or not. I can share my code in case you want to reproduce the error. I would be very happy to get around this error so I can finish that large computation.

Thanks!!

-Abraham

[Daniel R. Grayson]

I have access to a machine with 64GB of RAM.  How big was the process when it gave that error?