Reputation: 85
I'm running a program that fails with an abort trap (error code 12). To debug the error, I'm running valgrind, but every time I run it valgrind itself crashes with a "Killed" message and nothing else (the end message heap summary etc. does not show up). I can't find anything on google or SO about this and I'm hoping someone can shed some light on what causes this error and how to debug it.
Any help is appreciated!
Upvotes: 7
Views: 6262
Reputation: 20818
This can also happen if your machine is using non-standard memory allocation, for example using a coprocessor (GPU, FPGA) that has it's own physical memory that is accessed in the memory space of the processor.
In our case, the vendor had to fix their framework to work with valgrind.
Upvotes: 0
Reputation: 9220
I think your process is probably getting killed by the kernel's Out Of Memory system - when the system runs out of memory that will hard kill a process in a way that can't be trapped which is why valgrind is exiting without saying anything.
Presumably your program uses a lot of memory and, when run under valgrind where extra memory is needed to track the definedness of the memory, there is not enough memory available and the (doubtless very large) valgrind process is getting killed.
If I'm right then you are going to need a machine with more memory, or at least with more swap, to be able to run your program under valgrind.
Upvotes: 7