why does control+c not work to stop a run away calculation

I am solving a linear problem p=A\b where A is large, rectangular, and sparse. Prior to R2015b I was able to stop this calculation with a control+c when a mistake was made in the input (too large for the RAM or a badly scaled matrix element, etc.) and the calculation went into essentially an infinite loop. Now I am routinely having to kill the matlab process and restart matlab when things go wrong. Is there a change in the behavior of "\" calculations? What can be done to avoid the run away problem? BTW - sometimes the calculation terminates with the (correct) assessment that the problem exceeds available memory - but then at other times, it fails to identify the problem and runs until paging slows the calculation to a snails pace or all application memory is exhausted and my mac goes into the suspended app mode.

Answer 1

[Walter Roberson]

Any call into the BLAS or LINPAK or similar libraries is not under the control of the control-C handler until the call returns.