I have a user of my Theia library reporting a segmentation fault in the SparseQR solver. The debugger reveals that the segfault occurs in the solve() function at this line. This seems to be a very odd since I ensure info() == Success right beforehand. For some more context of the code: lhs is a M x N matrix, rhs is a M x 3 matrix, and solution is a N x 3 matrix. The stack trace of the debugger can be found here.

Any thoughts here? I am not even sure how to debug this further...

Your user seems to be using a pretty old version, so first update to latest 3.2.8 release. The problem might have already been solved.

If the problem is still there, then this should be quite easy to debug as this is not a segfault but an assertion saying that we taking a submatrix with invalid ranges.

Upgrading to the latest Eigen version fixed it -- thanks!