This looks like a bug in readline 8.0 Program received signal SIGSEGV, Segmentation fault. 0x00007ffff76637da in _rl_update_final () at ../display.c:2972 2972 botline_length = VIS_LLEN(_rl_vis_botlin) - woff;
Once the macros are expanded this is attempting to deference the NULL pointer line_state_visible->lbreaks. This pointer is initialized in rl_display, but in the password entry case the standard rl_display is overridden and it is never initialized. This was ok in version 7.0 as full_lines is set to zero and VIS_LLEN was not called. Relevant diff below. _rl_update_final () 2968 full_lines = 1; } _rl_move_vert (_rl_vis_botlin); • woff = W_OFFSET(_rl_vis_botlin, wrap_offset); • botline_length = VIS_LLEN(_rl_vis_botlin) - woff; /* If we've wrapped lines, remove the final xterm line-wrap flag. */ • if (full_lines && _rl_term_autowrap && (VIS_LLEN(_rl_vis_botlin) == _rl_screenwidth)) • if (full_lines && _rl_term_autowrap && botline_length == _rl_screenwidth)
signature.asc
Description: OpenPGP digital signature
