Segfault when making screen too narrow

[tompickering]

Narrowing the screen eventually triggers a segfault. This does not appear to be replicable when run within gdb, but seems likely to be down to layout code (probably formatting lines in the 'pages' window).

[colinnewell]

Reproduced this one and got a stack trace,

#0  0x0000000000405cfa in string_clean_buffer (buf=0x2251090 ".\214\"\002", 
    src=0x2108fe0 "listens on a socket for Valgrind commentary\n", len=1)
    at win/win.c:226
#1  0x00000000004045b5 in win_page_show (win=0x22440e0) at win/pages.c:72
#2  0x0000000000404801 in draw_win_pages () at win/pages.c:100
#3  0x0000000000405bf0 in win_draw_all () at win/win.c:202
#4  0x0000000000402542 in draw_windows (r=49, c=11) at draw.c:46
#5  0x000000000040256e in draw_screen () at draw.c:52
#6  0x0000000000402329 in main (argc=1, argv=0x7ffdf89c7078) at mantra.c:85

The line of code that looks suspicious and is obviously mentioned by the stack trace is this,

buf[len-3] = buf[len-2] = buf[len-1] = '.';