🚀 go-pugleaf

In article <9005161953.AA11842@ray.sun.com> oakhill!ray!jay@CS.UTEXAS.EDU (Jay Norwood) writes:
>I'm working with 3.5 coffread.c.  In the function getfilename, there is
>a section for reading in long filenames:
>  if (aux_entry->x_file.x_foff != 0)
>    strcpy (buffer, stringtab + aux_entry->x_file.x_foff);
>---------------
>Since the first filename symbol may very well be the first symbol
>which deposits a string in the string table (if the filename is longer
>than FILNMLEN), and since the x_foff is relative to the start of the
>string table, it is very possible to have a valid long filename at the
>same time as x_foff==0.  In other words this is a very risky test.
>Wouldn't it be better to just test x_file.x_fname[0] for a 0 in order to
>determine whether or not to use x_foff?
>Jay Norwood

Ummm... I take back what I said before.  It was only (at best) partly right.

Actually, I (now) seem to recall that the first four bytes of a coff
string table are occupied by a one work long count of the total number of
bytes in the string table.  If that's true (and somebody please check
my dim recollection on this) then the lawest numbered *actual* meaningful
index into the table is (obviously) 4 because bytes 0,1,2,3 are used.

// rfg