Problem with FreeBSD's version of SED

[keoki <keoki () techie com>]

A problem exist in FreeBSD's version of SED, that 
when you specify last line and do not get contents of 
hold area and then delete the first part of a pattern 
space against a file which contains two lines or more, 
it results in a Segmentation fault "coredump". 
This is the following two instructions explained above. 

sed '$!g; D' file 

And "file" containing two or more lines. 

When you specify last line and GET the contents of 
the hold area and then delete the first part of the 
pattern space, it results in a Segmentation fault as 
well. 

Which is the following two instructions, 

sed '$g; D' file 

[keoki@jeff:~/test] for i in 1 2; do echo "blah" >> 
example; done 
[keoki@jeff:~/test] sed '$!g; D' example 
Segmentation fault (core dumped) 
[keoki@jeff:~/test] sed '$g; D' example 
Segmentation fault (core dumped) 
[keoki@jeff:~/test] 


Then I get the following using gdb on sed.core 

[keoki@jeff:~/test] gdb sed sed.core 
#0 0x280da416 in memmove () from /usr/lib/libc.so.4 
(gdb) 

Then at address 0x280da416, which is 
0x280da416 <memmove+26>: repz movsl %ds:(%
esi),%es:(%edi) 

I believe the problem resides in proccess.c 

Which is located in /usr/src/usr.bin/sed/ 

and at line 145 of proccess.c, which is memmove
(ps, p + 1, psl); in the following case statement, 

case 'D': 
if (pd) 
goto new; 
if ((p = memchr(ps, '\n', psl - 1)) == NULL) { 
pd = 1; 
goto new; 
} else { 
psl -= (p + 1) - ps; 
memmove(ps, p + 1, psl); 
goto top; 
} 


I haven't tested on NetBSD or openBSD's version of 
sed. I only have tested on FreeBSD. Also note, sed !
= gsed. And this problem doesn't exist in gsed, which 
is what most linux distros afaik use as "sed". 


-- keoki