[2119] in Kerberos-V5-bugs

home help back first fref pref prev next nref lref last post

Linux

daemon@ATHENA.MIT.EDU (Tom Hamilton)
Mon Jul 29 15:26:33 1996

Date: Mon, 29 Jul 1996 14:26:15 -0500 (EST)
To: krb5-bugs@MIT.EDU
From: Tom Hamilton <thamilt@indiana.edu>

Ok, I have just started on this mailing list, and I couldn't find an

archive of mail on MIT passed 1992 where the word Linux was mentioned

one time.  I would love it if someone has a patch kit for linux. 
Given

the appearance of "rootkits" used to break into systems I am trying to
beef

up a few computers....


#1 -----------


The one compile flaw I found for Linux centered around the definition
of

fd_set which (apparently on most other systems is a struct)  This only

cropped up in /src/appl/gssftp/ftp/ftp.c


Linux declares the following in <<linux/posix_types.h>


typedef struct {

        unsigned long fds_bits [__FDSET_LONGS];

} __kernel_fd_set;


and <<linux/types.h> declares

<bold>

typedef __kernel_fd_set         fd_set;


</bold>So I just had to go through ftp.d replacing "struct fd_set" with
fd_set

for it to compile (And add #include <<linux/types.h> at the top of
ftp.c>



#2 -----------<bold>


</bold>The make file in /src/appl/gssftp/ftp has the following
lines<bold>


prefix=/usr/local

ADMIN_BINDIR = $(INSTALL_EXEC_PREFIX)/sbin


</bold>I am getting around to saying that after I did a make install
everything

installed in /usr/local/sbin instead of "/usr/local/admin" as stated on
Page 25

of the Kerberos "installation.ps" Dated March 20, 1996.


Any cause for alaram ?


#3 -----------------------


When I did "make check" the encryption tests and comparisons seemed to
go fine

until a line appearing something like "SHS"<bold>


here... (results of make check)


...

>7 tests passed successfully for MD5.

>./t_cksum "this is a test"

>correct MD5 checksum("this is a test") =
>b33594b9ba342c5a89ae108e0c50542e29ef139

>abee9caa3

>old MD5 checksum("this is a test") = f3e6358b3548f822f92f54f96acd7418

>Verify succeeded for "this is a test"

>Compatible checksum verify succeeded for "this is a test"

>Verify of bad checksum OK for "this is a test"

>Compatible checksum verify of altered checksum OK for "this is a
test"

>1 tests passed successfully for MD5 checksum

>make[3]: Leaving directory
>`/usr/src/athena/krb5-beta6/src/lib/crypto/md5'

>checking in lib/crypto/sha...

>make[3]: Entering directory
>`/usr/src/athena/krb5-beta6/src/lib/crypto/sha'

>./t_shs -x

>Running SHS test 1 ...

>SHS test 1 failed

>make[3]: *** [check-unix] Error 255

>make[3]: Leaving directory
>`/usr/src/athena/krb5-beta6/src/lib/crypto/sha'

>make[2]: *** [check-unix] Error 1

>make[2]: Leaving directory
`/usr/src/athena/krb5-beta6/src/lib/crypto'

>make[1]: *** [check-unix] Error 1

>make[1]: Leaving directory `/usr/src/athena/krb5-beta6/src/lib'

>make: *** [check-unix] Error 1


</bold>IF the SHS tests are part of the tests that use rlogin (as
mentionetoon page 8) 

then it might be a good idea to specifically mention this, or describe
what kind

of setup needs to be done for this to work.<bold>


</bold>#4 -----------------------<bold>


</bold>Next Big problem


After compiling  "/src/admin/create/kdb5_create.c" when I run
kdb5_create

I get a nasty "Segmentation fault".  No problem, I thought I would
just

put in a few printfs and track it down.


Here's the kicker for me, because I have never seen anything like it.


Even if I backup the original kd5_create.c and replace it with...


void main ( void )

{

  printf ("Testing\n");

}


I still will get a segmentation faults when run after compiling.

(ps. yes I did erase the binary which was originally produced)


When I do a make after this change I get a 


>make

>cc -L./../../lib -L./../../lib  -o kdb5_create kdb5_create.o  -lkadm
-lkdb5    -

>lkrb5 -ldbm -ldbm -lcrypto  -lcom_err


Am I to conclude that one of these libs that are being linked in is
causing this.


I mean I have Never seen a program just "segment fault" with out (ME)
doing

something goofy like have uninitialized pointers or forgetting to
convieniently

update what ever data structure I am working with....


Am I to conclude that I will need to dig a little deeper to get this to
work.. ?


Any help you can shed would be appreciated...


-Tom Hamilton



<bold>



</bold>

-+=+---+=+---+=+---+=+---+=+---+=+---+=+---+=+---+=+---+=+---+=+---+=+-


Tom Hamilton / Computing Services |

Indiana University East           | Very funny Scotty,

thamilt@indiana.edu               | Now beam down my clothes.          
 

root@ravenloft.iue.indiana.edu    |           -Capt Kirk 


home help back first fref pref prev next nref lref last post