[Passwdsecure-devel] RE: SEWL passwdsecure -- other note

[Harris, R. <ha...@ta...>]

Hi Andrew,

Looking at the truss output the problem seem symtematic of a buffer overrun
or a stray pointer being accessed.  I've had a look at authprocess.c and
found a couple of strong contenders.

You've changed the way MD5File works so it returns the MD5 data as a hex
string, rather than the raw MD5 data.  I assume you've done this for the
purposes of debugging so you can see the checksum more easily.  You've made
a few mistakes in doing this.

Firstly, in MD5File (if the digest is calculated correctly) you return
buffer2, unfortunately buffer2 is on the stack so as soon as you return from
MD5File the memory where you are pointing to will no longer exist.  Later on
you then call free on the returned value, this will cause all sorts of
confusion!  Also digest is being created on the heap but you don't return a
pointer to it so it creates a memory leak.

To do it your way MD5File should be:

#ifdef USE_MD5
char* MD5File(char* filename)
{
	FILE* file;
	if ((file = fopen (filename, "rb")) == NULL)
	{

#ifdef DEBUG
		fprintf(stderr, "Problem opening file: %s ...\n", filename);
#endif

		return (char*)NULL;
	}
	else {
		MD5_CTX context;
		int len;
		unsigned char buffer[1024];
		unsigned char digest[16];
		unsigned int i;
		char* hexStringDigest;

		MD5Init (&context);
		while (len = fread (buffer, 1, 1024, file))
			MD5Update (&context, buffer, len);
		MD5Final (digest, &context);

		fclose (file);

		/* 33 chars, 32 for hexString, and 1 for NULL */
		hexStringDigest = (char*)malloc(sizeof(char)*33);
		checkMalloc((void*)hexStringDigest);

		for (i = 0; i < 16; i++)
		{
			sprintf(hexStringDigest + i*2, "%02x", digest[i] );
		}

		return hexStringDigest;
	}
}
#endif

I havn't got a machine to compile this on to test but it looks all right.

The other big problem is with md5thusfar.  At line 144 you define md5thusfar
as a char* and then (at line 220) you call strncpy to it before it is
initialised.  It might be pointing at anything at this stage.

You need to change line 220 from:
				strncpy(md5thusfar, file[0], 32);
to:
				md5thusfar = (char*)malloc(sizeof(char)*33);
				checkMalloc((void*)md5thusfar);
				strncpy(md5thusfar, file[0], 32);
				md5thusfar[32] = '\0';

Note the last line, if file[0] is 32 or more characters long then md5thusfar
won't finish with a null, so trouble will ensue.

That's all of the problems I can see in authprocess.c at the moment.  If you
fix those problems you should be in with a much better shot of getting it to
work.

Regards,
Richard

PS: As a formatting point when you have something like:
<tab>func(first_arg,
<tab>     second_arg);
You shouldn't change it to
<tab>func(first_arg,
<tab><tab>second_arg);
As the arguments won't line up if someone is using different sized tab
stops.  You should always tab to specify indents and space to line up a
command the spreads multiple lines.  The same applies for if/while
conditions and for statements.

PPS: I've made you an admin of the sourceforge project.


-----Original Message-----
From: Andrew McGlashan [mailto:ae...@bi...]
Sent: Thursday, 9 May 2002 10:45 AM
To: Harris, Richard
Cc: Bill Lucas; Mario Piazzese
Subject: Fw: SEWL passwdsecure -- other note
Importance: High


Richard,

I have made this email high priority as SE Water has a number of uses that
will need to use the passwdsecure tool.  Bill has even suggested that I be
listed as a maintainer to try to progress this quicker; I would really like
some more help.

Okay, I have looked at FILECMP again -- I broke it with my former MD5CHECK
works.  I have since re-worked that but I still have some problems.  The
FILEICMP looks okay after all, although I haven't actually tested it; I
didn't look at it closely enough to understand it before.

The problem I am having now with the re-worked FILECMP / MD5CHECK section
having been done has to do with the getPsinfo function; it is dropping out
without completing all checks.  I don't know why it dumps out, but here is
the statement that is failing:

    sizeRead = fread(buf+sizeReadTotal, 1, sizeof(psinfo_t)-sizeReadTotal,
fileinfo);

Other information (from the non-Linux code):
> Checking file types / sizes...
>
> int      sizeRead         /* always 336 if successful */
> int      sizeReadTotal    /* always 0 */
>
>
> unsigned char buf[sizeof(psinfo_t)]
>

Now it fails after a number of iterations.  I did remove a single check from
one require group and it got through to an additional require group.  I was
going to include the output file with the debug information, but I don't
have it with me right now.  Let me know if you need it and I will endeavour
to get it to you.

I did adjust the following in the getPsinfo function:
    from    FILE* fileinfo;
    to    FILE *fileinfo;
I assume that was simply a typo. The change didn't help or seem to effect
anything.  I also made sure that every fopen had an fclose; I thought that
maybe there were too many files open or something like that.  Maybe there is
still something in this as it was quite late when I packed it in last night.

The Makefile in the attached source has a DEBUG entry added to the CFLAGS
line, so it always makes the debug version for now.  My 'do_conf' doesn't
include the MD5 directory, however, it uses the found OpenSSL libraries and
thus the configure defines USE_MD5.

AndrewM

Andrew McGlashan AACS
Director

Mobile: 04 2574 1827

Affinity Vision Australia Pty Ltd
ABN 91 099 273 855

This message contains privileged and confidential information intended only
for the use of the addressee named above. If you are not the intended
recipient of this message you are hereby notified that you must not
disseminate, copy or take any action in reliance on it.  If you have
received this message in error please notify Affinity Vision Australia Pty
Ltd at ae...@bi...  Any views expressed in this message are those of the
individual sender, except where the sender specifically states them to be
the views of Affinity Vision Australia Pty Ltd

----- Original Message -----
From: <And...@se...>
To: <ae...@bi...>
Sent: Thursday, May 09, 2002 1:12 AM
Subject: SEWL passwdsecure -- other note


>
> The following statement fails consistently for unknown reasons.....,
> but only after so many reads...
>
> sizeRead = fread(buf+sizeReadTotal, 1, sizeof(psinfo_t)-sizeReadTotal,
> fileinfo);
>
>
> Checking file types / sizes...
>
> int      sizeRead         /* always 336 if successful */
> int      sizeReadTotal    /* always 0 */
>
>
> unsigned char buf[sizeof(psinfo_t)]
>


***********************************************************************************
The information in this e-mail message and any files transmitted with it 
are intended to be confidential and for the use of only the individual or 
entity to whom they are addressed. The message and files may be
protected by legal professional privilege, or other legal rules. The 
confidentiality of and privilege applying to this message and 
files is not waived if this message or files has been sent to you by mistake. 
If the reader of this message or files is not the intended recipient, you are 
notified that retention, distribution or copying of this message and files are 
strictly prohibited.  If you receive this message or files in error, please 
notify us immediately by telephone or return e-mail and delete all copies 
from your computer system. It is the recipient's responsibility to check this 
message and files for viruses.

Thank you.
***********************************************************************************