A specific application doesn't recognize a DLL entry point - c

I have something weird...
I am trying to do a proxy DLL for an old game (based on LithTech engine, abandonware, Windows 98 application), so I used wrappit, made the DLL, made it compiling, but... Well...
The DLL itself seems to be OK. At least the DllMain gets executed when testing with LOADDLL from OllyDbg...
But when I insert it into the game folder instead of the original DLL, it loads the DLL, and then it doesn't execute the DLL entry point DllMain! In some builds (Visual Studio 2005) it just says nothing. In other builds (Visual Studio 2010) it raises the c0000139 exception (which is "Entry point not found"). But how can it be that entry point is not found when it loads fine with LOADDLL in ollydbg, which means that the entry oint is in place?
Please help me!
UPD:
Posting the dumpbin /exports output:
The original DLL:
Microsoft (R) COFF/PE Dumper Version 10.00.30319.01
Copyright (C) Microsoft Corporation. All rights reserved.
Dump of file IMUSIC25_.DLL
File Type: DLL
Section contains the following exports for IMUSIC25.DLL
00000000 characteristics
33713027 time date stamp Thu May 08 05:45:11 1997
0.00 version
1 ordinal base
29 number of functions
29 number of names
ordinal hint RVA name
6 0 0000FD8B DllCanUnloadNow
7 1 0000FDF9 DllGetClassObject
8 2 0000FD85 DllRegisterServer
9 3 0000FD88 DllUnregisterServer
1 4 000105BD _AllocAAEngine2#8
2 5 00010507 _AllocAAEngine#4
3 6 00007101 _AllocAALoader#4
4 7 000074FC _AllocAALoaderLite#4
5 8 00010675 _AllocAAMIDIIn#4
10 9 000106A1 _GetFinalMIDISink#4
11 A 0001127F _LoadBandFile#12
12 B 000115F9 _LoadMotifFile#12
13 C 00011601 _LoadPatternFile#12
14 D 0001123C _LoadPersonalityByName#12
15 E 000111F9 _LoadPersonalityFile#12
16 F 000118F7 _LoadPersonalityFromMemory#12
17 10 0001131E _LoadSectionFile#12
18 11 000114BB _LoadSongFile#12
19 12 000111B6 _LoadStyleByName#12
20 13 00011173 _LoadStyleFile#12
21 14 0001182C _LoadStyleFromMemory#12
22 15 0001155A _LoadTemplateFile#12
23 16 000102B1 _MusicEngineSimpleInit95#12
24 17 00010497 _MusicEngineSimpleInit#12
25 18 00010066 _MusicEngineSimpleInitNT#12
26 19 000119C2 _Panic#4
27 1A 00011465 _SaveSectionAsMIDIFile#8
28 1B 000113E1 _SaveSectionFile#8
29 1C 00010692 _SetAADebug#4
Summary
4000 .data
1000 .idata
1000 .rdata
2000 .reloc
1000 .rsrc
1B000 .text
My crafted DLL:
Microsoft (R) COFF/PE Dumper Version 10.00.30319.01
Copyright (C) Microsoft Corporation. All rights reserved.
Dump of file IMUSIC25.DLL
File Type: DLL
Section contains the following exports for imusic25.dll
00000000 characteristics
52F816DB time date stamp Mon Feb 10 04:01:31 2014
0.00 version
1 ordinal base
30 number of functions
30 number of names
ordinal hint RVA name
6 0 00001020 DllCanUnloadNow = ___E__0__#0
7 1 00001030 DllGetClassObject = ___E__1__#0
8 2 00001040 DllRegisterServer = ___E__2__#0
9 3 00001050 DllUnregisterServer = ___E__3__#0
2 4 00001060 _AllocAAEngine = ___E__5__#0
1 5 00001130 _AllocAAEngine2 = ___E__4__#0
3 6 00001070 _AllocAALoader = ___E__6__#0
4 7 00001080 _AllocAALoaderLite = ___E__7__#0
5 8 00001090 _AllocAAMIDIIn = ___E__8__#0
30 9 00001000 _DllMain#12 = _DllMain#12
10 A 000010A0 _GetFinalMIDISink = ___E__9__#0
11 B 000010B0 _LoadBandFile = ___E__10__#0
12 C 000010C0 _LoadMotifFile = ___E__11__#0
13 D 000010D0 _LoadPatternFile = ___E__12__#0
14 E 000010E0 _LoadPersonalityByName = ___E__13__#0
15 F 000010F0 _LoadPersonalityFile = ___E__14__#0
16 10 00001100 _LoadPersonalityFromMemory = ___E__15__#0
17 11 00001130 _LoadSectionFile = ___E__4__#0
18 12 00001110 _LoadSongFile = ___E__17__#0
19 13 00001120 _LoadStyleByName = ___E__18__#0
20 14 00001130 _LoadStyleFile = ___E__4__#0
21 15 00001150 _LoadStyleFromMemory = ___E__20__#0
22 16 00001160 _LoadTemplateFile = ___E__21__#0
24 17 00001180 _MusicEngineSimpleInit = ___E__23__#0
23 18 00001170 _MusicEngineSimpleInit95 = ___E__22__#0
25 19 00001190 _MusicEngineSimpleInitNT = ___E__24__#0
26 1A 000011A0 _Panic = ___E__25__#0
27 1B 000011B0 _SaveSectionAsMIDIFile = ___E__26__#0
28 1C 000011C0 _SaveSectionFile = ___E__27__#0
29 1D 000011D0 _SetAADebug = ___E__28__#0
Summary
1000 .data
1000 .rdata
1000 .rsrc
1000 .text
UPD: I checked the "GetLastError" on the failing LoadLibraryA call - it is 0x1E7 (487) - ERROR_INVALID_ADDRESS

Well, I managed to fix that by changing the definition of the functions and making them like in the original one.
For that, you need to set the function type to WINAPI (actually it is __stdcall but it seems that the compiler/linker somehow differentiates them? Because the mangling behaviour was different between __stdcall and WINAPI) and define the function parameters - you can calculate the number of needed ones from the last part of the mangled name (which is actually not present in the resulting DLL with the DEF file): #4 means one 4 bytes sized parameter (example - int), #8 means two, etc. You don't even need to know which is exactly the parameter type if you are working with wrappit generated functions - you just need to adjust the size to make mangling correct.
Next, you need to define the relation between the mangled names and ordinals in the DEF file.
There is probably a better and easier approach of using #pragma comment(linker, "/EXPORT:Function=Function#4,#1"), as per C++ DLL Export: Decorated/Mangled names (#wqw's answer) but I haven't tried it.
Also I needed to set up the relocations in project settings - that was probably the cause of "ERROR_INVALID_ADDRESS".
Case closed.

Related

What is a correct name for an operation that turns 3-column long table into a compact "2D" table with variable number of columns?

For example, from this table
row
col
val
0
A
32
0
B
31
0
C
35
1
A
30
1
B
29
1
C
29
2
A
15
2
B
14
2
D
18
3
A
34
3
B
39
3
C
34
3
D
35
it should produce this table:
A
B
C
D
0
32
31
35
1
30
29
29
2
15
14
18
3
34
39
34
35
Is there some official, canonical (or at least popular specific unambiguous) term for such operation (or its reverse)?
I am trying to find (or implement & publish) a tool that transforms CSV this way, but am unsure what to search for (or how to name it).
The term is pivot.
Some databases have native support for pivot, eg SQL Server's PIVOT (and even UNPIVOT) keywords.
For most databases you must craft a query that does the job.

How do I perform shell sort using sequence {3,2,1}?

Suppose I have an array:
30 20 29 19 28 18 27 17 26 16 25 15 24 14 23 13 22 12 21 11
I am not understanding how to do the shell sort using sequence 3:
Would I simply do this:
30 20 29 19 28 18 27 17 | 26 16 25 15 24 14 |23 13 22 12 21 11
Where I split it into 3 parts and sort the respective parts? And then do the same with 2 sorting after, except split into halves? What is the right way to do this? Can someone please explain?
If you look at your array and number the locations
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
30 20 29 19 28 18 27 17 26 16 25 15 24 14 23 13 22 12 21 11
In a shell sort, what you do is start with a skip number (in your case 3) so to make the first "list" you take a number and skip. With 3 this would be 1st, 4th, 7th etc.
So you would have a list of
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
30 19 27 16 24 13 21
and a second list of
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
20 28 17 25 14 22 11
The 3rd list is the remaining items.
For the next round you do it with one less... so items at odd number locations and items at even number locations.
In response to comment below
A Shell sort is an in-place sort — that means you don't remove the items to new lists or create any new data structures. You are using the array to "treat" items which are "far apart" in terms of array locations as next to each other. You don't actually make new lists or new arrays (that is why I showed my diagrams as I did); you just look at these locations.
Why?
Because it means that when you start (for example with 3) you are moving stuff farther) -- eg the 13 that starts at location 16 gets moved to location 1 in the first pass. Then as you reduce the number you start doing more local changes. This means you gain an advantage over a typical bubble sort. Still not good — but MUCH better than a bubble sort.

Same column of different files into the same new file

I have multiple folders Case-1, Case-2....Case-N and they all have a file named PPD. I want to extract all 2nd columns and put them into one file named 123.dat.
It seems that I cannot use awk in a for loop.
case=$1
for (( i = 1; i <= $case ; i ++ ))
do
file=Case-$i
cp $file/PPD temp$i.dat
awk 'FNR==1{f++}{a[f,FNR]=$2}
END
{for(x=1;x<=FNR;x++)
{for(y=1;y<ARGC;y++)
printf("%s ",a[y,x]);print ""} }'
temp$i.dat >> 123.dat
done
Now 123.dat only has the date of the last PPD in Case-N
I know I can use join(I used that command before) if every PPD file has at least one column the same, but it turns out to be extremely slow if I have lots of Case folders
Maybe
eval paste $(printf ' <(cut -f2 %s)' Case-*/PPD)
There is probably a limit to how many process substitutions you can perform in one go. I did this with 20 columns and it was fine. Process substitutions are a Bash feature, so not portable to other Bourne-compatible shells in general.
The wildcard will be expanded in alphabetical order. If you want the cases in numerical order, maybe use case-[1-9] case-[1-9][0-9] case-[1-9][0-9][0-9] to force the expansion to get the single digits first, then the double digits, etc.
The interaction between the outer shell script and inner awk invocation aren't working the way you expect.
Every time through the loop, the shell script calls awk a new time, which means that f will be unset, and then that first clause will set it to 1. It will never become 2. That is, you are starting a new awk process for each iteration through the outer loop, and awk is starting from scratch each time.
There are other ways to structure your code, but as a minimal tweak, you can pass in the number $i to the awk invocation using the -v option, e.g. awk -v i="$i" ....
Note that there are better ways to structure your overall solution, as other answerers have already suggested; I meant this response to be an answer the question, "Why doesn't this work?" and not "Please rewrite this code."
The below AWK program can help you.
#!/usr/bin/awk -f
BEGIN {
# Defaults
nrecord=1
nfiles=0
}
BEGINFILE {
# Check if the input file is accessible,
# if not skip the file and print error.
if (ERRNO != "") {
print("Error: ",FILENAME, ERRNO)
nextfile
}
}
{
# Check if the file is accessed for the first time
# if so then increment nfiles. This is to keep count of
# number of files processed.
if ( FNR == 1 ) {
nfiles++
} else if (FNR > nrecord) {
# Fetching the maximum size of the record processed so far.
nrecord=FNR
}
# Fetch the second column from the file.
array[nfiles,FNR]=$2
}
END {
# Iterate through the array and print the records.
for (i=1; i<=nrecord; i++) {
for (j=1; j<=nfiles; j++) {
printf("%5s", array[j,i])
}
print ""
}
}
Output:
$ ./get.awk Case-*/PPD
1 11 21
2 12 22
3 13 23
4 14 24
5 15 25
6 16 26
7 17 27
8 18 28
9 19 29
10 20 30
Here the Case*/PPD expands to Case-1/PPD, Case-2/PPD, Case-3/PPD and so on. Below are the source files for which the output was generated.
$ cat Case-1/PPD
1 1 1 1
2 2 2 2
3 3 3 3
4 4 4 4
5 5 5 5
6 6 6 6
7 7 7 7
8 8 8 8
9 9 9 9
10 10 10 10
$ cat Case-2/PPD
11 11 11 11
12 12 12 12
13 13 13 13
14 14 14 14
15 15 15 15
16 16 16 16
17 17 17 17
18 18 18 18
19 19 19 19
20 20 20 20
$ cat Case-3/PPD
21 21 21 21
22 22 22 22
23 23 23 23
24 24 24 24
25 25 25 25
26 26 26 26
27 27 27 27
28 28 28 28
29 29 29 29
30 30 30 30

Trouble reading a data set into R

I am new to R and I am trying to read in a data set. The data set is here:
http://petitlien.fr/myfiles
(The above link will expand to a GMX File storage folder link and click on Guest access to retrieve the file.)
The file named mydata.log has 32 entries with no header and it consists of 2 columns which are delimited by spaces.
I am trying the powerful command scan
test.frame<-scan(file="mydata.log",sep= "", nlines=32,blank.lines.skip=TRUE)
The above just read the first 3 rows:
head(test.frame)
[1] 0.0000 0.0000 144.3210 0.3400 159.4070 0.8925
I have tried also read.table:
test.frame<-read.table(file="mydata.log",sep= "", nrows=32,blank.lines.skip=TRUE)
This one reads the first 6 lines only as shown below:
names(test.frame)
[1] "V1" "V2"
> head(test.frame)
V1 V2
1 0.000 0.0000
2 144.321 0.3400
3 159.407 0.8925
4 198.413 0.9450
5 222.557 0.9975
6 235.464 1.0500
Does someone know how to read this data set properly?
A related question: Can I control the number of significant digits or perhaps decimal places in the data being read in?
Thanks a lot...
This line of your code works perfectly:
test.frame<-read.table(file="mydata.log",sep= "", nrows=32,blank.lines.skip=TRUE)
The reason why you only get 6 lines in your output is because you are using head. To view all lines, just enter the name of your object:
> test.frame
V1 V2
1 0.000 0.0000
2 144.321 0.3400
3 159.407 0.8925
4 198.413 0.9450
5 222.557 0.9975
6 235.464 1.0500
7 296.918 1.1025
8 346.773 1.1550
9 442.955 1.2075
10 694.879 1.2600
11 892.436 1.3125
12 1492.970 1.3650
13 2916.960 1.4175
14 3596.060 1.4700
15 5278.950 1.5225
16 7480.730 1.5750
17 12259.800 1.6275
18 14032.600 1.6800
19 19565.600 1.7325
20 31427.700 1.7850
21 58221.400 1.8375
22 92283.900 1.9900
23 165601.000 1.9425
24 165703.000 1.9950
25 213925.000 2.8750
26 260381.000 2.1000
27 312701.000 2.1525
28 370853.000 2.2050
29 479303.000 2.2575
30 487265.000 2.3100
31 545225.000 2.3625
32 703186.000 2.4150
Here is an easy way to see how many rows you have (useful when you have many observations):
nrow(test.frame)
[1] 32
As for the number of digits, see the round command. To look at the documentation for a command, enter a ? and then the command, in this case a function: ?round
#note that you do not have to put "digits=2", you can just put "2", but this way is clearer
> rounded_test.frame <- round(test.frame, digits=2)
> rounded_test.frame
V1 V2
1 0.00 0.00
2 144.32 0.34
3 159.41 0.89
4 198.41 0.94
5 222.56 1.00
6 235.46 1.05
7 296.92 1.10
8 346.77 1.16
9 442.95 1.21
10 694.88 1.26
11 892.44 1.31
12 1492.97 1.36
13 2916.96 1.42
14 3596.06 1.47
15 5278.95 1.52
16 7480.73 1.57
17 12259.80 1.63
18 14032.60 1.68
19 19565.60 1.73
20 31427.70 1.78
21 58221.40 1.84
22 92283.90 1.99
23 165601.00 1.94
24 165703.00 2.00
25 213925.00 2.88
26 260381.00 2.10
27 312701.00 2.15
28 370853.00 2.21
29 479303.00 2.26
30 487265.00 2.31
31 545225.00 2.36
32 703186.00 2.42
Note in the above I created a new object instead of replacing the current one. If you want to replace the current one and lose the data forever (until you reload the dataset of course!), then you can use this line instead:
test.frame <- round(test.frame, digits=2)
If you don't really want to compress your numbers, you might just be interested in viewing the rounded numbers. You can do this the following command:
print(test.frame,digits=2)
Instead of nrow() as suggested, I would recommend str() ("structure") that gives you more useful information about your data set (class of variables etc). It's also a bit less cryptic....:)

XOR File Decryption

So I have to decrypt a .txt file that is crypted with XOR code and with a repeated password that is unknown, and the goal is to discover the message.
Here are the things that I already know because of the professor:
First I need to find the length of the unknown password
The message has been altered and it doesn't have spaces (this may add a bit more difficulty because the space character has the highest frequency in a message)
Any ideas on how to solve this?
thx in advanced :)
First you need to find out the length of the password. You do this by assessing the Index of Coincidence or Kappa-test. XOR the ciphertext with itself shifted 1 step and count the number of characters that are the same (value 0). You get the Kappa value by dividing the result with the total number of characters minus 1. Shift one more time and again calculate the Kappa value. Shift the ciphertext as many times as needed until you discover the password length. If the length is 4 you should see something similar to this:
Offset Hits
-------------------------
1 2.68695%
2 2.36399%
3 3.79009%
4 6.74012%
5 3.6953%
6 1.81582%
7 3.82744%
8 6.03504%
9 3.60273%
10 1.98052%
11 3.83241%
12 6.5627%
As you see the Kappa value is significantly higher on multiples of 4 (4, 8 and 12) than the others. This suggests that the length of the password is 4.
Now that you have the password length you should again XOR the cipher text with itself but now you shift by multiples of the length. Why? Since the ciphertext looks like this:
THISISTHEPLAINTEXT <- Plaintext
PASSPASSPASSPASSPA <- Password
------------------
EJKELDOSOSKDOWQLAG <- Ciphertext
When two values which are the same are XOR:ed the result is 0:
EJKELDOSOSKDOWQLAG <- Ciphertext
EJKELDOSOSKDOWQLAG <- Ciphertext shifted 4.
Is in reality:
THISISTHEPLAINTEXT <- Plaintext
PASSPASSPASSPASSPA <- Password
THISISTHEPLAINTEXT <- Plaintext
PASSPASSPASSPASSPA <- Password
Which is:
THISISTHEPLAINTEXT <- Plaintext
THISISTHEPLAINTEXT <- Plaintext
As you see the password "disappears" and the plaintext is XOR:ed with itself.
So what can we do now then? You wrote that the spaces are removed. This makes it a bit harder to get the plaintext or password. But not at all impossible.
The following table shows the ciphertext values for all english characters:
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
A 0
B 3 0
C 2 1 0
D 5 6 7 0
E 4 7 6 1 0
F 7 4 5 2 3 0
G 6 5 4 3 2 1 0
H 9 10 11 12 13 14 15 0
I 8 11 10 13 12 15 14 1 0
J 11 8 9 14 15 12 13 2 3 0
K 10 9 8 15 14 13 12 3 2 1 0
L 13 14 15 8 9 10 11 4 5 6 7 0
M 12 15 14 9 8 11 10 5 4 7 6 1 0
N 15 12 13 10 11 8 9 6 7 4 5 2 3 0
O 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
P 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 0
Q 16 19 18 21 20 23 22 25 24 27 26 29 28 31 30 1 0
R 19 16 17 22 23 20 21 26 27 24 25 30 31 28 29 2 3 0
S 18 17 16 23 22 21 20 27 26 25 24 31 30 29 28 3 2 1 0
T 21 22 23 16 17 18 19 28 29 30 31 24 25 26 27 4 5 6 7 0
U 20 23 22 17 16 19 18 29 28 31 30 25 24 27 26 5 4 7 6 1 0
V 23 20 21 18 19 16 17 30 31 28 29 26 27 24 25 6 7 4 5 2 3 0
W 22 21 20 19 18 17 16 31 30 29 28 27 26 25 24 7 6 5 4 3 2 1 0
X 25 26 27 28 29 30 31 16 17 18 19 20 21 22 23 8 9 10 11 12 13 14 15 0
Y 24 27 26 29 28 31 30 17 16 19 18 21 20 23 22 9 8 11 10 13 12 15 14 1 0
Z 27 24 25 30 31 28 29 18 19 16 17 22 23 20 21 10 11 8 9 14 15 12 13 2 3 0
What does this mean then? If an A and a B is XOR:ed then the resulting value is 3. E and P will result in 21. Etc. OK but how will this help you?
Remember that the plaintext is XOR:ed with itself shifted by multiples of the password length. For each value you can check the above table and determine what combinations that position could have. Lets say the value is 25 then the two characters that resulted in the value 25 could be one of the following combinations:(I-P), (H-Q), (K-R), (J-S), (M-T), (L-U), (O-V), (N-W), (A-X) or (C-Z). But which one? Now you do more shifts and look up the corresponding values in the table again for each position. Next time the value might be 7 and since you already have a list of possible character combinations you only check against them. At the next two shifts the values are 3 and 1. Now you can determine that the character is W since that is the only common character in each shift, (N-W), (P-W), (T-W), (V-W). You can do this for most positions.
You will not get all the plaintext but you will get enough characters to discover the password. Take the known characters and XOR them in the correct position in the ciphertext. This will yield the password. The number of known characters you need atleast is the number of characters in the password if they are at the "correct" positions in regards to the password.
Good luck!
you should look at cracking a vigenere chiffre, especially at auto-correlation. The latter will help you finding out the length of the password and the rest is usually just bruteforcing on the normal distribution of letters (where the most common one is the letter e in the english language).
Although spaces are the most common characters and make decryptions like this easy, the other character also have different frequencies. For example, see this Wikipedia article. If you've got enough encrypted text and the password length isn't too large, it might just be enough to find out the most common bytes in the encrypted text. They will most likely be the encrypted versions of e that has the highest frequency in english texts.
This alone won't give you the decrypted text, but it's very likely you can find out the password length and (part of) the password itself with it. For example, let's assume the most frequent encrypted bytes are
w x m z y
with almost the same frequency and there's a significant drop in frequency after the last one. This will tell you two things:
The password length most likely is 5, because statistically, all encrypted e will be equally likely. EDIT: OK, this isn't correct, it will be 5 or above because the password can contain the same character multiple times.
The password will be some permutation of (w x m z y XOR e e e e e) - you can use the byte offsets modulo the password length to get the correct permutation.
EDIT: The same character occuring in the password multiple times makes things a bit harder, but you'll most likely be able to identify those because as I said, encrypted versions of e will cluster around frequency f - now if the character occurs n times, it will have a frequency near n*f.
The most common three letter trigram in English (assuming the language is probably English) is "the". Place "the" at all possible points on your cyphertext to derive a possible 3 characters of the key. Try each possible key fragment at all other possible positions on the cyphertext and see what you get. For example, "qzg" is unlikely to be correct, but "fen" could be. Look at the spacing between possible positions to derive the key length. With a key length and a key fragment you can place a lot more of the key.
As Lars said, look at ways of decrypting Vigenère, which is effectively what you have here.

Resources