-
Notifications
You must be signed in to change notification settings - Fork 2.1k
/
INSTALL
203 lines (152 loc) · 8.43 KB
/
INSTALL
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
Installing John the Ripper.
First of all, most likely you do not need to install John the Ripper
system-wide. Instead, after you extract the distribution archive and
possibly compile the source code (see below), you may simply enter the
"run" directory and invoke John from there.
System-wide installation is also supported, but it is intended for use
by packagers of John for *BSD "ports", Linux distributions, etc., rather
than by end-users. (If you're in fact preparing a package of John,
please refer to the JOHN_SYSTEMWIDE setting in src/params.h or the
Jumbo/autoconf-specific README-DISTROS.)
You may have obtained the source code or a "binary" (pre-compiled)
distribution of John the Ripper. On Unix-like systems, it is typical
to get the source code and compile it into "binary" executables right
on the system you intend to run John on. On Windows, however, it is
typical to get a binary distribution which is ready for use.
The following instructions apply to the source code distribution of
John only. If you have a binary distribution, then there's nothing
for you to compile and you can start using John right away.
Also, the instructions here are mostly generic (not very OS specific).
Please refer to INSTALL-FEDORA, INSTALL-UBUNTU, or INSTALL-WINDOWS for
instructions specific to those systems.
Requirements.
Much functionality of JtR jumbo requires OpenSSL. To get all of the related
functionality, you need OpenSSL 1.0.1 or later. With slightly older OpenSSL,
you'll receive partial functionality. You can also build without OpenSSL at
all by passing the "--without-openssl" option to "./configure" (see below),
but the functionality of such build will be greatly reduced.
For building with OpenSSL, not only the runtime libraries are needed but also
"development" stuff like header files. This is often a separate package, so,
e.g., for Ubuntu you need "libssl-dev" as well as "libssl".
Some helper tools are written in Python. A couple helper tools (vncpcap2john
and SIPdump) need libpcap. A couple of formats need extra libraries in order
to get included:
mozilla libnss
krb5-18/23 libkrb5
wowsrp libgmp (will build without it, but is slower then)
Again, you also need e.g. libnss-dev, libkrb5-dev and libgmp-dev in order to
build.
Some formats also need libz and libbz2 (and their "development" components) for
full functionality.
Compiling the sources on a Unix-like system.
Enter the directory into which you extracted the source code
distribution of John. Enter the "src" subdirectory.
cd src
This version of Jumbo has autoconf - that is, it supports the very common
chain of:
./configure && make
Pass a CFLAGS variable in case you need to enable/disable compiler options
different than default. For example, here's how to configure for just AVX
on a host that is AVX2-capable (default is to find all SIMD features you've
got and enable them):
./configure CFLAGS="-g -O2 -mno-avx2"
Note that whenever you pass CFLAGS like this, the default "-g -O2" will go
away, so add that too (like was done above).
You might want to add "-s" to make, for suppressing normal output and just
print warnings and errors. For a multi-core host you can also add e.g. "-j4"
for using 4 processes (hopefully on different cores) in parallel when building.
This can be written together, as in:
./configure && make -sj4
That -j option does not work well with the "clean" target so if you want to
use that, do it separately:
./configure && make -s clean && make -sj4
./configure should find and enable any extra stuff you have, including OMP,
OpenCL, and extra libraries mentioned above. The only exception is MPI
which needs the "--enable-mpi" option to be supplied to configure (nearly
all users should use the --fork option instead of MPI so wont need this). If
something is missing, it's likely installed in a non-standard location so
you'd need to pass parameters:
./configure LDFLAGS=-L/opt/lib CPPFLAGS=-I/opt/include && make -sj4
Note that it's important to use CPPFLAGS and not CFLAGS in the example above,
so that you don't override our compiler optimization flags in CFLAGS.
If you have a broken pkg-config (eg. claiming there is no OpenSSL even though
there is one) you can disable its use within configure:
./configure --disable-pkg-config
You can also disable certain features you don't want, eg:
./configure --disable-openmp && make -sj4
If everything goes well, this will create the executables for John and
its related utilities under "../run/". You can change directory to
there and start John, like this:
cd ../run
./john --test
Alternatively, you may copy the entire "run" directory to anywhere you
like and use John from there.
A note on moving binaries between systems.
With the "generic" make target, certain machine hardware performance
parameters are detected at compile time. Additionally, some OS-specific
make targets tell the C compiler to generate and optimize code for the
machine's specific CPU type (this currently applies to C compilers other
than gcc only). If you then move the binary executable to a different
machine, you might not get the best performance or the program might
not run at all if the CPU lacks features that the C compiler assumed it
would have. Thus, it is recommended to recompile John on each system if
you use one of these make targets.
Since Linux and *BSD distributions' packages of John typically use make
targets other than "generic" and since they typically use gcc, they are
usually not affected by this potential problem.
OMP fallback build.
Even a non-distro "native" build can gain from having OMP fallback. This
means you will have a "john" binary that has OpenMP support, but that it
automatically (quitely and seamlessly) will switch to a "john-non-omp"
binary when applicable. This gains performance and has some other benfits
too. Here's how to do it for your "home build":
./configure --disable-openmp &&
make -s clean && make -sj4 && mv ../run/john ../run/john-non-omp &&
./configure CPPFLAGS='-DOMP_FALLBACK -DOMP_FALLBACK_BINARY="\"john-non-omp\""' &&
make -s clean && make -sj4 && echo All Done
Then you just run "./john" (from run directory) as usual.
Optimal interleaving factors (intrinsics builds).
There is a script "testparas.pl" in the src directory, that can be used
to test for optimal performance. Run it on an idle system. It will finish
with printing a full "./configure" command line to use. For most people,
it's actually overkill. Spend the time learning tricks instead!
Optimal build on OS X.
Using OS X, you can install Xcode (free in App Store) and then its "command
line tools" and after that a normal build should work fine. However, using
native gcc (which is really clang) results in suboptimal performance and some
formats are disabled due to ancient OpenSSL. Also, clang doesn't have any
OpenMP support.
Here's how to make the best possible of your hardware. There are alternatives
that probably work fine but these instructions are for "Homebrew":
1. Install Xcode's command-line tools:
xcode-select --install
You might need to re-run that command after an update of macOS, if
you get "error: C compiler cannot create executables" when building.
2. Install Homebrew, by following the instructions given here:
https://brew.sh
3. Install Homebrew's gcc and OpenSSL, and some libs:
brew install gcc openssl libpcap
4. Consider adding homebrew directories such as /usr/local/bin and/or
/opt/homebrew/bin to your $PATH.
5. Configure, possibly adding a CC option for pointing to a specific gcc and
using whatever LDFLAGS and CPPFLAGS was recommended when you installed
Homebrew's OpenSSL:
./configure CC="/usr/local/bin/gcc-13" \
LDFLAGS="-L/usr/local/opt/openssl/lib " \
CPPFLAGS="-I/usr/local/opt/openssl/include"
6. Clean old files and make:
make -s clean && make -sj4
After the above, you should have a fairly optimal build with AVX/ASIMD or
whatever extra features your CPU has got.
If you get weird problems including things like "error: unknown type name
'dispatch_block_t'" on 10.10 Yosemite, you might need to apply a patch for
the system headers (at your own risk, as always, but backups are created).
From John's "src" directory:
sudo patch -b -N -p0 < unused/Yosemite.patch
The patch is not needed for 10.11 "El Capitan" or later!
NOTE: The above command will create backup files. If you ever want to restore
everything as it were originally:
cd /usr/include/dispatch
sudo mv -vi object.h.orig object.h
sudo mv -vi queue.h.orig queue.h
cd -