README.ENGINE

   1   ENGINE
   2   ======
   3
   4   With OpenSSL 0.9.6, a new component was added to support alternative
   5   cryptography implementations, most commonly for interfacing with external
   6   crypto devices (eg. accelerator cards). This component is called ENGINE,
   7   and its presence in OpenSSL 0.9.6 (and subsequent bug-fix releases)
   8   caused a little confusion as 0.9.6** releases were rolled in two
   9   versions, a "standard" and an "engine" version. In development for 0.9.7,
  10   the ENGINE code has been merged into the main branch and will be present
  11   in the standard releases from 0.9.7 forwards.
  12
  13   There are currently built-in ENGINE implementations for the following
  14   crypto devices:
  15
  16       o Microsoft CryptoAPI
  17       o VIA Padlock
  18       o nCipher CHIL
  19
  20   In addition, dynamic binding to external ENGINE implementations is now
  21   provided by a special ENGINE called "dynamic". See the "DYNAMIC ENGINE"
  22   section below for details.
  23
  24   At this stage, a number of things are still needed and are being worked on:
  25
  26       1 Integration of EVP support.
  27       2 Configuration support.
  28       3 Documentation!
  29
  30 1 With respect to EVP, this relates to support for ciphers and digests in
  31   the ENGINE model so that alternative implementations of existing
  32   algorithms/modes (or previously unimplemented ones) can be provided by
  33   ENGINE implementations.
  34
  35 2 Configuration support currently exists in the ENGINE API itself, in the
  36   form of "control commands". These allow an application to expose to the
  37   user/admin the set of commands and parameter types a given ENGINE
  38   implementation supports, and for an application to directly feed string
  39   based input to those ENGINEs, in the form of name-value pairs. This is an
  40   extensible way for ENGINEs to define their own "configuration" mechanisms
  41   that are specific to a given ENGINE (eg. for a particular hardware
  42   device) but that should be consistent across *all* OpenSSL-based
  43   applications when they use that ENGINE. Work is in progress (or at least
  44   in planning) for supporting these control commands from the CONF (or
  45   NCONF) code so that applications using OpenSSL's existing configuration
  46   file format can have ENGINE settings specified in much the same way.
  47   Presently however, applications must use the ENGINE API itself to provide
  48   such functionality. To see first hand the types of commands available
  49   with the various compiled-in ENGINEs (see further down for dynamic
  50   ENGINEs), use the "engine" openssl utility with full verbosity, ie;
  51        openssl engine -vvvv
  52
  53 3 Documentation? Volunteers welcome! The source code is reasonably well
  54   self-documenting, but some summaries and usage instructions are needed -
  55   moreover, they are needed in the same POD format the existing OpenSSL
  56   documentation is provided in. Any complete or incomplete contributions
  57   would help make this happen.
  58
  59   STABILITY & BUG-REPORTS
  60   =======================
  61
  62   What already exists is fairly stable as far as it has been tested, but
  63   the test base has been a bit small most of the time. For the most part,
  64   the vendors of the devices these ENGINEs support have contributed to the
  65   development and/or testing of the implementations, and *usually* (with no
  66   guarantees) have experience in using the ENGINE support to drive their
  67   devices from common OpenSSL-based applications. Bugs and/or inexplicable
  68   behaviour in using a specific ENGINE implementation should be sent to the
  69   author of that implementation (if it is mentioned in the corresponding C
  70   file), and in the case of implementations for commercial hardware
  71   devices, also through whatever vendor support channels are available.  If
  72   none of this is possible, or the problem seems to be something about the
  73   ENGINE API itself (ie. not necessarily specific to a particular ENGINE
  74   implementation) then you should mail complete details to the relevant
  75   OpenSSL mailing list. For a definition of "complete details", refer to
  76   the OpenSSL "README" file. As for which list to send it to;
  77
  78      openssl-users: if you are *using* the ENGINE abstraction, either in an
  79           pre-compiled application or in your own application code.
  80
  81      openssl-dev: if you are discussing problems with OpenSSL source code.
  82
  83   USAGE
  84   =====
  85
  86   The default "openssl" ENGINE is always chosen when performing crypto
  87   operations unless you specify otherwise. You must actively tell the
  88   openssl utility commands to use anything else through a new command line
  89   switch called "-engine". Also, if you want to use the ENGINE support in
  90   your own code to do something similar, you must likewise explicitly
  91   select the ENGINE implementation you want.
  92
  93   Depending on the type of hardware, system, and configuration, "settings"
  94   may need to be applied to an ENGINE for it to function as expected/hoped.
  95   The recommended way of doing this is for the application to support
  96   ENGINE "control commands" so that each ENGINE implementation can provide
  97   whatever configuration primitives it might require and the application
  98   can allow the user/admin (and thus the hardware vendor's support desk
  99   also) to provide any such input directly to the ENGINE implementation.
 100   This way, applications do not need to know anything specific to any
 101   device, they only need to provide the means to carry such user/admin
 102   input through to the ENGINE in question. Ie. this connects *you* (and
 103   your helpdesk) to the specific ENGINE implementation (and device), and
 104   allows application authors to not get buried in hassle supporting
 105   arbitrary devices they know (and care) nothing about.
 106
 107   A new "openssl" utility, "openssl engine", has been added in that allows
 108   for testing and examination of ENGINE implementations. Basic usage
 109   instructions are available by specifying the "-?" command line switch.
 110
 111   DYNAMIC ENGINES
 112   ===============
 113
 114   The new "dynamic" ENGINE provides a low-overhead way to support ENGINE
 115   implementations that aren't pre-compiled and linked into OpenSSL-based
 116   applications. This could be because existing compiled-in implementations
 117   have known problems and you wish to use a newer version with an existing
 118   application. It could equally be because the application (or OpenSSL
 119   library) you are using simply doesn't have support for the ENGINE you
 120   wish to use, and the ENGINE provider (eg. hardware vendor) is providing
 121   you with a self-contained implementation in the form of a shared-library.
 122   The other use-case for "dynamic" is with applications that wish to
 123   maintain the smallest foot-print possible and so do not link in various
 124   ENGINE implementations from OpenSSL, but instead leaves you to provide
 125   them, if you want them, in the form of "dynamic"-loadable
 126   shared-libraries. It should be possible for hardware vendors to provide
 127   their own shared-libraries to support arbitrary hardware to work with
 128   applications based on OpenSSL 0.9.7 or later. If you're using an
 129   application based on 0.9.7 (or later) and the support you desire is only
 130   announced for versions later than the one you need, ask the vendor to
 131   backport their ENGINE to the version you need.
 132
 133   How does "dynamic" work?
 134   ------------------------
 135     The dynamic ENGINE has a special flag in its implementation such that
 136     every time application code asks for the 'dynamic' ENGINE, it in fact
 137     gets its own copy of it. As such, multi-threaded code (or code that
 138     multiplexes multiple uses of 'dynamic' in a single application in any
 139     way at all) does not get confused by 'dynamic' being used to do many
 140     independent things. Other ENGINEs typically don't do this so there is
 141     only ever 1 ENGINE structure of its type (and reference counts are used
 142     to keep order). The dynamic ENGINE itself provides absolutely no
 143     cryptographic functionality, and any attempt to "initialise" the ENGINE
 144     automatically fails. All it does provide are a few "control commands"
 145     that can be used to control how it will load an external ENGINE
 146     implementation from a shared-library. To see these control commands,
 147     use the command-line;
 148
 149        openssl engine -vvvv dynamic
 150
 151     The "SO_PATH" control command should be used to identify the
 152     shared-library that contains the ENGINE implementation, and "NO_VCHECK"
 153     might possibly be useful if there is a minor version conflict and you
 154     (or a vendor helpdesk) is convinced you can safely ignore it.
 155     "ID" is probably only needed if a shared-library implements
 156     multiple ENGINEs, but if you know the engine id you expect to be using,
 157     it doesn't hurt to specify it (and this provides a sanity check if
 158     nothing else). "LIST_ADD" is only required if you actually wish the
 159     loaded ENGINE to be discoverable by application code later on using the
 160     ENGINE's "id". For most applications, this isn't necessary - but some
 161     application authors may have nifty reasons for using it. The "LOAD"
 162     command is the only one that takes no parameters and is the command
 163     that uses the settings from any previous commands to actually *load*
 164     the shared-library ENGINE implementation. If this command succeeds, the
 165     (copy of the) 'dynamic' ENGINE will magically morph into the ENGINE
 166     that has been loaded from the shared-library. As such, any control
 167     commands supported by the loaded ENGINE could then be executed as per
 168     normal. Eg. if ENGINE "foo" is implemented in the shared-library
 169     "libfoo.so" and it supports some special control command "CMD_FOO", the
 170     following code would load and use it (NB: obviously this code has no
 171     error checking);
 172
 173        ENGINE *e = ENGINE_by_id("dynamic");
 174        ENGINE_ctrl_cmd_string(e, "SO_PATH", "/lib/libfoo.so", 0);
 175        ENGINE_ctrl_cmd_string(e, "ID", "foo", 0);
 176        ENGINE_ctrl_cmd_string(e, "LOAD", NULL, 0);
 177        ENGINE_ctrl_cmd_string(e, "CMD_FOO", "some input data", 0);
 178
 179     For testing, the "openssl engine" utility can be useful for this sort
 180     of thing. For example the above code excerpt would achieve much the
 181     same result as;
 182
 183        openssl engine dynamic \
 184                  -pre SO_PATH:/lib/libfoo.so \
 185                  -pre ID:foo \
 186                  -pre LOAD \
 187                  -pre "CMD_FOO:some input data"
 188
 189     Or to simply see the list of commands supported by the "foo" ENGINE;
 190
 191        openssl engine -vvvv dynamic \
 192                  -pre SO_PATH:/lib/libfoo.so \
 193                  -pre ID:foo \
 194                  -pre LOAD
 195
 196     Applications that support the ENGINE API and more specifically, the
 197     "control commands" mechanism, will provide some way for you to pass
 198     such commands through to ENGINEs. As such, you would select "dynamic"
 199     as the ENGINE to use, and the parameters/commands you pass would
 200     control the *actual* ENGINE used. Each command is actually a name-value
 201     pair and the value can sometimes be omitted (eg. the "LOAD" command).
 202     Whilst the syntax demonstrated in "openssl engine" uses a colon to
 203     separate the command name from the value, applications may provide
 204     their own syntax for making that separation (eg. a win32 registry
 205     key-value pair may be used by some applications). The reason for the
 206     "-pre" syntax in the "openssl engine" utility is that some commands
 207     might be issued to an ENGINE *after* it has been initialised for use.
 208     Eg. if an ENGINE implementation requires a smart-card to be inserted
 209     during initialisation (or a PIN to be typed, or whatever), there may be
 210     a control command you can issue afterwards to "forget" the smart-card
 211     so that additional initialisation is no longer possible. In
 212     applications such as web-servers, where potentially volatile code may
 213     run on the same host system, this may provide some arguable security
 214     value. In such a case, the command would be passed to the ENGINE after
 215     it has been initialised for use, and so the "-post" switch would be
 216     used instead. Applications may provide a different syntax for
 217     supporting this distinction, and some may simply not provide it at all
 218     ("-pre" is almost always what you're after, in reality).
 219
 220   How do I build a "dynamic" ENGINE?
 221   ----------------------------------
 222     This question is trickier - currently OpenSSL bundles various ENGINE
 223     implementations that are statically built in, and any application that
 224     calls the "ENGINE_load_builtin_engines()" function will automatically
 225     have all such ENGINEs available (and occupying memory). Applications
 226     that don't call that function have no ENGINEs available like that and
 227     would have to use "dynamic" to load any such ENGINE - but on the other
 228     hand such applications would only have the memory footprint of any
 229     ENGINEs explicitly loaded using user/admin provided control commands.
 230     The main advantage of not statically linking ENGINEs and only using
 231     "dynamic" for hardware support is that any installation using no
 232     "external" ENGINE suffers no unnecessary memory footprint from unused
 233     ENGINEs. Likewise, installations that do require an ENGINE incur the
 234     overheads from only *that* ENGINE once it has been loaded.
 235
 236     Sounds good? Maybe, but currently building an ENGINE implementation as
 237     a shared-library that can be loaded by "dynamic" isn't automated in
 238     OpenSSL's build process. It can be done manually quite easily however.
 239     Such a shared-library can either be built with any OpenSSL code it
 240     needs statically linked in, or it can link dynamically against OpenSSL
 241     if OpenSSL itself is built as a shared library. The instructions are
 242     the same in each case, but in the former (statically linked any
 243     dependencies on OpenSSL) you must ensure OpenSSL is built with
 244     position-independent code ("PIC"). The default OpenSSL compilation may
 245     already specify the relevant flags to do this, but you should consult
 246     with your compiler documentation if you are in any doubt.
 247
 248     This example will show building the "atalla" ENGINE in the
 249     crypto/engine/ directory as a shared-library for use via the "dynamic"
 250     ENGINE.
 251     1) "cd" to the crypto/engine/ directory of a pre-compiled OpenSSL
 252        source tree.
 253     2) Recompile at least one source file so you can see all the compiler
 254        flags (and syntax) being used to build normally. Eg;
 255            touch hw_atalla.c ; make
 256        will rebuild "hw_atalla.o" using all such flags.
 257     3) Manually enter the same compilation line to compile the
 258        "hw_atalla.c" file but with the following two changes;
 259          (a) add "-DENGINE_DYNAMIC_SUPPORT" to the command line switches,
 260          (b) change the output file from "hw_atalla.o" to something new,
 261              eg. "tmp_atalla.o"
 262     4) Link "tmp_atalla.o" into a shared-library using the top-level
 263        OpenSSL libraries to resolve any dependencies. The syntax for doing
 264        this depends heavily on your system/compiler and is a nightmare
 265        known well to anyone who has worked with shared-library portability
 266        before. 'gcc' on Linux, for example, would use the following syntax;
 267           gcc -shared -o dyn_atalla.so tmp_atalla.o -L../.. -lcrypto
 268     5) Test your shared library using "openssl engine" as explained in the
 269        previous section. Eg. from the top-level directory, you might try;
 270           apps/openssl engine -vvvv dynamic \
 271               -pre SO_PATH:./crypto/engine/dyn_atalla.so -pre LOAD
 272        If the shared-library loads successfully, you will see both "-pre"
 273        commands marked as "SUCCESS" and the list of control commands
 274        displayed (because of "-vvvv") will be the control commands for the
 275        *atalla* ENGINE (ie. *not* the 'dynamic' ENGINE). You can also add
 276        the "-t" switch to the utility if you want it to try and initialise
 277        the atalla ENGINE for use to test any possible hardware/driver
 278        issues.
 279
 280   PROBLEMS
 281   ========
 282
 283   It seems like the ENGINE part doesn't work too well with CryptoSwift on Win32.
 284   A quick test done right before the release showed that trying "openssl speed
 285   -engine cswift" generated errors. If the DSO gets enabled, an attempt is made
 286   to write at memory address 0x00000002.
 287