boogie(1) compiler for the Boogie programming language


boogie [options] file.bpl ...


boogie is a compiler for Microsoft Research's Boogie programming language, an imperative compiler intermediate language with built-in specification constructs. The integrated static analyzer can verify functional correctness as part of the compilation process.


boogie accepts a number of different types of options.

File macros

A number of boogie options accept a file argument, which may contain the following macros:
Expands to the last filename specified on the command line.
Expands to the concatenation of strings given by /logPrefix options.
Expands to the current time.

General options

Control printing of command-line arguments. Accepted values for n are:
Suppress printing of command-line arguments.
(default) Print command-line arguments during BPL print and prover log.
Print command-line arguments during BPL print and prover log, and also to standard output.
Suppress printing of the version number and copyright message.
Wait for a carriage return from the keyboard before exiting.
Also produce output in XML format to file.

Boogie options

Multiple .bpl files supplied on the command line are concatenated into one Boogie program.
Control printing of enhanced error messages. Accepted values for n are:
(default) Do not print enhanced error messages.
Print Z3 error model enhanced error messages.
Unroll loops, following up to n back edges (and then some).
Save the model in BVD format to the specified file.
Parse only.
Parse and resolve only.
Skip any implementation with resolution or type checking errors.
Print Boogie program to the specified file after parsing it. Specify /print:- to print to standard output.
Print the control flow graph of each implementation in dot(1) format to files named
With /print, desugar calls.
Control printing of Z3's error model. Accepted values for n are:
(default) Do not print Z3's error model.
Print Z3's error model.
Print Z3's error model and reverse mappings.
Print Z3's error model in a more human-readable way.
With /print, print Z3's error model to file instead of standard output.
With /print, desugar all structured statements.
Print augmented information that uniquely identifies variables.
Only check procedures matched by the pattern p. This option may be specified multiple times to match multiple patterns. The pattern p matches the whole procedure name (e.g., a pattern "foo" will only match a procedure called "foo", not one called "fooBar"). The pattern p may contain * wildcards which match any character zero or more times. For example, the pattern "ab*d" would match "abd", "abcd", and "abccd", but not "Aabd" or "abdD". The pattern "*ab*d*" would match "abd", "abcd", "abccd", "Abd", and "abdD".
Enable sound loop unrolling.
When parsing, use the base name of the file for tokens instead of the path supplied on the command line.

Inference options

Instrument inferred invariants as asserts to be checked by the theorem prover.
Perform procedure contract inference.
Use abstract interpretation to infer invariants. domains may be any of the following:
constant propagation
dynamic type
stronger intervals (cannot be combined with other domains)
polyhedra for linear inequalities
trivial bottom/top lattice (cannot be combined with other domains)
You may also specify the following options as pseudo-domains:
debug statistics
number of iterations before applying a widen (default: 0)
The default is /infer:i. With /infer (and no domains), all domains will be used.
Control when inferred invariants are instrumented. Accepted values for flag are:
(default) Instrument inferred invariants only at the beginning of loop headers.
Instrument inferred invariants at the beginning and end of every block. This mode is intended for use in debugging abstract domains.
Turn off the default inference, and override any previous /infer flags.
Print Boogie program after it has been instrumented with invariants.

Debugging and tracing options

Launch and break into the debugger.
Print debug output during translation.
Blurt out various debug trace information. Implies /tracePOs.
Output information about the number of proof obligations.
Output timing information at certain points in the pipeline.

Verification condition generation options

Include free loop invariants as assumptions in checking contexts. Usually, a free loop invariant (or assume statement in that position) is ignored in checking contexts (like other free things).
Translate Boogie's A ==> B into prover's A ==> A && B.
Control when to coalesce blocks. Accepted values for n are:
Do not coalesce blocks.
(default) Coalesce blocks.
Use the specified fixed point engine for inference.
Infer the least number of constants (whose names are prefixed by prefix) that need to be set to true for the program to be correct. Implies /stratifiedInline:1.
Use the specified inlining strategy for procedures with the :inline attribute. Accepted strategies are none, assume (the default), assert, and spec.
Use the lazy inlining algorithm.
Control when and how to perform live variable analysis. Accepted values for n are:
Do not perform live variable analysis.
(default) Perform live variable analysis.
Perform interprocedural live variable analysis.
Do not abstract map types in the encoding. This is an experimental feature which will not do the right thing if the program uses polymorphism.
Skip verification condition generation and invocation of the theorem prover.
Print the implementation after inlining calls to procedures with the :inline attribute.
Set the recursion bound for stratified inlining to n. By default, n is 500.
In the verification condition, generate an auxiliary symbol, elsewhere defined to be +, instead of +.
Control whether to remove empty blocks during verification condition generation. Accepted values for n are:
Do not remove empty blocks.
(default) Remove empty blocks.
Run the soundness smoke test: try to stick assert false; in some places in the BPL and see if we can still prove it.
Set the timeout, in seconds, for a single theorem prover invocation during the smoke test. By default, n is 10.
Use the stratified inlining algorithm.
Control when subsumption is applied to asserted conditions. Accepted values for n are:
Never apply subsumption.
Do not apply subsumption for quantifiers.
(default) Always apply subsumption.
Print debug output during verification condition generation.
Control how to encode types when sending the verification condition to the the theorem prover. Allowed methods are:
none (unsound)
(default) predicates
Specify the verification condition variety. Accepted varieties are:
flat block
(default) DAG
nested block reach
nested block
flat block reach
Verify each input program separately.
Verify several program snapshots (named filename.v0.bpl to filename.vN.bpl), possibly using verification result caching. Accepted values for n are:
(default) Do not use any verification result caching.
Use basic verification result caching.
Use advanced verification result caching.
Use advanced verification result caching, and report errors according to the new source locations for errors and their related locations (but not /errorTrace and CaptureState locations).

Verification condition splitting

Try to verify n verification conditions at once. Defaults to 1.
For the nth split, dump and split.n.bpl. Note that this affects error reporting.
Set the timeout, in seconds, for the single last assertion in keep-going mode. By default, n is 30.
Set the timeout, in seconds, for a single theorem prover invocation in keep-going mode, except for the final single-assertion case. By default, n is 1.
Like /vcsCores:n, where n is the machine's processor count multiplied by f and rounded to the nearest integer. f must be in the range [0.0, 3.0]. This will never set n less than 1.
Verification conditions will not be split unless the cost of a verification condition exceeds f. f defaults to 2000.0. This does not apply in the keep-going mode after the first round of splitting.
If n is set to more than 1, this activates keep-going mode, where after the first round of splitting, verification conditions that time out are split into n pieces and retried until either proving them is successful or there is only one assertion on a single path and it times out. (In such a case, boogie reports an error for that assertion). By default, n is 1 (that is, keep-going mode is disabled).
Set the maximal number of verification conditions generated per method. In keep-going mode, this only applies to the first round. By default, n is 1.
/vcsPathCostMult:f1, /vcsAssumeMult:f2
Controls the cost of a block. Block cost is computed according to the formula
(assert-cost + f2 × assume-cost) × (1 + f1 × entering-paths)
where f1 defaults to 1.0 and f2 defaults to 0.01. The cost of a single assertion or assumption is always 1.0.
Sets a scale factor which boogie will multiply by the number of paths in a block if more than one path join at a block. This is intended to reflect the fact that the prover will learn something on one path before proceeding to the next. By default, f is 0.8.
If the best path split of a verification condition of cost A is into verification conditions of cost B and C, then the split is applied if Af × (B + C). Otherwise, assertion splitting will be applied. By default, f is 0.5 (that is, always do path splitting if possible). Increase f to do less path splitting and more assertion splitting.

Prover options

Limit the number of errors produced for each procedure. By default, n is 5, but some provers may only support 1.
Control whether or not trace labels appear in the error output. Accepted values for n are:
Print no trace labels in the error output.
(default) Print useful trace labels in error output.
Print all trace labels in error output.
Define the expansion of the macro @PREFIX@.
/p:key[:value/], /proverOpt:key[:value/]
Provide a prover-specific option.
Set the platform type and location. ptype may be v11, v2, or cli1, and location should be the platform libraries directory.
Use theorem prover p. p may be a full path to a prover DLL, or it may be one of the following standard provers:
This implies /vc:n and /vcBrackets:1.
(default) Use the SMTLib2 format, and call Z3.
Z3 with the Simplify format.
Z3 with the managed (CLI) API.
Log input for the theorem prover.
In addition to the standard file name macros, file can use the @PROC@ macro, which causes boogie to generate one prover log file per verification condition, expanding @PROC@ to the name of the procedure that the verification condition is for.
Append (do not overwrite) the specified prover log file.
Limit the prover to n megabytes of virtual memory before forcing it to restart. n defaults to 100.
Set the time, in seconds, between closing the stream to the prover and killing the prover process. n defaults to 0.
Control warning output from the prover. Accepted values for n are:
(default) Don't print warning output from the prover.
Print warnings to standard output.
Print warnings to standard error.
Restart the prover after each query.
Limit the number of seconds spent trying to verify each procedure.
Control whether or not odd-charactered identifier names will be bracketed with pipe characters ('|'). Accepted values for n are:
(default) Do not bracket odd-charactered identifier names.
Bracket odd-charactered identifier names.

Prover options (Simplify)

Set Simplify's matching depth limit.

Prover options (Z3)

Use Z3's native array theory, as opposed to axioms. Implies /monomorphize.
Output a model in SMTLib2 format.
Set the path to the Z3 executable. On Debian systems, this defaults to /usr/bin/z3.
Configure use of LETs in Z3. Accepted values for n are:
Do not use LETs.
Use only LET TERM.
(default) Use any LET.
Report multiple counterexamples for each error.
Set an additional Z3 option.
Generate a multi-sorted verification condition that makes use of Z3 types.


Boogie is copyright © 2003-2015 Microsoft Corporation and licensed under the Microsoft Public License <>.

This manual page is copyright © 2013, 2015-2016 Benjamin Barenblat and licensed under the Apache License, Version 2.0.