Mono(Mono 2.5) Mono(Mono 2.5)
NAMEmono - Mono's ECMA-CLI native code generator (Just-in-Time and Ahead-
of-Time)
SYNOPSISmono [options] file [arguments...]
DESCRIPTIONmono is a runtime implementation of the ECMA Common Language Infra‐
structure. This can be used to run ECMA and .NET applications.
The runtime contains a native code generator that transforms the Common
Intermediate Language into native code.
The code generator can operate in two modes: just in time compilation
(JIT) or ahead of time compilation (AOT). Since code can be dynami‐
cally loaded, the runtime environment and the JIT are always present,
even if code is compiled ahead of time.
The runtime loads the specified file and optionally passes the argu‐
ments to it. The file is an ECMA assembly. They typically have a .exe
or .dll extension.
The runtime provides a number of configuration options for running
applications, for developing and debugging, and for testing and debug‐
ging the runtime itself.
PORTABILITY
On Unix-based systems, Mono provides a mechanism to emulate the Win‐
dows-style file access, this includes providing a case insensitive view
of the file system, directory separator mapping (from \ to /) and
stripping the drive letters.
This functionality is enabled by setting the MONO_IOMAP environment
variable to one of all, drive and case.
See the description for MONO_IOMAP in the environment variables section
for more details.
RUNTIME OPTIONS
The following options are available:
--aot, --aot[=options]
This option is used to precompile the CIL code in the specified
assembly to native code. The generated code is stored in a file
with the extension .so. This file will be automatically picked
up by the runtime when the assembly is executed. Ahead-of-Time
compilation is most useful if you use it in combination with the
-O=all,-shared flag which enables all of the optimizations in
the code generator to be performed. Some of those optimizations
are not practical for Just-in-Time compilation since they might
be very time consuming. Unlike the .NET Framework, Ahead-of-
Time compilation will not generate domain independent code: it
generates the same code that the Just-in-Time compiler would
produce. Since most applications use a single domain, this is
fine. If you want to optimize the generated code for use in
multi-domain applications, consider using the -O=shared flag.
This pre-compiles the methods, but the original assembly is
still required to execute as this one contains the metadata and
exception information which is not available on the generated
file. When precompiling code, you might want to compile with
all optimizations (-O=all). Pre-compiled code is position inde‐
pendent code. Pre compilation is just a mechanism to reduce
startup time, increase code sharing across multiple mono pro‐
cesses and avoid just-in-time compilation program startup costs.
The original assembly must still be present, as the metadata is
contained there. AOT code typically can not be moved from one
computer to another (CPU-specific optimizations that are
detected at runtime) so you should not try to move the pre-gen‐
erated assemblies or package the pre-generated assemblies for
deployment. A few options are available as a parameter to the
--aot command line option. The options are separated by com‐
mas, and more than one can be specified:
bind-to-runtime-version
If specified, forces the generated AOT files to be bound
to the runtime version of the compiling Mono. This will
prevent the AOT files from being consumed by a different
Mono runtime. full This is currently an experimental
feature as it is not complete. This instructs Mono to
precompile code that has historically not been precom‐
piled with AOT.
outfile=[filename]
Instructs the AOT compiler to save the output to the
specified file.
write-symbols
Instructs the AOT compiler to emit debug symbol informa‐
tion.
save-temps,keep-temps
Instructs the AOT compiler to keep temporary files.
threads=[number]
This is an experimental option for the AOT compiler to
use multiple threads when compiling the methods.
nodebug
Instructs the AOT compiler to not output any debugging
information.
ntrampolines=[number]
When compiling in full aot mode, the method trampolines
must be precreated in the AOT image. You can add addi‐
tional method trampolines with this argument. Defaults
to 1024.
nrgctx-trampolines=[number]
When compiling in full aot mode, the generic sharing
trampolines must be precreated in the AOT image. You can
add additional method trampolines with this argument.
Defaults to 1024.
nimt-trampolines=[number]
When compiling in full aot mode, the IMT trampolines must
be precreated in the AOT image. You can add additional
method trampolines with this argument. Defaults to 128.
print-skipped-methods
If the AOT compiler cannot compile a method for any rea‐
son, enabling this flag will output the skipped methods
to the console.
autoreg
The AOT compiler will emit a (ELF only) library initial‐
izer to automatically register the aot compiled module
with the runtime. This is only useful in static mode
asmonly
Instructs the AOT compiler to output assembly code
instead of an object file.
soft-debug
This instructs the compiler to generate sequence point
checks that allow Mono's soft debugger to debug applica‐
tions even on systems where it is not possible to set
breakpoints or to single step (certain hardware configu‐
rations like the cell phones and video gaming consoles).
static Create an ELF object file (.o) which can be statically
linked into an executable when embedding the mono run‐
time. When this option is used, the object file needs to
be registered with the embedded runtime using the
mono_aot_register_module function which takes as its
argument the mono_aot_module_<ASSEMBLY NAME>_info global
symbol from the object file:
extern void *mono_aot_module_hello_info;
mono_aot_register_module (mono_aot_module_hello_info);
For more information about AOT, see: http://www.mono-
project.com/AOT
--attach=[options]
Currently the only option supported by this command line argu‐
ment is disable which disables the attach functionality.
--full-aot
This is an experimental flag that instructs the Mono runtime to
not generate any code at runtime and depend exclusively on the
code generated from using mono --aot=full previously. This is
useful for platforms that do not permit dynamic code generation.
Notice that this feature will abort execution at runtime if a
codepath in your program, or Mono's class libraries attempts to
generate code dynamically. You should test your software
upfront and make sure that you do not use any dynamic features.
--config filename
Load the specified configuration file instead of the default
one(s). The default files are /etc/mono/config and ~/.mono/con‐
fig or the file specified in the MONO_CONFIG environment vari‐
able, if set. See the mono-config(5) man page for details on
the format of this file.
--desktop
Configures the virtual machine to be better suited for desktop
applications. Currently this sets the GC system to avoid
expanding the heap as much as possible at the expense of slowing
down garbage collection a bit.
--help, -h
Displays usage instructions.
--optimize=MODE, -O=MODE
MODE is a comma separated list of optimizations. They also
allow optimizations to be turned off by prefixing the optimiza‐
tion name with a minus sign. In general, Mono has been tuned to
use the default set of flags, before using these flags for a
deployment setting, you might want to actually measure the bene‐
fits of using them. The following optimizations are imple‐
mented:
all Turn on all optimizations
peephole Peephole postpass
branch Branch optimizations
inline Inline method calls
cfold Constant folding
consprop Constant propagation
copyprop Copy propagation
deadce Dead code elimination
linears Linear scan global reg allocation
cmov Conditional moves [arch-dependency]
shared Emit per-domain code
sched Instruction scheduling
intrins Intrinsic method implementations
tailc Tail recursion and tail calls
loop Loop related optimizations
fcmov Fast x86 FP compares [arch-dependency]
leaf Leaf procedures optimizations
aot Usage of Ahead Of Time compiled code
precomp Precompile all methods before executing Main
abcrem Array bound checks removal
ssapre SSA based Partial Redundancy Elimination
sse2 SSE2 instructions on x86 [arch-dependency]
gshared Enable generic code sharing.
For example, to enable all the optimization but dead code elimi‐
nation and inlining, you can use:
-O=all,-deadce,-inline
The flags that are flagged with [arch-dependency] indicate that
the given option if used in combination with Ahead of Time com‐
pilation (--aot flag) would produce pre-compiled code that will
depend on the current CPU and might not be safely moved to
another computer.
--runtime=VERSION
Mono supports different runtime versions. The version used
depends on the program that is being run or on its configuration
file (named program.exe.config). This option can be used to
override such autodetection, by forcing a different runtime ver‐
sion to be used. Note that this should only be used to select a
later compatible runtime version than the one the program was
compiled against. A typical usage is for running a 1.1 program
on a 2.0 version:
mono --runtime=v2.0.50727 program.exe
--security, --security=mode
Activate the security manager, a currently experimental feature
in Mono and it is OFF by default. The new code verifier can be
enabled with this option as well.
Using security without parameters is equivalent as calling it
with the "cas" parameter.
The following modes are supported:
cas This allows mono to support declarative security
attributes, e.g. execution of Code Access Security (CAS)
or non-CAS demands.
core-clr
Enables the core-clr security system, typically used for
Moonlight/Silverlight applications. It provides a much
simpler security system than CAS, see http://www.mono-
project.com/Moonlight for more details and links to the
descriptions of this new system.
validil
Enables the new verifier and performs basic verification
for code validity. In this mode, unsafe code and
P/Invoke are allowed. This mode provides a better safety
guarantee but it is still possible for managed code to
crash Mono.
verifiable
Enables the new verifier and performs full verification
of the code being executed. It only allows verifiable
code to be executed. Unsafe code is not allowed but
P/Invoke is. This mode should not allow managed code to
crash mono. The verification is not as strict as ECMA
335 standard in order to stay compatible with the MS run‐
time.
The security system acts on user code: code contained in mscor‐
lib or the global assembly cache is always trusted.
--server
Configures the virtual machine to be better suited for server
operations (currently, a no-op).
--verify-all
Verifies mscorlib and assemblies in the global assembly cache
for valid IL, and all user code for IL verifiability.
This is different from --security's verifiable or validil in
that these options only check user code and skip mscorlib and
assemblies located on the global assembly cache.
-V, --version
Prints JIT version information (system configuration, release
number and branch names if available).
DEVELOPMENT OPTIONS
The following options are used to help when developing a JITed applica‐
tion.
--debug, --debug=OPTIONS
Turns on the debugging mode in the runtime. If an assembly was
compiled with debugging information, it will produce line number
information for stack traces.
The optional OPTIONS argument is a comma separated list of
debugging options. These options are turned off by default
since they generate much larger and slower code at runtime.
The following options are supported:
casts Produces a detailed error when throwing a InvalidCastEx‐
ception. This option needs to be enabled as this gener‐
ates more verbose code at execution time.
mdb-optimizations
Disable some JIT optimizations which are usually only
disabled when running inside the debugger. This can be
helpful if you want to attach to the running process with
mdb.
gdb Generate and register debugging information with gdb.
This is only supported on some platforms, and only when
using gdb 7.0 or later.
--profile[=profiler[:profiler_args]]
Turns on profiling. For more information about profiling appli‐
cations and code coverage see the sections "PROFILING" and "CODE
COVERAGE" below.
--trace[=expression]
Shows method names as they are invoked. By default all methods
are traced. The trace can be customized to include or exclude
methods, classes or assemblies. A trace expression is a comma
separated list of targets, each target can be prefixed with a
minus sign to turn off a particular target. The words `pro‐
gram', `all' and `disabled' have special meaning. `program'
refers to the main program being executed, and `all' means all
the method calls. The `disabled' option is used to start up
with tracing disabled. It can be enabled at a later point in
time in the program by sending the SIGUSR2 signal to the run‐
time. Assemblies are specified by their name, for example, to
trace all calls in the System assembly, use:
mono --trace=System app.exe
Classes are specified with the T: prefix. For example, to trace
all calls to the System.String class, use:
mono --trace=T:System.String app.exe
And individual methods are referenced with the M: prefix, and
the standard method notation:
mono --trace=M:System.Console:WriteLine app.exe
As previously noted, various rules can be specified at once:
mono --trace=T:System.String,T:System.Random app.exe
You can exclude pieces, the next example traces calls to Sys‐
tem.String except for the System.String:Concat method.
mono --trace=T:System.String,-M:System.String:Concat
Finally, namespaces can be specified using the N: prefix:
mono --trace=N:System.Xml
--no-x86-stack-align
Don't align stack frames on the x86 architecture. By default,
Mono aligns stack frames to 16 bytes on x86, so that local
floating point and SIMD variables can be properly aligned. This
option turns off the alignment, which usually saves one intruc‐
tion per call, but might result in significantly lower floating
point and SIMD performance.
JIT MAINTAINER OPTIONS
The maintainer options are only used by those developing the runtime
itself, and not typically of interest to runtime users or developers.
--break method
Inserts a breakpoint before the method whose name is `method'
(namespace.class:methodname). Use `Main' as method name to
insert a breakpoint on the application's main method.
--breakonex
Inserts a breakpoint on exceptions. This allows you to debug
your application with a native debugger when an exception is
thrown.
--compile name
This compiles a method (namespace.name:methodname), this is used
for testing the compiler performance or to examine the output of
the code generator.
--compileall
Compiles all the methods in an assembly. This is used to test
the compiler performance or to examine the output of the code
generator
--graph=TYPE METHOD
This generates a postscript file with a graph with the details
about the specified method (namespace.name:methodname). This
requires `dot' and ghostview to be installed (it expects Ghost‐
view to be called "gv"). The following graphs are available:
cfg Control Flow Graph (CFG)
dtree Dominator Tree
code CFG showing code
ssa CFG showing code after SSA translation
optcode CFG showing code after IR optimizations
Some graphs will only be available if certain optimizations are
turned on.
--ncompile
Instruct the runtime on the number of times that the method
specified by --compile (or all the methods if --compileall is
used) to be compiled. This is used for testing the code genera‐
tor performance.
--stats
Displays information about the work done by the runtime during
the execution of an application.
--wapi=hps|semdel
Perform maintenance of the process shared data. semdel will
delete the global semaphore. hps will list the currently used
handles.
-v, --verbose
Increases the verbosity level, each time it is listed, increases
the verbosity level to include more information (including, for
example, a disassembly of the native code produced, code selec‐
tor info etc.).
ATTACH SUPPORT
The Mono runtime allows external processes to attach to a running
process and load assemblies into the running program. To attach to
the process, a special protocol is implemented in the Mono.Management
assembly.
With this support it is possible to load assemblies that have an entry
point (they are created with -target:exe or -target:winexe) to be
loaded and executed in the Mono process.
The code is loaded into the root domain, and it starts execution on the
special runtime attach thread. The attached program should create
its own threads and return after invocation.
This support allows for example debugging applications by having the
csharp shell attach to running processes.
PROFILING
The mono runtime includes a profiler that can be used to explore vari‐
ous performance related problems in your application. The profiler is
activated by passing the --profile command line argument to the Mono
runtime, the format is:
--profile[=profiler[:profiler_args]]
Mono has a built-in profiler called 'default' (and is also the default
if no arguments are specified), but developers can write custom profil‐
ers, see the section "CUSTOM PROFILERS" for more details.
If a profiler is not specified, the default profiler is used. The pro‐
filer_args is a profiler-specific string of options for the profiler
itself. The default profiler accepts the following options 'alloc' to
profile memory consumption by the application; 'time' to profile the
time spent on each routine; 'jit' to collect time spent JIT-compiling
methods and 'stat' to perform sample statistical profiling. If no
options are provided the default is 'alloc,time,jit'.
By default the profile data is printed to stdout: to change this, use
the 'file=filename' option to output the data to filename. For exam‐
ple:
mono--profile program.exe
That will run the program with the default profiler and will do time
and allocation profiling.
mono --profile=default:stat,alloc,file=prof.out program.exe
Will do sample statistical profiling and allocation profiling on pro‐
gram.exe. The profile data is put in prof.out. Note that the statisti‐
cal profiler has a very low overhead and should be the preferred pro‐
filer to use (for better output use the full path to the mono binary
when running and make sure you have installed the addr2line utility
that comes from the binutils package).
LOGGING PROFILER
The logging profiler is a general purpose profiler that can track many
different kinds of events and logs those into a file as the program
executes. This is different than previous profilers in Mono that kept
the information in memory and rendered a summary of the results when
the program shut down.
Using the logging profiler means that useful information can be gath‐
ered for long-running applications, applications that terminate abor‐
mally (crashes, segfaults, complicated tear down processes) or when no
data about the shutdown is required.
The data collected by the running threads is kept independent of each
other to minimize the runtime overhead and the information dumped into
the log at regular intervals.
A sample use is very simple:
$ mono --profile=logging program.exe
$ mprof-decoder program.mprof
In the above example the logging profiler is used in its default mode
that merely records GC statistics for the execution of program.exe.
The profiling data collected is stored in the file program.mprof. The
mprof-decoder tool is then used to analyze the data.
You can instruct the logging profiler to record different one or more
sets of events. These are the modes supported:
Statistical Profiling (stat) the program instruction pointer is
periodically sampled to determine where the program is spending
most of its time. Statistical profiling has a very low impact
on a running application and it is very useful to get a general
picture of where time is being spent on a program.
If call chains are requested, for each sample the profiler gets
a partial stack trace (limited by the specified depth) so that
caller-callee information is available.
Instrumenting: each method enter and exit is logged with a time‐
stamp; further processing of the data can show the methods that
took the longer to execute, with complete accounting for callers
and callees. However, this way of profiling is rather intrusive
and slows down the application significantly.
Allocation: each allocation is logged.
Allocation summary: shows, for each collection, a summary of the
heap contents broken down by class (for each class the number of
allocated and released objects is given, together with their
aggregated size in bytes).
Heap snapshot mode: dumps the whole heap contents at every col‐
lection (or at user specified collections). It is also possible
to request a collection and snapshot dump with a signal.
Moreover, other events can be logged and analyzed, like jit time for
each method, load and unload for assemblies, modules and and individual
classes, and appdomain and thread creation and destruction.
This profiler is activated passing the --profile=logging option to the
mono runtime, and is controlled attaching further options, like --pro‐
file=logging:statistical for doing statistical profiling (multiple
options are separated by commas).
As a quick primer, here are a few examples of the most common usage
modes:
To perform statistical profiling:
mono --profile=logging:stat program.exe
To perform statistical profiling, inspecting call chains up to depth 8:
mono --profile=logging:stat=8 program.exe
To profile allocations (by default the call stack will be analized for
each allocation, producing detailed caller method attribution inforna‐
tion):
mono --profile=logging:allocations program.exe
To profile garbage collection activity at a high level (collection time
and objects freed at each collection for each class are reported, but
heap snapshots are not saved to disk):
mono --profile=logging:allocations-summary program.exe
To perform heap profiling taking heap snapshots:
mono --profile=logging:heap=all program.exe
To write the resulting data to a different file:
mono --profile=logging:output=mydata.mprof program.exe
Then you would need to invoke the decoder mprof-decoder(1) on the out‐
put file to see the profiling results, or to examine heap snapshots and
allocations in detail mprof-heap-viewer(1).
The operating modes described above are the default ones, and are suf‐
ficient to use the profiler.
To further customize the profiler behavior there are more options,
described below.
These options can be individually enabled and disabled prefixing them
with an (optional) '+' character or a '-' character. For instance, the
"allocations" option by default records also the full call stack at
each allocation. If only the caller is wanted, one should use "alloca‐
tions,-save-allocation-stack", or to disable call tracking completely
(making the profiler less intrusive) "allocations,-save-allocation-
caller,-save-allocation-stack". In practice the "allocation" option by
default behaves like "allocations,save-allocation-caller,save-alloca‐
tion-stack", but the user can tune this to his needs.
These are all the available options, organized by category:
Execution profiling modes
statistical, stat or s
Performs statistical profiling. This is a lightweight
profiling mechanism and it has a much lower overhead than
the enter-leave profiling as it works by sampling where
the program is spending its time by using a timer. If
specified with s=<number>, also inspect call chains up to
level <number>.
enter-leave, calls or c
Measure the time spent inside each method call, this is
done by logging the time when a method enters and when
the method leaves. This can be a time consuming opera‐
tion.
jit, j Collect information about time spent by the JIT engine
compiling methods.
Allocation profiling modes
allocations, alloc or a
Collect information about each allocation (object class
and size). By default this also implies "+save-alloca‐
tion-caller" and "+save-allocation-stack".
save-allocation-caller, sac
Save the direct caller of each allocation. The profiler
filters out wrapper methods, and also recognizes if the
allocation has been performed by the runtime while jit‐
ting a method.
save-allocation-stack, sas
Save the full managed execution stack at each allocation.
While the "sac" option saves the direct caller, this one
records the whole stack trace. Note that in the call
stack the wrapper methods are not filtered out. Anyway
the "sac" and "sas" options can be combined, and the
decoder will attribute the allocation to the correct
method even if the wrapper is at the top of the stack
trace.
allocations-summary or as
At each collection dump a summary of the heap contents
(for each class, the number and collective size of all
live and freed heap objects). This very lightweight com‐
pared to full heap snapshots.
unreachable, free or f
Performs a lightweight profile of the garbage collector.
On each collection performed by the GC, the list of
unreachable objects is recorded, and for each object the
class and size is provided. This information can be used
to compute the heap size broken down by class (combined
with "a" can give the same information of "as", but the
log file contains info about each individual object,
while in "as" the processing is done directly at runtime
and the log file contains only the summarized data broken
down by class).
gc or g
Measure the time spent in each collection, and also trace
heap resizes.
heap-shot[=ARG], heap[=ARG] or h[=ARH]
Performs full heap profiling. In this case on each col‐
lection a full heap snapshot is recorded to disk. Inside
the snapshots, each object reference is still represented
so that it's possible to investigate who is responsible
for keeping objects alive.
If the value of ARG is all, a heap snapshot is taken at each
collection.
If the value is an integer n, a snapshot will be taken at the
first n collections (like setting gcd=n );
If no additional argument is given to the heap option, the only
way to take heap snapshots is to requeste them using the runtime
socket based command interface described below (see "Profiler
activity control").
Heap profiling also enables full allocation profiling (with call
stacks), and each allocation can be related to its corresponding
object in the snapshots, enabling investigations like "find all
objects of a given class allocated by a given method and still
live at a given collection, and then find all objects referenc‐
ing them".
This kind of heap snapshot analysis is performed using the
mprof-heap-viewer(1) application.
The number of heap snapshots taken (and the moment in which they
are taken) can be further customized with the following options:
gc-dumps=N, gc-d=N, gcd=N
states the number of snapshots that must be dumped (since
the application starts). Zero means no dumps at all, -1
means dump at all collections.
These options exist because it can happen that the user wants to
investigate
what happens during collections but without forcing a
collection using the command interface, because forcing a
collection alters the program behavior. Of course it is
possible to simply take a snapshot at every collection,
but in some workloads this is could not be feasible (too
much data). So we have this "garbage collection dumps"
counter to control how many snapshots to take.
Profiler activity control
output=FILE, out=FILE or o=FILE
Use this option to provide the output file name for the
profile log. If this option is not specified, it will
default to "<program-name>.mprof".
output-suffix=SUFFIX, suffix=SUFFIX or os=SUFFIX: makes
the output file name equals to "<program-name>-SUF‐
FIX.mprof".
start-enabled or se: start with the profiler active
(which is the default).
start-disabled or sd: start with the profiler inactive.
force-accurate-timer (or fac): the profiler by default uses
rtdsc to acquire timestamps for frequent events, but this
can be imprecise; using this option you force the use of
"gettimeofday" at every event, which is more accurate but
much slower.
command-port=port or cp=port (where port is an integer between
1024 nd 65535): Choose a TCP port where the profiler will
listen for user commands. The protocol is ASCII based
and line oriented (one line per command), and the pro‐
filer answers with one line containing either "OK" or
"ERROR" to each received command.
The user can telnet to this port and give commands manually, or
a GUI can use this facility to control the profiler at runtime.
The available commands are:
enable: Enables the profiler.
disable: Disables the profiler.
heap-snapshot: Takes a heap snapshot now (forces a full garbage
collection).
heap-snapshot-counter=arg: Set the counter of the next heap
snapshots
that must be taken, where arg can be "all" (take a snap‐
shot at every collection), "none" (do not take snap‐
shots), or an integer "n" (take a heap snapshot for the
next "n" collections).
Internal buffer sizes
per-thread-buffer-size=N, tbs=N
Use to specify the number of events that a thread buffer
can hold. When the thread buffer is full, a log block
is written to disk. This defaults to tbs=10000.
statistical-thread-buffer-size=N, sbs=N
The number of statistical samples that are held in memory
before they are dumped to disk (the system does double-
buffering and the statistical samples are written by a
helper thread, so the statistical profiler never stops
and is able to profile the profiler itself). This
defaults to sbs=10000.
write-buffer-size, wbs
Specifies the size in bytes of the internal write buf‐
fers. This defaults to wbs=1024.
In its current state, this profiler can also perform heap analysis like
the HeapShot profiler, but there is no UI to process this information.
Another known issue is that if the timer is not strictly monotonic
(like rtdsc), differences between times can underflow (they are handled
as unsigned integers) and weird numbers can show up in the logs.
Finally, it can happen that when exceptions are thrown the profiler
temporarily loses track of the execution stack and misattributes the
caller for a few allocations (and method execution time).
The output file contains compressed events, to process the data you
should use tools like the "Mono.Profiler" tool provided on the Mono SVN
repository.
More explanations are provided here: "http://www.mono-project.com/Log‐
gingProfiler".
EXTERNAL PROFILERS
There are a number of external profilers that have been developed for
Mono, we will update this section to contain the profilers.
The heap Shot profiler can track all live objects, and references to
these objects, and includes a GUI tool, this is our recommended pro‐
filer. To install you must download the profiler from Mono's SVN:
svn co svn://anonsvn.mono-project.com/source/trunk/heap-shot
cd heap-shot
./autogen
make
make install
See the included documentation for details on using it.
The Live Type profiler shows at every GC iteration all of the live
objects of a given type. To install you must download the profiler
from Mono's SVN:
svn co svn://anonsvn.mono-project.com/source/trunk/heap-prof
cd heap-prof
./autogen
make
make install
To use the profiler, execute:
mono --profile=desc-heap program.exe
The output of this profiler looks like this:
Checkpoint at 102 for heap-resize
System.MonoType : 708
System.Threading.Thread : 352
System.String : 3230
System.String[] : 104
Gnome.ModuleInfo : 112
System.Object[] : 160
System.Collections.Hashtable : 96
System.Int32[] : 212
System.Collections.Hashtable+Slot[] : 296
System.Globalization.CultureInfo : 108
System.Globalization.NumberFormatInfo : 144
The first line describes the iteration number for the GC, in this case
checkpoint 102.
Then on each line the type is displayed as well as the number of bytes
that are being consumed by live instances of this object.
The AOT profiler is used to feed back information to the AOT compiler
about how to order code based on the access patterns for pages. To use
it, use:
mono --profile=aot program.exe
The output of this profile can be fed back into Mono's AOT compiler to
order the functions on the disk to produce precompiled images that have
methods in sequential pages.
CUSTOM PROFILERS
Mono provides a mechanism for loading other profiling modules which in
the form of shared libraries. These profiling modules can hook up to
various parts of the Mono runtime to gather information about the code
being executed.
To use a third party profiler you must pass the name of the profiler to
Mono, like this:
mono --profile=custom program.exe
In the above sample Mono will load the user defined profiler from the
shared library `mono-profiler-custom.so'. This profiler module must be
on your dynamic linker library path.
A list of other third party profilers is available from Mono's web site
(www.mono-project.com/Performance_Tips)
Custom profiles are written as shared libraries. The shared library
must be called `mono-profiler-NAME.so' where `NAME' is the name of your
profiler.
For a sample of how to write your own custom profiler look in the Mono
source tree for in the samples/profiler.c.
CODE COVERAGE
Mono ships with a code coverage module. This module is activated by
using the Mono --profile=cov option. The format is: --pro‐
file=cov[:assembly-name[/namespace]] test-suite.exe
By default code coverage will default to all the assemblies loaded, you
can limit this by specifying the assembly name, for example to perform
code coverage in the routines of your program use, for example the fol‐
lowing command line limits the code coverage to routines in the "demo"
assembly:
mono --profile=cov:demo demo.exe
Notice that the assembly-name does not include the extension.
You can further restrict the code coverage output by specifying a
namespace:
mono --profile=cov:demo/My.Utilities demo.exe
Which will only perform code coverage in the given assembly and names‐
pace.
Typical output looks like this:
Not covered: Class:.ctor ()
Not covered: Class:A ()
Not covered: Driver:.ctor ()
Not covered: Driver:method ()
Partial coverage: Driver:Main ()
offset 0x000a
The offsets displayed are IL offsets.
A more powerful coverage tool is available in the module `monocov'.
See the monocov(1) man page for details.
DEBUGGING AIDS
To debug managed applications, you can use the mdb command, a command
line debugger.
It is possible to obtain a stack trace of all the active threads in
Mono by sending the QUIT signal to Mono, you can do this from the com‐
mand line, like this:
kill -QUIT pid
Where pid is the Process ID of the Mono process you want to examine.
The process will continue running afterwards, but its state is not
guaranteed.
Important: this is a last-resort mechanism for debugging applications
and should not be used to monitor or probe a production application.
The integrity of the runtime after sending this signal is not guaran‐
teed and the application might crash or terminate at any given point
afterwards.
The --debug=casts option can be used to get more detailed information
for Invalid Cast operations, it will provide information about the
types involved.
You can use the MONO_LOG_LEVEL and MONO_LOG_MASK environment variables
to get verbose debugging output about the execution of your application
within Mono.
The MONO_LOG_LEVEL environment variable if set, the logging level is
changed to the set value. Possible values are "error", "critical",
"warning", "message", "info", "debug". The default value is "error".
Messages with a logging level greater then or equal to the log level
will be printed to stdout/stderr.
Use "info" to track the dynamic loading of assemblies.
Use the MONO_LOG_MASK environment variable to limit the extent of the
messages you get: If set, the log mask is changed to the set value.
Possible values are "asm" (assembly loader), "type", "dll" (native
library loader), "gc" (garbage collector), "cfg" (config file loader),
"aot" (precompiler) and "all". The default value is "all". Changing
the mask value allows you to display only messages for a certain compo‐
nent. You can use multiple masks by comma separating them. For example
to see config file messages and assembly loader messages set you mask
to "asm,cfg".
The following is a common use to track down problems with P/Invoke:
$ MONO_LOG_LEVEL="debug" MONO_LOG_MASK="dll" mono glue.exe
SERIALIZATION
Mono's XML serialization engine by default will use a reflection-based
approach to serialize which might be slow for continuous processing
(web service applications). The serialization engine will determine
when a class must use a hand-tuned serializer based on a few parameters
and if needed it will produce a customized C# serializer for your types
at runtime. This customized serializer then gets dynamically loaded
into your application.
You can control this with the MONO_XMLSERIALIZER_THS environment vari‐
able.
The possible values are `no' to disable the use of a C# customized
serializer, or an integer that is the minimum number of uses before the
runtime will produce a custom serializer (0 will produce a custom seri‐
alizer on the first access, 50 will produce a serializer on the 50th
use). Mono will fallback to an interpreted serializer if the serializer
generation somehow fails. This behavior can be disabled by setting the
option `nofallback' (for example: MONO_XMLSERIALIZER_THS=0,nofallback).
ENVIRONMENT VARIABLES
GC_DONT_GC
Turns off the garbage collection in Mono. This should be only
used for debugging purposes
LVM_COUNT
When Mono is compiled with LLVM support, this instructs the run‐
time to stop using LLVM after the specified number of methods
are JITed. This is a tool used in diagnostics to help isolate
problems in the code generation backend. For example
LLVM_COUNT=10 would only compile 10 methods with LLVM and then
switch to the Mono JIT engine. LLVM_COUNT=0 would disable the
LLVM engine altogether.
MONO_AOT_CACHE
If set, this variable will instruct Mono to ahead-of-time com‐
pile new assemblies on demand and store the result into a cache
in ~/.mono/aot-cache.
MONO_CFG_DIR
If set, this variable overrides the default system configuration
directory ($PREFIX/etc). It's used to locate machine.config
file.
MONO_COM
Sets the style of COM interop. If the value of this variable is
"MS" Mono will use string marhsalling routines from the
liboleaut32 for the BSTR type library, any other values will use
the mono-builtin BSTR string marshalling.
MONO_CONFIG
If set, this variable overrides the default runtime configura‐
tion file ($PREFIX/etc/mono/config). The --config command line
options overrides the environment variable.
MONO_DEBUG
If set, enables some features of the runtime useful for debug‐
ging. This variable should contain a comma separated list of
debugging options. Currently, the following options are sup‐
ported:
break-on-unverified
If this variable is set, when the Mono VM runs into a
verification problem, instead of throwing an exception it
will break into the debugger. This is useful when debug‐
ging verifier problems
collect-pagefault-stats
Collects information about pagefaults. This is used
internally to track the number of page faults produced to
load metadata. To display this information you must use
this option with "--stats" command line option.
dont-free-domains
This is an Optimization for multi-AppDomain applications
(most commonly ASP.NET applications). Due to internal
limitations Mono, Mono by default does not use typed
allocations on multi-appDomain applications as they could
leak memory when a domain is unloaded. Although this is
a fine default, for applications that use more than on
AppDomain heavily (for example, ASP.NET applications) it
is worth trading off the small leaks for the increased
performance (additionally, since ASP.NET applications are
not likely going to unload the application domains on
production systems, it is worth using this feature).
handle-sigint
Captures the interrupt signal (Control-C) and displays a
stack trace when pressed. Useful to find out where the
program is executing at a given point. This only dis‐
plays the stack trace of a single thread.
keep-delegates
This option will leak delegate trampolines that are no
longer referenced as to present the user with more infor‐
mation about a delegate misuse. Basically a delegate
instance might be created, passed to unmanaged code, and
no references kept in managed code, which will garbage
collect the code. With this option it is possible to
track down the source of the problems.
no-gdb-backtrace
This option will disable the GDB backtrace emitted by the
runtime after a SIGSEGV or SIGABRT in unmanaged code.
suspend-on-sigsegv
This option will suspend the program when a native
SIGSEGV is received. This is useful for debugging
crashes which do not happen under gdb, since a live
process contains more information than a core file.
MONO_DISABLE_AIO
If set, tells mono NOT to attempt using native asynchronous I/O
services. In that case, a default select/poll implementation is
used. Currently only epoll() is supported.
MONO_DISABLE_MANAGED_COLLATION
If this environment variable is `yes', the runtime uses unman‐
aged collation (which actually means no culture-sensitive colla‐
tion). It internally disables managed collation functionality
invoked via the members of System.Globalization.CompareInfo
class. Collation is enabled by default.
MONO_EGD_SOCKET
For platforms that do not otherwise have a way of obtaining ran‐
dom bytes this can be set to the name of a file system socket on
which an egd or prngd daemon is listening.
MONO_EVENTLOG_TYPE
Sets the type of event log provider to use (for System.Diagnos‐
tics.EventLog). Possible values are:
local[:path]
Persists event logs and entries to the local file system.
The directory in which to persist the event logs, event
sources and entries can be specified as part of the
value. If the path is not explicitly set, it defaults to
"/var/lib/mono/eventlog" on unix and "%APPDATA%no\vent‐
log" on Windows.
win32 Uses the native win32 API to write events and registers
event logs and event sources in the registry. This is
only available on Windows. On Unix, the directory per‐
mission for individual event log and event source direc‐
tories is set to 777 (with +t bit) allowing everyone to
read and write event log entries while only allowing
entries to be deleted by the user(s) that created them.
null Silently discards any events.
The default is "null" on Unix (and versions of Windows before
NT), and "win32" on Windows NT (and higher).
MONO_EXTERNAL_ENCODINGS
If set, contains a colon-separated list of text encodings to try
when turning externally-generated text (e.g. command-line argu‐
ments or filenames) into Unicode. The encoding names come from
the list provided by iconv, and the special case
"default_locale" which refers to the current locale's default
encoding.
When reading externally-generated text strings UTF-8 is tried
first, and then this list is tried in order with the first suc‐
cessful conversion ending the search. When writing external
text (e.g. new filenames or arguments to new processes) the
first item in this list is used, or UTF-8 if the environment
variable is not set.
The problem with using MONO_EXTERNAL_ENCODINGS to process your
files is that it results in a problem: although its possible to
get the right file name it is not necessarily possible to open
the file. In general if you have problems with encodings in
your filenames you should use the "convmv" program.
MONO_GAC_PREFIX
Provides a prefix the runtime uses to look for Global Assembly
Caches. Directories are separated by the platform path separa‐
tor (colons on unix). MONO_GAC_PREFIX should point to the top
directory of a prefixed install. Or to the directory provided in
the gacutil /gacdir command. Example: /home/user‐
name/.mono:/usr/local/mono/
MONO_IOMAP
Enables some filename rewriting support to assist badly-written
applications that hard-code Windows paths. Set to a colon-sepa‐
rated list of "drive" to strip drive letters, or "case" to do
case-insensitive file matching in every directory in a path.
"all" enables all rewriting methods. (Backslashes are always
mapped to slashes if this variable is set to a valid option.)
For example, this would work from the shell:
MONO_IOMAP=drive:case
export MONO_IOMAP
If you are using mod_mono to host your web applications, you can
use the MonoIOMAP directive instead, like this:
MonoIOMAP <appalias> all
See mod_mono(8) for more details.
MONO_MANAGED_WATCHER
If set to "disabled", System.IO.FileSystemWatcher will use a
file watcher implementation which silently ignores all the
watching requests. If set to any other value, Sys‐
tem.IO.FileSystemWatcher will use the default managed implemen‐
tation (slow). If unset, mono will try to use inotify, FAM,
Gamin, kevent under Unix systems and native API calls on Win‐
dows, falling back to the managed implementation on error.
MONO_NO_SMP
If set causes the mono process to be bound to a single proces‐
sor. This may be useful when debugging or working around race
conditions.
MONO_PATH
Provides a search path to the runtime where to look for library
files. This is a tool convenient for debugging applications,
but should not be used by deployed applications as it breaks the
assembly loader in subtle ways. Directories are separated by
the platform path separator (colons on unix). Example:
/home/username/lib:/usr/local/mono/lib Alternative solutions to
MONO_PATH include: installing libraries into the Global Assembly
Cache (see gacutil(1)) or having the dependent libraries side-
by-side with the main executable. For a complete description of
recommended practices for application deployment, see
http://www.mono-project.com/Guidelines:Application_Deployment
MONO_RTC
Experimental RTC support in the statistical profiler: if the
user has the permission, more accurate statistics are gathered.
The MONO_RTC value must be restricted to what the Linux rtc
allows: power of two from 64 to 8192 Hz. To enable higher fre‐
quencies like 4096 Hz, run as root:
echo 4096 > /proc/sys/dev/rtc/max-user-freq
For example:
MONO_RTC=4096 mono --profiler=default:stat program.exe
MONO_NO_TLS
Disable inlining of thread local accesses. Try setting this if
you get a segfault early on in the execution of mono.
MONO_SHARED_DIR
If set its the directory where the ".wapi" handle state is
stored. This is the directory where the Windows I/O Emulation
layer stores its shared state data (files, events, mutexes,
pipes). By default Mono will store the ".wapi" directory in the
users's home directory.
MONO_SHARED_HOSTNAME
Uses the string value of this variable as a replacement for the
host name when creating file names in the ".wapi" directory.
This helps if the host name of your machine is likely to be
changed when a mono application is running or if you have a
.wapi directory shared among several different computers. Mono
typically uses the hostname to create the files that are used to
share state across multiple Mono processes. This is done to
support home directories that might be shared over the network.
MONO_STRICT_IO_EMULATION
If set, extra checks are made during IO operations. Currently,
this includes only advisory locks around file writes.
MONO_DISABLE_SHM
If set, disables the shared memory files used for cross-process
handles: process have only private handles. This means that
process and thread handles are not available to other processes,
and named mutexes, named events and named semaphores are not
visible between processes. This is can also be enabled by
default by passing the "--disable-shared-handles" option to con‐
figure.
MONO_THEME
The name of the theme to be used by Windows.Forms. Available
themes today include "clearlooks", "nice" and "win32". The
default is "win32".
MONO_TLS_SESSION_CACHE_TIMEOUT
The time, in seconds, that the SSL/TLS session cache will keep
it's entry to avoid a new negotiation between the client and a
server. Negotiation are very CPU intensive so an application-
specific custom value may prove useful for small embedded sys‐
tems. The default is 180 seconds.
MONO_THREADS_PER_CPU
The maximum number of threads in the general threadpool will be
20 + (MONO_THREADS_PER_CPU * number of CPUs). The default value
for this variable is 10.
MONO_XMLSERIALIZER_THS
Controls the threshold for the XmlSerializer to produce a custom
serializer for a given class instead of using the Reflection-
based interpreter. The possible values are `no' to disable the
use of a custom serializer or a number to indicate when the
XmlSerializer should start serializing. The default value is
50, which means that the a custom serializer will be produced on
the 50th use.
MONO_XMLSERIALIZER_DEBUG
Set this value to 1 to prevent the serializer from removing the
temporary files that are created for fast serialization; This
might be useful when debugging.
MONO_ASPNET_INHIBIT_SETTINGSMAP
Mono contains a feature which allows modifying settings in the
.config files shipped with Mono by using config section mappers.
The mappers and the mapping rules are defined in the $pre‐
fix/etc/mono/2.0/settings.map file and, optionally, in the set‐
tings.map file found in the top-level directory of your ASP.NET
application. Both files are read by System.Web on application
startup, if they are found at the above locations. If you don't
want the mapping to be performed you can set this variable in
your environment before starting the application and no action
will be taken.
MONO_MESSAGING_PROVIDER
Mono supports a plugin model for its implementation of Sys‐
tem.Messaging making it possible to support a variety of messag‐
ing implementations (e.g. AMQP, ActiveMQ). To specify which
messaging implementation is to be used the evironement variable
needs to be set to the full class name for the provider. E.g.
to use the RabbitMQ based AMQP implementation the variable
should be set to:
Mono.Messaging.RabbitMQ.RabbitMQMessagingProvider,Mono.Messaging.RabbitMQ
ENVIRONMENT VARIABLES FOR DEBUGGING
MONO_ASPNET_NODELETE
If set to any value, temporary source files generated by ASP.NET
support classes will not be removed. They will be kept in the
user's temporary directory.
MONO_LOG_LEVEL
The logging level, possible values are `error', `critical',
`warning', `message', `info' and `debug'. See the DEBUGGING
section for more details.
MONO_LOG_MASK
Controls the domain of the Mono runtime that logging will apply
to. If set, the log mask is changed to the set value. Possible
values are "asm" (assembly loader), "type", "dll" (native
library loader), "gc" (garbage collector), "cfg" (config file
loader), "aot" (precompiler) and "all". The default value is
"all". Changing the mask value allows you to display only mes‐
sages for a certain component. You can use multiple masks by
comma separating them. For example to see config file messages
and assembly loader messages set you mask to "asm,cfg".
MONO_TRACE
Used for runtime tracing of method calls. The format of the
comma separated trace options is:
[-]M:method name
[-]N:namespace
[-]T:class name
[-]all
[-]program
disabled Trace output off upon start.
You can toggle trace output on/off sending a SIGUSR2 signal to
the program.
MONO_TRACE_LISTENER
If set, enables the System.Diagnostics.DefaultTraceListener,
which will print the output of the System.Diagnostics Trace and
Debug classes. It can be set to a filename, and to Console.Out
or Console.Error to display output to standard output or stan‐
dard error, respectively. If it's set to Console.Out or Con‐
sole.Error you can append an optional prefix that will be used
when writing messages like this: Console.Error:MyProgramName.
See the System.Diagnostics.DefaultTraceListener documentation
for more information.
MONO_XEXCEPTIONS
This throws an exception when a X11 error is encountered; by
default a message is displayed but execution continues
MONO_XSYNC
This is used in the System.Windows.Forms implementation when
running with the X11 backend. This is used to debug problems in
Windows.Forms as it forces all of the commands send to X11
server to be done synchronously. The default mode of operation
is asynchronous which makes it hard to isolate the root of cer‐
tain problems.
MONO_GENERIC_SHARING
This environment variable controls the kind of generic sharing
used. This variable is used by internal JIT developers and
should not be changed in production. Do not use it. The vari‐
able controls which classes will have generic code sharing
enabled. Permissible values are:
all All generated code can be shared.
collections
Only the classes in System.Collections.Generic will have
its code shared (this is the default value).
corlib Only code in corlib will have its code shared.
none No generic code sharing will be performed.
Generic code sharing by default only applies to collections. The Mono
JIT by default turns this on.
MONO_XDEBUG
When the the MONO_XDEBUG env var is set, debugging info for JIT‐
ted code is emitted into a shared library, loadable into gdb.
This enables, for example, to see managed frame names on gdb
backtraces.
MONO_VERBOSE_METHOD
Enables the maximum JIT verbosity for the specified method. This
is very helpfull to diagnose a miscompilation problems of a spe‐
cific method.
VALGRIND
If you want to use Valgrind, you will find the file `mono.supp' useful,
it contains the suppressions for the GC which trigger incorrect warn‐
ings. Use it like this:
valgrind --suppressions=mono.supp mono ...
DTRACE
On some platforms, Mono can expose a set of DTrace probes (also known
as user-land statically defined, USDT Probes).
They are defined in the file `mono.d'.
ves-init-begin, ves-init-end
Begin and end of runtime initialization.
method-compile-begin, method-compile-end
Begin and end of method compilation. The probe arguments are
class name, method name and signature, and in case of method-
compile-end success or failure of compilation.
gc-begin, gc-end
Begin and end of Garbage Collection.
To verify the availability of the probes, run:
dtrace -P mono'$target' -l -c monoFILES
On Unix assemblies are loaded from the installation lib directory. If
you set `prefix' to /usr, the assemblies will be located in /usr/lib.
On Windows, the assemblies are loaded from the directory where mono and
mint live.
~/.mono/aot-cache
The directory for the ahead-of-time compiler demand creation
assemblies are located.
/etc/mono/config, ~/.mono/config
Mono runtime configuration file. See the mono-config(5) manual
page for more information.
~/.config/.mono/certs, /usr/share/.mono/certs
Contains Mono certificate stores for users / machine. See the
certmgr(1) manual page for more information on managing certifi‐
cate stores and the mozroots(1) page for information on how to
import the Mozilla root certificates into the Mono certificate
store.
~/.mono/assemblies/ASSEMBLY/ASSEMBLY.config
Files in this directory allow a user to customize the configura‐
tion for a given system assembly, the format is the one
described in the mono-config(5) page.
~/.config/.mono/keypairs, /usr/share/.mono/keypairs
Contains Mono cryptographic keypairs for users / machine. They
can be accessed by using a CspParameters object with DSACryp‐
toServiceProvider and RSACryptoServiceProvider classes.
~/.config/.isolatedstorage, ~/.local/share/.isolatedstorage,
/usr/share/.isolatedstorage
Contains Mono isolated storage for non-roaming users, roaming
users and local machine. Isolated storage can be accessed using
the classes from the System.IO.IsolatedStorage namespace.
<assembly>.config
Configuration information for individual assemblies is loaded by
the runtime from side-by-side files with the .config files, see
the http://www.mono-project.com/Config for more information.
Web.config, web.config
ASP.NET applications are configured through these files, the
configuration is done on a per-directory basis. For more infor‐
mation on this subject see the http://www.mono-project.com/Con‐
fig_system.web page.
MAILING LISTS
Mailing lists are listed at the http://www.mono-project.com/Mail‐
ing_Lists
WEB SITE
http://www.mono-project.com
SEE ALSOcertmgr(1), csharp(1), mcs(1), mdb(1), monocov(1), monodis(1), mono-
config(5), mozroots(1), pdb2mdb(1), xsp(1), mod_mono(8).
For more information on AOT: http://www.mono-project.com/AOT
For ASP.NET-related documentation, see the xsp(1) manual page
Mono(Mono 2.5)
