This dumps gzipped memory reports for this process and its child
processes. If a file of the given name exists, it will be overwritten.
Sample output, annotated with comments for explanatory purposes.
{
// The version number of the format, which will be incremented each time
// backwards-incompatible changes are made. A mandatory integer.
“version”: 1
// Equal to nsIMemoryReporterManager::hasMozMallocUsableSize. A
// mandatory boolean.
“hasMozMallocUsableSize”: true,
// The memory reports. A mandatory array.
“reports”: [
// The properties correspond to the arguments of
// nsIHandleReportCallback::callback. Every one is mandatory.
{“process”:”Main Process (pid 12345)”, “path”:”explicit/foo/bar”,
“kind”:1, “units”:0, “amount”:2000000, “description”:”Foo bar.”},
{“process”:”Main Process (pid 12345)”, “path”:”heap-allocated”,
“kind”:1, “units”:0, “amount”:3000000, “description”:”Heap allocated.”},
{“process”:”Main Process (pid 12345)”, “path”:”vsize”,
“kind”:1, “units”:0, “amount”:10000000, “description”:”Vsize.”}
]
}
| aFilename | The output file. |
| aFinishDumping | The callback called on completion. |
| aFinishDumpingData | The environment for the callback. |
| aAnonymize | Should the reports be anonymized? |
Similar to dumpMemoryReportsToNamedFile, this method dumps gzipped memory
reports for this process and its child processes to files in the tmp
directory called memory-reports-
If DMD is enabled, this method also dumps gzipped DMD output for this
process and its child processes to files in the tmp directory called
dmd-
If the identifier is empty, the implementation may set it arbitrarily
and use that new value for its own dump and the dumps of its child
processes. For example, the implementation may set |aIdentifier| to the
number of seconds since the epoch.
| aIdentifier | this identifier will appear in the filename of our about:memory dump and those of our children. |
| aAnonymize | Should the reports be anonymized? |
| aMinimizeMemoryUsage | indicates whether we should run a series of gc/cc's in an attempt to reduce our memory usage before collecting our memory report. |
Dump GC and CC logs to files in the OS’s temp directory (or in
$MOZ_CC_LOG_DIRECTORY, if that environment variable is specified).
If aIdentifier is empty, the implementation may set it to an
arbitrary value; for example, it may set aIdentifier to the number
of seconds since the epoch.
All-traces logs are much bigger than the alternative, but they may be
helpful when trying to understand why a particular object is alive. For
example, a non-traces-log will skip references held by an active
document; if your object is being held alive by such a document, you
probably want to see those references.
| aIdentifier | If aIdentifier is non-empty, this string will appear in the filenames of the logs we create (both for this process and, if aDumpChildProcesses is true, for our child processes). |
| aDumpAllTraces | indicates whether we should run an all-traces CC log. An all-traces log visits all objects currently eligible for cycle collection, while a non-all-traces log avoids visiting some objects which we know are reachable. |
| aDumpChildProcesses | indicates whether we should call DumpGCAndCCLogsToFile in our child processes. If so, the child processes will dump their children, and so on. |
Like |dumpGCAndCCLogsToFile|, but sends the logs to the given log
sink object instead of accessing the filesystem directly, and
dumps the current process only.