diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-19 00:47:55 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-19 00:47:55 +0000 |
commit | 26a029d407be480d791972afb5975cf62c9360a6 (patch) | |
tree | f435a8308119effd964b339f76abb83a57c29483 /xpcom/base/nsIMemoryReporter.idl | |
parent | Initial commit. (diff) | |
download | firefox-26a029d407be480d791972afb5975cf62c9360a6.tar.xz firefox-26a029d407be480d791972afb5975cf62c9360a6.zip |
Adding upstream version 124.0.1.upstream/124.0.1
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'xpcom/base/nsIMemoryReporter.idl')
-rw-r--r-- | xpcom/base/nsIMemoryReporter.idl | 615 |
1 files changed, 615 insertions, 0 deletions
diff --git a/xpcom/base/nsIMemoryReporter.idl b/xpcom/base/nsIMemoryReporter.idl new file mode 100644 index 0000000000..05b5701519 --- /dev/null +++ b/xpcom/base/nsIMemoryReporter.idl @@ -0,0 +1,615 @@ +/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */ +/* vim: set ts=8 sts=2 et sw=2 tw=80: */ +/* This Source Code Form is subject to the terms of the Mozilla Public + * License, v. 2.0. If a copy of the MPL was not distributed with this + * file, You can obtain one at http://mozilla.org/MPL/2.0/. */ + +#include "nsISupports.idl" +%{C++ +#include <stdio.h> +%} + +interface mozIDOMWindowProxy; +interface nsIRunnable; +interface nsISimpleEnumerator; +[ptr] native FILE(FILE); + +/* + * Memory reporters measure Firefox's memory usage. They are primarily used to + * generate the about:memory page. You should read + * https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Memory_reporting + * before writing a memory reporter. + */ + +[scriptable, function, uuid(62ef0e1c-dbd6-11e3-aa75-3c970e9f4238)] +interface nsIHandleReportCallback : nsISupports +{ + /* + * The arguments to the callback are as follows. + * + * + * |process| The name of the process containing this reporter. Each + * reporter initially has "" in this field, indicating that it applies to the + * current process. (This is true even for reporters in a child process.) + * When a reporter from a child process is copied into the main process, the + * copy has its 'process' field set appropriately. + * + * + * |path| The path that this memory usage should be reported under. Paths + * are '/'-delimited, eg. "a/b/c". + * + * Each reporter can be viewed as representing a leaf node in a tree. + * Internal nodes of the tree don't have reporters. So, for example, the + * reporters "explicit/a/b", "explicit/a/c", "explicit/d/e", and + * "explicit/d/f" define this tree: + * + * explicit + * |--a + * | |--b [*] + * | \--c [*] + * \--d + * |--e [*] + * \--f [*] + * + * Nodes marked with a [*] have a reporter. Notice that the internal + * nodes are implicitly defined by the paths. + * + * Nodes within a tree should not overlap measurements, otherwise the + * parent node measurements will be double-counted. So in the example + * above, |b| should not count any allocations counted by |c|, and vice + * versa. + * + * All nodes within each tree must have the same units. + * + * If you want to include a '/' not as a path separator, e.g. because the + * path contains a URL, you need to convert each '/' in the URL to a '\'. + * Consumers of the path will undo this change. Any other '\' character + * in a path will also be changed. This is clumsy but hasn't caused any + * problems so far. + * + * The paths of all reporters form a set of trees. Trees can be + * "degenerate", i.e. contain a single entry with no '/'. + * + * + * |kind| There are three kinds of memory reporters. + * + * - HEAP: reporters measuring memory allocated by the heap allocator, + * e.g. by calling malloc, calloc, realloc, memalign, operator new, or + * operator new[]. Reporters in this category must have units + * UNITS_BYTES. + * + * - NONHEAP: reporters measuring memory which the program explicitly + * allocated, but does not live on the heap. Such memory is commonly + * allocated by calling one of the OS's memory-mapping functions (e.g. + * mmap, VirtualAlloc, or vm_allocate). Reporters in this category + * must have units UNITS_BYTES. + * + * - OTHER: reporters which don't fit into either of these categories. + * They can have any units. + * + * The kind only matters for reporters in the "explicit" tree; + * aboutMemory.js uses it to calculate "heap-unclassified". + * + * + * |units| The units on the reporter's amount. One of the following. + * + * - BYTES: The amount contains a number of bytes. + * + * - COUNT: The amount is an instantaneous count of things currently in + * existence. For instance, the number of tabs currently open would have + * units COUNT. + * + * - COUNT_CUMULATIVE: The amount contains the number of times some event + * has occurred since the application started up. For instance, the + * number of times the user has opened a new tab would have units + * COUNT_CUMULATIVE. + * + * The amount returned by a reporter with units COUNT_CUMULATIVE must + * never decrease over the lifetime of the application. + * + * - PERCENTAGE: The amount contains a fraction that should be expressed as + * a percentage. NOTE! The |amount| field should be given a value 100x + * the actual percentage; this number will be divided by 100 when shown. + * This allows a fractional percentage to be shown even though |amount| is + * an integer. E.g. if the actual percentage is 12.34%, |amount| should + * be 1234. + * + * Values greater than 100% are allowed. + * + * + * |amount| The numeric value reported by this memory reporter. Accesses + * can fail if something goes wrong when getting the amount. + * + * + * |description| A human-readable description of this memory usage report. + */ + void callback(in ACString process, in AUTF8String path, in int32_t kind, + in int32_t units, in int64_t amount, + in AUTF8String description, in nsISupports data); +}; + +/* + * An nsIMemoryReporter reports one or more memory measurements via a + * callback function which is called once for each measurement. + * + * An nsIMemoryReporter that reports a single measurement is sometimes called a + * "uni-reporter". One that reports multiple measurements is sometimes called + * a "multi-reporter". + * + * aboutMemory.js is the most important consumer of memory reports. It + * places the following constraints on reports. + * + * - All reports within a single sub-tree must have the same units. + * + * - There may be an "explicit" tree. If present, it represents + * non-overlapping regions of memory that have been explicitly allocated with + * an OS-level allocation (e.g. mmap/VirtualAlloc/vm_allocate) or a + * heap-level allocation (e.g. malloc/calloc/operator new). Reporters in + * this tree must have kind HEAP or NONHEAP, units BYTES. + * + * It is preferred, but not required, that report descriptions use complete + * sentences (i.e. start with a capital letter and end with a period, or + * similar). + */ +[scriptable, uuid(92a36db1-46bd-4fe6-988e-47db47236d8b)] +interface nsIMemoryReporter : nsISupports +{ + /* + * Run the reporter. + * + * If |anonymize| is true, the memory reporter should anonymize any + * privacy-sensitive details in memory report paths, by replacing them with a + * string such as "<anonymized>". Anonymized memory reports may be sent + * automatically via crash reports or telemetry. + * + * The following things are considered privacy-sensitive. + * + * - Content domains and URLs, and information derived from them. + * - Content data, such as strings. + * - Details about content code, such as filenames, function names or stack + * traces. + * - Details about or data from the user's system, such as filenames. + * - Running apps. + * + * In short, anything that could identify parts of the user's browsing + * history is considered privacy-sensitive. + * + * The following thing are not considered privacy-sensitive. + * + * - Chrome domains and URLs. + * - Information about installed extensions. + */ + void collectReports(in nsIHandleReportCallback callback, + in nsISupports data, + in boolean anonymize); + + /* + * Kinds. See the |kind| comment in nsIHandleReportCallback. + */ + const int32_t KIND_NONHEAP = 0; + const int32_t KIND_HEAP = 1; + const int32_t KIND_OTHER = 2; + + /* + * Units. See the |units| comment in nsIHandleReportCallback. + */ + const int32_t UNITS_BYTES = 0; + const int32_t UNITS_COUNT = 1; + const int32_t UNITS_COUNT_CUMULATIVE = 2; + const int32_t UNITS_PERCENTAGE = 3; +}; + +[scriptable, function, uuid(548b3909-c04d-4ca6-8466-b8bee3837457)] +interface nsIFinishReportingCallback : nsISupports +{ + void callback(in nsISupports data); +}; + +[scriptable, function, uuid(1a80cd0f-0d9e-4397-be69-68ad28fe5175)] +interface nsIHeapAllocatedCallback : nsISupports +{ + void callback(in int64_t bytesAllocated); +}; + +[scriptable, builtinclass, uuid(2998574d-8993-407a-b1a5-8ad7417653e1)] +interface nsIMemoryReporterManager : nsISupports +{ + /* + * Initialize. + */ + [must_use] void init(); + + /* + * Register the given nsIMemoryReporter. The Manager service will hold a + * strong reference to the given reporter, and will be responsible for freeing + * the reporter at shutdown. You may manually unregister the reporter with + * unregisterStrongReporter() at any point. + */ + void registerStrongReporter(in nsIMemoryReporter reporter); + void registerStrongAsyncReporter(in nsIMemoryReporter reporter); + + /* + * Like registerReporter, but the Manager service will hold a weak reference + * via a raw pointer to the given reporter. The reporter should be + * unregistered before shutdown. + * You cannot register JavaScript components with this function! Always + * register your JavaScript components with registerStrongReporter(). + */ + void registerWeakReporter(in nsIMemoryReporter reporter); + void registerWeakAsyncReporter(in nsIMemoryReporter reporter); + + /* + * Unregister the given memory reporter, which must have been registered with + * registerStrongReporter(). You normally don't need to unregister your + * strong reporters, as nsIMemoryReporterManager will take care of that at + * shutdown. + */ + void unregisterStrongReporter(in nsIMemoryReporter reporter); + + /* + * Unregister the given memory reporter, which must have been registered with + * registerWeakReporter(). + */ + void unregisterWeakReporter(in nsIMemoryReporter reporter); + + /* + * These functions should only be used for testing purposes. + */ + void blockRegistrationAndHideExistingReporters(); + void unblockRegistrationAndRestoreOriginalReporters(); + void registerStrongReporterEvenIfBlocked(in nsIMemoryReporter aReporter); + + /* + * Get memory reports for the current process and all child processes. + * |handleReport| is called for each report, and |finishReporting| is called + * once all reports have been handled. + * + * |finishReporting| is called even if, for example, some child processes + * fail to report back. However, calls to this method will silently and + * immediately abort -- and |finishReporting| will not be called -- if a + * previous getReports() call is still in flight, i.e. if it has not yet + * finished invoking |finishReporting|. The silent abort is because the + * in-flight request will finish soon, and the caller would very likely just + * catch and ignore any error anyway. + * + * If |anonymize| is true, it indicates that the memory reporters should + * anonymize any privacy-sensitive data (see above). + */ + void getReports(in nsIHandleReportCallback handleReport, + in nsISupports handleReportData, + in nsIFinishReportingCallback finishReporting, + in nsISupports finishReportingData, + in boolean anonymize); + + /* + * As above, but: If |minimizeMemoryUsage| is true, then each process will + * minimize its memory usage (see the |minimizeMemoryUsage| method) before + * gathering its report. If DMD is enabled and |DMDDumpIdent| is non-empty + * then write a DMD report to a file in the usual temporary directory (see + * |dumpMemoryInfoToTempDir| in |nsIMemoryInfoDumper|.) + */ + [noscript] void + getReportsExtended(in nsIHandleReportCallback handleReport, + in nsISupports handleReportData, + in nsIFinishReportingCallback finishReporting, + in nsISupports finishReportingData, + in boolean anonymize, + in boolean minimizeMemoryUsage, + in AString DMDDumpIdent); + + /* + * As above, but if DMD is enabled and |DMDFile| is non-null then + * write a DMD report to that file and close it. + */ + [noscript] void + getReportsForThisProcessExtended(in nsIHandleReportCallback handleReport, + in nsISupports handleReportData, + in boolean anonymize, + in FILE DMDFile, + in nsIFinishReportingCallback finishReporting, + in nsISupports finishReportingData); + + /* + * Called by an asynchronous memory reporter upon completion. + */ + [noscript] void endReport(); + + /* + * The memory reporter manager, for the most part, treats reporters + * registered with it as a black box. However, there are some + * "distinguished" amounts (as could be reported by a memory reporter) that + * the manager provides as attributes, because they are sufficiently + * interesting that we want external code (e.g. telemetry) to be able to rely + * on them. + * + * Note that these are not reporters and so getReports() does not look at + * them. However, distinguished amounts can be embedded in a reporter. + * + * Access to these attributes can fail. In particular, some of them are not + * available on all platforms. + * + * If you add a new distinguished amount, please update + * toolkit/components/aboutmemory/tests/test_memoryReporters.xul. + * + * |vsize| (UNITS_BYTES) The virtual size, i.e. the amount of address space + * taken up. + * + * |vsizeMaxContiguous| (UNITS_BYTES) The size of the largest contiguous + * block of virtual memory. + * + * |resident| (UNITS_BYTES) The resident size (a.k.a. RSS or physical memory + * used). + * + * |residentFast| (UNITS_BYTES) This is like |resident|, but on Mac OS + * |resident| can purge pages, which is slow. It also affects the result of + * |residentFast|, and so |resident| and |residentFast| should not be used + * together. + * + * |residentPeak| (UNITS_BYTES) The peak resident size. + * + * |residentUnique| (UNITS_BYTES) The unique set size (a.k.a. USS). + * + * |heapAllocated| (UNITS_BYTES) Memory mapped by the heap allocator. + * + * |heapOverheadFraction| (UNITS_PERCENTAGE) In the heap allocator, this is + * the fraction of committed heap bytes that are overhead. Like all + * UNITS_PERCENTAGE measurements, its amount is multiplied by 100x so it can + * be represented by an int64_t. + * + * |JSMainRuntimeGCHeap| (UNITS_BYTES) Size of the main JS runtime's GC + * heap. + * + * |JSMainRuntimeTemporaryPeak| (UNITS_BYTES) Peak size of the transient + * storage in the main JSRuntime. + * + * |JSMainRuntimeCompartments{System,User}| (UNITS_COUNT) The number of + * {system,user} compartments in the main JS runtime. + * + * |JSMainRuntimeRealms{System,User}| (UNITS_COUNT) The number of + * {system,user} realms in the main JS runtime. + * + * |imagesContentUsedUncompressed| (UNITS_BYTES) Memory used for decoded + * raster images in content. + * + * |storageSQLite| (UNITS_BYTES) Memory used by SQLite. + * + * |lowMemoryEventsPhysical| (UNITS_COUNT_CUMULATIVE) + * The number of low-physical-memory events that have occurred since the + * process started. + * + * |ghostWindows| (UNITS_COUNT) A cached value of the number of ghost + * windows. This should have been updated within the past 60s. + * + * |pageFaultsHard| (UNITS_COUNT_CUMULATIVE) The number of hard (a.k.a. + * major) page faults that have occurred since the process started. + */ + [must_use] readonly attribute int64_t vsize; + [must_use] readonly attribute int64_t vsizeMaxContiguous; + [must_use] readonly attribute int64_t resident; + [must_use] readonly attribute int64_t residentFast; + [must_use] readonly attribute int64_t residentPeak; + [must_use] readonly attribute int64_t residentUnique; + + [must_use] readonly attribute int64_t heapAllocated; + [must_use] readonly attribute int64_t heapOverheadFraction; + + [must_use] readonly attribute int64_t JSMainRuntimeGCHeap; + [must_use] readonly attribute int64_t JSMainRuntimeTemporaryPeak; + [must_use] readonly attribute int64_t JSMainRuntimeCompartmentsSystem; + [must_use] readonly attribute int64_t JSMainRuntimeCompartmentsUser; + [must_use] readonly attribute int64_t JSMainRuntimeRealmsSystem; + [must_use] readonly attribute int64_t JSMainRuntimeRealmsUser; + + [must_use] readonly attribute int64_t imagesContentUsedUncompressed; + + [must_use] readonly attribute int64_t storageSQLite; + + [must_use] readonly attribute int64_t lowMemoryEventsPhysical; + + [must_use] readonly attribute int64_t ghostWindows; + + [must_use] readonly attribute int64_t pageFaultsHard; + + /* + * This attribute indicates if moz_malloc_usable_size() works. + */ + [infallible] readonly attribute boolean hasMozMallocUsableSize; + + /* + * These attributes indicate DMD's status. "Enabled" means enabled at + * build-time. + */ + [infallible] readonly attribute boolean isDMDEnabled; + [infallible] readonly attribute boolean isDMDRunning; + + /* + * Run a series of GC/CC's in an attempt to minimize the application's memory + * usage. When we're finished doing this for the current process, we invoke + * the given runnable if it's not null. We do not wait for any child processes + * that might be doing their own minimization via child-mmu-request to finish. + */ + [must_use] void minimizeMemoryUsage(in nsIRunnable callback); + + /* + * Measure the memory that is known to be owned by this tab, split up into + * several broad categories. Note that this will be an underestimate of the + * true number, due to imperfect memory reporter coverage (corresponding to + * about:memory's "heap-unclassified"), and due to some memory shared between + * tabs not being counted. + * + * The time taken for the measurement (split into JS and non-JS parts) is + * also returned. + */ + [must_use] + void sizeOfTab(in mozIDOMWindowProxy window, + out int64_t jsObjectsSize, out int64_t jsStringsSize, + out int64_t jsOtherSize, out int64_t domSize, + out int64_t styleSize, out int64_t otherSize, + out int64_t totalSize, + out double jsMilliseconds, out double nonJSMilliseconds); +}; + +%{C++ + +#include "js/TypeDecls.h" +#include "nsString.h" +#include "nsTArray.h" + +class nsPIDOMWindowOuter; + +namespace mozilla { + +// All the following registration/unregistration functions don't use +// [[nodiscard]] because ignoring failures is common and reasonable. + +// Register a memory reporter. The manager service will hold a strong +// reference to this reporter. +XPCOM_API(nsresult) RegisterStrongMemoryReporter(nsIMemoryReporter* aReporter); +XPCOM_API(nsresult) RegisterStrongAsyncMemoryReporter(nsIMemoryReporter* aReporter); + +// Register a memory reporter. The manager service will hold a weak reference +// to this reporter. +XPCOM_API(nsresult) RegisterWeakMemoryReporter(nsIMemoryReporter* aReporter); +XPCOM_API(nsresult) RegisterWeakAsyncMemoryReporter(nsIMemoryReporter* aReporter); + +// Unregister a strong memory reporter. +XPCOM_API(nsresult) UnregisterStrongMemoryReporter(nsIMemoryReporter* aReporter); + +// Unregister a weak memory reporter. +XPCOM_API(nsresult) UnregisterWeakMemoryReporter(nsIMemoryReporter* aReporter); + +// The memory reporter manager provides access to several distinguished +// amounts via attributes. Some of these amounts are provided by Gecko +// components that cannot be accessed directly from XPCOM code. So we provide +// the following functions for those components to be registered with the +// manager. + +typedef int64_t (*InfallibleAmountFn)(); + +#define DECL_REGISTER_DISTINGUISHED_AMOUNT(kind, name) \ + nsresult Register##name##DistinguishedAmount(kind##AmountFn aAmountFn); +#define DECL_UNREGISTER_DISTINGUISHED_AMOUNT(name) \ + nsresult Unregister##name##DistinguishedAmount(); + +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, JSMainRuntimeGCHeap) +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, JSMainRuntimeTemporaryPeak) +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, JSMainRuntimeCompartmentsSystem) +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, JSMainRuntimeCompartmentsUser) +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, JSMainRuntimeRealmsSystem) +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, JSMainRuntimeRealmsUser) + +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, ImagesContentUsedUncompressed) +DECL_UNREGISTER_DISTINGUISHED_AMOUNT(ImagesContentUsedUncompressed) + +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, StorageSQLite) +DECL_UNREGISTER_DISTINGUISHED_AMOUNT(StorageSQLite) + +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, LowMemoryEventsPhysical) + +DECL_REGISTER_DISTINGUISHED_AMOUNT(Infallible, GhostWindows) + +#undef DECL_REGISTER_DISTINGUISHED_AMOUNT +#undef DECL_UNREGISTER_DISTINGUISHED_AMOUNT + +// Likewise for per-tab measurement. + +typedef nsresult (*JSSizeOfTabFn)(JSObject* aObj, + size_t* aJsObjectsSize, + size_t* aJsStringSize, + size_t* aJsPrivateSize, + size_t* aJsOtherSize); +typedef nsresult (*NonJSSizeOfTabFn)(nsPIDOMWindowOuter* aWindow, + size_t* aDomSize, + size_t* aStyleSize, + size_t* aOtherSize); + +nsresult RegisterJSSizeOfTab(JSSizeOfTabFn aSizeOfTabFn); +nsresult RegisterNonJSSizeOfTab(NonJSSizeOfTabFn aSizeOfTabFn); + +} + +#if defined(MOZ_DMD) +#if !defined(MOZ_MEMORY) +#error "MOZ_DMD requires MOZ_MEMORY" +#endif + +#include "DMD.h" + +#define MOZ_REPORT(ptr) mozilla::dmd::Report(ptr) +#define MOZ_REPORT_ON_ALLOC(ptr) mozilla::dmd::ReportOnAlloc(ptr) + +#else + +#define MOZ_REPORT(ptr) +#define MOZ_REPORT_ON_ALLOC(ptr) + +#endif // defined(MOZ_DMD) + +// Functions generated via this macro should be used by all traversal-based +// memory reporters. Such functions return |moz_malloc_size_of(ptr)|; this +// will always be zero on some obscure platforms. +// +// You might be wondering why we have a macro that creates multiple functions +// that differ only in their name, instead of a single MallocSizeOf function. +// It's mostly to help with DMD integration, though it sometimes also helps +// with debugging and temporary ad hoc profiling. The function name chosen +// doesn't matter greatly, but it's best to make it similar to the path used by +// the relevant memory reporter(s). +#define MOZ_DEFINE_MALLOC_SIZE_OF(fn) \ + static size_t fn(const void* aPtr) \ + { \ + MOZ_REPORT(aPtr); \ + return moz_malloc_size_of(aPtr); \ + } + +// This is an alternative to MOZ_DEFINE_MALLOC_SIZE_OF that defines a +// MallocSizeOf function that can handle interior pointers. +#ifdef MOZ_MEMORY + +#include "mozmemory.h" + +#define MOZ_DEFINE_MALLOC_ENCLOSING_SIZE_OF(fn) \ + static size_t fn(const void* aPtr) \ + { \ + jemalloc_ptr_info_t info; \ + jemalloc_ptr_info(aPtr, &info); \ + MOZ_REPORT(info.addr); \ + return jemalloc_ptr_is_live(&info) ? info.size : 0; \ + } + +#else + +#define MOZ_DEFINE_MALLOC_ENCLOSING_SIZE_OF(fn) \ + static size_t fn(const void* aPtr) \ + { \ + return 0; \ + } + +#endif + +// Functions generated by the next two macros should be used by wrapping +// allocators that report heap blocks as soon as they are allocated and +// unreport them as soon as they are freed. Such allocators are used in cases +// where we have third-party code that we cannot modify. The two functions +// must always be used in tandem. +#define MOZ_DEFINE_MALLOC_SIZE_OF_ON_ALLOC(fn) \ + static size_t fn(const void* aPtr) \ + { \ + MOZ_REPORT_ON_ALLOC(aPtr); \ + return moz_malloc_size_of(aPtr); \ + } +#define MOZ_DEFINE_MALLOC_SIZE_OF_ON_FREE(fn) \ + static size_t fn(const void* aPtr) \ + { \ + return moz_malloc_size_of(aPtr); \ + } + +// This macro assumes the presence of appropriate |aHandleReport| and |aData| +// variables. The (void) is there because we should always ignore the return +// value of the callback, because callback failures aren't fatal. +#define MOZ_COLLECT_REPORT(path, kind, units, amount, description) \ + (void)aHandleReport->Callback(""_ns, nsLiteralCString(path), \ + kind, units, amount, \ + nsLiteralCString(description), aData) + +%} |