gecko-dev/storage/QuotaVFS.cpp

622 строки
19 KiB
C++

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-
* vim: sw=2 ts=2 et lcs=trail\:.,tab\:>~ :
* 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 "sqlite3.h"
#include "mozilla/dom/quota/PersistenceType.h"
#include "mozilla/dom/quota/QuotaManager.h"
#include "mozilla/dom/quota/QuotaObject.h"
#include "mozilla/dom/quota/ResultExtensions.h"
#include "nsDirectoryServiceDefs.h"
#include "nsEscape.h"
#include "mozilla/StaticPrefs_storage.h"
#if defined(XP_WIN) || defined(XP_UNIX)
# include "mozilla/StaticPrefs_dom.h"
#endif
// The last VFS version for which this file has been updated.
#define LAST_KNOWN_VFS_VERSION 3
// The last io_methods version for which this file has been updated.
#define LAST_KNOWN_IOMETHODS_VERSION 3
namespace {
using namespace mozilla;
using namespace mozilla::dom::quota;
struct QuotaFile {
// Base class. Must be first
sqlite3_file base;
// quota object for this file
RefPtr<QuotaObject> quotaObject;
// The chunk size for this file. See the documentation for
// sqlite3_file_control() and FCNTL_CHUNK_SIZE.
int fileChunkSize;
// This contains the vfs that actually does work
sqlite3_file pReal[1];
};
already_AddRefed<QuotaObject> GetQuotaObjectFromName(const char* zName) {
MOZ_ASSERT(zName);
const char* directoryLockIdParam =
sqlite3_uri_parameter(zName, "directoryLockId");
if (!directoryLockIdParam) {
return nullptr;
}
nsresult rv;
const int64_t directoryLockId =
nsDependentCString(directoryLockIdParam).ToInteger64(&rv);
MOZ_RELEASE_ASSERT(NS_SUCCEEDED(rv));
QuotaManager* quotaManager = QuotaManager::Get();
MOZ_ASSERT(quotaManager);
return quotaManager->GetQuotaObject(directoryLockId,
NS_ConvertUTF8toUTF16(zName));
}
void MaybeEstablishQuotaControl(const char* zName, QuotaFile* pFile,
int flags) {
MOZ_ASSERT(pFile);
MOZ_ASSERT(!pFile->quotaObject);
if (!(flags & (SQLITE_OPEN_URI | SQLITE_OPEN_WAL))) {
return;
}
pFile->quotaObject = GetQuotaObjectFromName(zName);
}
/*
** Close a QuotaFile.
*/
int QuotaClose(sqlite3_file* pFile) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xClose(p->pReal);
if (rc == SQLITE_OK) {
delete p->base.pMethods;
p->base.pMethods = nullptr;
p->quotaObject = nullptr;
#ifdef DEBUG
p->fileChunkSize = 0;
#endif
}
return rc;
}
/*
** Read data from a QuotaFile.
*/
int QuotaRead(sqlite3_file* pFile, void* zBuf, int iAmt, sqlite_int64 iOfst) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xRead(p->pReal, zBuf, iAmt, iOfst);
return rc;
}
/*
** Return the current file-size of a QuotaFile.
*/
int QuotaFileSize(sqlite3_file* pFile, sqlite_int64* pSize) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xFileSize(p->pReal, pSize);
return rc;
}
/*
** Write data to a QuotaFile.
*/
int QuotaWrite(sqlite3_file* pFile, const void* zBuf, int iAmt,
sqlite_int64 iOfst) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
if (p->quotaObject) {
MOZ_ASSERT(INT64_MAX - iOfst >= iAmt);
if (!p->quotaObject->MaybeUpdateSize(iOfst + iAmt, /* aTruncate */ false)) {
return SQLITE_FULL;
}
}
rc = p->pReal->pMethods->xWrite(p->pReal, zBuf, iAmt, iOfst);
if (p->quotaObject && rc != SQLITE_OK) {
NS_WARNING(
"xWrite failed on a quota-controlled file, attempting to "
"update its current size...");
sqlite_int64 currentSize;
if (QuotaFileSize(pFile, &currentSize) == SQLITE_OK) {
DebugOnly<bool> res =
p->quotaObject->MaybeUpdateSize(currentSize, /* aTruncate */ true);
MOZ_ASSERT(res);
}
}
return rc;
}
/*
** Truncate a QuotaFile.
*/
int QuotaTruncate(sqlite3_file* pFile, sqlite_int64 size) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
if (p->quotaObject) {
if (p->fileChunkSize > 0) {
// Round up to the smallest multiple of the chunk size that will hold all
// the data.
size =
((size + p->fileChunkSize - 1) / p->fileChunkSize) * p->fileChunkSize;
}
if (!p->quotaObject->MaybeUpdateSize(size, /* aTruncate */ true)) {
return SQLITE_FULL;
}
}
rc = p->pReal->pMethods->xTruncate(p->pReal, size);
if (p->quotaObject) {
if (rc == SQLITE_OK) {
#ifdef DEBUG
// Make sure xTruncate set the size exactly as we calculated above.
sqlite_int64 newSize;
MOZ_ASSERT(QuotaFileSize(pFile, &newSize) == SQLITE_OK);
MOZ_ASSERT(newSize == size);
#endif
} else {
NS_WARNING(
"xTruncate failed on a quota-controlled file, attempting to "
"update its current size...");
if (QuotaFileSize(pFile, &size) == SQLITE_OK) {
DebugOnly<bool> res =
p->quotaObject->MaybeUpdateSize(size, /* aTruncate */ true);
MOZ_ASSERT(res);
}
}
}
return rc;
}
/*
** Sync a QuotaFile.
*/
int QuotaSync(sqlite3_file* pFile, int flags) {
QuotaFile* p = (QuotaFile*)pFile;
return p->pReal->pMethods->xSync(p->pReal, flags);
}
/*
** Lock a QuotaFile.
*/
int QuotaLock(sqlite3_file* pFile, int eLock) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xLock(p->pReal, eLock);
return rc;
}
/*
** Unlock a QuotaFile.
*/
int QuotaUnlock(sqlite3_file* pFile, int eLock) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xUnlock(p->pReal, eLock);
return rc;
}
/*
** Check if another file-handle holds a RESERVED lock on a QuotaFile.
*/
int QuotaCheckReservedLock(sqlite3_file* pFile, int* pResOut) {
QuotaFile* p = (QuotaFile*)pFile;
int rc = p->pReal->pMethods->xCheckReservedLock(p->pReal, pResOut);
return rc;
}
/*
** File control method. For custom operations on a QuotaFile.
*/
int QuotaFileControl(sqlite3_file* pFile, int op, void* pArg) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
// Hook SQLITE_FCNTL_SIZE_HINT for quota-controlled files and do the necessary
// work before passing to the SQLite VFS.
if (op == SQLITE_FCNTL_SIZE_HINT && p->quotaObject) {
sqlite3_int64 hintSize = *static_cast<sqlite3_int64*>(pArg);
sqlite3_int64 currentSize;
rc = QuotaFileSize(pFile, &currentSize);
if (rc != SQLITE_OK) {
return rc;
}
if (hintSize > currentSize) {
rc = QuotaTruncate(pFile, hintSize);
if (rc != SQLITE_OK) {
return rc;
}
}
}
rc = p->pReal->pMethods->xFileControl(p->pReal, op, pArg);
// Grab the file chunk size after the SQLite VFS has approved.
if (op == SQLITE_FCNTL_CHUNK_SIZE && rc == SQLITE_OK) {
p->fileChunkSize = *static_cast<int*>(pArg);
}
#ifdef DEBUG
if (op == SQLITE_FCNTL_SIZE_HINT && p->quotaObject && rc == SQLITE_OK) {
sqlite3_int64 hintSize = *static_cast<sqlite3_int64*>(pArg);
if (p->fileChunkSize > 0) {
hintSize = ((hintSize + p->fileChunkSize - 1) / p->fileChunkSize) *
p->fileChunkSize;
}
sqlite3_int64 currentSize;
MOZ_ASSERT(QuotaFileSize(pFile, &currentSize) == SQLITE_OK);
MOZ_ASSERT(currentSize >= hintSize);
}
#endif
return rc;
}
/*
** Return the sector-size in bytes for a QuotaFile.
*/
int QuotaSectorSize(sqlite3_file* pFile) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xSectorSize(p->pReal);
return rc;
}
/*
** Return the device characteristic flags supported by a QuotaFile.
*/
int QuotaDeviceCharacteristics(sqlite3_file* pFile) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xDeviceCharacteristics(p->pReal);
return rc;
}
/*
** Shared-memory operations.
*/
int QuotaShmLock(sqlite3_file* pFile, int ofst, int n, int flags) {
QuotaFile* p = (QuotaFile*)pFile;
return p->pReal->pMethods->xShmLock(p->pReal, ofst, n, flags);
}
int QuotaShmMap(sqlite3_file* pFile, int iRegion, int szRegion, int isWrite,
void volatile** pp) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xShmMap(p->pReal, iRegion, szRegion, isWrite, pp);
return rc;
}
void QuotaShmBarrier(sqlite3_file* pFile) {
QuotaFile* p = (QuotaFile*)pFile;
p->pReal->pMethods->xShmBarrier(p->pReal);
}
int QuotaShmUnmap(sqlite3_file* pFile, int delFlag) {
QuotaFile* p = (QuotaFile*)pFile;
int rc;
rc = p->pReal->pMethods->xShmUnmap(p->pReal, delFlag);
return rc;
}
int QuotaFetch(sqlite3_file* pFile, sqlite3_int64 iOff, int iAmt, void** pp) {
QuotaFile* p = (QuotaFile*)pFile;
MOZ_ASSERT(p->pReal->pMethods->iVersion >= 3);
return p->pReal->pMethods->xFetch(p->pReal, iOff, iAmt, pp);
}
int QuotaUnfetch(sqlite3_file* pFile, sqlite3_int64 iOff, void* pResOut) {
QuotaFile* p = (QuotaFile*)pFile;
MOZ_ASSERT(p->pReal->pMethods->iVersion >= 3);
return p->pReal->pMethods->xUnfetch(p->pReal, iOff, pResOut);
}
int QuotaOpen(sqlite3_vfs* vfs, const char* zName, sqlite3_file* pFile,
int flags, int* pOutFlags) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
int rc;
QuotaFile* p = (QuotaFile*)pFile;
MaybeEstablishQuotaControl(zName, p, flags);
rc = orig_vfs->xOpen(orig_vfs, zName, p->pReal, flags, pOutFlags);
if (rc != SQLITE_OK) return rc;
if (p->pReal->pMethods) {
sqlite3_io_methods* pNew = new sqlite3_io_methods;
const sqlite3_io_methods* pSub = p->pReal->pMethods;
memset(pNew, 0, sizeof(*pNew));
// If the io_methods version is higher than the last known one, you should
// update this VFS adding appropriate IO methods for any methods added in
// the version change.
pNew->iVersion = pSub->iVersion;
MOZ_ASSERT(pNew->iVersion <= LAST_KNOWN_IOMETHODS_VERSION);
pNew->xClose = QuotaClose;
pNew->xRead = QuotaRead;
pNew->xWrite = QuotaWrite;
pNew->xTruncate = QuotaTruncate;
pNew->xSync = QuotaSync;
pNew->xFileSize = QuotaFileSize;
pNew->xLock = QuotaLock;
pNew->xUnlock = QuotaUnlock;
pNew->xCheckReservedLock = QuotaCheckReservedLock;
pNew->xFileControl = QuotaFileControl;
pNew->xSectorSize = QuotaSectorSize;
pNew->xDeviceCharacteristics = QuotaDeviceCharacteristics;
if (pNew->iVersion >= 2) {
// Methods added in version 2.
pNew->xShmMap = pSub->xShmMap ? QuotaShmMap : nullptr;
pNew->xShmLock = pSub->xShmLock ? QuotaShmLock : nullptr;
pNew->xShmBarrier = pSub->xShmBarrier ? QuotaShmBarrier : nullptr;
pNew->xShmUnmap = pSub->xShmUnmap ? QuotaShmUnmap : nullptr;
}
if (pNew->iVersion >= 3) {
// Methods added in version 3.
// SQLite 3.7.17 calls these methods without checking for nullptr first,
// so we always define them. Verify that we're not going to call
// nullptrs, though.
MOZ_ASSERT(pSub->xFetch);
pNew->xFetch = QuotaFetch;
MOZ_ASSERT(pSub->xUnfetch);
pNew->xUnfetch = QuotaUnfetch;
}
pFile->pMethods = pNew;
}
return rc;
}
int QuotaDelete(sqlite3_vfs* vfs, const char* zName, int syncDir) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
int rc;
RefPtr<QuotaObject> quotaObject;
if (StringEndsWith(nsDependentCString(zName), "-wal"_ns)) {
quotaObject = GetQuotaObjectFromName(zName);
}
rc = orig_vfs->xDelete(orig_vfs, zName, syncDir);
if (rc == SQLITE_OK && quotaObject) {
MOZ_ALWAYS_TRUE(quotaObject->MaybeUpdateSize(0, /* aTruncate */ true));
}
return rc;
}
int QuotaAccess(sqlite3_vfs* vfs, const char* zName, int flags, int* pResOut) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xAccess(orig_vfs, zName, flags, pResOut);
}
int QuotaFullPathname(sqlite3_vfs* vfs, const char* zName, int nOut,
char* zOut) {
#if defined(XP_WIN)
// SQLite uses GetFullPathnameW which also normailizes file path. If a file
// component ends with a dot, it would be removed. However, it's not desired.
//
// And that would result SQLite uses wrong database and quotaObject.
// Note that we are safe to avoid the GetFullPathnameW call for \\?\ prefixed
// paths.
// And note that this hack will be removed once the issue is fixed directly in
// SQLite.
// zName that starts with "//?/" is the case when a file URI was passed and
// zName that starts with "\\?\" is the case when a normal path was passed
// (not file URI).
if (StaticPrefs::dom_quotaManager_overrideXFullPathname() &&
((zName[0] == '/' && zName[1] == '/' && zName[2] == '?' &&
zName[3] == '/') ||
(zName[0] == '\\' && zName[1] == '\\' && zName[2] == '?' &&
zName[3] == '\\'))) {
MOZ_ASSERT(nOut >= vfs->mxPathname);
MOZ_ASSERT(static_cast<size_t>(nOut) > strlen(zName));
size_t index = 0;
while (zName[index] != '\0') {
if (zName[index] == '/') {
zOut[index] = '\\';
} else {
zOut[index] = zName[index];
}
index++;
}
zOut[index] = '\0';
return SQLITE_OK;
}
#elif defined(XP_UNIX)
// SQLite canonicalizes (resolves path components) file paths on Unix which
// doesn't work well with file path sanity checks in quota manager. This is
// especially a problem on mac where /var is a symlink to /private/var.
// Since QuotaVFS is used only by quota clients which never access databases
// outside of PROFILE/storage, we override Unix xFullPathname with own
// implementation that doesn't do any canonicalization.
if (StaticPrefs::dom_quotaManager_overrideXFullPathnameUnix()) {
if (nOut < 0) {
// Match the return code used by SQLite's xFullPathname implementation
// here and below.
return SQLITE_CANTOPEN;
}
QM_TRY_INSPECT(
const auto& path, ([&zName]() -> Result<nsString, nsresult> {
NS_ConvertUTF8toUTF16 name(zName);
if (name.First() == '/') {
return name;
}
QM_TRY_INSPECT(const auto& file,
MOZ_TO_RESULT_INVOKE_TYPED(nsCOMPtr<nsIFile>,
NS_GetSpecialDirectory,
NS_OS_CURRENT_WORKING_DIR));
QM_TRY(MOZ_TO_RESULT(file->Append(name)));
QM_TRY_RETURN(
MOZ_TO_RESULT_INVOKE_MEMBER_TYPED(nsString, file, GetPath));
}()),
SQLITE_CANTOPEN);
QM_TRY_INSPECT(const auto& quotaFile, QM_NewLocalFile(path),
SQLITE_CANTOPEN);
QM_TRY_INSPECT(
const auto& quotaPath,
MOZ_TO_RESULT_INVOKE_MEMBER_TYPED(nsString, quotaFile, GetPath),
SQLITE_CANTOPEN);
NS_ConvertUTF16toUTF8 sqlitePath(quotaPath);
if (sqlitePath.Length() > (unsigned int)nOut) {
return SQLITE_CANTOPEN;
}
nsCharTraits<char>::copy(zOut, sqlitePath.get(), sqlitePath.Length());
zOut[sqlitePath.Length()] = '\0';
return SQLITE_OK;
}
#endif
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xFullPathname(orig_vfs, zName, nOut, zOut);
}
void* QuotaDlOpen(sqlite3_vfs* vfs, const char* zFilename) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xDlOpen(orig_vfs, zFilename);
}
void QuotaDlError(sqlite3_vfs* vfs, int nByte, char* zErrMsg) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
orig_vfs->xDlError(orig_vfs, nByte, zErrMsg);
}
void (*QuotaDlSym(sqlite3_vfs* vfs, void* pHdle, const char* zSym))(void) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xDlSym(orig_vfs, pHdle, zSym);
}
void QuotaDlClose(sqlite3_vfs* vfs, void* pHandle) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
orig_vfs->xDlClose(orig_vfs, pHandle);
}
int QuotaRandomness(sqlite3_vfs* vfs, int nByte, char* zOut) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xRandomness(orig_vfs, nByte, zOut);
}
int QuotaSleep(sqlite3_vfs* vfs, int microseconds) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xSleep(orig_vfs, microseconds);
}
int QuotaCurrentTime(sqlite3_vfs* vfs, double* prNow) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xCurrentTime(orig_vfs, prNow);
}
int QuotaGetLastError(sqlite3_vfs* vfs, int nBuf, char* zBuf) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xGetLastError(orig_vfs, nBuf, zBuf);
}
int QuotaCurrentTimeInt64(sqlite3_vfs* vfs, sqlite3_int64* piNow) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xCurrentTimeInt64(orig_vfs, piNow);
}
static int QuotaSetSystemCall(sqlite3_vfs* vfs, const char* zName,
sqlite3_syscall_ptr pFunc) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xSetSystemCall(orig_vfs, zName, pFunc);
}
static sqlite3_syscall_ptr QuotaGetSystemCall(sqlite3_vfs* vfs,
const char* zName) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xGetSystemCall(orig_vfs, zName);
}
static const char* QuotaNextSystemCall(sqlite3_vfs* vfs, const char* zName) {
sqlite3_vfs* orig_vfs = static_cast<sqlite3_vfs*>(vfs->pAppData);
return orig_vfs->xNextSystemCall(orig_vfs, zName);
}
} // namespace
namespace mozilla {
namespace storage {
const char* GetQuotaVFSName() { return "quotavfs"; }
UniquePtr<sqlite3_vfs> ConstructQuotaVFS(const char* aBaseVFSName) {
MOZ_ASSERT(aBaseVFSName);
if (sqlite3_vfs_find(GetQuotaVFSName()) != nullptr) {
return nullptr;
}
sqlite3_vfs* vfs = sqlite3_vfs_find(aBaseVFSName);
if (!vfs) {
return nullptr;
}
auto qvfs = MakeUnique<sqlite3_vfs>();
memset(qvfs.get(), 0, sizeof(::sqlite3_vfs));
// If the VFS version is higher than the last known one, you should update
// this VFS adding appropriate methods for any methods added in the version
// change.
qvfs->iVersion = vfs->iVersion;
MOZ_ASSERT(vfs->iVersion <= LAST_KNOWN_VFS_VERSION);
qvfs->szOsFile = static_cast<int>(sizeof(QuotaFile) - sizeof(sqlite3_file) +
vfs->szOsFile);
qvfs->mxPathname = vfs->mxPathname;
qvfs->zName = GetQuotaVFSName();
qvfs->pAppData = vfs;
qvfs->xOpen = QuotaOpen;
qvfs->xDelete = QuotaDelete;
qvfs->xAccess = QuotaAccess;
qvfs->xFullPathname = QuotaFullPathname;
qvfs->xDlOpen = QuotaDlOpen;
qvfs->xDlError = QuotaDlError;
qvfs->xDlSym = QuotaDlSym;
qvfs->xDlClose = QuotaDlClose;
qvfs->xRandomness = QuotaRandomness;
qvfs->xSleep = QuotaSleep;
qvfs->xCurrentTime = QuotaCurrentTime;
qvfs->xGetLastError = QuotaGetLastError;
if (qvfs->iVersion >= 2) {
// Methods added in version 2.
qvfs->xCurrentTimeInt64 = QuotaCurrentTimeInt64;
}
if (qvfs->iVersion >= 3) {
// Methods added in version 3.
qvfs->xSetSystemCall = QuotaSetSystemCall;
qvfs->xGetSystemCall = QuotaGetSystemCall;
qvfs->xNextSystemCall = QuotaNextSystemCall;
}
return qvfs;
}
already_AddRefed<QuotaObject> GetQuotaObjectForFile(sqlite3_file* pFile) {
MOZ_ASSERT(pFile);
QuotaFile* p = (QuotaFile*)pFile;
RefPtr<QuotaObject> result = p->quotaObject;
return result.forget();
}
} // namespace storage
} // namespace mozilla