mirror of
https://github.com/mozilla/gecko-dev.git
synced 2024-11-01 06:35:42 +00:00
348 lines
12 KiB
C++
348 lines
12 KiB
C++
/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
|
/* vim:set ts=2 sw=2 sts=2 et cindent: */
|
|
/* 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 "FileBlockCache.h"
|
|
#include "VideoUtils.h"
|
|
#include "prio.h"
|
|
#include <algorithm>
|
|
|
|
namespace mozilla {
|
|
|
|
nsresult FileBlockCache::Open(PRFileDesc* aFD)
|
|
{
|
|
NS_ASSERTION(NS_IsMainThread(), "Only call on main thread");
|
|
NS_ENSURE_TRUE(aFD != nullptr, NS_ERROR_FAILURE);
|
|
{
|
|
MonitorAutoLock mon(mFileMonitor);
|
|
mFD = aFD;
|
|
}
|
|
{
|
|
MonitorAutoLock mon(mDataMonitor);
|
|
nsresult res = NS_NewThread(getter_AddRefs(mThread),
|
|
nullptr,
|
|
MEDIA_THREAD_STACK_SIZE);
|
|
mIsOpen = NS_SUCCEEDED(res);
|
|
return res;
|
|
}
|
|
}
|
|
|
|
FileBlockCache::FileBlockCache()
|
|
: mFileMonitor("MediaCache.Writer.IO.Monitor"),
|
|
mFD(nullptr),
|
|
mFDCurrentPos(0),
|
|
mDataMonitor("MediaCache.Writer.Data.Monitor"),
|
|
mIsWriteScheduled(false),
|
|
mIsOpen(false)
|
|
{
|
|
MOZ_COUNT_CTOR(FileBlockCache);
|
|
}
|
|
|
|
FileBlockCache::~FileBlockCache()
|
|
{
|
|
NS_ASSERTION(!mIsOpen, "Should Close() FileBlockCache before destroying");
|
|
{
|
|
// Note, mThread will be shutdown by the time this runs, so we won't
|
|
// block while taking mFileMonitor.
|
|
MonitorAutoLock mon(mFileMonitor);
|
|
if (mFD) {
|
|
PRStatus prrc;
|
|
prrc = PR_Close(mFD);
|
|
if (prrc != PR_SUCCESS) {
|
|
NS_WARNING("PR_Close() failed.");
|
|
}
|
|
mFD = nullptr;
|
|
}
|
|
}
|
|
MOZ_COUNT_DTOR(FileBlockCache);
|
|
}
|
|
|
|
|
|
void FileBlockCache::Close()
|
|
{
|
|
NS_ASSERTION(NS_IsMainThread(), "Only call on main thread");
|
|
MonitorAutoLock mon(mDataMonitor);
|
|
|
|
mIsOpen = false;
|
|
|
|
if (mThread) {
|
|
// We must shut down the thread in another runnable. This is called
|
|
// while we're shutting down the media cache, and nsIThread::Shutdown()
|
|
// can cause events to run before it completes, which could end up
|
|
// opening more streams, while the media cache is shutting down and
|
|
// releasing memory etc! Also note we close mFD in the destructor so
|
|
// as to not disturb any IO that's currently running.
|
|
nsCOMPtr<nsIRunnable> event = new ShutdownThreadEvent(mThread);
|
|
mThread = nullptr;
|
|
NS_DispatchToMainThread(event);
|
|
}
|
|
}
|
|
|
|
nsresult FileBlockCache::WriteBlock(uint32_t aBlockIndex, const uint8_t* aData)
|
|
{
|
|
MonitorAutoLock mon(mDataMonitor);
|
|
|
|
if (!mIsOpen)
|
|
return NS_ERROR_FAILURE;
|
|
|
|
// Check if we've already got a pending write scheduled for this block.
|
|
mBlockChanges.EnsureLengthAtLeast(aBlockIndex + 1);
|
|
bool blockAlreadyHadPendingChange = mBlockChanges[aBlockIndex] != nullptr;
|
|
mBlockChanges[aBlockIndex] = new BlockChange(aData);
|
|
|
|
if (!blockAlreadyHadPendingChange || !mChangeIndexList.Contains(aBlockIndex)) {
|
|
// We either didn't already have a pending change for this block, or we
|
|
// did but we didn't have an entry for it in mChangeIndexList (we're in the process
|
|
// of writing it and have removed the block's index out of mChangeIndexList
|
|
// in Run() but not finished writing the block to file yet). Add the blocks
|
|
// index to the end of mChangeIndexList to ensure the block is written as
|
|
// as soon as possible.
|
|
mChangeIndexList.PushBack(aBlockIndex);
|
|
}
|
|
NS_ASSERTION(mChangeIndexList.Contains(aBlockIndex), "Must have entry for new block");
|
|
|
|
EnsureWriteScheduled();
|
|
|
|
return NS_OK;
|
|
}
|
|
|
|
void FileBlockCache::EnsureWriteScheduled()
|
|
{
|
|
mDataMonitor.AssertCurrentThreadOwns();
|
|
|
|
if (!mIsWriteScheduled) {
|
|
mThread->Dispatch(this, NS_DISPATCH_NORMAL);
|
|
mIsWriteScheduled = true;
|
|
}
|
|
}
|
|
|
|
nsresult FileBlockCache::Seek(int64_t aOffset)
|
|
{
|
|
mFileMonitor.AssertCurrentThreadOwns();
|
|
|
|
if (mFDCurrentPos != aOffset) {
|
|
int64_t result = PR_Seek64(mFD, aOffset, PR_SEEK_SET);
|
|
if (result != aOffset) {
|
|
NS_WARNING("Failed to seek media cache file");
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
mFDCurrentPos = result;
|
|
}
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult FileBlockCache::ReadFromFile(int64_t aOffset,
|
|
uint8_t* aDest,
|
|
int32_t aBytesToRead,
|
|
int32_t& aBytesRead)
|
|
{
|
|
mFileMonitor.AssertCurrentThreadOwns();
|
|
|
|
nsresult res = Seek(aOffset);
|
|
if (NS_FAILED(res)) return res;
|
|
|
|
aBytesRead = PR_Read(mFD, aDest, aBytesToRead);
|
|
if (aBytesRead <= 0)
|
|
return NS_ERROR_FAILURE;
|
|
mFDCurrentPos += aBytesRead;
|
|
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult FileBlockCache::WriteBlockToFile(int32_t aBlockIndex,
|
|
const uint8_t* aBlockData)
|
|
{
|
|
mFileMonitor.AssertCurrentThreadOwns();
|
|
|
|
nsresult rv = Seek(BlockIndexToOffset(aBlockIndex));
|
|
if (NS_FAILED(rv)) return rv;
|
|
|
|
int32_t amount = PR_Write(mFD, aBlockData, BLOCK_SIZE);
|
|
if (amount < BLOCK_SIZE) {
|
|
NS_WARNING("Failed to write media cache block!");
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
mFDCurrentPos += BLOCK_SIZE;
|
|
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult FileBlockCache::MoveBlockInFile(int32_t aSourceBlockIndex,
|
|
int32_t aDestBlockIndex)
|
|
{
|
|
mFileMonitor.AssertCurrentThreadOwns();
|
|
|
|
uint8_t buf[BLOCK_SIZE];
|
|
int32_t bytesRead = 0;
|
|
if (NS_FAILED(ReadFromFile(BlockIndexToOffset(aSourceBlockIndex),
|
|
buf,
|
|
BLOCK_SIZE,
|
|
bytesRead))) {
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
return WriteBlockToFile(aDestBlockIndex, buf);
|
|
}
|
|
|
|
nsresult FileBlockCache::Run()
|
|
{
|
|
MonitorAutoLock mon(mDataMonitor);
|
|
NS_ASSERTION(!NS_IsMainThread(), "Don't call on main thread");
|
|
NS_ASSERTION(!mChangeIndexList.IsEmpty(), "Only dispatch when there's work to do");
|
|
NS_ASSERTION(mIsWriteScheduled, "Should report write running or scheduled.");
|
|
|
|
while (!mChangeIndexList.IsEmpty()) {
|
|
if (!mIsOpen) {
|
|
// We've been closed, abort, discarding unwritten changes.
|
|
mIsWriteScheduled = false;
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
// Process each pending change. We pop the index out of the change
|
|
// list, but leave the BlockChange in mBlockChanges until the change
|
|
// is written to file. This is so that any read which happens while
|
|
// we drop mDataMonitor to write will refer to the data's source in
|
|
// memory, rather than the not-yet up to date data written to file.
|
|
// This also ensures we will insert a new index into mChangeIndexList
|
|
// when this happens.
|
|
|
|
// Hold a reference to the change, in case another change
|
|
// overwrites the mBlockChanges entry for this block while we drop
|
|
// mDataMonitor to take mFileMonitor.
|
|
int32_t blockIndex = mChangeIndexList.PopFront();
|
|
nsRefPtr<BlockChange> change = mBlockChanges[blockIndex];
|
|
NS_ABORT_IF_FALSE(change,
|
|
"Change index list should only contain entries for blocks with changes");
|
|
{
|
|
MonitorAutoUnlock unlock(mDataMonitor);
|
|
MonitorAutoLock lock(mFileMonitor);
|
|
if (change->IsWrite()) {
|
|
WriteBlockToFile(blockIndex, change->mData.get());
|
|
} else if (change->IsMove()) {
|
|
MoveBlockInFile(change->mSourceBlockIndex, blockIndex);
|
|
}
|
|
}
|
|
// If a new change has not been made to the block while we dropped
|
|
// mDataMonitor, clear reference to the old change. Otherwise, the old
|
|
// reference has been cleared already.
|
|
if (mBlockChanges[blockIndex] == change) {
|
|
mBlockChanges[blockIndex] = nullptr;
|
|
}
|
|
}
|
|
|
|
mIsWriteScheduled = false;
|
|
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult FileBlockCache::Read(int64_t aOffset,
|
|
uint8_t* aData,
|
|
int32_t aLength,
|
|
int32_t* aBytes)
|
|
{
|
|
MonitorAutoLock mon(mDataMonitor);
|
|
|
|
if (!mFD || (aOffset / BLOCK_SIZE) > INT32_MAX)
|
|
return NS_ERROR_FAILURE;
|
|
|
|
int32_t bytesToRead = aLength;
|
|
int64_t offset = aOffset;
|
|
uint8_t* dst = aData;
|
|
while (bytesToRead > 0) {
|
|
int32_t blockIndex = static_cast<int32_t>(offset / BLOCK_SIZE);
|
|
int32_t start = offset % BLOCK_SIZE;
|
|
int32_t amount = std::min(BLOCK_SIZE - start, bytesToRead);
|
|
|
|
// If the block is not yet written to file, we can just read from
|
|
// the memory buffer, otherwise we need to read from file.
|
|
int32_t bytesRead = 0;
|
|
nsRefPtr<BlockChange> change = mBlockChanges[blockIndex];
|
|
if (change && change->IsWrite()) {
|
|
// Block isn't yet written to file. Read from memory buffer.
|
|
const uint8_t* blockData = change->mData.get();
|
|
memcpy(dst, blockData + start, amount);
|
|
bytesRead = amount;
|
|
} else {
|
|
if (change && change->IsMove()) {
|
|
// The target block is the destination of a not-yet-completed move
|
|
// action, so read from the move's source block from file. Note we
|
|
// *don't* follow a chain of moves here, as a move's source index
|
|
// is resolved when MoveBlock() is called, and the move's source's
|
|
// block could be have itself been subject to a move (or write)
|
|
// which happened *after* this move was recorded.
|
|
blockIndex = mBlockChanges[blockIndex]->mSourceBlockIndex;
|
|
}
|
|
// Block has been written to file, either as the source block of a move,
|
|
// or as a stable (all changes made) block. Read the data directly
|
|
// from file.
|
|
nsresult res;
|
|
{
|
|
MonitorAutoUnlock unlock(mDataMonitor);
|
|
MonitorAutoLock lock(mFileMonitor);
|
|
res = ReadFromFile(BlockIndexToOffset(blockIndex) + start,
|
|
dst,
|
|
amount,
|
|
bytesRead);
|
|
}
|
|
NS_ENSURE_SUCCESS(res,res);
|
|
}
|
|
dst += bytesRead;
|
|
offset += bytesRead;
|
|
bytesToRead -= bytesRead;
|
|
}
|
|
*aBytes = aLength - bytesToRead;
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult FileBlockCache::MoveBlock(int32_t aSourceBlockIndex, int32_t aDestBlockIndex)
|
|
{
|
|
NS_ASSERTION(NS_IsMainThread(), "Only call on main thread");
|
|
MonitorAutoLock mon(mDataMonitor);
|
|
|
|
if (!mIsOpen)
|
|
return NS_ERROR_FAILURE;
|
|
|
|
mBlockChanges.EnsureLengthAtLeast(std::max(aSourceBlockIndex, aDestBlockIndex) + 1);
|
|
|
|
// The source block's contents may be the destination of another pending
|
|
// move, which in turn can be the destination of another pending move,
|
|
// etc. Resolve the final source block, so that if one of the blocks in
|
|
// the chain of moves is overwritten, we don't lose the reference to the
|
|
// contents of the destination block.
|
|
int32_t sourceIndex = aSourceBlockIndex;
|
|
BlockChange* sourceBlock = nullptr;
|
|
while ((sourceBlock = mBlockChanges[sourceIndex]) &&
|
|
sourceBlock->IsMove()) {
|
|
sourceIndex = sourceBlock->mSourceBlockIndex;
|
|
}
|
|
|
|
if (mBlockChanges[aDestBlockIndex] == nullptr ||
|
|
!mChangeIndexList.Contains(aDestBlockIndex)) {
|
|
// Only add another entry to the change index list if we don't already
|
|
// have one for this block. We won't have an entry when either there's
|
|
// no pending change for this block, or if there is a pending change for
|
|
// this block and we're in the process of writing it (we've popped the
|
|
// block's index out of mChangeIndexList in Run() but not finished writing
|
|
// the block to file yet.
|
|
mChangeIndexList.PushBack(aDestBlockIndex);
|
|
}
|
|
|
|
// If the source block hasn't yet been written to file then the dest block
|
|
// simply contains that same write. Resolve this as a write instead.
|
|
if (sourceBlock && sourceBlock->IsWrite()) {
|
|
mBlockChanges[aDestBlockIndex] = new BlockChange(sourceBlock->mData.get());
|
|
} else {
|
|
mBlockChanges[aDestBlockIndex] = new BlockChange(sourceIndex);
|
|
}
|
|
|
|
EnsureWriteScheduled();
|
|
|
|
NS_ASSERTION(mChangeIndexList.Contains(aDestBlockIndex),
|
|
"Should have scheduled block for change");
|
|
|
|
return NS_OK;
|
|
}
|
|
|
|
} // End namespace mozilla.
|