input/buffering: pass offset to Read() and eliminate Seek()

src/input/BufferedInputStream.cxx

@@ -46,11 +46,10 @@ BufferedInputStream::Check()
 }
 
 void
-BufferedInputStream::Seek(std::unique_lock<Mutex> &lock,
+BufferedInputStream::Seek(std::unique_lock<Mutex> &,
 			  offset_type new_offset)
 {
-	BufferingInputStream::Seek(lock, new_offset);
-	InputStream::offset = new_offset;
+	offset = new_offset;
 }
 
 bool
@@ -62,14 +61,14 @@ BufferedInputStream::IsEOF() noexcept
 bool
 BufferedInputStream::IsAvailable() noexcept
 {
-	return BufferingInputStream::IsAvailable();
+	return BufferingInputStream::IsAvailable(offset);
 }
 
 size_t
 BufferedInputStream::Read(std::unique_lock<Mutex> &lock,
 			  void *ptr, size_t s)
 {
-	size_t nbytes = BufferingInputStream::Read(lock, ptr, s);
+	size_t nbytes = BufferingInputStream::Read(lock, offset, ptr, s);
 	InputStream::offset += nbytes;
 	return nbytes;
 }

src/input/BufferingInputStream.cxx

@@ -55,42 +55,25 @@ BufferingInputStream::Check()
 		input->Check();
 }
 
-void
-BufferingInputStream::Seek(std::unique_lock<Mutex> &lock, size_t new_offset)
+bool
+BufferingInputStream::IsAvailable(size_t offset) noexcept
 {
-	if (new_offset >= size()) {
-		offset = new_offset;
-		return;
-	}
-
-	auto r = buffer.Read(new_offset);
-	if (r.HasData()) {
-		/* nice, we already have some data at the desired
-		   offset and this method call is a no-op */
-		offset = new_offset;
-		return;
-	}
-
-	seek_offset = new_offset;
-	seek = true;
-	wake_cond.notify_one();
+	if (offset >= size())
+		return true;
 
-	client_cond.wait(lock, [this]{ return !seek; });
+	if (buffer.Read(offset).HasData())
+		return true;
 
-	if (seek_error)
-		std::rethrow_exception(std::exchange(seek_error, {}));
+	/* if no data is available now, make sure it will be soon */
+	if (want_offset == INVALID_OFFSET)
+		want_offset = offset;
 
-	offset = new_offset;
-}
-
-bool
-BufferingInputStream::IsAvailable() noexcept
-{
-	return offset == size() || buffer.Read(offset).HasData();
+	return false;
 }
 
 size_t
-BufferingInputStream::Read(std::unique_lock<Mutex> &lock, void *ptr, size_t s)
+BufferingInputStream::Read(std::unique_lock<Mutex> &lock, size_t offset,
+			   void *ptr, size_t s)
 {
 	if (offset >= size())
 		return 0;
@@ -101,13 +84,15 @@ BufferingInputStream::Read(std::unique_lock<Mutex> &lock, void *ptr, size_t s)
 			/* yay, we have some data */
 			size_t nbytes = std::min(s, r.defined_buffer.size);
 			memcpy(ptr, r.defined_buffer.data, nbytes);
-			offset += nbytes;
 			return nbytes;
 		}
 
 		if (error)
 			std::rethrow_exception(error);
 
+		if (want_offset == INVALID_OFFSET)
+			want_offset = offset;
+
 		client_cond.wait(lock);
 	}
 }
@@ -132,27 +117,13 @@ inline void
 BufferingInputStream::RunThreadLocked(std::unique_lock<Mutex> &lock)
 {
 	while (!stop) {
-		if (seek) {
-			try {
-				input->Seek(lock, seek_offset);
-			} catch (...) {
-				seek_error = std::current_exception();
-			}
+		if (want_offset != INVALID_OFFSET) {
+			assert(want_offset < size());
 
-			seek = false;
-			client_cond.notify_all();
-		} else if (offset != input->GetOffset() && !IsAvailable()) {
-			/* a past Seek() call was a no-op because data
-			   was already available at that position, but
-			   now we've reached a new position where
-			   there is no more data in the buffer, and
-			   our input is reading somewhere else (maybe
-			   stuck at the end of the file); to find a
-			   way out, we now seek our input to our
-			   reading position to be able to fill our
-			   buffer */
-
-			input->Seek(lock, offset);
+			const size_t seek_offset = want_offset;
+			want_offset = INVALID_OFFSET;
+			if (!buffer.Read(seek_offset).HasData())
+				input->Seek(lock, seek_offset);
 		} else if (input->IsEOF()) {
 			/* our input has reached its end: prepare
 			   reading the first remaining hole */
@@ -170,27 +141,13 @@ BufferingInputStream::RunThreadLocked(std::unique_lock<Mutex> &lock)
 			auto w = buffer.Write(read_offset);
 
 			if (w.empty()) {
-				if (IsAvailable()) {
-					/* we still have enough data
-					   for the next Read() - seek
-					   to the first hole */
-
-					size_t new_offset = FindFirstHole();
-					if (new_offset == INVALID_OFFSET)
-						/* the file has been
-						   read completely */
-						break;
-
-					input->Seek(lock, new_offset);
-				} else {
-					/* we need more data at our
-					   current position, because
-					   the next Read() will stall
-					   - seek our input to our
-					   offset to prepare filling
-					   the buffer from there */
-					input->Seek(lock, offset);
-				}
+				size_t new_offset = FindFirstHole();
+				if (new_offset == INVALID_OFFSET)
+					/* the file has been read
+					   completely */
+					break;
+
+				input->Seek(lock, new_offset);
 
 				continue;
 			}

src/input/BufferingInputStream.hxx

@@ -55,11 +55,9 @@ class BufferingInputStream : InputStreamHandler {
 
 	SparseBuffer<uint8_t> buffer;
 
-	bool stop = false, seek = false;
+	bool stop = false;
 
-	size_t offset = 0;
-
-	size_t seek_offset;
+	size_t want_offset = INVALID_OFFSET;
 
 	std::exception_ptr error, seek_error;
 
@@ -78,9 +76,9 @@ public:
 	}
 
 	void Check();
-	void Seek(std::unique_lock<Mutex> &lock, size_t new_offset);
-	bool IsAvailable() noexcept;
-	size_t Read(std::unique_lock<Mutex> &lock, void *ptr, size_t size);
+	bool IsAvailable(size_t offset) noexcept;
+	size_t Read(std::unique_lock<Mutex> &lock, size_t offset,
+		    void *ptr, size_t size);
 
 protected:
 	virtual void OnBufferAvailable() noexcept {}