blob: 467a3f1d6a9487b1351ea470cf3eb1f5529153db (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
|
#pragma once
#include <Core/Types.h>
#include <optional>
#include <memory>
#include <future>
#include <boost/noncopyable.hpp>
#include <Common/Stopwatch.h>
#include <Common/Priority.h>
namespace DB
{
/** Interface for asynchronous reads from file descriptors.
* It can abstract Linux AIO, io_uring or normal reads from separate thread pool,
* and also reads from non-local filesystems.
* The implementation not necessarily to be efficient for large number of small requests,
* instead it should be ok for moderate number of sufficiently large requests
* (e.g. read 1 MB of data 50 000 times per seconds; BTW this is normal performance for reading from page cache).
* For example, this interface may not suffice if you want to serve 10 000 000 of 4 KiB requests per second.
* This interface is fairly limited.
*/
class IAsynchronousReader : private boost::noncopyable
{
public:
/// For local filesystems, the file descriptor is simply integer
/// but it can be arbitrary opaque object for remote filesystems.
struct IFileDescriptor
{
virtual ~IFileDescriptor() = default;
};
using FileDescriptorPtr = std::shared_ptr<IFileDescriptor>;
struct LocalFileDescriptor : public IFileDescriptor
{
explicit LocalFileDescriptor(int fd_) : fd(fd_) {}
int fd;
};
/// Read from file descriptor at specified offset up to size bytes into buf.
/// Some implementations may require alignment and it is responsibility of
/// the caller to provide conforming requests.
struct Request
{
FileDescriptorPtr descriptor;
size_t offset = 0;
size_t size = 0;
char * buf = nullptr;
Priority priority;
size_t ignore = 0;
};
struct Result
{
/// size
/// Less than requested amount of data can be returned.
/// If size is zero - the file has ended.
/// (for example, EINTR must be handled by implementation automatically)
size_t size = 0;
/// offset
/// Optional. Useful when implementation needs to do ignore().
size_t offset = 0;
std::unique_ptr<Stopwatch> execution_watch = {};
operator std::tuple<size_t &, size_t &>() { return {size, offset}; }
};
/// Submit request and obtain a handle. This method don't perform any waits.
/// If this method did not throw, the caller must wait for the result with 'wait' method
/// or destroy the whole reader before destroying the buffer for request.
/// The method can be called concurrently from multiple threads.
virtual std::future<Result> submit(Request request) = 0;
virtual void wait() = 0;
/// Destructor must wait for all not completed request and ignore the results.
/// It may also cancel the requests.
virtual ~IAsynchronousReader() = default;
};
using AsynchronousReaderPtr = std::shared_ptr<IAsynchronousReader>;
}
|
