genode/repos/os/src/lib/vfs/block_file_system.h

452 lines
13 KiB
C
Raw Normal View History

/*
* \brief Block device file system
* \author Josef Soentgen
* \author Norman Feske
* \date 2013-12-20
*/
/*
* Copyright (C) 2013-2017 Genode Labs GmbH
*
* This file is part of the Genode OS framework, which is distributed
* under the terms of the GNU Affero General Public License version 3.
*/
#ifndef _INCLUDE__VFS__BLOCK_FILE_SYSTEM_H_
#define _INCLUDE__VFS__BLOCK_FILE_SYSTEM_H_
#include <base/allocator_avl.h>
#include <block_session/connection.h>
#include <vfs/single_file_system.h>
namespace Vfs { class Block_file_system; }
class Vfs::Block_file_system : public Single_file_system
{
private:
Vfs::Env &_env;
typedef Genode::String<64> Label;
Label _label;
/*
* Serialize access to packet stream of the block session
*/
2020-02-20 15:47:08 +01:00
Mutex _mutex { };
char *_block_buffer;
unsigned _block_buffer_count;
Genode::Allocator_avl _tx_block_alloc { &_env.alloc() };
Block::Connection<> _block {
_env.env(), &_tx_block_alloc, 128*1024, _label.string() };
Block::Session::Info const _info { _block.info() };
Block::Session::Tx::Source *_tx_source;
bool _writeable;
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
Genode::Signal_receiver _signal_receiver { };
Genode::Signal_context _signal_context { };
Genode::Signal_context_capability _source_submit_cap;
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
/*
* Noncopyable
*/
Block_file_system(Block_file_system const &);
Block_file_system &operator = (Block_file_system const &);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
class Block_vfs_handle : public Single_vfs_handle
{
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
private:
Genode::Allocator &_alloc;
Label &_label;
2020-02-20 15:47:08 +01:00
Mutex &_mutex;
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
char *_block_buffer;
unsigned &_block_buffer_count;
Genode::Allocator_avl &_tx_block_alloc;
Block::Connection<> &_block;
Genode::size_t const _block_size;
Block::sector_t const _block_count;
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
Block::Session::Tx::Source *_tx_source;
bool const _writeable;
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
Genode::Signal_receiver &_signal_receiver;
Genode::Signal_context &_signal_context;
Genode::Signal_context_capability &_source_submit_cap;
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
/*
* Noncopyable
*/
Block_vfs_handle(Block_vfs_handle const &);
Block_vfs_handle &operator = (Block_vfs_handle const &);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
file_size _block_io(file_size nr, void *buf, file_size sz,
bool write, bool bulk = false)
{
Block::Packet_descriptor::Opcode op;
op = write ? Block::Packet_descriptor::WRITE : Block::Packet_descriptor::READ;
file_size packet_size = bulk ? sz : _block_size;
file_size packet_count = bulk ? (sz / _block_size) : 1;
Block::Packet_descriptor packet;
/* sanity check */
if (packet_count > _block_buffer_count) {
packet_size = _block_buffer_count * _block_size;
packet_count = _block_buffer_count;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
while (true) {
try {
2020-02-20 15:47:08 +01:00
Mutex::Guard guard(_mutex);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
packet = _block.alloc_packet(packet_size);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
break;
} catch (Block::Session::Tx::Source::Packet_alloc_failed) {
if (!_tx_source->ready_to_submit())
_signal_receiver.wait_for_signal();
else {
if (packet_count > 1) {
packet_size /= 2;
packet_count /= 2;
}
}
}
}
2020-02-20 15:47:08 +01:00
Mutex::Guard guard(_mutex);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
Block::Packet_descriptor p(packet, op, nr, packet_count);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
if (write)
Genode::memcpy(_tx_source->packet_content(p), buf, packet_size);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
_tx_source->submit_packet(p);
p = _tx_source->get_acked_packet();
if (!p.succeeded()) {
Genode::error("Could not read block(s)");
_tx_source->release_packet(p);
return 0;
}
if (!write)
Genode::memcpy(buf, _tx_source->packet_content(p), packet_size);
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
_tx_source->release_packet(p);
return packet_size;
}
public:
Block_vfs_handle(Directory_service &ds,
File_io_service &fs,
Genode::Allocator &alloc,
Label &label,
2020-02-20 15:47:08 +01:00
Mutex &mutex,
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
char *block_buffer,
unsigned &block_buffer_count,
Genode::Allocator_avl &tx_block_alloc,
Block::Connection<> &block,
Genode::size_t block_size,
Block::sector_t block_count,
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
Block::Session::Tx::Source *tx_source,
bool writeable,
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
Genode::Signal_receiver &signal_receiver,
Genode::Signal_context &signal_context,
Genode::Signal_context_capability &source_submit_cap)
: Single_vfs_handle(ds, fs, alloc, 0),
_alloc(alloc),
_label(label),
2020-02-20 15:47:08 +01:00
_mutex(mutex),
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
_block_buffer(block_buffer),
_block_buffer_count(block_buffer_count),
_tx_block_alloc(tx_block_alloc),
_block(block),
_block_size(block_size),
_block_count(block_count),
_tx_source(tx_source),
_writeable(writeable),
_signal_receiver(signal_receiver),
_signal_context(signal_context),
_source_submit_cap(source_submit_cap)
{ }
Read_result read(char *dst, file_size count,
file_size &out_count) override
{
file_size seek_offset = seek();
file_size read = 0;
while (count > 0) {
file_size displ = 0;
file_size length = 0;
file_size nbytes = 0;
file_size blk_nr = seek_offset / _block_size;
displ = seek_offset % _block_size;
if ((displ + count) > _block_size)
length = (_block_size - displ);
else
length = count;
/*
* We take a shortcut and read the blocks all at once if the
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
* offset is aligned on a block boundary and the count is a
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
* multiple of the block size, e.g. 4K reads will be read at
* once.
*
* XXX this is quite hackish because we have to omit partial
* blocks at the end.
*/
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
if (displ == 0 && !(count < _block_size)) {
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
file_size bytes_left = count - (count % _block_size);
nbytes = _block_io(blk_nr, dst + read, bytes_left, false, true);
if (nbytes == 0) {
Genode::error("error while reading block:", blk_nr, " from block device");
return READ_ERR_INVALID;
}
read += nbytes;
count -= nbytes;
seek_offset += nbytes;
continue;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
nbytes = _block_io(blk_nr, _block_buffer, _block_size, false);
if ((unsigned)nbytes != _block_size) {
Genode::error("error while reading block:", blk_nr, " from block device");
return READ_ERR_INVALID;
}
Genode::memcpy(dst + read, _block_buffer + displ, length);
read += length;
count -= length;
seek_offset += length;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
out_count = read;
return READ_OK;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
Write_result write(char const *buf, file_size count,
file_size &out_count) override
{
if (!_writeable) {
Genode::error("block device is not writeable");
return WRITE_ERR_INVALID;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
file_size seek_offset = seek();
file_size written = 0;
while (count > 0) {
file_size displ = 0;
file_size length = 0;
file_size nbytes = 0;
file_size blk_nr = seek_offset / _block_size;
displ = seek_offset % _block_size;
if ((displ + count) > _block_size)
length = (_block_size - displ);
else
length = count;
/*
* We take a shortcut and write as much as possible without
* using the block buffer if the offset is aligned on a block
* boundary and the count is a multiple of the block size,
* e.g. 4K writes will be written at once.
*
* XXX this is quite hackish because we have to omit partial
* blocks at the end.
*/
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
if (displ == 0 && !(count < _block_size)) {
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
file_size bytes_left = count - (count % _block_size);
nbytes = _block_io(blk_nr, (void*)(buf + written),
bytes_left, true, true);
if (nbytes == 0) {
Genode::error("error while write block:", blk_nr, " to block device");
return WRITE_ERR_INVALID;
}
written += nbytes;
count -= nbytes;
seek_offset += nbytes;
continue;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
/*
* The offset is not aligned on a block boundary. Therefore
* we need to read the block to the block buffer first and
* put the buffer content at the right offset before we can
* write the whole block back. In addition if length is less
* than block size, we also have to read the block first.
*/
if (displ > 0 || length < _block_size)
_block_io(blk_nr, _block_buffer, _block_size, false);
Genode::memcpy(_block_buffer + displ, buf + written, length);
nbytes = _block_io(blk_nr, _block_buffer, _block_size, true);
if ((unsigned)nbytes != _block_size) {
Genode::error("error while writing block:", blk_nr, " to block_device");
return WRITE_ERR_INVALID;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
written += length;
count -= length;
seek_offset += length;
}
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
out_count = written;
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
return WRITE_OK;
}
Sync_result sync() override
{
/*
* Just in case bail if we cannot submit the packet.
* (Since the plugin operates in a synchronous fashion
* that should not happen.)
*/
if (!_tx_source->ready_to_submit()) {
Genode::error("vfs_block: could not sync blocks");
return SYNC_ERR_INVALID;
}
Block::Packet_descriptor p =
Block::Session::sync_all_packet_descriptor(
_block.info(), Block::Session::Tag { 0 });
_tx_source->submit_packet(p);
p = _tx_source->get_acked_packet();
_tx_source->release_packet(p);
if (!p.succeeded()) {
Genode::error("vfs_block: syncing blocks failed");
return SYNC_ERR_INVALID;
}
return SYNC_OK;
}
bool read_ready() override { return true; }
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
};
public:
Block_file_system(Vfs::Env &env, Genode::Xml_node config)
:
Single_file_system(Node_type::CONTINUOUS_FILE, name(),
Node_rwx::rw(), config),
_env(env),
_label(config.attribute_value("label", Label())),
_block_buffer(0),
_block_buffer_count(config.attribute_value("block_buffer_count", 1UL)),
_tx_source(_block.tx()),
_writeable(_info.writeable),
_source_submit_cap(_signal_receiver.manage(&_signal_context))
{
_block_buffer = new (_env.alloc())
char[_block_buffer_count * _info.block_size];
_block.tx_channel()->sigh_ready_to_submit(_source_submit_cap);
}
~Block_file_system()
{
_signal_receiver.dissolve(&_signal_context);
destroy(_env.alloc(), _block_buffer);
}
2017-02-07 19:03:23 +01:00
static char const *name() { return "block"; }
char const *type() override { return "block"; }
/*********************************
** Directory service interface **
*********************************/
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
Open_result open(char const *path, unsigned,
Vfs_handle **out_handle,
Allocator &alloc) override
{
if (!_single_file(path))
return OPEN_ERR_UNACCESSIBLE;
try {
*out_handle = new (alloc) Block_vfs_handle(*this, *this, alloc,
2020-02-20 15:47:08 +01:00
_label, _mutex,
_block_buffer,
_block_buffer_count,
_tx_block_alloc,
_block,
_info.block_size,
_info.block_count,
_tx_source,
_info.writeable,
_signal_receiver,
_signal_context,
_source_submit_cap);
return OPEN_OK;
}
catch (Genode::Out_of_ram) { return OPEN_ERR_OUT_OF_RAM; }
catch (Genode::Out_of_caps) { return OPEN_ERR_OUT_OF_CAPS; }
VFS: nonblocking interface The VFS library can be used in single-threaded or multi-threaded environments and depending on that, signals are handled by the same thread which uses the VFS library or possibly by a different thread. If a VFS plugin needs to block to wait for a signal, there is currently no way which works reliably in both environments. For this reason, this commit makes the interface of the VFS library nonblocking, similar to the File_system session interface. The most important changes are: - Directories are created and opened with the 'opendir()' function and the directory entries are read with the recently introduced 'queue_read()' and 'complete_read()' functions. - Symbolic links are created and opened with the 'openlink()' function and the link target is read with the 'queue_read()' and 'complete_read()' functions and written with the 'write()' function. - The 'write()' function does not wait for signals anymore. This can have the effect that data written by a VFS library user has not been processed by a file system server yet when the library user asks for the size of the file or closes it (both done with RPC functions at the file system server). For this reason, a user of the VFS library should request synchronization before calling 'stat()' or 'close()'. To make sure that a file system server has processed all write request packets which a client submitted before the synchronization request, synchronization is now requested at the file system server with a synchronization packet instead of an RPC function. Because of this change, the synchronization interface of the VFS library is now split into 'queue_sync()' and 'complete_sync()' functions. Fixes #2399
2017-08-15 20:51:53 +02:00
}
Stat_result stat(char const *path, Stat &out) override
{
Stat_result const result = Single_file_system::stat(path, out);
out.size = _info.block_count * _info.block_size;
return result;
}
/********************************
** File I/O service interface **
********************************/
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
Ftruncate_result ftruncate(Vfs_handle *, file_size) override
{
return FTRUNCATE_OK;
}
Follow practices suggested by "Effective C++" The patch adjust the code of the base, base-<kernel>, and os repository. To adapt existing components to fix violations of the best practices suggested by "Effective C++" as reported by the -Weffc++ compiler argument. The changes follow the patterns outlined below: * A class with virtual functions can no longer publicly inherit base classed without a vtable. The inherited object may either be moved to a member variable, or inherited privately. The latter would be used for classes that inherit 'List::Element' or 'Avl_node'. In order to enable the 'List' and 'Avl_tree' to access the meta data, the 'List' must become a friend. * Instead of adding a virtual destructor to abstract base classes, we inherit the new 'Interface' class, which contains a virtual destructor. This way, single-line abstract base classes can stay as compact as they are now. The 'Interface' utility resides in base/include/util/interface.h. * With the new warnings enabled, all member variables must be explicitly initialized. Basic types may be initialized with '='. All other types are initialized with braces '{ ... }' or as class initializers. If basic types and non-basic types appear in a row, it is nice to only use the brace syntax (also for basic types) and align the braces. * If a class contains pointers as members, it must now also provide a copy constructor and assignment operator. In the most cases, one would make them private, effectively disallowing the objects to be copied. Unfortunately, this warning cannot be fixed be inheriting our existing 'Noncopyable' class (the compiler fails to detect that the inheriting class cannot be copied and still gives the error). For now, we have to manually add declarations for both the copy constructor and assignment operator as private class members. Those declarations should be prepended with a comment like this: /* * Noncopyable */ Thread(Thread const &); Thread &operator = (Thread const &); In the future, we should revisit these places and try to replace the pointers with references. In the presence of at least one reference member, the compiler would no longer implicitly generate a copy constructor. So we could remove the manual declaration. Issue #465
2017-12-21 15:42:15 +01:00
Ioctl_result ioctl(Vfs_handle *, Ioctl_opcode opcode, Ioctl_arg,
Ioctl_out &out) override
{
switch (opcode) {
case IOCTL_OP_DIOCGMEDIASIZE:
out.diocgmediasize.size = _info.block_count * _info.block_size;
return IOCTL_OK;
default:
Genode::warning("invalid ioctl request ", (int)opcode);
break;
}
/* never reached */
return IOCTL_ERR_INVALID;
}
};
#endif /* _INCLUDE__VFS__BLOCK_FILE_SYSTEM_H_ */