sapling/eden/fs/fuse/Dispatcher.h

367 lines
9.8 KiB
C
Raw Normal View History

/*
* Copyright (c) 2016-present, Facebook, Inc.
* All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree. An additional grant
* of patent rights can be found in the PATENTS file in the same directory.
*
*/
#pragma once
#include <folly/Portability.h>
#include <folly/Range.h>
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
#include <sys/statvfs.h>
#include "eden/fs/fuse/FileHandleMap.h"
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
#include "eden/fs/fuse/FuseTypes.h"
#include "eden/fs/utils/PathFuncs.h"
namespace folly {
template <class T, class Tag, class AccessMode>
class ThreadLocal;
template <class T>
class Future;
}; // namespace folly
namespace facebook {
namespace eden {
namespace fusell {
#define FUSELL_NOT_IMPL() \
do { \
LOG_FIRST_N(ERROR, 1) << __PRETTY_FUNCTION__ << " not implemented"; \
folly::throwSystemErrorExplicit(ENOSYS, __PRETTY_FUNCTION__); \
} while (0)
class Dispatcher;
class EdenStats;
class EdenStatsTag;
class RequestData;
class FileHandle;
class DirHandle;
class MountPoint;
using ThreadLocalEdenStats = folly::ThreadLocal<EdenStats, EdenStatsTag, void>;
class Dispatcher {
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fuse_init_out connInfo_;
ThreadLocalEdenStats* stats_{nullptr};
FileHandleMap fileHandles_;
public:
virtual ~Dispatcher();
explicit Dispatcher(ThreadLocalEdenStats* stats);
ThreadLocalEdenStats* getStats() const;
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
const fuse_init_out& getConnInfo() const;
FileHandleMap& getFileHandles();
// delegates to FileHandleMap::getGenericFileHandle
std::shared_ptr<FileHandleBase> getGenericFileHandle(uint64_t fh);
// delegates to FileHandleMap::getFileHandle
std::shared_ptr<FileHandle> getFileHandle(uint64_t fh);
// delegates to FileHandleMap::getDirHandle
std::shared_ptr<DirHandle> getDirHandle(uint64_t dh);
/**
* Called during filesystem mounting. It informs the filesystem
* of kernel capabilities and provides an opportunity to poke some
* flags and limits in the conn_info to report capabilities back
* to the kernel
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual void initConnection(const fuse_init_out& out);
/**
* Called when fuse is tearing down the session
*/
virtual void destroy();
/**
* Lookup a directory entry by name and get its attributes
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<fuse_entry_out> lookup(
fusell::InodeNumber parent,
PathComponentPiece name);
/**
* Forget about an inode
*
* The nlookup parameter indicates the number of lookups
* previously performed on this inode.
*
* If the filesystem implements inode lifetimes, it is recommended
* that inodes acquire a single reference on each lookup, and lose
* nlookup references on each forget.
*
* The filesystem may ignore forget calls, if the inodes don't
* need to have a limited lifetime.
*
* On unmount it is not guaranteed, that all referenced inodes
* will receive a forget message.
*
* @param ino the inode number
* @param nlookup the number of lookups to forget
*/
virtual folly::Future<folly::Unit> forget(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber ino,
unsigned long nlookup);
/**
* The stat information and the cache TTL for the kernel
*
* The timeout value is measured in seconds and indicates how long
* the kernel side of the FUSE will cache the values in the
* struct stat before calling getattr() again to refresh it.
*/
struct Attr {
struct stat st;
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
uint64_t timeout_seconds;
explicit Attr(
const struct stat& st,
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
uint64_t timeout = std::numeric_limits<uint64_t>::max());
fuse_attr_out asFuseAttr() const;
};
/**
* Get file attributes
*
* @param ino the inode number
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<Attr> getattr(fusell::InodeNumber ino);
/**
* Set file attributes
*
* In the 'attr' argument only members indicated by the 'to_set'
* bitmask contain valid values. Other members contain undefined
* values.
*
* @param ino the inode number
* @param attr the attributes
* @param to_set bit mask of attributes which should be set
*
* Changed in version 2.5:
* file information filled in for ftruncate
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<Attr> setattr(
fusell::InodeNumber ino,
const fuse_setattr_in& attr);
/**
* Read symbolic link
*
* @param ino the inode number
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<std::string> readlink(fusell::InodeNumber ino);
/**
* Create file node
*
* Create a regular file, character device, block device, fifo or
* socket node.
*
* @param parent inode number of the parent directory
* @param name to create
* @param mode file type and mode with which to create the new file
* @param rdev the device number (only valid if created file is a device)
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<fuse_entry_out> mknod(
fusell::InodeNumber parent,
PathComponentPiece name,
mode_t mode,
dev_t rdev);
/**
* Create a directory
*
* @param parent inode number of the parent directory
* @param name to create
* @param mode with which to create the new file
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<fuse_entry_out>
mkdir(fusell::InodeNumber parent, PathComponentPiece name, mode_t mode);
/**
* Remove a file
*
* @param parent inode number of the parent directory
* @param name to remove
*/
virtual folly::Future<folly::Unit> unlink(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber parent,
PathComponentPiece name);
/**
* Remove a directory
*
* @param parent inode number of the parent directory
* @param name to remove
*/
virtual folly::Future<folly::Unit> rmdir(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber parent,
PathComponentPiece name);
/**
* Create a symbolic link
*
* @param parent inode number of the parent directory
* @param name to create
* @param link the contents of the symbolic link
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<fuse_entry_out> symlink(
fusell::InodeNumber parent,
PathComponentPiece name,
folly::StringPiece link);
/**
* Rename a file
*
* @param parent inode number of the old parent directory
* @param name old name
* @param newparent inode number of the new parent directory
* @param newname new name
*/
virtual folly::Future<folly::Unit> rename(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber parent,
PathComponentPiece name,
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber newparent,
PathComponentPiece newname);
/**
* Create a hard link
*
* @param ino the old inode number
* @param newparent inode number of the new parent directory
* @param newname new name to create
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<fuse_entry_out> link(
fusell::InodeNumber ino,
fusell::InodeNumber newparent,
PathComponentPiece newname);
/**
* Open a file
*
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
* open(2) flags (with the exception of O_CREAT, O_EXCL, O_NOCTTY and
* O_TRUNC) are available in the flags parameter.
*/
virtual folly::Future<std::shared_ptr<FileHandle>> open(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber ino,
int flags);
/**
* Open a directory
*
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
* open(2) flags are available in the flags parameter.
*/
virtual folly::Future<std::shared_ptr<DirHandle>> opendir(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber ino,
int flags);
/**
* Get file system statistics
*
* @param ino the inode number, zero means "undefined"
*/
virtual folly::Future<struct fuse_kstatfs> statfs(fusell::InodeNumber ino);
/**
* Set an extended attribute
*/
virtual folly::Future<folly::Unit> setxattr(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber ino,
folly::StringPiece name,
folly::StringPiece value,
int flags);
/**
* Get an extended attribute
*/
virtual folly::Future<std::string> getxattr(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber ino,
folly::StringPiece name);
static const int kENOATTR;
/**
* List extended attribute names
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<std::vector<std::string>> listxattr(
fusell::InodeNumber ino);
/**
* Remove an extended attribute
*
* @param ino the inode number
* @param name of the extended attribute
*/
virtual folly::Future<folly::Unit> removexattr(
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fusell::InodeNumber ino,
folly::StringPiece name);
/**
* Check file access permissions
*
* This will be called for the access() system call. If the
* 'default_permissions' mount option is given, this method is not
* called.
*
* This method is not called under Linux kernel versions 2.4.x
*
* Introduced in version 2.5
*
* @param ino the inode number
* @param mask requested access mode
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<folly::Unit> access(fusell::InodeNumber ino, int mask);
struct Create {
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
fuse_entry_out entry;
std::shared_ptr<FileHandle> fh;
};
/**
* Create and open a file
*
* If the file does not exist, first create it with the specified
* mode, and then open it.
*
* Open flags (with the exception of O_NOCTTY) are available in
* fi->flags.
*
* If this method is not implemented or under Linux kernel
* versions earlier than 2.6.15, the mknod() and open() methods
* will be called instead.
*
* Introduced in version 2.5
*
* @param parent inode number of the parent directory
* @param name to create
* @param mode file type and mode with which to create the new file
*/
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
virtual folly::Future<Create> create(
fusell::InodeNumber parent,
PathComponentPiece name,
mode_t mode,
int flags);
/**
* Map block index within file to block index within device
*
* Note: This makes sense only for block device backed filesystems
* mounted with the 'blkdev' option
*
* Introduced in version 2.6
*
* @param ino the inode number
* @param blocksize unit of block index
* @param idx block index within file
*/
virtual folly::Future<uint64_t>
remove dep on libfuse Summary: This serves a few purposes: 1. We can avoid some conditional code inside eden if we know that we have a specific fuse_kernel.h header implementation. 2. We don't have to figure out a way to propagate the kernel capabilities through the graceful restart process. 3. libfuse3 removed the channel/session hooks that we've been using thus far to interject ourselves for mounting and graceful restarting, so we were already effectively the walking dead here. 4. We're now able to take advtange of the latest aspects of the fuse kernel interface without being tied to the implementation of libfuse2 or libfuse3. We're interested in the readdirplus functionality and will look at enabling that in a future diff. This may make some things slightly harder for the more immediate macOS port but I belive that we're in a much better place overall. This diff is relatively mechanical and sadly is (unavoidably) large. The main aspects of this diff are: 1. The `fuse_ino_t` type was provided by libfuse so we needed to replace it with our own definition. This has decent penetration throughout the codebase. 2. The confusing `fuse_file_info` type that was multi-purpose and had fields that were sometimes *in* parameters and sometimes *out* parameters has been removed and replaced with a simpler *flags* parameter that corresponds to the `open(2)` flags parameter. The *out* portions are subsumed by existing file handle metadata methods. 3. The fuse parameters returned from variations of the `LOOKUP` opcode now return the fuse kernel type for this directly. I suspect that we may need to introduce a compatibility type when we revisit the macOS port, but this at least makes this diff slightly simpler. You'll notice that some field and symbol name prefixes vary as a result of this. 4. Similarly for `setattr`, libfuse separated the kernel data into two parameters that were a little awkward to use; we're now just passing the kernel data through and this, IMO, makes the interface slightly more understandable. 5. The bulk of the code from `Dispatcher.cpp` that shimmed the libfuse callbacks into the C++ virtual methods has been removed and replaced by a `switch` statement based dispatcher in `FuseChannel`. I'm not married to this being `switch` based and may revise this to be driven by an `unordered_map` of opcode -> dispatcher method defined in `FuseChannel`. Regardless, `Dispatcher.cpp` is now much slimmer and should be easier to replace by rolling it together into `EdenDispatcher.cpp` in the future should we desire to do so. 6. This diff disables dispatching `poll` and `ioctl` calls. We didn't make use of them and their interfaces are a bit fiddly. 7. `INTERRUPT` is also disabled here. I will re-enable it in a follow-up diff where I can also revise how we track outstanding requests for graceful shutdown. 8. I've imported `fuse_kernel.h` from libfuse. This is included under the permissive 2-clause BSD license that it allows for exactly this integration purpose. Reviewed By: simpkins Differential Revision: D6576472 fbshipit-source-id: 7cb088af5e06fe27bf22a1bed295c18c17d8006c
2018-01-03 03:25:03 +03:00
bmap(fusell::InodeNumber ino, size_t blocksize, uint64_t idx);
};
} // namespace fusell
} // namespace eden
} // namespace facebook