1 /* Copyright (c) 2015-2019. The SimGrid Team. All rights reserved. */
3 /* This program is free software; you can redistribute it and/or modify it
4 * under the terms of the license (GNU LGPL) which comes with this package. */
6 #include "simgrid/plugins/file_system.h"
7 #include "simgrid/s4u/Actor.hpp"
8 #include "src/surf/HostImpl.hpp"
9 #include "src/surf/xml/platf_private.hpp"
10 #include "xbt/config.hpp"
13 #include <boost/algorithm/string.hpp>
14 #include <boost/algorithm/string/join.hpp>
15 #include <boost/algorithm/string/split.hpp>
19 XBT_LOG_NEW_DEFAULT_CATEGORY(s4u_file, "S4U files");
20 int sg_storage_max_file_descriptors = 1024;
24 simgrid::xbt::Extension<Disk, FileSystemDiskExt> FileSystemDiskExt::EXTENSION_ID;
25 simgrid::xbt::Extension<Storage, FileSystemStorageExt> FileSystemStorageExt::EXTENSION_ID;
26 simgrid::xbt::Extension<Host, FileDescriptorHostExt> FileDescriptorHostExt::EXTENSION_ID;
28 File::File(const std::string& fullpath, void* userdata) : File(fullpath, Host::current(), userdata){};
30 File::File(const std::string& fullpath, sg_host_t host, void* userdata) : fullpath_(fullpath), userdata_(userdata)
32 // this cannot fail because we get a xbt_die if the mountpoint does not exist
33 if (not host->get_mounted_storages().empty()) {
34 Storage* st = nullptr;
35 size_t longest_prefix_length = 0;
36 XBT_DEBUG("Search for storage name for '%s' on '%s'", fullpath_.c_str(), host->get_cname());
38 for (auto const& mnt : host->get_mounted_storages()) {
39 XBT_DEBUG("See '%s'", mnt.first.c_str());
40 mount_point_ = fullpath_.substr(0, mnt.first.length());
42 if (mount_point_ == mnt.first && mnt.first.length() > longest_prefix_length) {
43 /* The current mount name is found in the full path and is bigger than the previous*/
44 longest_prefix_length = mnt.first.length();
48 if (longest_prefix_length > 0) { /* Mount point found, split fullpath_ into mount_name and path+filename*/
49 mount_point_ = fullpath_.substr(0, longest_prefix_length);
50 path_ = fullpath_.substr(longest_prefix_length, fullpath_.length());
52 xbt_die("Can't find mount point for '%s' on '%s'", fullpath_.c_str(), host->get_cname());
56 if (not host->get_disks().empty()) {
58 size_t longest_prefix_length = 0;
59 for (auto const& disk : host->get_disks()) {
60 const char* current_mount_str = disk->get_property("mount");
61 if (current_mount_str) {
62 std::string current_mount = std::string(current_mount_str);
63 mount_point_ = fullpath_.substr(0, current_mount.length());
64 if (mount_point_ == current_mount && current_mount.length() > longest_prefix_length) {
65 /* The current mount name is found in the full path and is bigger than the previous*/
66 longest_prefix_length = current_mount.length();
69 if (longest_prefix_length > 0) { /* Mount point found, split fullpath_ into mount_name and path+filename*/
70 mount_point_ = fullpath_.substr(0, longest_prefix_length);
71 if (mount_point_ == std::string("/"))
74 path_ = fullpath_.substr(longest_prefix_length, fullpath_.length());
75 XBT_DEBUG("%s + %s", mount_point_.c_str(), path_.c_str());
77 xbt_die("Can't find mount point for '%s' on '%s'", fullpath_.c_str(), host->get_cname());
83 // assign a file descriptor id to the newly opened File
84 FileDescriptorHostExt* ext = host->extension<simgrid::s4u::FileDescriptorHostExt>();
85 if (ext->file_descriptor_table == nullptr) {
86 ext->file_descriptor_table.reset(new std::vector<int>(sg_storage_max_file_descriptors));
87 std::iota(ext->file_descriptor_table->rbegin(), ext->file_descriptor_table->rend(), 0); // Fill with ..., 1, 0.
89 xbt_assert(not ext->file_descriptor_table->empty(), "Too much files are opened! Some have to be closed.");
90 desc_id = ext->file_descriptor_table->back();
91 ext->file_descriptor_table->pop_back();
93 XBT_DEBUG("\tOpen file '%s'", path_.c_str());
94 std::map<std::string, sg_size_t>* content;
96 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
99 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
101 // if file does not exist create an empty file
102 auto sz = content->find(path_);
103 if (sz != content->end()) {
107 content->insert({path_, size_});
108 XBT_DEBUG("File '%s' was not found, file created.", path_.c_str());
114 Host::current()->extension<simgrid::s4u::FileDescriptorHostExt>()->file_descriptor_table->push_back(desc_id);
120 XBT_INFO("File Descriptor information:\n"
121 "\t\tFull path: '%s'\n"
123 "\t\tMount point: '%s'\n"
124 "\t\tStorage Id: '%s'\n"
125 "\t\tStorage Type: '%s'\n"
126 "\t\tFile Descriptor Id: %d",
127 get_path(), size_, mount_point_.c_str(), local_storage_->get_cname(), local_storage_->get_type(), desc_id);
129 XBT_INFO("File Descriptor information:\n"
130 "\t\tFull path: '%s'\n"
132 "\t\tMount point: '%s'\n"
133 "\t\tDisk Id: '%s'\n"
134 "\t\tHost Id: '%s'\n"
135 "\t\tFile Descriptor Id: %d",
136 get_path(), size_, mount_point_.c_str(), local_disk_->get_cname(), local_disk_->get_host()->get_cname(),
140 sg_size_t File::read(sg_size_t size)
142 if (size_ == 0) /* Nothing to read, return */
144 sg_size_t read_size = 0;
146 if (local_storage_) {
147 /* Find the host where the file is physically located and read it */
148 Host* host = local_storage_->get_host();
149 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_storage_->get_cname());
150 // if the current position is close to the end of the file, we may not be able to read the requested size
151 read_size = local_storage_->read(std::min(size, size_ - current_position_));
152 current_position_ += read_size;
154 if (host->get_name() != Host::current()->get_name() && read_size > 0) {
155 /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
156 XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", host->get_cname(), read_size);
157 host->send_to(Host::current(), read_size);
162 /* Find the host where the file is physically located and read it */
163 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_disk_->get_cname());
164 // if the current position is close to the end of the file, we may not be able to read the requested size
165 read_size = local_disk_->read(std::min(size, size_ - current_position_));
166 current_position_ += read_size;
172 /** @brief Write into a file (local or remote)
174 * @param size of the file to write
175 * @return the number of bytes successfully write or -1 if an error occurred
177 sg_size_t File::write(sg_size_t size, int write_inside)
179 if (size == 0) /* Nothing to write, return */
181 sg_size_t write_size = 0;
183 if (local_storage_) {
184 /* Find the host where the file is physically located (remote or local)*/
185 Host* host = local_storage_->get_host();
187 if (host->get_name() != Host::current()->get_name()) {
188 /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
189 XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", host->get_cname(), size);
190 Host::current()->send_to(host, size);
193 XBT_DEBUG("WRITE %s on disk '%s'. size '%llu/%llu' '%llu:%llu'", get_path(), local_storage_->get_cname(), size,
194 size_, sg_storage_get_size_used(local_storage_), sg_storage_get_size(local_storage_));
195 // If the storage is full before even starting to write
196 if (sg_storage_get_size_used(local_storage_) >= sg_storage_get_size(local_storage_))
198 if (write_inside == 0) {
199 /* Substract the part of the file that might disappear from the used sized on the storage element */
200 local_storage_->extension<FileSystemStorageExt>()->decr_used_size(size_ - current_position_);
201 write_size = local_storage_->write(size);
202 local_storage_->extension<FileSystemStorageExt>()->incr_used_size(write_size);
203 current_position_ += write_size;
204 size_ = current_position_;
206 write_size = local_storage_->write(size);
207 current_position_ += write_size;
208 if (current_position_ > size_)
209 size_ = current_position_;
211 std::map<std::string, sg_size_t>* content = local_storage_->extension<FileSystemStorageExt>()->get_content();
213 content->erase(path_);
214 content->insert({path_, size_});
218 XBT_DEBUG("WRITE %s on disk '%s'. size '%llu/%llu' '%llu:%llu'", get_path(), local_disk_->get_cname(), size, size_,
219 sg_disk_get_size_used(local_disk_), sg_disk_get_size(local_disk_));
220 // If the storage is full before even starting to write
221 if (sg_disk_get_size_used(local_disk_) >= sg_disk_get_size(local_disk_))
223 if (write_inside == 0) {
224 /* Substract the part of the file that might disappear from the used sized on the storage element */
225 local_disk_->extension<FileSystemDiskExt>()->decr_used_size(size_ - current_position_);
226 write_size = local_disk_->write(size);
227 local_disk_->extension<FileSystemDiskExt>()->incr_used_size(write_size);
228 current_position_ += write_size;
229 size_ = current_position_;
231 write_size = local_disk_->write(size);
232 current_position_ += write_size;
233 if (current_position_ > size_)
234 size_ = current_position_;
236 std::map<std::string, sg_size_t>* content = local_disk_->extension<FileSystemDiskExt>()->get_content();
238 content->erase(path_);
239 content->insert({path_, size_});
244 sg_size_t File::size()
249 void File::seek(sg_offset_t offset)
251 current_position_ = offset;
254 void File::seek(sg_offset_t offset, int origin)
258 current_position_ = offset;
261 current_position_ += offset;
264 current_position_ = size_ + offset;
271 sg_size_t File::tell()
273 return current_position_;
276 void File::move(const std::string& fullpath)
278 /* Check if the new full path is on the same mount point */
279 if (fullpath.compare(0, mount_point_.length(), mount_point_) == 0) {
280 std::map<std::string, sg_size_t>* content;
282 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
284 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
285 auto sz = content->find(path_);
286 if (sz != content->end()) { // src file exists
287 sg_size_t new_size = sz->second;
288 content->erase(path_);
289 std::string path = fullpath.substr(mount_point_.length(), fullpath.length());
290 content->insert({path.c_str(), new_size});
291 XBT_DEBUG("Move file from %s to %s, size '%llu'", path_.c_str(), fullpath.c_str(), new_size);
293 XBT_WARN("File %s doesn't exist", path_.c_str());
296 XBT_WARN("New full path %s is not on the same mount point: %s.", fullpath.c_str(), mount_point_.c_str());
302 /* Check if the file is on local storage */
303 std::map<std::string, sg_size_t>* content;
304 const char* name = nullptr;
305 if (local_storage_) {
306 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
307 name = local_storage_->get_cname();
310 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
311 name = local_disk_->get_cname();
314 if (content->find(path_) == content->end()) {
315 XBT_WARN("File %s is not on disk %s. Impossible to unlink", path_.c_str(), name);
318 XBT_DEBUG("UNLINK %s on disk '%s'", path_.c_str(), name);
321 local_storage_->extension<FileSystemStorageExt>()->decr_used_size(size_);
324 local_disk_->extension<FileSystemDiskExt>()->decr_used_size(size_);
326 // Remove the file from storage
327 content->erase(fullpath_);
333 int File::remote_copy(sg_host_t host, const char* fullpath)
335 /* Find the host where the file is physically located and read it */
337 if (local_storage_) {
338 src_host = local_storage_->get_host();
339 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_storage_->get_cname());
343 src_host = local_disk_->get_host();
344 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_disk_->get_cname());
348 // if the current position is close to the end of the file, we may not be able to read the requested size
349 sg_size_t read_size = 0;
351 read_size = local_storage_->read(size_);
353 read_size = local_disk_->read(size_);
355 current_position_ += read_size;
357 Host* dst_host = host;
358 if (local_storage_) {
359 /* Find the host that owns the storage where the file has to be copied */
360 Storage* storage_dest = nullptr;
361 size_t longest_prefix_length = 0;
363 for (auto const& elm : host->get_mounted_storages()) {
364 std::string mount_point = std::string(fullpath).substr(0, elm.first.size());
365 if (mount_point == elm.first && elm.first.length() > longest_prefix_length) {
366 /* The current mount name is found in the full path and is bigger than the previous*/
367 longest_prefix_length = elm.first.length();
368 storage_dest = elm.second;
372 if (storage_dest != nullptr) {
373 /* Mount point found, retrieve the host the storage is attached to */
374 dst_host = storage_dest->get_host();
376 XBT_WARN("Can't find mount point for '%s' on destination host '%s'", fullpath, host->get_cname());
382 size_t longest_prefix_length = 0;
383 Disk* dst_disk = nullptr;
385 for (auto const& disk : host->get_disks()) {
386 const char* current_mount_str = disk->get_property("mount");
387 if (current_mount_str) {
388 std::string current_mount = std::string(current_mount_str);
389 std::string mount_point = std::string(fullpath).substr(0, current_mount.length());
390 if (mount_point == current_mount && current_mount.length() > longest_prefix_length) {
391 /* The current mount name is found in the full path and is bigger than the previous*/
392 longest_prefix_length = current_mount.length();
398 if (dst_disk == nullptr) {
399 XBT_WARN("Can't find mount point for '%s' on destination host '%s'", fullpath, host->get_cname());
404 XBT_DEBUG("Initiate data transfer of %llu bytes between %s and %s.", read_size, src_host->get_cname(),
405 dst_host->get_cname());
406 src_host->send_to(dst_host, read_size);
408 /* Create file on remote host, write it and close it */
409 File* fd = new File(fullpath, dst_host, nullptr);
410 if (local_storage_) {
411 sg_size_t write_size = fd->local_storage_->write(read_size);
412 fd->local_storage_->extension<FileSystemStorageExt>()->incr_used_size(write_size);
413 (*(fd->local_storage_->extension<FileSystemStorageExt>()->get_content()))[path_] = size_;
416 fd->write(read_size);
421 int File::remote_move(sg_host_t host, const char* fullpath)
423 int res = remote_copy(host, fullpath);
428 FileSystemDiskExt::FileSystemDiskExt(simgrid::s4u::Disk* ptr)
430 const char* size_str = ptr->get_property("size");
431 const char* content_str = ptr->get_property("content");
432 size_ = size_str ? surf_parse_get_size(size_str, "disk size", ptr->get_name()) : 0;
434 content_.reset(parse_content(content_str));
437 FileSystemStorageExt::FileSystemStorageExt(simgrid::s4u::Storage* ptr)
439 content_.reset(parse_content(ptr->get_impl()->content_name_));
440 size_ = ptr->get_impl()->size_;
443 std::map<std::string, sg_size_t>* FileSystemDiskExt::parse_content(const std::string& filename)
445 if (filename.empty())
448 std::map<std::string, sg_size_t>* parse_content = new std::map<std::string, sg_size_t>();
450 std::ifstream* fs = surf_ifsopen(filename);
453 std::vector<std::string> tokens;
455 std::getline(*fs, line);
457 if (line.length() > 0) {
458 boost::split(tokens, line, boost::is_any_of(" \t"), boost::token_compress_on);
459 xbt_assert(tokens.size() == 2, "Parse error in %s: %s", filename.c_str(), line.c_str());
460 sg_size_t size = std::stoull(tokens.at(1));
463 parse_content->insert({tokens.front(), size});
465 } while (not fs->eof());
467 return parse_content;
470 std::map<std::string, sg_size_t>* FileSystemStorageExt::parse_content(const std::string& filename)
472 if (filename.empty())
475 std::map<std::string, sg_size_t>* parse_content = new std::map<std::string, sg_size_t>();
477 std::ifstream* fs = surf_ifsopen(filename);
480 std::vector<std::string> tokens;
482 std::getline(*fs, line);
484 if (line.length() > 0) {
485 boost::split(tokens, line, boost::is_any_of(" \t"), boost::token_compress_on);
486 xbt_assert(tokens.size() == 2, "Parse error in %s: %s", filename.c_str(), line.c_str());
487 sg_size_t size = std::stoull(tokens.at(1));
490 parse_content->insert({tokens.front(), size});
492 } while (not fs->eof());
494 return parse_content;
499 using simgrid::s4u::FileDescriptorHostExt;
500 using simgrid::s4u::FileSystemDiskExt;
501 using simgrid::s4u::FileSystemStorageExt;
503 static void on_disk_creation(simgrid::s4u::Disk& d)
505 d.extension_set(new FileSystemDiskExt(&d));
507 static void on_storage_creation(simgrid::s4u::Storage& st)
509 st.extension_set(new FileSystemStorageExt(&st));
512 static void on_host_creation(simgrid::s4u::Host& host)
514 host.extension_set<FileDescriptorHostExt>(new FileDescriptorHostExt());
517 /* **************************** Public interface *************************** */
518 void sg_storage_file_system_init()
520 sg_storage_max_file_descriptors = 1024;
521 simgrid::config::bind_flag(sg_storage_max_file_descriptors, "storage/max_file_descriptors",
522 "Maximum number of concurrently opened files per host. Default is 1024");
524 if (not FileSystemStorageExt::EXTENSION_ID.valid()) {
525 FileSystemStorageExt::EXTENSION_ID = simgrid::s4u::Storage::extension_create<FileSystemStorageExt>();
526 simgrid::s4u::Storage::on_creation.connect(&on_storage_creation);
529 if (not FileSystemDiskExt::EXTENSION_ID.valid()) {
530 FileSystemDiskExt::EXTENSION_ID = simgrid::s4u::Disk::extension_create<FileSystemDiskExt>();
531 simgrid::s4u::Disk::on_creation.connect(&on_disk_creation);
534 if (not FileDescriptorHostExt::EXTENSION_ID.valid()) {
535 FileDescriptorHostExt::EXTENSION_ID = simgrid::s4u::Host::extension_create<FileDescriptorHostExt>();
536 simgrid::s4u::Host::on_creation.connect(&on_host_creation);
540 sg_file_t sg_file_open(const char* fullpath, void* data)
542 return new simgrid::s4u::File(fullpath, data);
545 sg_size_t sg_file_read(sg_file_t fd, sg_size_t size)
547 return fd->read(size);
550 sg_size_t sg_file_write(sg_file_t fd, sg_size_t size)
552 return fd->write(size);
555 void sg_file_close(sg_file_t fd)
560 const char* sg_file_get_name(sg_file_t fd)
562 xbt_assert((fd != nullptr), "Invalid file descriptor");
563 return fd->get_path();
566 sg_size_t sg_file_get_size(sg_file_t fd)
571 void sg_file_dump(sg_file_t fd)
576 void* sg_file_get_data(sg_file_t fd)
578 return fd->get_userdata();
581 void sg_file_set_data(sg_file_t fd, void* data)
583 fd->set_userdata(data);
587 * @brief Set the file position indicator in the sg_file_t by adding offset bytes
588 * to the position specified by origin (either SEEK_SET, SEEK_CUR, or SEEK_END).
590 * @param fd : file object that identifies the stream
591 * @param offset : number of bytes to offset from origin
592 * @param origin : Position used as reference for the offset. It is specified by one of the following constants defined
593 * in \<stdio.h\> exclusively to be used as arguments for this function (SEEK_SET = beginning of file,
594 * SEEK_CUR = current position of the file pointer, SEEK_END = end of file)
596 void sg_file_seek(sg_file_t fd, sg_offset_t offset, int origin)
598 fd->seek(offset, origin);
601 sg_size_t sg_file_tell(sg_file_t fd)
606 void sg_file_move(sg_file_t fd, const char* fullpath)
611 void sg_file_unlink(sg_file_t fd)
618 * @brief Copy a file to another location on a remote host.
619 * @param file : the file to move
620 * @param host : the remote host where the file has to be copied
621 * @param fullpath : the complete path destination on the remote host
622 * @return If successful, the function returns 0. Otherwise, it returns -1.
624 int sg_file_rcopy(sg_file_t file, sg_host_t host, const char* fullpath)
626 return file->remote_copy(host, fullpath);
630 * @brief Move a file to another location on a remote host.
631 * @param file : the file to move
632 * @param host : the remote host where the file has to be moved
633 * @param fullpath : the complete path destination on the remote host
634 * @return If successful, the function returns 0. Otherwise, it returns -1.
636 int sg_file_rmove(sg_file_t file, sg_host_t host, const char* fullpath)
638 return file->remote_move(host, fullpath);
641 sg_size_t sg_disk_get_size_free(sg_disk_t d)
643 return d->extension<FileSystemDiskExt>()->get_size() - d->extension<FileSystemDiskExt>()->get_used_size();
646 sg_size_t sg_disk_get_size_used(sg_disk_t d)
648 return d->extension<FileSystemDiskExt>()->get_used_size();
651 sg_size_t sg_disk_get_size(sg_disk_t d)
653 return d->extension<FileSystemDiskExt>()->get_size();
656 sg_size_t sg_storage_get_size_free(sg_storage_t st)
658 return st->extension<FileSystemStorageExt>()->get_size() - st->extension<FileSystemStorageExt>()->get_used_size();
661 sg_size_t sg_storage_get_size_used(sg_storage_t st)
663 return st->extension<FileSystemStorageExt>()->get_used_size();
666 sg_size_t sg_storage_get_size(sg_storage_t st)
668 return st->extension<FileSystemStorageExt>()->get_size();
671 xbt_dict_t sg_storage_get_content(sg_storage_t storage)
673 std::map<std::string, sg_size_t>* content = storage->extension<simgrid::s4u::FileSystemStorageExt>()->get_content();
674 // Note: ::operator delete is ok here (no destructor called) since the dict elements are of POD type sg_size_t.
675 xbt_dict_t content_as_dict = xbt_dict_new_homogeneous(::operator delete);
677 for (auto const& entry : *content) {
678 sg_size_t* psize = new sg_size_t;
679 *psize = entry.second;
680 xbt_dict_set(content_as_dict, entry.first.c_str(), psize, nullptr);
682 return content_as_dict;
685 xbt_dict_t sg_host_get_storage_content(sg_host_t host)
687 xbt_assert((host != nullptr), "Invalid parameters");
688 xbt_dict_t contents = xbt_dict_new_homogeneous(nullptr);
689 for (auto const& elm : host->get_mounted_storages())
690 xbt_dict_set(contents, elm.first.c_str(), sg_storage_get_content(elm.second), nullptr);