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 std::string current_mount = disk->extension<FileSystemDiskExt>()->get_mount_point();
61 mount_point_ = fullpath_.substr(0, current_mount.length());
62 if (mount_point_ == current_mount && current_mount.length() > longest_prefix_length) {
63 /* The current mount name is found in the full path and is bigger than the previous*/
64 longest_prefix_length = current_mount.length();
67 if (longest_prefix_length > 0) { /* Mount point found, split fullpath_ into mount_name and path+filename*/
68 mount_point_ = fullpath_.substr(0, longest_prefix_length);
69 if (mount_point_ == std::string("/"))
72 path_ = fullpath_.substr(longest_prefix_length, fullpath_.length());
73 XBT_DEBUG("%s + %s", mount_point_.c_str(), path_.c_str());
75 xbt_die("Can't find mount point for '%s' on '%s'", fullpath_.c_str(), host->get_cname());
80 // assign a file descriptor id to the newly opened File
81 FileDescriptorHostExt* ext = host->extension<simgrid::s4u::FileDescriptorHostExt>();
82 if (ext->file_descriptor_table == nullptr) {
83 ext->file_descriptor_table.reset(new std::vector<int>(sg_storage_max_file_descriptors));
84 std::iota(ext->file_descriptor_table->rbegin(), ext->file_descriptor_table->rend(), 0); // Fill with ..., 1, 0.
86 xbt_assert(not ext->file_descriptor_table->empty(), "Too much files are opened! Some have to be closed.");
87 desc_id = ext->file_descriptor_table->back();
88 ext->file_descriptor_table->pop_back();
90 XBT_DEBUG("\tOpen file '%s'", path_.c_str());
91 std::map<std::string, sg_size_t>* content;
93 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
96 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
98 // if file does not exist create an empty file
99 auto sz = content->find(path_);
100 if (sz != content->end()) {
104 content->insert({path_, size_});
105 XBT_DEBUG("File '%s' was not found, file created.", path_.c_str());
111 Host::current()->extension<simgrid::s4u::FileDescriptorHostExt>()->file_descriptor_table->push_back(desc_id);
117 XBT_INFO("File Descriptor information:\n"
118 "\t\tFull path: '%s'\n"
120 "\t\tMount point: '%s'\n"
121 "\t\tStorage Id: '%s'\n"
122 "\t\tStorage Type: '%s'\n"
123 "\t\tFile Descriptor Id: %d",
124 get_path(), size_, mount_point_.c_str(), local_storage_->get_cname(), local_storage_->get_type(), desc_id);
126 XBT_INFO("File Descriptor information:\n"
127 "\t\tFull path: '%s'\n"
129 "\t\tMount point: '%s'\n"
130 "\t\tDisk Id: '%s'\n"
131 "\t\tHost Id: '%s'\n"
132 "\t\tFile Descriptor Id: %d",
133 get_path(), size_, mount_point_.c_str(), local_disk_->get_cname(), local_disk_->get_host()->get_cname(),
137 sg_size_t File::read(sg_size_t size)
139 if (size_ == 0) /* Nothing to read, return */
141 sg_size_t read_size = 0;
143 if (local_storage_) {
144 /* Find the host where the file is physically located and read it */
145 Host* host = local_storage_->get_host();
146 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_storage_->get_cname());
147 // if the current position is close to the end of the file, we may not be able to read the requested size
148 read_size = local_storage_->read(std::min(size, size_ - current_position_));
149 current_position_ += read_size;
151 if (host->get_name() != Host::current()->get_name() && read_size > 0) {
152 /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
153 XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", host->get_cname(), read_size);
154 host->send_to(Host::current(), read_size);
159 /* Find the host where the file is physically located and read it */
160 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_disk_->get_cname());
161 // if the current position is close to the end of the file, we may not be able to read the requested size
162 read_size = local_disk_->read(std::min(size, size_ - current_position_));
163 current_position_ += read_size;
169 /** @brief Write into a file (local or remote)
171 * @param size of the file to write
172 * @return the number of bytes successfully write or -1 if an error occurred
174 sg_size_t File::write(sg_size_t size, int write_inside)
176 if (size == 0) /* Nothing to write, return */
178 sg_size_t write_size = 0;
180 if (local_storage_) {
181 /* Find the host where the file is physically located (remote or local)*/
182 Host* host = local_storage_->get_host();
184 if (host->get_name() != Host::current()->get_name()) {
185 /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
186 XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", host->get_cname(), size);
187 Host::current()->send_to(host, size);
190 XBT_DEBUG("WRITE %s on disk '%s'. size '%llu/%llu' '%llu:%llu'", get_path(), local_storage_->get_cname(), size,
191 size_, sg_storage_get_size_used(local_storage_), sg_storage_get_size(local_storage_));
192 // If the storage is full before even starting to write
193 if (sg_storage_get_size_used(local_storage_) >= sg_storage_get_size(local_storage_))
195 if (write_inside == 0) {
196 /* Substract the part of the file that might disappear from the used sized on the storage element */
197 local_storage_->extension<FileSystemStorageExt>()->decr_used_size(size_ - current_position_);
198 write_size = local_storage_->write(size);
199 local_storage_->extension<FileSystemStorageExt>()->incr_used_size(write_size);
200 current_position_ += write_size;
201 size_ = current_position_;
203 write_size = local_storage_->write(size);
204 current_position_ += write_size;
205 if (current_position_ > size_)
206 size_ = current_position_;
208 std::map<std::string, sg_size_t>* content = local_storage_->extension<FileSystemStorageExt>()->get_content();
210 content->erase(path_);
211 content->insert({path_, size_});
215 XBT_DEBUG("WRITE %s on disk '%s'. size '%llu/%llu' '%llu:%llu'", get_path(), local_disk_->get_cname(), size, size_,
216 sg_disk_get_size_used(local_disk_), sg_disk_get_size(local_disk_));
217 // If the storage is full before even starting to write
218 if (sg_disk_get_size_used(local_disk_) >= sg_disk_get_size(local_disk_))
220 if (write_inside == 0) {
221 /* Substract the part of the file that might disappear from the used sized on the storage element */
222 local_disk_->extension<FileSystemDiskExt>()->decr_used_size(size_ - current_position_);
223 write_size = local_disk_->write(size);
224 local_disk_->extension<FileSystemDiskExt>()->incr_used_size(write_size);
225 current_position_ += write_size;
226 size_ = current_position_;
228 write_size = local_disk_->write(size);
229 current_position_ += write_size;
230 if (current_position_ > size_)
231 size_ = current_position_;
233 std::map<std::string, sg_size_t>* content = local_disk_->extension<FileSystemDiskExt>()->get_content();
235 content->erase(path_);
236 content->insert({path_, size_});
241 sg_size_t File::size()
246 void File::seek(sg_offset_t offset)
248 current_position_ = offset;
251 void File::seek(sg_offset_t offset, int origin)
255 current_position_ = offset;
258 current_position_ += offset;
261 current_position_ = size_ + offset;
268 sg_size_t File::tell()
270 return current_position_;
273 void File::move(const std::string& fullpath)
275 /* Check if the new full path is on the same mount point */
276 if (fullpath.compare(0, mount_point_.length(), mount_point_) == 0) {
277 std::map<std::string, sg_size_t>* content;
279 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
281 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
282 auto sz = content->find(path_);
283 if (sz != content->end()) { // src file exists
284 sg_size_t new_size = sz->second;
285 content->erase(path_);
286 std::string path = fullpath.substr(mount_point_.length(), fullpath.length());
287 content->insert({path.c_str(), new_size});
288 XBT_DEBUG("Move file from %s to %s, size '%llu'", path_.c_str(), fullpath.c_str(), new_size);
290 XBT_WARN("File %s doesn't exist", path_.c_str());
293 XBT_WARN("New full path %s is not on the same mount point: %s.", fullpath.c_str(), mount_point_.c_str());
299 /* Check if the file is on local storage */
300 std::map<std::string, sg_size_t>* content;
301 const char* name = nullptr;
302 if (local_storage_) {
303 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
304 name = local_storage_->get_cname();
307 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
308 name = local_disk_->get_cname();
311 if (content->find(path_) == content->end()) {
312 XBT_WARN("File %s is not on disk %s. Impossible to unlink", path_.c_str(), name);
315 XBT_DEBUG("UNLINK %s on disk '%s'", path_.c_str(), name);
318 local_storage_->extension<FileSystemStorageExt>()->decr_used_size(size_);
321 local_disk_->extension<FileSystemDiskExt>()->decr_used_size(size_);
323 // Remove the file from storage
324 content->erase(fullpath_);
330 int File::remote_copy(sg_host_t host, const char* fullpath)
332 /* Find the host where the file is physically located and read it */
334 if (local_storage_) {
335 src_host = local_storage_->get_host();
336 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_storage_->get_cname());
340 src_host = local_disk_->get_host();
341 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_disk_->get_cname());
345 // if the current position is close to the end of the file, we may not be able to read the requested size
346 sg_size_t read_size = 0;
348 read_size = local_storage_->read(size_);
350 read_size = local_disk_->read(size_);
352 current_position_ += read_size;
354 Host* dst_host = host;
355 if (local_storage_) {
356 /* Find the host that owns the storage where the file has to be copied */
357 Storage* storage_dest = nullptr;
358 size_t longest_prefix_length = 0;
360 for (auto const& elm : host->get_mounted_storages()) {
361 std::string mount_point = std::string(fullpath).substr(0, elm.first.size());
362 if (mount_point == elm.first && elm.first.length() > longest_prefix_length) {
363 /* The current mount name is found in the full path and is bigger than the previous*/
364 longest_prefix_length = elm.first.length();
365 storage_dest = elm.second;
369 if (storage_dest != nullptr) {
370 /* Mount point found, retrieve the host the storage is attached to */
371 dst_host = storage_dest->get_host();
373 XBT_WARN("Can't find mount point for '%s' on destination host '%s'", fullpath, host->get_cname());
379 size_t longest_prefix_length = 0;
380 Disk* dst_disk = nullptr;
382 for (auto const& disk : host->get_disks()) {
383 std::string current_mount = disk->extension<FileSystemDiskExt>()->get_mount_point();
384 std::string mount_point = std::string(fullpath).substr(0, current_mount.length());
385 if (mount_point == current_mount && current_mount.length() > longest_prefix_length) {
386 /* The current mount name is found in the full path and is bigger than the previous*/
387 longest_prefix_length = current_mount.length();
392 if (dst_disk == nullptr) {
393 XBT_WARN("Can't find mount point for '%s' on destination host '%s'", fullpath, host->get_cname());
398 XBT_DEBUG("Initiate data transfer of %llu bytes between %s and %s.", read_size, src_host->get_cname(),
399 dst_host->get_cname());
400 src_host->send_to(dst_host, read_size);
402 /* Create file on remote host, write it and close it */
403 File* fd = new File(fullpath, dst_host, nullptr);
404 if (local_storage_) {
405 sg_size_t write_size = fd->local_storage_->write(read_size);
406 fd->local_storage_->extension<FileSystemStorageExt>()->incr_used_size(write_size);
407 (*(fd->local_storage_->extension<FileSystemStorageExt>()->get_content()))[path_] = size_;
410 fd->write(read_size);
415 int File::remote_move(sg_host_t host, const char* fullpath)
417 int res = remote_copy(host, fullpath);
422 FileSystemDiskExt::FileSystemDiskExt(simgrid::s4u::Disk* ptr)
424 const char* size_str = ptr->get_property("size");
426 size_ = surf_parse_get_size(size_str, "disk size", ptr->get_name());
428 const char* current_mount_str = ptr->get_property("mount");
429 if (current_mount_str)
430 mount_point_ = std::string(current_mount_str);
432 mount_point_ = std::string("/");
434 const char* content_str = ptr->get_property("content");
436 content_.reset(parse_content(content_str));
439 FileSystemStorageExt::FileSystemStorageExt(simgrid::s4u::Storage* ptr)
441 content_.reset(parse_content(ptr->get_impl()->content_name_));
442 size_ = ptr->get_impl()->size_;
445 std::map<std::string, sg_size_t>* FileSystemDiskExt::parse_content(const std::string& filename)
447 if (filename.empty())
450 std::map<std::string, sg_size_t>* parse_content = new std::map<std::string, sg_size_t>();
452 std::ifstream* fs = surf_ifsopen(filename);
455 std::vector<std::string> tokens;
457 std::getline(*fs, line);
459 if (line.length() > 0) {
460 boost::split(tokens, line, boost::is_any_of(" \t"), boost::token_compress_on);
461 xbt_assert(tokens.size() == 2, "Parse error in %s: %s", filename.c_str(), line.c_str());
462 sg_size_t size = std::stoull(tokens.at(1));
465 parse_content->insert({tokens.front(), size});
467 } while (not fs->eof());
469 return parse_content;
472 std::map<std::string, sg_size_t>* FileSystemStorageExt::parse_content(const std::string& filename)
474 if (filename.empty())
477 std::map<std::string, sg_size_t>* parse_content = new std::map<std::string, sg_size_t>();
479 std::ifstream* fs = surf_ifsopen(filename);
482 std::vector<std::string> tokens;
484 std::getline(*fs, line);
486 if (line.length() > 0) {
487 boost::split(tokens, line, boost::is_any_of(" \t"), boost::token_compress_on);
488 xbt_assert(tokens.size() == 2, "Parse error in %s: %s", filename.c_str(), line.c_str());
489 sg_size_t size = std::stoull(tokens.at(1));
492 parse_content->insert({tokens.front(), size});
494 } while (not fs->eof());
496 return parse_content;
501 using simgrid::s4u::FileDescriptorHostExt;
502 using simgrid::s4u::FileSystemDiskExt;
503 using simgrid::s4u::FileSystemStorageExt;
505 static void on_disk_creation(simgrid::s4u::Disk& d)
507 d.extension_set(new FileSystemDiskExt(&d));
509 static void on_storage_creation(simgrid::s4u::Storage& st)
511 st.extension_set(new FileSystemStorageExt(&st));
514 static void on_host_creation(simgrid::s4u::Host& host)
516 host.extension_set<FileDescriptorHostExt>(new FileDescriptorHostExt());
519 /* **************************** Public interface *************************** */
520 void sg_storage_file_system_init()
522 sg_storage_max_file_descriptors = 1024;
523 simgrid::config::bind_flag(sg_storage_max_file_descriptors, "storage/max_file_descriptors",
524 "Maximum number of concurrently opened files per host. Default is 1024");
526 if (not FileSystemStorageExt::EXTENSION_ID.valid()) {
527 FileSystemStorageExt::EXTENSION_ID = simgrid::s4u::Storage::extension_create<FileSystemStorageExt>();
528 simgrid::s4u::Storage::on_creation.connect(&on_storage_creation);
531 if (not FileSystemDiskExt::EXTENSION_ID.valid()) {
532 FileSystemDiskExt::EXTENSION_ID = simgrid::s4u::Disk::extension_create<FileSystemDiskExt>();
533 simgrid::s4u::Disk::on_creation.connect(&on_disk_creation);
536 if (not FileDescriptorHostExt::EXTENSION_ID.valid()) {
537 FileDescriptorHostExt::EXTENSION_ID = simgrid::s4u::Host::extension_create<FileDescriptorHostExt>();
538 simgrid::s4u::Host::on_creation.connect(&on_host_creation);
542 sg_file_t sg_file_open(const char* fullpath, void* data)
544 return new simgrid::s4u::File(fullpath, data);
547 sg_size_t sg_file_read(sg_file_t fd, sg_size_t size)
549 return fd->read(size);
552 sg_size_t sg_file_write(sg_file_t fd, sg_size_t size)
554 return fd->write(size);
557 void sg_file_close(sg_file_t fd)
562 const char* sg_file_get_name(sg_file_t fd)
564 xbt_assert((fd != nullptr), "Invalid file descriptor");
565 return fd->get_path();
568 sg_size_t sg_file_get_size(sg_file_t fd)
573 void sg_file_dump(sg_file_t fd)
578 void* sg_file_get_data(sg_file_t fd)
580 return fd->get_userdata();
583 void sg_file_set_data(sg_file_t fd, void* data)
585 fd->set_userdata(data);
589 * @brief Set the file position indicator in the sg_file_t by adding offset bytes
590 * to the position specified by origin (either SEEK_SET, SEEK_CUR, or SEEK_END).
592 * @param fd : file object that identifies the stream
593 * @param offset : number of bytes to offset from origin
594 * @param origin : Position used as reference for the offset. It is specified by one of the following constants defined
595 * in \<stdio.h\> exclusively to be used as arguments for this function (SEEK_SET = beginning of file,
596 * SEEK_CUR = current position of the file pointer, SEEK_END = end of file)
598 void sg_file_seek(sg_file_t fd, sg_offset_t offset, int origin)
600 fd->seek(offset, origin);
603 sg_size_t sg_file_tell(sg_file_t fd)
608 void sg_file_move(sg_file_t fd, const char* fullpath)
613 void sg_file_unlink(sg_file_t fd)
620 * @brief Copy a file to another location on a remote host.
621 * @param file : the file to move
622 * @param host : the remote host where the file has to be copied
623 * @param fullpath : the complete path destination on the remote host
624 * @return If successful, the function returns 0. Otherwise, it returns -1.
626 int sg_file_rcopy(sg_file_t file, sg_host_t host, const char* fullpath)
628 return file->remote_copy(host, fullpath);
632 * @brief Move a file to another location on a remote host.
633 * @param file : the file to move
634 * @param host : the remote host where the file has to be moved
635 * @param fullpath : the complete path destination on the remote host
636 * @return If successful, the function returns 0. Otherwise, it returns -1.
638 int sg_file_rmove(sg_file_t file, sg_host_t host, const char* fullpath)
640 return file->remote_move(host, fullpath);
643 sg_size_t sg_disk_get_size_free(sg_disk_t d)
645 return d->extension<FileSystemDiskExt>()->get_size() - d->extension<FileSystemDiskExt>()->get_used_size();
648 sg_size_t sg_disk_get_size_used(sg_disk_t d)
650 return d->extension<FileSystemDiskExt>()->get_used_size();
653 sg_size_t sg_disk_get_size(sg_disk_t d)
655 return d->extension<FileSystemDiskExt>()->get_size();
658 const char* sg_disk_get_mount_point(sg_disk_t d)
660 return d->extension<FileSystemDiskExt>()->get_mount_point();
663 sg_size_t sg_storage_get_size_free(sg_storage_t st)
665 return st->extension<FileSystemStorageExt>()->get_size() - st->extension<FileSystemStorageExt>()->get_used_size();
668 sg_size_t sg_storage_get_size_used(sg_storage_t st)
670 return st->extension<FileSystemStorageExt>()->get_used_size();
673 sg_size_t sg_storage_get_size(sg_storage_t st)
675 return st->extension<FileSystemStorageExt>()->get_size();
678 xbt_dict_t sg_storage_get_content(sg_storage_t storage)
680 std::map<std::string, sg_size_t>* content = storage->extension<simgrid::s4u::FileSystemStorageExt>()->get_content();
681 // Note: ::operator delete is ok here (no destructor called) since the dict elements are of POD type sg_size_t.
682 xbt_dict_t content_as_dict = xbt_dict_new_homogeneous(::operator delete);
684 for (auto const& entry : *content) {
685 sg_size_t* psize = new sg_size_t;
686 *psize = entry.second;
687 xbt_dict_set(content_as_dict, entry.first.c_str(), psize, nullptr);
689 return content_as_dict;
692 xbt_dict_t sg_host_get_storage_content(sg_host_t host)
694 xbt_assert((host != nullptr), "Invalid parameters");
695 xbt_dict_t contents = xbt_dict_new_homogeneous(nullptr);
696 for (auto const& elm : host->get_mounted_storages())
697 xbt_dict_set(contents, elm.first.c_str(), sg_storage_get_content(elm.second), nullptr);