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 path_ = fullpath_.substr(longest_prefix_length, fullpath_.length());
73 xbt_die("Can't find mount point for '%s' on '%s'", fullpath_.c_str(), host->get_cname());
79 // assign a file descriptor id to the newly opened File
80 FileDescriptorHostExt* ext = host->extension<simgrid::s4u::FileDescriptorHostExt>();
81 if (ext->file_descriptor_table == nullptr) {
82 ext->file_descriptor_table.reset(new std::vector<int>(sg_storage_max_file_descriptors));
83 std::iota(ext->file_descriptor_table->rbegin(), ext->file_descriptor_table->rend(), 0); // Fill with ..., 1, 0.
85 xbt_assert(not ext->file_descriptor_table->empty(), "Too much files are opened! Some have to be closed.");
86 desc_id = ext->file_descriptor_table->back();
87 ext->file_descriptor_table->pop_back();
89 XBT_DEBUG("\tOpen file '%s'", path_.c_str());
90 std::map<std::string, sg_size_t>* content;
92 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
95 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
97 // if file does not exist create an empty file
98 auto sz = content->find(path_);
99 if (sz != content->end()) {
103 content->insert({path_, size_});
104 XBT_DEBUG("File '%s' was not found, file created.", path_.c_str());
110 Host::current()->extension<simgrid::s4u::FileDescriptorHostExt>()->file_descriptor_table->push_back(desc_id);
116 XBT_INFO("File Descriptor information:\n"
117 "\t\tFull path: '%s'\n"
119 "\t\tMount point: '%s'\n"
120 "\t\tStorage Id: '%s'\n"
121 "\t\tStorage Type: '%s'\n"
122 "\t\tFile Descriptor Id: %d",
123 get_path(), size_, mount_point_.c_str(), local_storage_->get_cname(), local_storage_->get_type(), desc_id);
125 XBT_INFO("File Descriptor information:\n"
126 "\t\tFull path: '%s'\n"
128 "\t\tMount point: '%s'\n"
129 "\t\tDisk Id: '%s'\n"
130 "\t\tHost Id: '%s'\n"
131 "\t\tFile Descriptor Id: %d",
132 get_path(), size_, mount_point_.c_str(), local_disk_->get_cname(), local_disk_->get_host()->get_cname(),
136 sg_size_t File::read(sg_size_t size)
138 if (size_ == 0) /* Nothing to read, return */
140 sg_size_t read_size = 0;
142 if (local_storage_) {
143 /* Find the host where the file is physically located and read it */
144 Host* host = local_storage_->get_host();
145 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_storage_->get_cname());
146 // if the current position is close to the end of the file, we may not be able to read the requested size
147 read_size = local_storage_->read(std::min(size, size_ - current_position_));
148 current_position_ += read_size;
150 if (host->get_name() != Host::current()->get_name() && read_size > 0) {
151 /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
152 XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", host->get_cname(), read_size);
153 host->send_to(Host::current(), read_size);
158 /* Find the host where the file is physically located and read it */
159 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_disk_->get_cname());
160 // if the current position is close to the end of the file, we may not be able to read the requested size
161 read_size = local_disk_->read(std::min(size, size_ - current_position_));
162 current_position_ += read_size;
168 /** @brief Write into a file (local or remote)
170 * @param size of the file to write
171 * @return the number of bytes successfully write or -1 if an error occurred
173 sg_size_t File::write(sg_size_t size, int write_inside)
175 if (size == 0) /* Nothing to write, return */
177 sg_size_t write_size = 0;
179 if (local_storage_) {
180 /* Find the host where the file is physically located (remote or local)*/
181 Host* host = local_storage_->get_host();
183 if (host->get_name() != Host::current()->get_name()) {
184 /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
185 XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", host->get_cname(), size);
186 Host::current()->send_to(host, size);
189 XBT_DEBUG("WRITE %s on disk '%s'. size '%llu/%llu' '%llu:%llu'", get_path(), local_storage_->get_cname(), size,
190 size_, sg_storage_get_size_used(local_storage_), sg_storage_get_size(local_storage_));
191 // If the storage is full before even starting to write
192 if (sg_storage_get_size_used(local_storage_) >= sg_storage_get_size(local_storage_))
194 if (write_inside == 0) {
195 /* Substract the part of the file that might disappear from the used sized on the storage element */
196 local_storage_->extension<FileSystemStorageExt>()->decr_used_size(size_ - current_position_);
197 write_size = local_storage_->write(size);
198 local_storage_->extension<FileSystemStorageExt>()->incr_used_size(write_size);
199 current_position_ += write_size;
200 size_ = current_position_;
202 write_size = local_storage_->write(size);
203 current_position_ += write_size;
204 if (current_position_ > size_)
205 size_ = current_position_;
207 std::map<std::string, sg_size_t>* content = local_storage_->extension<FileSystemStorageExt>()->get_content();
209 content->erase(path_);
210 content->insert({path_, size_});
214 XBT_DEBUG("WRITE %s on disk '%s'. size '%llu/%llu' '%llu:%llu'", get_path(), local_disk_->get_cname(), size, size_,
215 sg_disk_get_size_used(local_disk_), sg_disk_get_size(local_disk_));
216 // If the storage is full before even starting to write
217 if (sg_disk_get_size_used(local_disk_) >= sg_disk_get_size(local_disk_))
219 if (write_inside == 0) {
220 /* Substract the part of the file that might disappear from the used sized on the storage element */
221 local_disk_->extension<FileSystemDiskExt>()->decr_used_size(size_ - current_position_);
222 write_size = local_disk_->write(size);
223 local_disk_->extension<FileSystemDiskExt>()->incr_used_size(write_size);
224 current_position_ += write_size;
225 size_ = current_position_;
227 write_size = local_disk_->write(size);
228 current_position_ += write_size;
229 if (current_position_ > size_)
230 size_ = current_position_;
232 std::map<std::string, sg_size_t>* content = local_disk_->extension<FileSystemDiskExt>()->get_content();
234 content->erase(path_);
235 content->insert({path_, size_});
240 sg_size_t File::size()
245 void File::seek(sg_offset_t offset)
247 current_position_ = offset;
250 void File::seek(sg_offset_t offset, int origin)
254 current_position_ = offset;
257 current_position_ += offset;
260 current_position_ = size_ + offset;
267 sg_size_t File::tell()
269 return current_position_;
272 void File::move(const std::string& fullpath)
274 /* Check if the new full path is on the same mount point */
275 if (fullpath.compare(0, mount_point_.length(), mount_point_) == 0) {
276 std::map<std::string, sg_size_t>* content;
278 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
280 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
281 auto sz = content->find(path_);
282 if (sz != content->end()) { // src file exists
283 sg_size_t new_size = sz->second;
284 content->erase(path_);
285 std::string path = fullpath.substr(mount_point_.length(), fullpath.length());
286 content->insert({path.c_str(), new_size});
287 XBT_DEBUG("Move file from %s to %s, size '%llu'", path_.c_str(), fullpath.c_str(), new_size);
289 XBT_WARN("File %s doesn't exist", path_.c_str());
292 XBT_WARN("New full path %s is not on the same mount point: %s.", fullpath.c_str(), mount_point_.c_str());
298 /* Check if the file is on local storage */
299 std::map<std::string, sg_size_t>* content;
300 const char* name = nullptr;
301 if (local_storage_) {
302 content = local_storage_->extension<FileSystemStorageExt>()->get_content();
303 name = local_storage_->get_cname();
306 content = local_disk_->extension<FileSystemDiskExt>()->get_content();
307 name = local_disk_->get_cname();
310 if (content->find(path_) == content->end()) {
311 XBT_WARN("File %s is not on disk %s. Impossible to unlink", path_.c_str(), name);
314 XBT_DEBUG("UNLINK %s on disk '%s'", path_.c_str(), name);
317 local_storage_->extension<FileSystemStorageExt>()->decr_used_size(size_);
320 local_disk_->extension<FileSystemDiskExt>()->decr_used_size(size_);
322 // Remove the file from storage
323 content->erase(fullpath_);
329 int File::remote_copy(sg_host_t host, const char* fullpath)
331 /* Find the host where the file is physically located and read it */
333 if (local_storage_) {
334 src_host = local_storage_->get_host();
335 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_storage_->get_cname());
339 src_host = local_disk_->get_host();
340 XBT_DEBUG("READ %s on disk '%s'", get_path(), local_disk_->get_cname());
344 // if the current position is close to the end of the file, we may not be able to read the requested size
345 sg_size_t read_size = 0;
347 read_size = local_storage_->read(size_);
349 read_size = local_disk_->read(size_);
351 current_position_ += read_size;
353 Host* dst_host = host;
354 if (local_storage_) {
355 /* Find the host that owns the storage where the file has to be copied */
356 Storage* storage_dest = nullptr;
357 size_t longest_prefix_length = 0;
359 for (auto const& elm : host->get_mounted_storages()) {
360 std::string mount_point = std::string(fullpath).substr(0, elm.first.size());
361 if (mount_point == elm.first && elm.first.length() > longest_prefix_length) {
362 /* The current mount name is found in the full path and is bigger than the previous*/
363 longest_prefix_length = elm.first.length();
364 storage_dest = elm.second;
368 if (storage_dest != nullptr) {
369 /* Mount point found, retrieve the host the storage is attached to */
370 dst_host = storage_dest->get_host();
372 XBT_WARN("Can't find mount point for '%s' on destination host '%s'", fullpath, host->get_cname());
378 size_t longest_prefix_length = 0;
379 Disk* dst_disk = nullptr;
381 for (auto const& disk : host->get_disks()) {
382 const char* current_mount_str = disk->get_property("mount");
383 if (current_mount_str) {
384 std::string current_mount = std::string(current_mount_str);
385 std::string mount_point = std::string(fullpath).substr(0, current_mount.length());
386 if (mount_point == current_mount && current_mount.length() > longest_prefix_length) {
387 /* The current mount name is found in the full path and is bigger than the previous*/
388 longest_prefix_length = current_mount.length();
394 if (dst_disk == nullptr) {
395 XBT_WARN("Can't find mount point for '%s' on destination host '%s'", fullpath, host->get_cname());
400 XBT_DEBUG("Initiate data transfer of %llu bytes between %s and %s.", read_size, src_host->get_cname(),
401 dst_host->get_cname());
402 src_host->send_to(dst_host, read_size);
404 /* Create file on remote host, write it and close it */
405 File* fd = new File(fullpath, dst_host, nullptr);
406 if (local_storage_) {
407 sg_size_t write_size = fd->local_storage_->write(read_size);
408 fd->local_storage_->extension<FileSystemStorageExt>()->incr_used_size(write_size);
409 (*(fd->local_storage_->extension<FileSystemStorageExt>()->get_content()))[path_] = size_;
412 fd->write(read_size);
417 int File::remote_move(sg_host_t host, const char* fullpath)
419 int res = remote_copy(host, fullpath);
424 FileSystemDiskExt::FileSystemDiskExt(simgrid::s4u::Disk* ptr)
426 const char* size_str = ptr->get_property("size");
427 const char* content_str = ptr->get_property("content");
428 size_ = size_str ? surf_parse_get_size(size_str, "disk size", ptr->get_name()) : 0;
430 content_.reset(parse_content(content_str));
433 FileSystemStorageExt::FileSystemStorageExt(simgrid::s4u::Storage* ptr)
435 content_.reset(parse_content(ptr->get_impl()->content_name_));
436 size_ = ptr->get_impl()->size_;
439 std::map<std::string, sg_size_t>* FileSystemDiskExt::parse_content(const std::string& filename)
441 if (filename.empty())
444 std::map<std::string, sg_size_t>* parse_content = new std::map<std::string, sg_size_t>();
446 std::ifstream* fs = surf_ifsopen(filename);
449 std::vector<std::string> tokens;
451 std::getline(*fs, line);
453 if (line.length() > 0) {
454 boost::split(tokens, line, boost::is_any_of(" \t"), boost::token_compress_on);
455 xbt_assert(tokens.size() == 2, "Parse error in %s: %s", filename.c_str(), line.c_str());
456 sg_size_t size = std::stoull(tokens.at(1));
459 parse_content->insert({tokens.front(), size});
461 } while (not fs->eof());
463 return parse_content;
466 std::map<std::string, sg_size_t>* FileSystemStorageExt::parse_content(const std::string& filename)
468 if (filename.empty())
471 std::map<std::string, sg_size_t>* parse_content = new std::map<std::string, sg_size_t>();
473 std::ifstream* fs = surf_ifsopen(filename);
476 std::vector<std::string> tokens;
478 std::getline(*fs, line);
480 if (line.length() > 0) {
481 boost::split(tokens, line, boost::is_any_of(" \t"), boost::token_compress_on);
482 xbt_assert(tokens.size() == 2, "Parse error in %s: %s", filename.c_str(), line.c_str());
483 sg_size_t size = std::stoull(tokens.at(1));
486 parse_content->insert({tokens.front(), size});
488 } while (not fs->eof());
490 return parse_content;
495 using simgrid::s4u::FileDescriptorHostExt;
496 using simgrid::s4u::FileSystemDiskExt;
497 using simgrid::s4u::FileSystemStorageExt;
499 static void on_disk_creation(simgrid::s4u::Disk& d)
501 d.extension_set(new FileSystemDiskExt(&d));
503 static void on_storage_creation(simgrid::s4u::Storage& st)
505 st.extension_set(new FileSystemStorageExt(&st));
508 static void on_host_creation(simgrid::s4u::Host& host)
510 host.extension_set<FileDescriptorHostExt>(new FileDescriptorHostExt());
513 /* **************************** Public interface *************************** */
514 void sg_storage_file_system_init()
516 sg_storage_max_file_descriptors = 1024;
517 simgrid::config::bind_flag(sg_storage_max_file_descriptors, "storage/max_file_descriptors",
518 "Maximum number of concurrently opened files per host. Default is 1024");
520 if (not FileSystemStorageExt::EXTENSION_ID.valid()) {
521 FileSystemStorageExt::EXTENSION_ID = simgrid::s4u::Storage::extension_create<FileSystemStorageExt>();
522 simgrid::s4u::Storage::on_creation.connect(&on_storage_creation);
525 if (not FileSystemDiskExt::EXTENSION_ID.valid()) {
526 FileSystemDiskExt::EXTENSION_ID = simgrid::s4u::Disk::extension_create<FileSystemDiskExt>();
527 simgrid::s4u::Disk::on_creation.connect(&on_disk_creation);
530 if (not FileDescriptorHostExt::EXTENSION_ID.valid()) {
531 FileDescriptorHostExt::EXTENSION_ID = simgrid::s4u::Host::extension_create<FileDescriptorHostExt>();
532 simgrid::s4u::Host::on_creation.connect(&on_host_creation);
536 sg_file_t sg_file_open(const char* fullpath, void* data)
538 return new simgrid::s4u::File(fullpath, data);
541 sg_size_t sg_file_read(sg_file_t fd, sg_size_t size)
543 return fd->read(size);
546 sg_size_t sg_file_write(sg_file_t fd, sg_size_t size)
548 return fd->write(size);
551 void sg_file_close(sg_file_t fd)
556 const char* sg_file_get_name(sg_file_t fd)
558 xbt_assert((fd != nullptr), "Invalid file descriptor");
559 return fd->get_path();
562 sg_size_t sg_file_get_size(sg_file_t fd)
567 void sg_file_dump(sg_file_t fd)
572 void* sg_file_get_data(sg_file_t fd)
574 return fd->get_userdata();
577 void sg_file_set_data(sg_file_t fd, void* data)
579 fd->set_userdata(data);
583 * @brief Set the file position indicator in the sg_file_t by adding offset bytes
584 * to the position specified by origin (either SEEK_SET, SEEK_CUR, or SEEK_END).
586 * @param fd : file object that identifies the stream
587 * @param offset : number of bytes to offset from origin
588 * @param origin : Position used as reference for the offset. It is specified by one of the following constants defined
589 * in \<stdio.h\> exclusively to be used as arguments for this function (SEEK_SET = beginning of file,
590 * SEEK_CUR = current position of the file pointer, SEEK_END = end of file)
592 void sg_file_seek(sg_file_t fd, sg_offset_t offset, int origin)
594 fd->seek(offset, origin);
597 sg_size_t sg_file_tell(sg_file_t fd)
602 void sg_file_move(sg_file_t fd, const char* fullpath)
607 void sg_file_unlink(sg_file_t fd)
614 * @brief Copy a file to another location on a remote host.
615 * @param file : the file to move
616 * @param host : the remote host where the file has to be copied
617 * @param fullpath : the complete path destination on the remote host
618 * @return If successful, the function returns 0. Otherwise, it returns -1.
620 int sg_file_rcopy(sg_file_t file, sg_host_t host, const char* fullpath)
622 return file->remote_copy(host, fullpath);
626 * @brief Move a file to another location on a remote host.
627 * @param file : the file to move
628 * @param host : the remote host where the file has to be moved
629 * @param fullpath : the complete path destination on the remote host
630 * @return If successful, the function returns 0. Otherwise, it returns -1.
632 int sg_file_rmove(sg_file_t file, sg_host_t host, const char* fullpath)
634 return file->remote_move(host, fullpath);
637 sg_size_t sg_disk_get_size_free(sg_disk_t d)
639 return d->extension<FileSystemDiskExt>()->get_size() - d->extension<FileSystemDiskExt>()->get_used_size();
642 sg_size_t sg_disk_get_size_used(sg_disk_t d)
644 return d->extension<FileSystemDiskExt>()->get_used_size();
647 sg_size_t sg_disk_get_size(sg_disk_t d)
649 return d->extension<FileSystemDiskExt>()->get_size();
652 sg_size_t sg_storage_get_size_free(sg_storage_t st)
654 return st->extension<FileSystemStorageExt>()->get_size() - st->extension<FileSystemStorageExt>()->get_used_size();
657 sg_size_t sg_storage_get_size_used(sg_storage_t st)
659 return st->extension<FileSystemStorageExt>()->get_used_size();
662 sg_size_t sg_storage_get_size(sg_storage_t st)
664 return st->extension<FileSystemStorageExt>()->get_size();
667 xbt_dict_t sg_storage_get_content(sg_storage_t storage)
669 std::map<std::string, sg_size_t>* content = storage->extension<simgrid::s4u::FileSystemStorageExt>()->get_content();
670 // Note: ::operator delete is ok here (no destructor called) since the dict elements are of POD type sg_size_t.
671 xbt_dict_t content_as_dict = xbt_dict_new_homogeneous(::operator delete);
673 for (auto const& entry : *content) {
674 sg_size_t* psize = new sg_size_t;
675 *psize = entry.second;
676 xbt_dict_set(content_as_dict, entry.first.c_str(), psize, nullptr);
678 return content_as_dict;
681 xbt_dict_t sg_host_get_storage_content(sg_host_t host)
683 xbt_assert((host != nullptr), "Invalid parameters");
684 xbt_dict_t contents = xbt_dict_new_homogeneous(nullptr);
685 for (auto const& elm : host->get_mounted_storages())
686 xbt_dict_set(contents, elm.first.c_str(), sg_storage_get_content(elm.second), nullptr);