LocalFileSystem.cc

Go to the documentation of this file.

#define _GNU_SOURCE 1
#define _FILE_OFFSET_BITS 64
#include "Utilities/StorageFactory/interface/LocalFileSystem.h"
#include "FWCore/MessageLogger/interface/MessageLogger.h"
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include <sys/param.h>
#if BSD
# include <sys/statvfs.h>
# include <sys/ucred.h>
# include <sys/mount.h>
#else
# include <mntent.h>
# include <sys/vfs.h>
#endif
#include <sys/stat.h>
#include <unistd.h>
#include <iostream>
#include <atomic>

#pragma GCC diagnostic ignored "-Wformat" // shut warning on '%z'

struct LocalFileSystem::FSInfo
{
  char      *fsname;    //< file system name
  char      *type;      //< file system type
  char      *dir;       //< mount point directory
  char      *origin = nullptr; //< mount origin
  dev_t     dev;        //< device id
  long      fstype;     //< file system id
  double    freespc;    //< free space in megabytes
  unsigned  local : 1;  //< flag for local device
  unsigned  bind : 1;   //< flag for bind mounts
  std::atomic<bool> checked ;   //< flag for valid dev, fstype
};

int
LocalFileSystem::readFSTypes(void)
{
  int ret = 0;

#if __linux__
  static const char procfs[] = "/proc/filesystems";
  FILE *fs = fopen(procfs, "r");
  if (! fs)
  {
    int nerr = errno;
    edm::LogWarning("LocalFileSystem::readFSTypes()")
      << "Cannot read '" << procfs << "': "
      << strerror(nerr) << " (error " << nerr << ")";
    return -1;
  }

  ssize_t nread;
  int line = 0;
  while (! feof(fs))
  {
    char *type = 0;
    char *fstype = 0;
    size_t len = 0;
    ++line;

    if ((nread = getdelim(&type, &len, '\t', fs)) == -1 && ! feof(fs))
    {
      fprintf(stderr, "%s:%d: %s (%zd; 1)\n",
          procfs, line, strerror(errno), nread);
      free(type);
      ret = -1;
      break;
    }

    if ((nread = getdelim(&fstype, &len, '\n', fs)) == -1 && ! feof(fs))
    {
      fprintf(stderr, "%s:%d: %s (%zd; 2)\n",
          procfs, line, strerror(errno), nread);
      free(type);
      free(fstype);
      ret = -1;
      break;
    }

    if (feof (fs))
    {
      free(type);
      free(fstype);
      break;
    }
    
    if (! strcmp(type, "nodev\t")
    || ! strcmp(fstype, "lustre\n")
    || ! strncmp(fstype, "fuse", 4))
    {
      free(type);
      free(fstype);
      continue;
    }

    assert(nread >= 1);
    fstype[nread-1] = 0;
    fstypes_.push_back(fstype);
    free(fstype);
    free(type);
  }

  fclose(fs);
#endif // __linux__

  return ret;
}

LocalFileSystem::FSInfo *
LocalFileSystem::initFSInfo(void *arg)
{
#if BSD
  struct statfs *m = static_cast<struct statfs *>(arg);
  size_t infolen = sizeof(struct FSInfo);
  size_t fslen = strlen(m->f_mntfromname) + 1;
  size_t dirlen = strlen(m->f_mntonname) + 1;
  size_t typelen = strlen(m->f_fstypename) + 1;
  size_t totlen = infolen + fslen + dirlen + typelen;
  FSInfo *i = (FSInfo *) malloc(totlen);
  char *p = (char *) i;
  i->fsname = strncpy(p += infolen, m->f_mntfromname, fslen);
  i->type = strncpy(p += fslen, m->f_fstypename, typelen);
  i->dir = strncpy(p += typelen, m->f_mntonname, dirlen);
  i->dev = m->f_fsid.val[0];
  i->fstype = m->f_type;
  i->freespc = 0;
  i->bind = 0;
  i->origin = nullptr;
  if (m->f_bsize > 0)
  {
    i->freespc = m->f_bavail;
    i->freespc *= m->f_bsize;
    i->freespc /= 1024. * 1024. * 1024.;
  } 
  /* FIXME: This incorrectly says that mounted disk images are local,
     even if it was mounted from a network server. The alternative is
     to walk up the device tree using either a) process IORegistry to
     get the device tree, which lists devices for disk images, and from
     there translate volume uuid to a mount point; b) parse output from
     'hdiutil info -plist' to determine image-path / dev-entry map. */
  i->local = ((m->f_flags & MNT_LOCAL) ? 1 : 0);
  i->checked = 1;
  return i;

#else // ! BSD
  mntent *m = static_cast<mntent *>(arg);
  size_t infolen = sizeof(struct FSInfo);
  size_t fslen = strlen(m->mnt_fsname) + 1;
  size_t dirlen = strlen(m->mnt_dir) + 1;
  size_t typelen = strlen(m->mnt_type) + 1;
  size_t originlen = strlen(m->mnt_fsname) + 1;
  size_t totlen = infolen + fslen + dirlen + typelen + originlen;
  FSInfo *i = (FSInfo *) malloc(totlen);
  char *p = (char *) i;
  i->fsname = strncpy(p += infolen, m->mnt_fsname, fslen);
  i->type = strncpy(p += fslen, m->mnt_type, typelen);
  i->dir = strncpy(p += typelen, m->mnt_dir, dirlen);
  i->origin = strncpy(p += dirlen, m->mnt_fsname, originlen);
  i->dev = -1;
  i->fstype = -1;
  i->freespc = 0;
  i->local = 0;
  i->checked = 0;
  i->bind = strstr(m->mnt_opts, "bind") != nullptr;

  for (size_t j = 0; j < fstypes_.size() && ! i->local; ++j)
    if (fstypes_[j] == i->type)
      i->local = 1;
#endif // BSD

  return i;
}

int
LocalFileSystem::initFSList(void)
{
#if BSD
  int rc;
  struct statfs *mtab = 0;
  if ((rc = getmntinfo(&mtab, MNT_NOWAIT)) < 0)
  {
    int nerr = errno;
    edm::LogWarning("LocalFileSystem::initFSList()")
      << "getmntinfo() failed: " << strerror(nerr)
      << " (error " << nerr << ")";
    return -1;
  }

  fs_.reserve(rc);
  for (int ix = 0; ix < rc; ++ix)
    fs_.push_back(initFSInfo(&mtab[ix]));

  free(mtab);
#else
  const char * const _PATH_MOUNTED_LINUX = "/proc/self/mounts";
  struct mntent *m;
  FILE *mtab = setmntent(_PATH_MOUNTED_LINUX, "r");
  if (! mtab)
  {
    int nerr = errno;
    edm::LogWarning("LocalFileSystem::initFSList()")
      << "Cannot read '" << _PATH_MOUNTED_LINUX << "': "
      << strerror(nerr) << " (error " << nerr << ")";
    return -1;
  }

  fs_.reserve(20);
  while ((m = getmntent(mtab)))
    fs_.push_back(initFSInfo(m));

  endmntent(mtab);
#endif

  return 0;
}

int
LocalFileSystem::statFSInfo(FSInfo *i) const
{
  int ret = 0;
  struct stat s;
  struct statfs sfs;

  if (! i->checked)
  {
    if (lstat(i->dir, &s) < 0)
    {
      i->checked = 1;

      int nerr = errno;
      if (nerr != ENOENT && nerr != EACCES)
        edm::LogWarning("LocalFileSystem::statFSInfo()")
      << "Cannot lstat('" << i->dir << "'): "
      << strerror(nerr) << " (error " << nerr << ")";
      return -1;
    }

    if (statfs(i->dir, &sfs) < 0)
    {
      i->checked = 1;
      int nerr = errno;
      edm::LogWarning("LocalFileSystem::statFSInfo()")
    << "Cannot statfs('" << i->dir << "'): "
    << strerror(nerr) << " (error " << nerr << ")";
      return -1;
    }

    i->dev = s.st_dev;
    i->fstype = sfs.f_type;
    if (sfs.f_bsize > 0)
    {
      i->freespc = sfs.f_bavail;
      i->freespc *= sfs.f_bsize;
      i->freespc /= 1024. * 1024. * 1024.;
    }
    i->checked = 1;
  }
  else if (i->fstype == -1)
  {
    errno = ENOENT;
    ret = -1;
  }

  return ret;
}

LocalFileSystem::FSInfo *
LocalFileSystem::findMount(const char *path, struct statfs *sfs, struct stat *s, std::vector<std::string> &prev_paths) const
{
  for (const auto & old_path : prev_paths)
  {
    if (!strcmp(old_path.c_str(), path))
    {
      edm::LogWarning("LocalFileSystem::findMount()")
        << "Found a loop in bind mounts; stopping evaluation.";
      return nullptr;
    }
  }

  FSInfo *best = 0;
  size_t bestlen = 0;
  size_t len = strlen(path);
  for (size_t i = 0; i < fs_.size(); ++i)
  {
    // First match simply against the file system path.  We don't
    // touch the file system until the path prefix matches.
    // When we have a path prefix match, check the file system if
    //   we don't have a best match candidate yet, OR
    //   this match is longer (more specific) than the previous best OR
    //   this match is the same length and the previous best isn't local
    // The final condition handles cases such as '/' that can appear twice
    // in the file system list, once as 'rootfs' and once as local fs.
    size_t fslen = strlen(fs_[i]->dir);
    if (! strncmp(fs_[i]->dir, path, fslen)
    && ((fslen == 1 && fs_[i]->dir[0] == '/')
        || len == fslen || path[fslen] == '/')
    && (! best || fslen > bestlen || (fslen == bestlen && !best->local)))
    {
      // Get the file system device and file system ids.
      if (statFSInfo(fs_[i]) < 0)
    return 0;

      // Check the path is on the same device / file system.  If this
      // fails, we found a better prefix match on path, but it's the
      // wrong device, so reset our idea of the best match: it can't
      // be the outer mount any more.  Not sure this is the right
      // thing to do with e.g. loop-back or union mounts.
      if (fs_[i]->dev != s->st_dev || fs_[i]->fstype != sfs->f_type)
      {
    best = 0;
    continue;
      }

      // OK this is better than anything else we found so far.
      best = fs_[i];
      bestlen = fslen;
    }
  }
  // In the case of a bind mount, try looking again at the source directory.
  if (best && best->bind && best->origin)
  {
    struct stat s2;
    struct statfs sfs2;
    char *fullpath = realpath(best->origin, 0);

    if (! fullpath)
      fullpath = strdup(best->origin);

    if (lstat(fullpath, &s2) < 0)
    {
      int nerr = errno;
      edm::LogWarning("LocalFileSystem::findMount()")
        << "Cannot lstat('" << fullpath << "' alias '"
        << path << "'): " << strerror(nerr) << " (error "
        << nerr << ")";
      free(fullpath);
      return best;
    }

    if (statfs(fullpath, &sfs2) < 0)
    {
      int nerr = errno;
      edm::LogWarning("LocalFileSystem::findMount()")
        << "Cannot statfs('" << fullpath << "' alias '"
        << path << "'): " << strerror(nerr) << " (error "
        << nerr << ")";
      free(fullpath);
      return best;
    }

    prev_paths.push_back(path);
    LocalFileSystem::FSInfo *new_best = findMount(fullpath, &sfs2, &s2, prev_paths);
    return new_best ? new_best : best;
  }

  return best;
}

bool
LocalFileSystem::isLocalPath(const std::string &path) const
{
  struct stat s;
  struct statfs sfs;
  char *fullpath = realpath(path.c_str(), 0);

  if (! fullpath)
    fullpath = strdup(path.c_str());

  if (lstat(fullpath, &s) < 0)
  {
    int nerr = errno;
    edm::LogWarning("LocalFileSystem::isLocalPath()")
      << "Cannot lstat('" << fullpath << "' alias '"
      << path << "'): " << strerror(nerr) << " (error "
      << nerr << ")";
    free(fullpath);
    return false;
  }
    
  if (statfs(fullpath, &sfs) < 0)
  {
    int nerr = errno;
    edm::LogWarning("LocalFileSystem::isLocalPath()")
      << "Cannot statfs('" << fullpath << "' alias '"
      << path << "'): " << strerror(nerr) << " (error "
      << nerr << ")";
    free(fullpath);
    return false;
  }

  std::vector<std::string> prev_paths;
  FSInfo *m = findMount(fullpath, &sfs, &s, prev_paths);
  free(fullpath);

  return m ? m->local : false;
}

std::pair<std::string, std::string>
LocalFileSystem::findCachePath(const std::vector<std::string> &paths,
                   double minFreeSpace) const
{
  struct stat s;
  struct statfs sfs;
  std::ostringstream warningst;
  warningst << "Cannot use lazy-download because:\n";

  for (size_t i = 0, e = paths.size(); i < e; ++i)
  {
    char *fullpath;
    const char *inpath = paths[i].c_str();
    const char *path = inpath;

    if (*path == '$')
    {
      char *p = getenv(path+1);
      if (p && *p)
    path = p;
      else if (! strcmp(path, "$TMPDIR"))
    path = "/tmp";
    }

    if (! (fullpath = realpath(path, 0)))
      fullpath = strdup(path);

#if 0
    std::cerr /* edm::LogInfo("LocalFileSystem") */
      << "Checking if '" << fullpath << "', from '"
      << inpath << "' is valid cache path with "
      << minFreeSpace << " free space" << std::endl;
#endif

    if (lstat(fullpath, &s) < 0)
    {
      int nerr = errno;
      if (nerr != ENOENT && nerr != EACCES)
        edm::LogWarning("LocalFileSystem::findCachePath()")
      << "Cannot lstat('" << fullpath << "', from '"
      << inpath << "'): " << strerror(nerr) << " (error "
      << nerr << ")";
      free(fullpath);
      continue;
    }
    
    if (statfs(fullpath, &sfs) < 0)
    {
      int nerr = errno;
      edm::LogWarning("LocalFileSystem::findCachePath()")
    << "Cannot statfs('" << fullpath << "', from '"
    << inpath << "'): " << strerror(nerr) << " (error "
    << nerr << ")";
      free(fullpath);
      continue;
    }

    std::vector<std::string> prev_paths;
    FSInfo *m = findMount(fullpath, &sfs, &s, prev_paths);
#if 0
    std::cerr /* edm::LogInfo("LocalFileSystem") */
      << "Candidate '" << fullpath << "': "
      << "found=" << (m ? 1 : 0)
      << " local=" << (m && m->local)
      << " free=" << (m ? m->freespc : 0)
      << " access=" << access(fullpath, W_OK)
      << std::endl;
#endif

    if (m
    && m->local
    && m->freespc >= minFreeSpace
    && access(fullpath, W_OK) == 0)
    {
      std::string result(fullpath);
      free(fullpath);
      return std::make_pair(result,std::string());
    }
    else if (m)
    {
      if (!m->local)
      {
        warningst << "- The mount " << fullpath << " is not local.\n";
      }
      else if (m->freespc < minFreeSpace)
      {
        warningst << " - The mount at " << fullpath << " has only " << m->freespc << " GB free; a minumum of " << minFreeSpace << " GB is required.\n";
      }
      else if (access(fullpath, W_OK))
      {
        warningst << " - The process has no permission to write into " << fullpath << "\n";
      }
    }

    free(fullpath);
  }

  std::string warning_str = warningst.str();
  if (warning_str.size())
  {
    warning_str = warning_str.substr(0, warning_str.size()-2);
  }
  
  return std::make_pair(std::string(), std::move(warning_str));
}

LocalFileSystem::LocalFileSystem(void)
{
  if (readFSTypes() < 0)
    return;

  if (initFSList() < 0)
    return;
}

LocalFileSystem::~LocalFileSystem(void)
{
  for (size_t i = 0, e = fs_.size(); i < e; ++i)
    free(fs_[i]);
}