Fix end-of-job slowdown for random IO with a random map
[fio.git] / filesetup.c
index 57928133369cdd520a1b89052cbc9362b773ca78..0454cc4007936611ac53bad147b7f06b2afd2faa 100644 (file)
@@ -3,6 +3,7 @@
 #include <string.h>
 #include <assert.h>
 #include <dirent.h>
+#include <libgen.h>
 #include <sys/stat.h>
 #include <sys/mman.h>
 #include <sys/types.h>
@@ -10,6 +11,7 @@
 #include "fio.h"
 #include "smalloc.h"
 #include "filehash.h"
+#include "os/os.h"
 
 static int root_warn;
 
@@ -64,6 +66,19 @@ static int extend_file(struct thread_data *td, struct fio_file *f)
                return 1;
        }
 
+#ifdef FIO_HAVE_FALLOCATE
+       if (td->o.fallocate && !td->o.fill_device) {
+               dprint(FD_FILE, "fallocate file %s size %llu\n", f->file_name,
+                                                       f->real_file_size);
+
+               r = posix_fallocate(f->fd, 0, f->real_file_size);
+               if (r > 0) {
+                       log_err("fio: posix_fallocate fails: %s\n",
+                                       strerror(r));
+               }
+       }
+#endif
+
        if (!new_layout)
                goto done;
 
@@ -78,16 +93,6 @@ static int extend_file(struct thread_data *td, struct fio_file *f)
                        td_verror(td, errno, "ftruncate");
                        goto err;
                }
-
-#ifdef FIO_HAVE_FALLOCATE
-               dprint(FD_FILE, "fallocate file %s, size %llu\n", f->file_name,
-                                                       f->real_file_size);
-               r = posix_fallocate(f->fd, 0, f->real_file_size);
-               if (r < 0) {
-                       log_err("fio: posix_fallocate fails: %s\n",
-                                       strerror(-r));
-               }
-#endif
        }
 
        b = malloc(td->o.max_bs[DDIR_WRITE]);
@@ -156,6 +161,9 @@ static int pre_read_file(struct thread_data *td, struct fio_file *f)
        unsigned int bs;
        char *b;
 
+       if (td->io_ops->flags & FIO_PIPEIO)
+               return 0;
+
        if (!fio_file_open(f)) {
                if (td->io_ops->open_file(td, f)) {
                        log_err("fio: cannot pre-read, failed to open file\n");
@@ -225,7 +233,7 @@ static int file_size(struct thread_data *td, struct fio_file *f)
 
 static int bdev_size(struct thread_data *td, struct fio_file *f)
 {
-       unsigned long long bytes;
+       unsigned long long bytes = 0;
        int r;
 
        if (td->io_ops->open_file(td, f)) {
@@ -234,9 +242,10 @@ static int bdev_size(struct thread_data *td, struct fio_file *f)
                return 1;
        }
 
-       r = blockdev_size(f->fd, &bytes);
+       r = blockdev_size(f, &bytes);
        if (r) {
                td_verror(td, r, "blockdev_size");
+               printf("fd is %d\n", f->fd);
                goto err;
        }
 
@@ -246,10 +255,46 @@ static int bdev_size(struct thread_data *td, struct fio_file *f)
        }
 
        f->real_file_size = bytes;
+       td->io_ops->close_file(td, f);
+       return 0;
+err:
+       td->io_ops->close_file(td, f);
+       return 1;
+}
+
+static int char_size(struct thread_data *td, struct fio_file *f)
+{
+#ifdef FIO_HAVE_CHARDEV_SIZE
+       unsigned long long bytes = 0;
+       int r;
+
+       if (td->io_ops->open_file(td, f)) {
+               log_err("fio: failed opening blockdev %s for size check\n",
+                       f->file_name);
+               return 1;
+       }
+
+       r = chardev_size(f, &bytes);
+       if (r) {
+               td_verror(td, r, "chardev_size");
+               goto err;
+       }
+
+       if (!bytes) {
+               log_err("%s: zero sized char device?\n", f->file_name);
+               goto err;
+       }
+
+       f->real_file_size = bytes;
+       td->io_ops->close_file(td, f);
        return 0;
 err:
        td->io_ops->close_file(td, f);
        return 1;
+#else
+       f->real_file_size = -1ULL;
+       return 0;
+#endif
 }
 
 static int get_file_size(struct thread_data *td, struct fio_file *f)
@@ -263,6 +308,8 @@ static int get_file_size(struct thread_data *td, struct fio_file *f)
                ret = file_size(td, f);
        else if (f->filetype == FIO_TYPE_BD)
                ret = bdev_size(td, f);
+       else if (f->filetype == FIO_TYPE_CHAR)
+               ret = char_size(td, f);
        else
                f->real_file_size = -1;
 
@@ -270,7 +317,7 @@ static int get_file_size(struct thread_data *td, struct fio_file *f)
                return ret;
 
        if (f->file_offset > f->real_file_size) {
-               log_err("%s: offset extends end (%Lu > %Lu)\n", td->o.name,
+               log_err("%s: offset extends end (%llu > %llu)\n", td->o.name,
                                        f->file_offset, f->real_file_size);
                return 1;
        }
@@ -299,12 +346,15 @@ static int __file_invalidate_cache(struct thread_data *td, struct fio_file *f,
        /*
         * FIXME: add blockdev flushing too
         */
-       if (f->mmap_ptr)
-               ret = madvise(f->mmap_ptr, f->mmap_sz, MADV_DONTNEED);
-       else if (f->filetype == FIO_TYPE_FILE) {
-               ret = fadvise(f->fd, off, len, POSIX_FADV_DONTNEED);
+       if (f->mmap_ptr) {
+               ret = posix_madvise(f->mmap_ptr, f->mmap_sz, POSIX_MADV_DONTNEED);
+#ifdef FIO_MADV_FREE
+               (void) posix_madvise(f->mmap_ptr, f->mmap_sz, FIO_MADV_FREE);
+#endif
+       } else if (f->filetype == FIO_TYPE_FILE) {
+               ret = posix_fadvise(f->fd, off, len, POSIX_FADV_DONTNEED);
        } else if (f->filetype == FIO_TYPE_BD) {
-               ret = blockdev_invalidate_cache(f->fd);
+               ret = blockdev_invalidate_cache(f);
                if (ret < 0 && errno == EACCES && geteuid()) {
                        if (!root_warn) {
                                log_err("fio: only root may flush block "
@@ -435,7 +485,7 @@ open_again:
                char buf[FIO_VERROR_SIZE];
                int __e = errno;
 
-               if (errno == EPERM && (flags & FIO_O_NOATIME)) {
+               if (__e == EPERM && (flags & FIO_O_NOATIME)) {
                        flags &= ~FIO_O_NOATIME;
                        goto open_again;
                }
@@ -493,6 +543,81 @@ static int get_file_sizes(struct thread_data *td)
        return err;
 }
 
+struct fio_mount {
+       struct flist_head list;
+       const char *base;
+       char __base[256];
+       unsigned int key;
+};
+
+/*
+ * Get free number of bytes for each file on each unique mount.
+ */
+static unsigned long long get_fs_free_counts(struct thread_data *td)
+{
+       struct flist_head *n, *tmp;
+       unsigned long long ret = 0;
+       struct fio_mount *fm;
+       FLIST_HEAD(list);
+       struct fio_file *f;
+       unsigned int i;
+
+       for_each_file(td, f, i) {
+               struct stat sb;
+               char buf[256];
+
+               if (f->filetype == FIO_TYPE_BD || f->filetype == FIO_TYPE_CHAR) {
+                       if (f->real_file_size != -1ULL)
+                               ret += f->real_file_size;
+                       continue;
+               } else if (f->filetype != FIO_TYPE_FILE)
+                       continue;
+
+               strcpy(buf, f->file_name);
+
+               if (stat(buf, &sb) < 0) {
+                       if (errno != ENOENT)
+                               break;
+                       strcpy(buf, ".");
+                       if (stat(buf, &sb) < 0)
+                               break;
+               }
+
+               fm = NULL;
+               flist_for_each(n, &list) {
+                       fm = flist_entry(n, struct fio_mount, list);
+                       if (fm->key == sb.st_dev)
+                               break;
+
+                       fm = NULL;
+               }
+
+               if (fm)
+                       continue;
+
+               fm = malloc(sizeof(*fm));
+               strcpy(fm->__base, buf);
+               fm->base = basename(fm->__base);
+               fm->key = sb.st_dev;
+               flist_add(&fm->list, &list);
+       }
+
+       flist_for_each_safe(n, tmp, &list) {
+               unsigned long long sz;
+
+               fm = flist_entry(n, struct fio_mount, list);
+               flist_del(&fm->list);
+
+               sz = get_fs_size(fm->base);
+               if (sz && sz != -1ULL)
+                       ret += sz;
+
+               free(fm);
+       }
+
+       return ret;
+}
+
 /*
  * Open the files and setup files sizes, creating files if necessary.
  */
@@ -533,6 +658,9 @@ int setup_files(struct thread_data *td)
                        total_size += f->real_file_size;
        }
 
+       if (td->o.fill_device)
+               td->fill_device_size = get_fs_free_counts(td);
+
        /*
         * device/file sizes are zero and no size given, punt
         */
@@ -607,7 +735,7 @@ int setup_files(struct thread_data *td)
                temp_stall_ts = 1;
                if (!terse_output)
                        log_info("%s: Laying out IO file(s) (%u file(s) /"
-                                " %LuMiB)\n", td->o.name, need_extend,
+                                " %lluMB)\n", td->o.name, need_extend,
                                        extend_size >> 20);
 
                for_each_file(td, f, i) {
@@ -620,7 +748,8 @@ int setup_files(struct thread_data *td)
                        fio_file_clear_extend(f);
                        if (!td->o.fill_device) {
                                old_len = f->real_file_size;
-                               extend_len = f->io_size + f->file_offset - old_len;
+                               extend_len = f->io_size + f->file_offset -
+                                               old_len;
                        }
                        f->real_file_size = (f->io_size + f->file_offset);
                        err = extend_file(td, f);
@@ -683,7 +812,7 @@ int init_random_map(struct thread_data *td)
                                (unsigned long long) td->o.rw_min_bs;
                num_maps = (blocks + BLOCKS_PER_MAP - 1) /
                                (unsigned long long) BLOCKS_PER_MAP;
-               f->file_map = smalloc(num_maps * sizeof(int));
+               f->file_map = smalloc(num_maps * sizeof(unsigned long));
                if (f->file_map) {
                        f->num_maps = num_maps;
                        continue;
@@ -709,8 +838,10 @@ void close_files(struct thread_data *td)
        struct fio_file *f;
        unsigned int i;
 
-       for_each_file(td, f, i)
-               td_io_close_file(td, f);
+       for_each_file(td, f, i) {
+               if (fio_file_open(f))
+                       td_io_close_file(td, f);
+       }
 }
 
 void close_and_free_files(struct thread_data *td)
@@ -726,16 +857,15 @@ void close_and_free_files(struct thread_data *td)
                        unlink(f->file_name);
                }
 
-               td_io_close_file(td, f);
+               if (fio_file_open(f))
+                       td_io_close_file(td, f);
+
                remove_file_hash(f);
 
                sfree(f->file_name);
                f->file_name = NULL;
-
-               if (f->file_map) {
-                       sfree(f->file_map);
-                       f->file_map = NULL;
-               }
+               sfree(f->file_map);
+               f->file_map = NULL;
                sfree(f);
        }
 
@@ -755,8 +885,10 @@ static void get_file_type(struct fio_file *f)
        else
                f->filetype = FIO_TYPE_FILE;
 
-       if (!lstat(f->file_name, &sb)) {
-               if (S_ISBLK(sb.st_mode))
+       if (!stat(f->file_name, &sb)) {
+               /* \\.\ is the device namespace in Windows, where every file is
+                * a block device */
+               if (S_ISBLK(sb.st_mode) || strncmp(f->file_name, "\\\\.\\", 4) == 0)
                        f->filetype = FIO_TYPE_BD;
                else if (S_ISCHR(sb.st_mode))
                        f->filetype = FIO_TYPE_CHAR;
@@ -779,8 +911,9 @@ int add_file(struct thread_data *td, const char *fname)
                log_err("fio: smalloc OOM\n");
                assert(0);
        }
-               
+
        f->fd = -1;
+       fio_file_reset(f);
 
        if (td->files_size <= td->files_index) {
                int new_size = td->o.nr_files + 1;
@@ -807,7 +940,7 @@ int add_file(struct thread_data *td, const char *fname)
                log_err("fio: smalloc OOM\n");
                assert(0);
        }
-       
+
        get_file_type(f);
 
        switch (td->o.file_lock_mode) {
@@ -834,6 +967,19 @@ int add_file(struct thread_data *td, const char *fname)
        return cur_files;
 }
 
+int add_file_exclusive(struct thread_data *td, const char *fname)
+{
+       struct fio_file *f;
+       unsigned int i;
+
+       for_each_file(td, f, i) {
+               if (!strcmp(f->file_name, fname))
+                       return i;
+       }
+
+       return add_file(td, fname);
+}
+
 void get_file(struct fio_file *f)
 {
        dprint(FD_FILE, "get file %s, ref=%d\n", f->file_name, f->references);
@@ -847,8 +993,10 @@ int put_file(struct thread_data *td, struct fio_file *f)
 
        dprint(FD_FILE, "put file %s, ref=%d\n", f->file_name, f->references);
 
-       if (!fio_file_open(f))
+       if (!fio_file_open(f)) {
+               assert(f->fd == -1);
                return 0;
+       }
 
        assert(f->references);
        if (--f->references)
@@ -865,6 +1013,7 @@ int put_file(struct thread_data *td, struct fio_file *f)
 
        td->nr_open_files--;
        fio_file_clear_open(f);
+       assert(f->fd == -1);
        return ret;
 }
 
@@ -1004,14 +1153,16 @@ void dup_files(struct thread_data *td, struct thread_data *org)
                        log_err("fio: smalloc OOM\n");
                        assert(0);
                }
-       
+               __f->fd = -1;
+               fio_file_reset(__f);
+
                if (f->file_name) {
                        __f->file_name = smalloc_strdup(f->file_name);
                        if (!__f->file_name) {
                                log_err("fio: smalloc OOM\n");
                                assert(0);
                        }
-       
+
                        __f->filetype = f->filetype;
                }