Fallocate on a file with size equal to the length

[CMCDragonkai]

Reposted from #326 (comment)

I'm using 0.7.2-1, and I noticed that if you run posix_fallocate on a file with the same size as the length specified, it returns with EBADF. This doesn't happen when I do it on tmpfs.

//usr/bin/env make -s "${0%.*}" && ./"${0%.*}" "$@"; s=$?; rm ./"${0%.*}"; exit $s

#include <fcntl.h>
#include <stdio.h>
#include <errno.h>
#include <string.h>

int main () {
  int fd = open("./randomfile", O_WRONLY | O_CREAT, S_IRUSR | S_IWUSR);
  if (fd == -1) {
    perror("open()");
  }
  int status = posix_fallocate(fd, 0, 100);
  if (status != 0) {
    printf("%s\n", strerror(status));
  }
  return 0;
}

Running the above on an empty or non-existent file works fine, as soon as you run it again, it fails with EBADF. This is bit strange behaviour.

[loli10K]

This also happens for files with different size:

root@linux:~# truncate -s 64m /var/tmp/issue-6860
root@linux:~# zpool create -m /issue-6860 issue-6860 /var/tmp/issue-6860
root@linux:~# cp -v posix_fallocate.c /issue-6860
‘posix_fallocate.c’ -> ‘/issue-6860/posix_fallocate.c’
root@linux:~# cd /issue-6860
root@linux:/issue-6860# cat posix_fallocate.c 
#include <fcntl.h>
#include <stdio.h>
#include <errno.h>
#include <string.h>
#include <stdlib.h>

int main (int argc, char **argv) {
  int len = atoi(argv[1]);
  int fd = open("./randomfile", O_WRONLY | O_CREAT, S_IRUSR | S_IWUSR);
  if (fd == -1) {
    perror("open()");
  }
  int status = posix_fallocate(fd, 0, len);
  if (status != 0) {
    printf("%s\n", strerror(status));
  }
  return 0;
}
root@linux:/issue-6860# gcc posix_fallocate.c -o posix_fallocate
root@linux:/issue-6860# ./posix_fallocate 100
root@linux:/issue-6860# ls -l randomfile 
-rw------- 1 root root 100 Nov 13 19:44 randomfile
root@linux:/issue-6860# ./posix_fallocate 50
Bad file descriptor
root@linux:/issue-6860# ls -l randomfile 
-rw------- 1 root root 100 Nov 13 19:44 randomfile
root@linux:/issue-6860# ./posix_fallocate 10
Bad file descriptor
root@linux:/issue-6860# ls -l randomfile 
-rw------- 1 root root 100 Nov 13 19:44 randomfile
root@linux:/issue-6860# ./posix_fallocate 1000
root@linux:/issue-6860# 

I don't think there's an issue here: fallocate(2) with mode=0 on tmpfs doesn't return EOPNOTSUPP so libc is free to use the "native" syscall.

On ZFS fallocate(2) with mode=0 is not supported, so posix_fallocate(3) tries to "emulate" it: the error EBADF is actually from pread(2), because we used O_WRONLY to open the file:

root@linux:/issue-6860# strace -f ./posix_fallocate 100
execve("./posix_fallocate", ["./posix_fallocate", "100"], [/* 16 vars */]) = 0
brk(0)                                  = 0x240f000
access("/etc/ld.so.nohwcap", F_OK)      = -1 ENOENT (No such file or directory)
mmap(NULL, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5c2b75a000
access("/etc/ld.so.preload", R_OK)      = -1 ENOENT (No such file or directory)
open("/etc/ld.so.cache", O_RDONLY|O_CLOEXEC) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=29412, ...}) = 0
mmap(NULL, 29412, PROT_READ, MAP_PRIVATE, 3, 0) = 0x7f5c2b752000
close(3)                                = 0
access("/etc/ld.so.nohwcap", F_OK)      = -1 ENOENT (No such file or directory)
open("/lib/x86_64-linux-gnu/libc.so.6", O_RDONLY|O_CLOEXEC) = 3
read(3, "\177ELF\2\1\1\3\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0P\34\2\0\0\0\0\0"..., 832) = 832
fstat(3, {st_mode=S_IFREG|0755, st_size=1738176, ...}) = 0
mmap(NULL, 3844640, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0x7f5c2b191000
mprotect(0x7f5c2b332000, 2097152, PROT_NONE) = 0
mmap(0x7f5c2b532000, 24576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x1a1000) = 0x7f5c2b532000
mmap(0x7f5c2b538000, 14880, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x7f5c2b538000
close(3)                                = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5c2b751000
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5c2b750000
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5c2b74f000
arch_prctl(ARCH_SET_FS, 0x7f5c2b750700) = 0
mprotect(0x7f5c2b532000, 16384, PROT_READ) = 0
mprotect(0x7f5c2b75c000, 4096, PROT_READ) = 0
munmap(0x7f5c2b752000, 29412)           = 0
open("./randomfile", O_WRONLY|O_CREAT, 0600) = 3
fallocate(3, 0, 0, 100)                 = -1 EOPNOTSUPP (Operation not supported)
fstat(3, {st_mode=S_IFREG|0600, st_size=1000, ...}) = 0
fstatfs(3, {f_type=0x2fc12fc1, f_bsize=131072, f_blocks=191, f_bfree=191, f_bavail=191, f_files=48958, f_ffree=48949, f_fsid={1485573181, 14169277}, f_namelen=255, f_frsize=131072}) = 0
pread(3, 0x7fffc374798f, 1, 99)         = -1 EBADF (Bad file descriptor)
fstat(1, {st_mode=S_IFCHR|0600, st_rdev=makedev(136, 0), ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5c2b759000
write(1, "Bad file descriptor\n", 20Bad file descriptor
)   = 20
exit_group(0)                           = ?
+++ exited with 0 +++
root@linux:/issue-6860# 

This is documented: http://man7.org/linux/man-pages/man3/posix_fallocate.3.html#NOTES

If fd has been opened with the O_APPEND or O_WRONLY flags, the
function will fail with the error EBADF.

Change

open("./randomfile", O_WRONLY | O_CREAT, S_IRUSR | S_IWUSR);

to

open("./randomfile", O_RDWR | O_CREAT, S_IRUSR | S_IWUSR);

in your reproducer to avoid this.

[CMCDragonkai]

Does tmpfs supply a "native syscall" for fallocate? Ok I see the issue now.

[loli10K]

On tmpfs we have no EOPNOTSUPP -> happy libc, no fallocate(2) "emulation" needed -> no pread(2) -> no EBADF:

root@linux:~# mount -t tmpfs tmpfs /mnt/tmpfs
root@linux:~# cp -fav posix_fallocate /mnt/tmpfs
‘posix_fallocate’ -> ‘/mnt/tmpfs/posix_fallocate’
root@linux:~# cd /mnt/tmpfs
root@linux:/mnt/tmpfs# strace -f -e trace=fallocate,pread ./posix_fallocate 100
fallocate(3, 0, 0, 100)                 = 0
+++ exited with 0 +++
root@linux:/mnt/tmpfs# strace -f -e trace=fallocate,pread ./posix_fallocate 100
fallocate(3, 0, 0, 100)                 = 0
+++ exited with 0 +++
root@linux:/mnt/tmpfs# strace -f -e trace=fallocate,pread ./posix_fallocate 50
fallocate(3, 0, 0, 50)                  = 0
+++ exited with 0 +++
root@linux:/mnt/tmpfs# strace -f -e trace=fallocate,pread ./posix_fallocate 10
fallocate(3, 0, 0, 10)                  = 0
+++ exited with 0 +++
root@linux:/mnt/tmpfs# 

@CMCDragonkai can we close this?

[CMCDragonkai]

Yep