https://github.com/0x2b3bfa0/ext4-imc
Bash script to calculate inode modes for the Ext4 filesystem.
https://github.com/0x2b3bfa0/ext4-imc
bash calculate-inode-modes command-line ext4-filesystem filesystem inode mount proof-of-concept
Last synced: 5 months ago
JSON representation
Bash script to calculate inode modes for the Ext4 filesystem.
- Host: GitHub
- URL: https://github.com/0x2b3bfa0/ext4-imc
- Owner: 0x2b3bfa0
- License: gpl-2.0
- Created: 2015-05-22T16:39:34.000Z (over 10 years ago)
- Default Branch: master
- Last Pushed: 2021-09-25T20:59:18.000Z (about 4 years ago)
- Last Synced: 2025-05-05T19:18:00.660Z (5 months ago)
- Topics: bash, calculate-inode-modes, command-line, ext4-filesystem, filesystem, inode, mount, proof-of-concept
- Language: Shell
- Homepage:
- Size: 33.2 KB
- Stars: 3
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Ext4 inode mode calculator
Bash script to calculate inode modes for the Ext4 filesystem. Originally made for an [AskUbuntu](https://askubuntu.com/questions/626634/convert-a-file-to-directory/626731) answer.
***
# Achieving the conversion
### Creating a test filesystem
In order to preserve our main filesystem from any possible damage after running this experiment, we're going to create a small filesystem inside a normal file for test purposes.
1. Create a zero-filled file called `test` with a size of 10 megabytes:
dd if=/dev/zero of=~/test bs=10M count=1
2. Create an Ext4 filesystem inside the file, as if it were a partition:
mkfs.ext4 ~/test
### Creating some files and directories
Now we hace a fully functional filesystem inside the `test` file, so we're going to create some files and directories inside it.
1. Mount the newly created filesystem inside `/mnt`:
sudo mount ~/test /mnt
2. Create a file and a directory:
sudo mkdir /mnt/folder
echo "contents" | sudo tee /mnt/file
3. Check the contents of the filesystem:
ls -l /mnt
Output should be something like this:
total 2
-rw-r--r-- 1 root root 0 may 21 18:53 file
drw-r--r-- 2 root root 1024 may 21 18:55 folder
4. Unmount the test filesystem:
sudo umount /mnt
### Swapping the file and the folder
1. Run `debugfs` against the `test` file with write permission (`-w` flag):
debugfs -w ~/test
2. Convert `file` into a folder:
* At the `debugfs` prompt, type this:
modify_inode file
* A prompt will appear asking you a mode; type this:
040644
* Keep pressing **return** to leave the remaining data as-is until the prompt appears again.
3. Convert `folder` into a file:
* At the `debugfs` prompt, type this:
modify_inode folder
* A prompt will appear asking you a mode; type this:
0100644
* Keep pressing **return** to leave the remaining data as-is until the prompt appears again.
4. To exit `debugfs` prompt, simply hit **q** and then **return**
### Checking the success of the operation
1. Mount the test filesystem again:
sudo mount ~/test /mnt
2. Check the filesystem contents:
ls -l /mnt
Now, it should show the file as if it were a directory and *vice versa*:
total 2
drw-r--r-- 1 root root 0 may 21 18:53 file
-rw-r--r-- 2 root root 1024 may 21 18:55 folder
---
# Script to calculate inode modes
#!/bin/bash
#### See https://ext4.wiki.kernel.org/index.php/Ext4_Disk_Layout#Inode_Table
## Terminal measures:
x="$(( $(tput cols) / 2 ))" # Width of the terminal
y="$(( $(tput lines) / 2 ))" # Height of the terminal
## File descriptors:
declare -A types # Declare an associative array with file descriptors
types[f]='0x8000' # File
types[l]='0xA000' # Link
types[s]='0xC000' # Socket
types[d]='0x4000' # Directory
types[p]='0x1000' # Named pipe
types[b]='0x6000' # Block device
types[c]='0x2000' # Character device
## Permissions:
declare -A permission # Declare an associative array with permissions
permission[user_S]='0x800' # UID
permission[user_s]='0x840' # UID and user can execute
permission[user_r]='0x100' # User can read
permission[user_w]='0x80' # User can write
permission[user_x]='0x40' # User can execute
permission[group_S]='0x400' # GID
permission[group_s]='0x408' # GID and group can execute
permission[group_r]='0x20' # Group can read
permission[group_w]='0x10' # Group can write
permission[group_x]='0x8' # Group can execute
permission[other_T]='0x200' # Sticky bit
permission[other_t]='0x201' # Sticky bit and other can execute
permission[other_r]='0x4' # Other can read
permission[other_w]='0x2' # Other can write
permission[other_x]='0x1' # Other can execute
## Cleanup function:
function cleanup() {
tput cvvis # Make the cursor visible
tput rmcup # Restore saved terminal contents
stty sane # Fix problems caused by read -s
exit 0 # Exit gracefully
}
## Function to print at a specified position:
function pprint() {
tput cup $1 $2
printf "${@:3}"
}
## Function to clear the notification area:
function reset() {
pprint $((y+2)) $((x-40)) ' %.0s' {1..25} # Print 25 spaces
}
## Function to notify something to the user:
function notify() {
reset # Clear the notification area
pprint $((y+2)) $((x-40)) "$@" # Print the notification text
}
## If the terminal is smaller than 100x8, exit gracefully (self-explainatory):
if [ $x -lt 50 ] || [ $y -lt 5 ]; then
echo 'Error, I need a minimum of 100x10 lines to run'
exit 0
fi
## Initialize the terminal:
trap cleanup EXIT SIGHUP SIGINT SIGTERM # Call cleanup function after receiving ^C
stty -echo cbreak # Put terminal in silent mode
tput smcup # Save terminal contents
tput civis # Make the cursor inisible
## Draw the big box:
printf '\033[1;37m' # Color
pprint $((y-3)) $((x-48)) '\u2500%.0s' {1..97} # Upper line
pprint $((y+4)) $((x-48)) '\u2500%.0s' {1..97} # Lower line
for ((i=4;i>-4;i--)); do # Sides:
pprint $((y+i)) $((x-49)) '\u2502' # Left line
pprint $((y+i)) $((x+49)) '\u2502' # Right line
done # End sides
pprint $((y-3)) $((x-49)) '\u256D' # Upper-left corner
pprint $((y+4)) $((x-49)) '\u2570' # Lower-left corner
pprint $((y-3)) $((x+49)) '\u256E' # Upper-right corner
pprint $((y+4)) $((x+49)) '\u256F' # Lower-right corner
## Draw the small box:
printf '\033[1;35m' # Color
pprint $((y+1)) $((x-10)) '\u2501%.0s' {1..10} # Upper line
pprint $((y+3)) $((x-10)) '\u2501%.0s' {1..10} # Lower line
pprint $((y+2)) $((x-11)) '\u2503' # Left line
pprint $((y+2)) $((x+00)) '\u2503' # Right line
pprint $((y+1)) $((x-11)) '\u250F' # Upper-left corner
pprint $((y+3)) $((x-11)) '\u2517' # Lower-left corner
pprint $((y+1)) $((x+00)) '\u2513' # Upper-right corner
pprint $((y+3)) $((x+00)) '\u251B' # Lower-right corner
## Print type help:
pprint $((y-2)) $((x-44)) '\033[0;37mInode type: \033[1;37mf\033[0;37mile, \033[1;37md\033[0;37mirectory, \033[1;37ml\033[0;37mink, named \033[1;37mp\033[0;37mipe, \033[1;37ms\033[0;37mocket, \033[1;37mc\033[0;37mharacter device or \033[1;37mb\033[0;37mlock device.'
## Print permission help:
pprint $((y-1)) $((x-40)) '\033[0;36mPermission (\033[1;32mu\033[0;32mser\033[0;36m, \033[1;33mg\033[0;33mroup\033[0;36m or \033[1;31mo\033[0;31mther\033[0;36m): \033[1;36mr\033[0;36mead, \033[1;36mw\033[0;36mrite, e\033[1;36mx\033[0;36mecute, \033[1;36mhyphen\033[0;36m or \033[1;36mspace\033[0;36m to skip.'
pprint $((y+0)) $((x+8)) 's\033[1;36mt\033[0;36micky bit and executable, '
pprint $((y+1)) $((x+8)) 's\033[1;36mT\033[0;36micky bit not executable, '
pprint $((y+2)) $((x+8)) '\033[1;36ms\033[0;36metuid/setgid and executable, '
pprint $((y+3)) $((x+8)) '\033[1;36mS\033[0;36metuid/setgid not executable. '
## Endless loop:
while :; do
## Clear the input area:
pprint $((y+2)) $((x-10)) '% *s\n' 10 # Print 16 spaces
## Print mask in the input area:
printf '\033[1;37m' # Color for the type
pprint $((y+2)) $((x-10)) '\u2588' # Block for the type
printf '\033[1;36m' # Color for the permision
pprint $((y+2)) $((x- 9)) '\u2588%.0s' {1..9} # Blocks for the permission
## Loop through all variables to make a proper input:
for var in type {user,group,other}_{r,w,x}; do
## Assign colors and regex to fields:
case "$var" in
(type) color='\033[1;37m'; regex='^[fdlpscb]$' ;;
(other_x) regex='^[-xtT]$' ;;&
(user_x|group_x) regex='^[-xsS]$' ;;&
(user_[rw]|group_[rw]|other_[rw]) regex="^[-${var: -1}]$";;&
(user*) color='\033[1;32m' ;;
(group*) color='\033[1;33m' ;;
(other*) color='\033[1;31m' ;;
esac
## Change the pointer position:
pprint $((y+3)) $(((x-10)+pointer)) "${color}\u2501" # Print the pointer on its new position
if (( pointer > 0 )); then # If the pointer is not in the first position:
pprint $((y+3)) $(((x-10)+(pointer-1))) '\033[1;35m\u2501' # Clear the old pointer
fi
## Infinite loop until there is a valid input for the current character:
while :; do
printf "$color" # Set the character color
IFS= read -rn 1 $var # Read a character (even if it's a space)
declare $var="${!var// /-}" # Convert spaces to hyphens.
if [[ "$var" == "type" ]]; then # If the current variable is type:
declare $var="${!var//-/f}" # Convert "-" to "f"
fi
if [[ "${!var}" =~ $regex ]]; then # If there is a valid input:
reset # Clear error notification if any
break # Exit from this loop
else # Else:
notify "\033[1;31mWrong input!" # Print the error message
fi
done
## Print the entered value:
pprint $((y+2)) $(((x-10)+pointer)) "${!var}"
## Sum the current permission:
((mode+=permission[${var%_*}_${!var}]))
## Increment the pointer:
((pointer++))
done
## Post-read:
unset pointer # Reset the pointer
pprint $((y+3)) $((x-1)) "\033[1;35m\u2501" # Clear the pointer
read -n 1 # Wait for Return or another character
## Sum file descriptor type:
((mode+=${types[$type]}))
## Final commands:
mode=$(printf "%o" $mode) # Convert mode to octal (before this was decimal)
notify "\033[1;32mOctal mode:\033[1;34m $mode" # Print the octal mode
unset mode # Reset the mode
done
[View script on GitHub](https://github.com/0x2b3bfa0/ext4-imc)
# Handicaps
* The folder doesn't open. You can't open it *unless* you put on it the "raw folder data" that contained it originally.
# Further reading
https://ext4.wiki.kernel.org/index.php/Ext4_Disk_Layout#Inode_Table
***
**Thanks to [@tallus](https://askubuntu.com/users/80756/tallus).** He gave me a great hint:
> debugfs has a modify_inode command that allows you to edit an inode
> directly which would allow you to set the file flag to a dir.