Find Swap Memory Usage by Processes in Linux

1. Overview

In this tutorial, we’ll discuss how we can check for processes that use swap memory. First, we’ll cover the /proc directory to see what it contains and how we can extract processes’ information from it. Afterward, we’ll write a shell script that will automatically extract the swap usage information used by the processes.

Last, we’ll use the smem tool as an alternative to check for processes that use the swap memory.

2. The /proc Directory

On Linux, proc is a special directory that contains information about the Linux kernel, its configuration, and processes loaded into the physical memory. The operating system creates this directory once we boot into the Linux machine. Moreover, the virtual files inside this directory have no physical size on the disk. The information inside these files is generated on the fly once we read them.

Now that we have an idea about the proc directory, let’s see what’s inside it:

$ ls -l /proc
total 0
dr-xr-xr-x  9 root    root                  0 Feb 17 15:35 1
dr-xr-xr-x  9 root    root                  0 Feb 17 15:35 10
dr-xr-xr-x  9 root    root                  0 Feb 17 15:35 100
...
dr-xr-xr-x  9 hey     hey                   0 Feb 17 22:12 34160
...

As we review the list of items inside this directory, we’ll notice that there are a lot of numbered directories, named directories, and other virtual files. However, we’re only concerned with the numbered directories because they correspond to each running process. The number of the directory is actually the PID of the process.

As we can see in the above snippet, it also lists the user processes. So, let’s list the items inside the “34160” directory to see what it contains:

$ ls -l /proc/34160

total 0
...
-r--r--r-- 1 hey hey 0 Feb 17 22:12 stat
-r--r--r-- 1 hey hey 0 Feb 17 22:12 statm
-r--r--r-- 1 hey hey 0 Feb 17 22:12 status
...

The virtual files inside the directory will contain a plethora of information regarding the process. But in our case, we’re interested in the status virtual file because it contains the memory usage information.

Since we’re only interested in the swap usage, we’ll grep the VmSwap field from the file:

$ cat /proc/34160/status | grep VmSwap
VmSwap:	       0 kB

Now that we know how to check for swap memory used by a process, we can write a shell script to print a list of processes using the swap memory.

2.1. Shell Script to List Processes using Swap Memory

The shell script will go through each numbered directory in the proc directory and print the swap usage it extracts from the status file. It’ll also print the name and PID of the process as well:

#!/bin/bash

overall=0
for status_file in /proc/[0-9]*/status; do
    swap_mem=$(grep VmSwap "$status_file" | awk '{ print $2 }')
    if [ "$swap_mem" ] && [ "$swap_mem" -gt 0 ]; then
        pid=$(grep Tgid "$status_file" | awk '{ print $2 }')
        name=$(grep Name "$status_file" | awk '{ print $2 }')
        printf "%s\t%s\t%s KB\n" "$pid" "$name" "$swap_mem"
    fi
    overall=$((overall+swap_mem))
done
printf "Total Swapped Memory: %14u KB\n" $overall

The script is pretty self-explanatory. Let’s run it in the terminal:

$ ./swp.sh
1658	Discord	    41448 KB
1681	Xwayland    15772 KB
18621	firefox	    213204 KB
18872	WebExtensions 60912 KB
...
Total Swapped Memory:        1177212 KB

We can easily customize the output further by using something like the column tool.

3. Using the smem Utility

The smem tool displays memory usage of processes. Apart from showing the RSS, PSS, and USS memory, it can also show the swap memory.

By default, it doesn’t ship with most Linux distributions. So, we’ll have to install it from our distribution’s official repository.

3.1. Installation

The smem utility will be available under the smem package name. We can use a package manager like yum or apt to install it.

For Debian and its derivatives:

# apt install smem

For Fedora, OpenSUSE, and RHEL:

# yum install smem

3.2. Usage

Once smem is installed, we can try it out in the terminal:

$ smem
  PID User     Command                         Swap      USS      PSS      RSS 
  494 hey      swaybg -o DVI-I-1 -i /home/    15136        4       86     2304 
 1665 hey      /opt/discord/Discord --type     8800        0       96      820 
  589 hey      /usr/lib/pulse/gsettings-he     1224       20      122     2372 
37348 hey      /bin/sh /usr/bin/android-st      204      164      180     1220 
...

Let’s say we want to print the top 10 processes sorted by swap memory usage in descending order. We can do so with the -s or –sort option:

$ smem -s swap -r | head -n10
  PID User     Command                         Swap      USS      PSS      RSS 
18943 hey      /usr/lib/firefox/firefox -c   385536   161836   165339   213496 
18621 hey      /usr/lib/firefox/firefox      225660   458788   479474   576556 
 1748 hey      /opt/discord/Discord --type    84652   130220   137640   153064 
 1658 hey      /opt/discord/Discord           76108    28336    34916    51324 
18872 hey      /usr/lib/firefox/firefox -c    65172   133824   135810   169480 
33599 hey      /usr/lib/firefox/firefox -c    55012   261264   266269   324880 
20578 hey      megasync                       52252    13272    15308    24040 
35605 hey      /usr/lib/firefox/firefox -c    46584   504264   507708   554040 
 1705 hey      /opt/discord/Discord --type    41236    13876    18735    29988

The -r option will reverse the output, which will print the list in descending order.

4. Conclusion

In this brief tutorial, we saw how we could view the swap memory used by processes in Linux. We accomplished that by viewing the processes’ information inside the proc directory and using the smem tool.