Skip to content
Tutorials Center
My learning

Topics

Linux Command Reference ยท Special Files and Directories

The /proc/meminfo file

Reference for /proc/meminfo: what MemTotal, MemFree, MemAvailable, Buffers, Cached and swap lines mean, the units, and how free and top calculate from it.

/proc/meminfo is the kernel's memory report. Each line is Name: value kB, generated when you read it. It is the data behind free, top and most monitoring tools. Reading it directly is handy in scripts, because the format is simple and needs no extra program.

The most important lesson is that free memory is not the number to watch. Linux uses spare RAM as a file cache, so MemFree is usually small on a healthy machine. MemAvailable estimates how much could be given to new programs without swapping, which is the figure to alert on.

Terminal
grep -E '^(MemTotal|MemAvailable):' /proc/meminfo

Try it

Bash Reading values from the real file

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with read or cat), it gets the next line from this box.

Bash Calculating usage from sample data

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with read or cat), it gets the next line from this box.

Syntax

All sizes are in kibibytes (1024 bytes), even though the suffix is kB. Selected lines:

Field Description
MemTotal Usable RAM (less than installed: the kernel reserves some).
MemFree RAM that is completely unused.
MemAvailable Estimate of memory available to start applications without swapping (free plus reclaimable cache). Added in kernel 3.14.
Buffers, Cached Memory used for block-device buffers and the page cache.
SwapCached Swapped-out data also still in RAM.
Active, Inactive Recently used and less recently used pages.
SwapTotal, SwapFree Size and free part of swap.
Dirty, Writeback Data waiting to be written, and being written, to disk.
AnonPages, Shmem, Slab Program memory, shared and tmpfs memory, kernel object caches.
CommitLimit, Committed_AS Overcommit accounting.
HugePages_*, Hugepagesize Huge-page counters.

Edge cases: some lines have no unit (HugePages_Total: 0). Inside containers the file may show the host's memory, not your cgroup limit; read /sys/fs/cgroup/memory.max for that.

Usage notes

  • Used memory is best computed as MemTotal - MemAvailable. Older tools used MemTotal - MemFree - Buffers - Cached, which gives a similar answer on most systems.
  • Low MemFree is normal. Do not "free" the cache; echo 3 > /proc/sys/vm/drop_caches only slows the system down while the cache refills.
  • Swap use alone is not a problem. Constant swap in and out (watch vmstat 1 columns si and so) is.
  • Units: divide by 1024 for MiB; free -h does that for you.
  • Per process: use /proc/PID/status (VmRSS) or smem; see /proc/PID.
  • Parsing: the key has a trailing colon, so match $1 == "MemTotal:" exactly, or MemFree also matches MemFreeSomething in a loose regular expression.
  • OOM killer: when MemAvailable reaches zero and swap is full, the kernel kills a process to recover; check dmesg or the journal.