Skip to content


Ethan Minwoo Oh edited this page Mar 7, 2022 · 26 revisions


A race condition was found in the way the Linux kernel's memory subsystem handled the copy-on-write (COW) breakage of private read-only memory mappings. All the information we have so far is included in this page.

The bug has existed since around 2.6.22 (released in 2007) and was fixed on Oct 18, 2016. List of patched versions here

There are proof of concepts available here.

Video Explanation

Explaining dirtyc0w local root exploit - CVE-2016-5195


  • An unprivileged local user could use this flaw to gain write access to otherwise read-only memory mappings and thus increase their privileges on the system.
  • This flaw allows an attacker with a local system account to modify on-disk binaries, bypassing the standard permission mechanisms that would prevent modification without an appropriate permission set.


        do_fault <- pte is not present
	  do_cow_fault <- FAULT_FLAG_WRITE
	      maybe_mkwrite(pte_mkdirty(entry), vma) <- mark the page dirty
	      						but keep it RO 
# Returns with 0 and retry
    (flags & FOLL_WRITE) && !pte_write(pte) <- retry fault
        FAULT_FLAG_WRITE && !pte_write
	    PageAnon() <- this is CoWed page already
	    reuse_swap_page <- page is exclusively ours
	      maybe_mkwrite <- dirty but RO again
	      ret = VM_FAULT_WRITE
((ret & VM_FAULT_WRITE) && !(vma->vm_flags & VM_WRITE)) <- we drop FOLL_WRITE
# Returns with 0 and retry as a read fault
cond_resched -> different thread will now unmap via madvise
  !pte_present && pte_none
        do_fault <- pte is not present
	  do_read_fault <- this is a read fault and we will get pagecache


  • The In The Wild exploit relied on writing to /proc/self/mem on one side of the race.
  • ptrace(PTRACE_POKEDATA) can write to readonly mappings.
  • The attack relies on racing the madvise(MADV_DONTNEED) system call while having the page of the executable mmapped in memory.

Commit messages

    commit 4ceb5db9757aaeadcf8fbbf97d76bd42aa4df0d6
    Author: Linus Torvalds <>
    Date:   Mon Aug 1 11:14:49 2005 -0700
    Fix get_user_pages() race for write access
    There's no real guarantee that handle_mm_fault() will always be able to
    break a COW situation - if an update from another thread ends up
    modifying the page table some way, handle_mm_fault() may end up
    requiring us to re-try the operation.
    That's normally fine, but get_user_pages() ended up re-trying it as a
    read, and thus a write access could in theory end up losing the dirty
    bit or be done on a page that had not been properly COW'ed.
    This makes get_user_pages() always retry write accesses as write
    accesses by making "follow_page()" require that a writable follow has
    the dirty bit set.  That simplifies the code and solves the race: if the
    COW break fails for some reason, we'll just loop around and try again.
    commit 19be0eaffa3ac7d8eb6784ad9bdbc7d67ed8e619
    Author: Linus Torvalds <>
    Date:   Thu Oct 13 20:07:36 2016 GMT
    This is an ancient bug that was actually attempted to be fixed once
    (badly) by me eleven years ago in commit 4ceb5db9757a ("Fix
    get_user_pages() race for write access") but that was then undone due to
    problems on s390 by commit f33ea7f404e5 ("fix get_user_pages bug").
    In the meantime, the s390 situation has long been fixed, and we can now
    fix it by checking the pte_dirty() bit properly (and do it better).  The
    s390 dirty bit was implemented in abf09bed3cce ("s390/mm: implement
    software dirty bits") which made it into v3.9.  Earlier kernels will
    have to look at the page state itself.
    Also, the VM has become more scalable, and what used a purely
    theoretical race back then has become easier to trigger.
    To fix it, we introduce a new internal FOLL_COW flag to mark the "yes,
    we already did a COW" rather than play racy games with FOLL_WRITE that
    is very fundamental, and then use the pte dirty flag to validate that
    the FOLL_COW flag is still valid.