CVE-2026-70651

In libvips before version 8.18.3, applications using custom sources and processing untrusted PPM images can trigger a max/min error in vips_source_read_to_memory. Using VIPS_MAX instead of VIPS_MIN when selecting the read size can cause up to 4032 bytes to be written beyond the allocated buffer, leading to memory corruption or process crash.

In libvips before version 8.18.3, the old-style Radiance RLE decoder can process a repeat marker at the beginning of a scanline and read q[-1] before any prior pixel exists. This can disclose four bytes of adjacent heap data when loading a crafted Radiance image.

libvips before 8.18.3, built with libultrahdr support, can incorrectly size an output buffer in uhdrsave.c when a pipeline enlarges an incoming JPEG to a very large output before encoding a gain map. This can cause a heap buffer over-read.

libvips before version 8.18.3 has a vulnerability that allows a crafted many-band TIFF to evade scanline validation and cause an integer overflow, leading to out-of-bounds reads/writes, possible data disclosure, and likely crashes. Remote code execution has not been demonstrated but cannot be ruled out.

Vulnerability in libvips up to version 8.18.1. The tiffload operation may incorrectly determine the number of channels in JPEG/JPEG2000 tiles within a TIFF image, leading to a buffer overflow. Fixed in version 8.18.2.

libvips versions up to 8.18.1 have a vulnerability in the EXIF decoder. Missing range verification of EXIF tag groups can cause a null pointer dereference and crash. Fixed in version 8.18.2.

libvips versions up to 8.18.0 on 32-bit systems have a vulnerability in the gifload operation. Incorrect dimension determination can lead to an integer overflow. Fixed in version 8.18.1.

Vulnerability in libvips up to version 8.18.0. The vipsload operation may incorrectly determine image dimensions, leading to an integer overflow and subsequent heap-based buffer overflow. Fixed in version 8.18.1.