The internet is broken right now.

FFmpeg through 9.0.2 contains an out-of-bounds read vulnerability in the SGI encoder's encode_frame() in libavcodec/sgienc.c, where unsigned size variables mis-handle negative linesize frames. Attackers can supply a video whose display matrix triggers autorotate vflip, so transcoding to SGI images reads pixel data far outside the frame and crashes the process.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds write vulnerability in v4l2_buffer_swframe_to_buf() in libavcodec/v4l2_buffers.c because cumulative plane offsets are not checked against the V4L2 buffer length. Attackers can supply crafted video whose dimensions or padding cause unsigned wraparound in v4l2_bufref_to_buf(), making memcpy() write past the mmaped driver buffer during V4L2 M2M encoding.

Memory safety OS & kernel

FFmpeg through 9.0.2 contains an out-of-bounds read and invalid free vulnerability in aribcaption_trans_ass_subtitle() in libavcodec/libaribcaption.c that frees subtitle rectangles using region_count instead of num_rects. Attackers can supply crafted ARIB captions whose later region exceeds the 8 KiB AVBPrint limit, triggering cleanup that frees out-of-bounds pointers and corrupts the heap.

Memory safety

FFmpeg 7.0 before 8.1.3 and 9.x before 9.0.2 contains a heap out-of-bounds write vulnerability in vk_av1_decode_slice() in libavcodec/vulkan_av1.c because the tile-count limit is checked only on entry. Attackers can supply a crafted AV1 stream with more than 256 tiles decoded via the Vulkan hardware accelerator to overflow tile_sizes and corrupt heap memory.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds read vulnerability in bitpacked_decode_uyvy422() in libavcodec/bitpacked_dec.c because it never verifies packets hold width * height * 2 bytes. Attackers can supply truncated bitpacked UYVY422 packets via crafted SDP/RFC 4175 RTP sessions or raw input to crash applications or expose adjacent heap memory.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds write vulnerability in flic_decode_frame_15_16BPP() in libavcodec/flicvideo.c that allows attackers to corrupt heap memory on big-endian builds. Attackers can supply a crafted 15/16 bpp FLIC/FLX file with an FLI_BRUN chunk whose non-terminating byte-swap loop writes packet-derived data past the frame buffer, crashing the process.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds read vulnerability in libxevd_receive_frame() in libavcodec/libxevd.c that fails to validate NAL unit length prefixes against remaining packet data. Attackers can supply crafted EVC media with oversized 4-byte NAL length prefixes, causing libxevd to read past the packet buffer, crashing the application or exposing adjacent heap memory.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds heap read vulnerability in tdsc_load_cursor() in libavcodec/tdsc.c, which stores cursor dimensions before validating them. Attackers can supply a crafted AVI file with oversized TDSC cursor dimensions so tdsc_paint_cursor() reads past the cursor buffer, crashing the decoder or leaking heap memory into frames.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds read vulnerability in the Musepack SV7 decoder because mpc7_decode_init() does not validate lastframelen against MPC_FRAME_SIZE. Attackers can supply a crafted Musepack SV7 file declaring an oversized final frame, causing consumers of decoded frames to read past the audio buffer, leading to crashes or heap memory disclosure.

Memory safety

A flaw was found in GStreamer gst-plugins-good (avidemux). In the superindex and subindex parsers, integer overflow can bypass bounds checks and cause out-of-bounds reads. Because avidemux is auto-plugged by playbin, decodebin, and gst-discoverer, opening or previewing a crafted AVI is sufficient to trigger the issue. Fixed upstream in gst-plugins-good 1.28.6 (GStreamer-SA-2026-0072).

Memory safety

stable-diffusion.cpp is a pure C/C++ library for running diffusion model (Stable Diffusion, Flux, Wan, Qwen Image, Z-Image, and more) inference. Prior to master-764-e22272e, the safetensors loader in src/model_io/safetensors_io.cpp trusts each tensor's data_offsets after checking only that nbytes equals end - begin. A crafted .safetensors file can declare a byte range beyond the file's data section, causing src/model_loader.cpp to consume an attacker-chosen length from an unchecked mapped-file offset. Loading the model can read adjacent process memory into tensor buffers or crash the process. This issue is fixed in master-764-e22272e.

Memory safety Library

The GPIO syscall handlers in drivers/gpio/gpio_handlers.c did not validate the pin argument supplied by a user-mode caller. z_vrfy_gpio_pin_configure(), z_vrfy_gpio_pin_get_config() and z_vrfy_gpio_pin_interrupt_configure() verified only the device object (K_SYSCALL_DRIVER_GPIO() / K_SYSCALL_OBJ()) and then passed the caller-supplied gpio_pin_t straight through to z_impl_gpio_pin_*() and on to the driver. The only pin check in include/zephyr/drivers/gpio.h was __ASSERT(gpio_port_pin_is_supported(cfg->port_pin_mask, pin), ...), which compiles to nothing unless CONFIG_ASSERT is enabled, so in a production build no check existed at all. GPIO drivers use pin as a shift count and as an array index on the documented assumption that it is a pin the controller supports. On STM32, for example, gpio_stm32_pin_interrupt_configure() turns pin into an EXTI line and stm32_gpio_intc_set_irq_callback() writes data->cb[line_num].cb and data->cb[line_num].data into a NUM_EXTI_LINES-entry array (default 16) in drivers/interrupt_controller/intc_gpio_stm32.c, guarded only by a compiled-out __ASSERT_NO_MSG(); the same function then reads exti_irq_table[line_num] and calls irq_enable() on the result. Other controllers index fixed-size per-pin configuration tables the same way — drivers/gpio/gpio_ite_it8xxx2_v2.c reads an MMIO base pointer out of wuc_base[8] with the caller's pin and performs a read-modify-write through it. The issue is reachable only in builds with CONFIG_USERSPACE=y (the file is compiled under that symbol), by a user-mode thread that has been granted access to a GPIO device object. Such a grant is intended to authorize control of that port's pins; the missing validation instead lets the thread drive a supervisor-mode out-of-bounds write of a function pointer and an associated void * past a kernel .bss array, and out-of-bounds reads that feed irq_enable() and MMIO accesses — an escape from the user-mode sandbox that can lead to kernel memory corruption, arbitrary code execution, or a crash. The fix validates pin against the controller's gpio_driver_config.port_pin_mask with gpio_port_pin_is_supported() in every pin-taking z_vrfy_ handler and returns -EINVAL otherwise, complemented by an unconditional pin >= GPIO_MAX_PINS_PER_PORT range check added to the shared z_impl_gpio_pin_*() inlines.

Memory safety OS & kernel

Tanium addressed a heap-based buffer overflow vulnerability in Index.

Memory safety

SDL_image through 3.4.8 and through 2.8.12 contains a stack out-of-bounds read vulnerability in IMG_LoadLBM_IO() in src/IMG_lbm.c that skips bitplane count validation when HAM mode is set. Attackers can supply a crafted ILBM file with more than 8 HAM bitplanes to read past the colormap stack array, causing crashes or disclosing stack memory.

Memory safety

CImg through 4.0.6 contains a heap out-of-bounds read vulnerability in the _load_bmp() function of CImg.h caused by an integer overflow in the row byte count. Attackers can supply a crafted BMP file with a large header width that wraps dx_bytes, causing crashes or heap memory disclosure into the decoded image.

Memory safety

sharp before 0.35.5 bundles a librsvg dependency affected by a memory-related vulnerability (CVE-2026-96889) that, under certain runtime-specific conditions, can lead to code execution on glibc-based Linux. An attacker can supply a crafted SVG image that, when decoded by sharp through librsvg, may result in arbitrary code execution.

Memory safety

A ThreadX module reaches a kernel object's delete service through the Module Manager's dispatch layer, which asked one question about the pointer: does the object lie outside the module's own code and data. That is the policy for *using* an object, and using an object a module does not own is supported and intended — `txm_module_object_pointer_get_extended` searches the system's created objects by name and hands back objects the application created and objects other modules created, so that a module can send to a shared queue, take a shared semaphore or get a shared mutex. Destroying one of those is not sharing it. A module could name any object it had a pointer to and have the privileged dispatcher delete it.

Memory safety OS & kernel

In the Linux kernel, the following vulnerability has been resolved: net/sched: act_ct: fix helper UAF due to extensions realloc While calling the helpers, a raw pointer to the extensions area is wired into expectations list: -> nf_ct_helper() -> helper->help() -> nf_ct_expect_related_report() -> nf_ct_expect_insert() -> hlist_add_head_rcu(&exp->lnode, &master_help->expectations) In case the connection is not confirmed yet, more extensions can be added afterwards with *_ext_add() calls reallocating the extension space and leaving the now invalid pointer in the expectations list that is later accessed while removing the expectation. Make sure that helpers are called at the end after all the other extensions are already added. Note that the helper rejection now leaves the mark and labels set, but that's not different from how the NAT was handled before or how the mark and the labels were handled on confirmation failure. And there are no atomicity guarantees provided by the API anyway.

Memory safety OS & kernel

GNU LibreDWG through 0.14.8612 contains a global out-of-bounds read vulnerability in the AcDs decoder acds_private() that allows attackers to read beyond a static table via an unchecked schema property type index. Attackers can supply a crafted R2004 or R2010+ DWG file with a large 'schdat' property type value to crash dwgread, dwg2dxf, or any application using LibreDWG.

Memory safety

FreeImage (danoli3 fork) through 3.19.23 contains a heap out-of-bounds read vulnerability in Unpack32Bits() in PluginPICT.cpp because rowBytes is never checked against the file-controlled PixMap width. Attackers can supply a crafted PICT file with a truncated or undersized rowBytes to over-read up to 64 KiB of heap, leaking memory into decoded pixels or crashing the process.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized stack memory disclosure vulnerability in old_codec23() of libavcodec/sanm.c that leaks stack bytes into decoded frames. Attackers can supply a crafted LucasArts SMUSH/SANM animation using codec 23 with an inline table to expose decoder process stack memory in output video.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized memory disclosure vulnerability in the CineForm HD decoder that allows attackers to leak heap memory via unzeroed idwt_tmp buffers in libavcodec/cfhd.c. Attackers can supply a crafted CFHD file with an ImageHeight tag exceeding coded band heights, causing uninitialized heap bytes to be filtered into decoded video output.

Memory safety

FFmpeg through 9.0.2 contains an integer overflow vulnerability in ff_prores_kostya_encode_init() that miscalculates frame_size_upper_bound using 32-bit arithmetic for large frame dimensions. Attackers can supply a high-resolution video transcoded with the prores_ks encoder, causing encode_frame() to write slices past an undersized packet buffer and corrupt heap memory.

Memory safety

FFmpeg before 8.1.3 and 9.x before 9.0.2 contains an out-of-bounds write vulnerability in the VVC decoder's sh_entry_points() function in libavcodec/vvc/ps.c. Attackers can supply a crafted VVC (H.266) bitstream whose SPS/PPS tile and WPP layout produces excess entry points, overflowing entry_point_start_ctu and corrupting decoder memory.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds memory access vulnerability in the encode_frame() function of libavcodec/huffyuvenc.c, because encode_init() never validates minimum frame dimensions for the median predictor. Attackers can supply a tiny video, such as a 2-pixel-wide or 1-2-pixel-tall frame, that causes negative sub_median_pred lengths and out-of-bounds reads and writes when transcoded to huffyuv or ffvhuff, crashing the encoder.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized memory use vulnerability in roq_dpcm_encode_frame() in libavcodec/roqaudioenc.c that encodes stale heap contents when flushed early. Attackers can supply short audio streams under eight frames for RoQ DPCM transcoding, causing DPCM-quantized uninitialized heap data from frame_buffer to leak into output packets.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized memory vulnerability in the DXA decoder in libavcodec/dxa.c that allows attackers to disclose heap memory via crafted DXA files. Attackers can supply a DXA video with a type 4 frame that copies the uninitialized decomp_buf buffer into output pictures, exposing leftover heap contents in decoded frames.

Memory safety

FFmpeg through 9.0.2 contains an information disclosure vulnerability in decode_mvc2() in libavcodec/mvcdec.c that leaks uninitialized stack memory through unchecked palette indices. Attackers can supply a crafted MVC2 stream declaring few colors but referencing higher palette indices, copying process stack contents into decoded RGB frames.

Memory safety

FFmpeg through 9.0.2 contains an out-of-bounds read vulnerability in the Cook decoder joint_decode() function because cook_decode_init() does not check js_subband_start against subbands. Attackers can supply a crafted RealMedia file whose extradata causes reads past decode_buffer_0, leaking pointers and decoder state into decoded audio to help defeat ASLR.

Memory safety

FFmpeg 6.1 through 9.0.2 contains an information disclosure vulnerability in the CamStudio decoder (libavcodec/cscd.c) that copies uninitialized heap memory into decoded frames when decompressed payloads are short. Attackers can supply a crafted AVI with a CSCD stream whose LZO or zlib frames decompress to fewer bytes than the frame size, recovering heap contents from the decoded or re-encoded output.

Memory safety

FFmpeg through 9.0.2 contains an information disclosure vulnerability in flashsv_decode_block() in libavcodec/flashsv.c that copies uninitialized heap memory into decoded frames when inflate output is short. Attackers can supply a crafted FLV with an FSV1 stream whose zlib block data decompresses short, then recover heap contents from the decoded or re-encoded output.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized memory read vulnerability in decode_hybrid() in libavcodec/flashsv.c that reads past decompressed data when decoding 15-bit color pixels. Attackers can supply a crafted Flash Screen Video 2 block whose inflated data ends mid-pixel to leak heap memory into decoded output frames.

Memory safety

FFmpeg through 9.0.2 contains an information disclosure vulnerability in the MSCC/SRGC decoder in libavcodec/mscc.c that outputs uninitialized heap bytes when RLE data ends early. Attackers can craft an MSCC or SRGC stream, such as in AVI, whose RLE data terminates early so the decoded frame exposes leftover heap memory to viewers of the output.

Memory safety

FFmpeg through 9.0.2 contains an integer overflow vulnerability in the OpenEXR decoder's decode_frame() in libavcodec/exr.c that miscalculates the tiled block count. Attackers can supply a crafted tiled OpenEXR image with large dimensions and small tiles to skip decoding and disclose uninitialized memory or stale frame data.

Memory safety

FFmpeg through 9.0.2 contains an information disclosure vulnerability in the VMnc decoder in libavcodec/vmnc.c that restores an unwritten cursor save area into later frames. Attackers can supply a crafted VMnc stream where a frame fails after loading a cursor shape, causing uninitialized heap bytes from screendta to appear in decoded output.

Memory safety

FFmpeg through 9.0.2 contains an information disclosure vulnerability in ff_mlz_decompression() in libavcodec/mlz.c, which accepts string codes beyond the dictionary's next free code. Attackers can supply a crafted MPEG-4 ALS file with malformed initial or out-of-range codes to copy uninitialized heap bytes into decoded audio output.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized memory disclosure vulnerability in targa_decode_rle() in libavcodec/targa.c caused by missing packet length checks. Attackers can supply truncated RLE-compressed TGA images so decoded frames contain leftover heap buffer data and replicated uninitialized stack bytes.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized memory disclosure vulnerability in libavcodec/tiertexseqv.c that allows attackers to leak stack contents by triggering failed RLE unpacking. Attackers can supply a crafted Tiertex SEQ file so seq_decode_op1() copies an uninitialized 64-byte stack block into decoded frames, exposing stack memory in visible output.

Memory safety

FFmpeg through 9.0.2 contains an uninitialized memory disclosure vulnerability in raw_decode() in libavcodec/rawdec.c that fails to verify packet stride covers a full row for 1 to 8 bits per pixel video. Attackers can supply crafted AVI, MOV or NUT files with truncated low bit depth rawvideo packets to leak uninitialized heap memory through decoded frame output.

Memory safety

stable-diffusion.cpp is a pure C/C++ library for running diffusion model (Stable Diffusion, Flux, Wan, Qwen Image, Z-Image, and more) inference. From master-742-1a13107 until master-763-4fcc6fe, IMatrixCollector::load_imatrix in src/runtime/imatrix.cpp reuses the values and counts vectors for entries with the same tensor name but resizes them only when they are empty. An untrusted .imatrix file can provide a small nval for the first entry and a larger nval for a second entry with the same name, causing the accumulation loop to write attacker-controlled floating-point and integer values beyond both heap buffers. Loading the file through the --imat-in conversion path can corrupt the heap or crash the process before a model is required. This issue is fixed in master-763-4fcc6fe.

Memory safety Library

An out-of-bounds read was addressed with improved bounds checking. This issue is fixed in macOS Golden Gate 27.2. An app may be able to cause unexpected system termination.

Memory safety OS & kernel Apple macOS

RPly through 1.1.4 contains an improper array index validation vulnerability in ply_read_header_property() that accepts list as a list property's length or value type. Attackers can supply a crafted PLY file that makes ply_read_list_property() call an out-of-bounds handler, crashing ASCII parsing or corrupting memory in binary parsing.

Memory safety

Netpbm through 11.16.00 contains a heap out-of-bounds write vulnerability in the rasttopnm Sun rasterfile run-length decoder pr_load_image in converter/other/rast.c. Attackers can supply a byte-encoded rasterfile with tiny dimensions and a long run-length body to write controlled bytes past md_image, causing memory corruption and crashes.

Memory safety

Schism Tracker through 20260524 contains a heap out-of-bounds write vulnerability in fmt_ult_load_song() in fmt/ult.c that fails to clamp the one-byte sample count to MAX_SAMPLES. Attackers can supply a crafted .ult module that overwrites the instruments pointer array, causing csf_destroy() to free attacker-chosen pointers and crash the application.

Memory safety Library

smpeg through 2.0.0 contains a heap out-of-bounds read vulnerability in MPEGsystem::FillBuffer() that allows attackers to read past the 16384-byte buffer by supplying a crafted system header stream table. Attackers can provide an MPEG file whose stream table bytes all set the high bit, which crashes applications opening it through SMPEG_new, SMPEG_new_rwops, or SMPEG_new_data.

Memory safety

smpeg through 2.0.0 contains an integer underflow vulnerability in stream_header() in MPEGsystem.cpp that allows attackers to wrap packet_size by declaring an undersized packet length. Attackers can supply a crafted MPEG file opened via SMPEG_new, SMPEG_new_rwops or SMPEG_new_data to trigger a reachable assertion or out-of-bounds copy, crashing the application.

Memory safety

LibRaw through 0.22.2 contains a heap-based out-of-bounds read in LibRaw::parseNikonMakernote() that allows attackers to read past a short ShotInfo buffer. Attackers can supply a crafted Nikon NEF/TIFF with a 5-byte ShotInfo tag declaring version 243 to crash applications decoding the image.

Memory safety Library

Game Music Emu (libgme) through 0.6.5 contains a global out-of-bounds read vulnerability in Classic_Emu::mute_voices_ that allows attackers to crash applications by supplying NSF files whose voice count exceeds the 16-entry voice-type table. Attackers can craft an NSF header with chip_flags 0x3F, opened through gme_open_data or gme_open_file, to read past the static array during loading and cause denial of service.

Memory safety

libTiMidity through 0.2.8 contains a stack out-of-bounds write vulnerability in the SoundFont loader that allows attackers to corrupt stack memory via unchecked generator operators in parse_gen. Attackers can supply a crafted .sf2 file with out-of-range pgen/igen operator values to write chosen 16-bit values near the stack Layer object, causing crashes and stack memory corruption.

Memory safety Library

libTiMidity through 0.2.8 contains an out-of-bounds access vulnerability in init_soundfont() in src/sndfont.c that allows attackers to corrupt memory by supplying unvalidated SF2 preset header bank and preset values. Attackers can craft a .sf2 file whose bank or preset indexes past the 128-entry tonebank or drumset arrays, causing heap memory reads, writes, and application crashes.

Memory safety