CVE-2026-72486

The CVE-2026-72486 entry concerns the Linux kernel mailbox for the mtk-adsp (SOF audio) where a use-after-free (UAF) can occur during device teardown. When the SOF driver fails to initialize (e.g., firmware boot timeout), devres unwinds and frees snd_sof_dev, while the mtk_adsp_mailbox shutdown c...

CVE-2026-72484

The CVE-2026-72484 entry documents a Linux kernel staging component (staging: most: video)double-free during video device registration. The root cause: comp_register_videodev() may release mdev->vdev via vdev->release() if video_register_device() fails, risking a second free when __video_re...

CVE-2026-72482

Summary of CVE-2026-72482 : In the Linux kernel’s gpib subsystem, a descriptor_busy reference counter used to pin gpib_descriptor across IO ioctl operations could be decremented twice in the command_ioctl() error path, causing the counter to underflow to -1 and enabling use-after-free on an activ...

CVE-2026-72483

In the Linux kernel, the max3421 USB hub control path (max3421_hub_control) improperly shifted port_status by 1 <= 32, this shift can cause shift-out-of-bounds behavior, leading to arbitrary bit corruption of port_status and potential bypass of internal checks. The fix rejects requests whose v...

CVE-2026-72481

The CVE-2026-72481 issue affects the Linux kernel AK8975 magnetometer driver under iio. It describes four instances where uninitialized kernel stack memory could be leaked when i2c_smbus_read_i2c_block_data_or_emulated() returns a value smaller than the buffer size. The fixed code adds validation...

CVE-2026-72479

CVE-2026-72479 affects the Linux kernel IIO MMA8452 accelerometer driver. The issue occurs in mma8452_read() when an I2C read call to i2c_smbus_read_i2c_block_data() fails but mma8452_set_runtime_pm_state() succeeds; in that path, mma8452_read() returns 0, causing mma8452_read_raw() to use an uni...

CVE-2026-72480

CVE-2026-72480 : In the Linux kernel, the iio: adc: xilinx-ams component had an out-of-bounds channel lookup in event handling. The function ams_event_to_channel() could return a pointer past the end of dev->channels when no matching scan_index was found, risking invalid memory access in ams_h...

CVE-2026-72477

The CVE-2026-72477 entry concerns the Linux kernel (fs/ntfs3) where a failure to rename could trigger an inconsistent inode state. The root issue was a WARN_ON path; it has been replaced with a call to _ntfs_bad_inode(), preventing further operations on the affected inode. This change appears to ...

CVE-2026-72478

The CVE describes a bound-checking fix in Linux kernel ntfs3 (fs/ntfs3) for run_get_highest_vcn(), which previously parsed a packed NTFS mapping-pairs buffer without length bounds and could read past the slab allocation, triggering a KASAN slab-out-of-bounds read on mount. The patch adds a run_bu...

CVE-2026-72476

CVE-2026-72476 affects the Linux kernel, specifically the dmaengine path. The issue arises in dma_release_channel() where a subsequent dma_chan_put() may call dma_device_put() and release the last reference to the DMA device if the provider is already gone, potentially freeing the device prematur...

CVE-2026-72475

The CVE-2026-72475 impact is in the Linux kernel DMA engine, specifically the dma-axi-dmac path. The issue arises from mismanagement of descriptors: a struct axi_dmac_desc is not fully freed along the failure path in axi_dmac_prep_peripheral_dma_vec() when calling axi_dmac_alloc_desc(). The mitig...

CVE-2026-72474

Summary (concrete details available) : CVE-2026-72474 concerns the Linux kernel’s dmaengine, specifically the dma-axi-dmac path. The issue arises on architectures such as Microblaze or arm64 where DMA_DIRECT_REMAP can cause dma_alloc_coherent() to remap memory (including vmalloc()). In the affect...

CVE-2026-72473

Affected software: Linux kernel (xprtrdma RPC path). Vulnerability: a race/lifecycle issue in rl_kref and send-side references within rpcrdma/xprt_rdma code caused improper freeing of request buffers while a DMA operation could still be active. Root cause: rl_kref previously gated two lifetimes w...

CVE-2026-72472

CVE-2026-72472 – Linux kernel NFS file lock traversal protection fix . The issue arises in the NFS portion of the Linux kernel where releasing flc_lock during file lock list traversal could trigger use-after-free (UAF). The mitigation switches to protecting the traversal with nfsi->rwsem, appl...

CVE-2026-72471

CVE-2026-72471: In the Linux kernel (ntfs3) vulnerability, a path where target VCN isn’t found could lead run_lookup() to false, causing a chain that results in attr_data_get_block() receiving a NULL and finally exit without initializing lcn. The fix prevents this by moving the clen check in ni_s...

CVE-2026-72470

The CVE-2026-72470 vulnerability affects the Linux kernel fs/ntfs3 sub-system. log_replay() initially allocates log->one_page_buf using the host’s PAGE_SIZE, but when an on-disk restart area specifies a larger page size, log->page_size is updated, while one_page_buf may remain undersized. S...

CVE-2026-72469

CVE-2026-72469 affects the Linux kernel’s xprtrdma/rpcrdma path, where an ADDR_CHANGE event can cause an ep kref underflow because rpcrdma_ep_put() would run without a matching rpcrdma_ep_get() when ESTABLISHED hasn’t occurred yet. The documents describe the lifecycle sequence, showing a mismatch...

CVE-2026-72468

Summary: CVE-2026-72468 affects the Linux kernel components xprtrdma/rpcrdma. A race between NFS/RDMA mount and RDMA device removal could trigger a NULL dereference in the kernel trace path. Affected context (from provided docs): rpcrdma_ep_create() and rpcrdma_create_id() interactions allow the ...

CVE-2026-72467

CVE-2026-72467 (Linux kernel, xprtrdma) : The vulnerability arises when frwr_wp_create() fails during connect and rpcrdma_xprt_connect() ignores the error, causing xprt_rdma_connect_worker() to incorrectly report success and set XPRT_CONNECTED. A subsequent NFS/RDMA read with a non‑4‑byte‑aligned...

CVE-2026-72466

Summary of CVE-2026-72466 : In the Linux kernel’s xprtrdma implementation, two path issues in rpcrdma_is_bcall() can leak the receive buffer and drain the Receive queue when decoding a reply. The speculative zero-byte peek uses xdr_inline_decode() without respecting xdr->end, allowing five sub...