Vendredi 11 septembre 2026

Linux

In the Linux kernel, the following vulnerability has been resolved: HID: mcp2221: clear rxbuf after I2C/SMBus transfer completes mcp_i2c_smbus_read() stores the caller-supplied buffer pointer in mcp->rxbuf for the duration of a transfer but never clears it when the transfer finishes or times out. Once the caller frees or reuses the buffer, mcp->rxbuf becomes a dangling pointer. A delayed or spurious MCP2221_I2C_GET_DATA report can then drive mcp2221_raw_event() to memcpy device data into the freed memory, causing a write use-after-free. Route all return paths through a single exit point that clears mcp->rxbuf and mcp->rxbuf_size, so that the existing !mcp->rxbuf guard in the raw_event handler can reject any report arriving after the transfer has ended.

Linux

In the Linux kernel, the following vulnerability has been resolved: HID: intel-thc-hid: intel-quickspi: validate report size before copy write_cmd_to_txdma() builds an output report in qsdev->report_buf, a heap buffer allocated in quickspi_alloc_report_buf() to the device-descriptor derived max_report_len (a few hundred bytes for a touch controller). It copies the caller-supplied report into that buffer: memcpy(write_buf->content, report_buf, report_buf_len); The HID core caps a report at HID_MAX_BUFFER_SIZE (16384) by default, and quickspi_hid_ll_driver does not set max_buffer_size, so the length reaches the driver unbounded. A hidraw SET_REPORT/SET_FEATURE ioctl carrying a report larger than max_report_len therefore overflows report_buf with attacker-controlled length and content. Record the report_buf allocation size and reject reports that do not fit before copying, matching the equivalent guard in the intel-quicki2c sibling (quicki2c_init_write_buf()) and the hid-goodix-spi fix. write_cmd_to_txdma() writes the output report header ahead of the content in the same buffer, so size the allocation to cover the header as well. That keeps the added bound from rejecting a maximum-sized report.

Linux

In the Linux kernel, the following vulnerability has been resolved: HID: intel-thc-hid: intel-quickspi: bound GET_REPORT response to the caller buffer quickspi_hid_raw_request() receives the caller's buffer length in len, but quickspi_get_report() never sees it and copies the whole device-supplied response into buf regardless: memcpy(buf, qsdev->report_buf, qsdev->report_len); qsdev->report_len comes from the input report the touch controller returns, while buf is sized to whatever the caller asked hidraw for through HIDIOCGFEATURE or HIDIOCGINPUT. A response larger than that overflows buf with device-controlled content. The intel-quicki2c sibling already passes the caller length down to quicki2c_get_report() and validates the response against it before the copy. Do the same here.

Linux

In the Linux kernel, the following vulnerability has been resolved: fs/ntfs3: validate dirty page table on log replay Each DIR_PAGE_ENTRY ends in a page_lcns[] array whose length is the on-disk lcns_follow field. check_rstbl() validates the table bookkeeping but never checks that this array fits in the entry, so a crafted lcns_follow lets the v0->v1 conversion memmove and later replay passes run off the entry. Add check_dp_table() to reject, right after check_rstbl(), any entry larger than its size claims via struct_size() (the same expression used to allocate these entries, so the check is overflow-safe by construction). All consumers can then trust lcns_follow as the real capacity. This covers every page_lcns[] access whose index is bounded by the entry itself (the conversion memmove, the HotFix store via find_dp(), and the self-bounded scan loops). Accesses whose index comes from the log record need a separate bound and are handled in a follow-up patch.

Linux

In the Linux kernel, the following vulnerability has been resolved: fs/ntfs3: fix info-leak on partial LZNT decompress in ni_read_frame() ni_read_frame() decompresses an LZNT $DATA frame into the vmapped target pages and then trusts decompress_lznt()'s return value: unc_size = decompress_lznt(frame_ondisk, ondisk_size, frame_mem, frame_size); if ((ssize_t)unc_size < 0) err = unc_size; else if (!unc_size || unc_size > frame_size) err = -EINVAL; decompress_lznt() stops as soon as the compressed stream is exhausted (e.g. a zero chunk header) and returns the number of bytes it actually wrote, which may be far less than frame_size. The bytes between unc_size and frame_size are never written. The only memset() that follows zeroes the region beyond i_valid; when the frame lies entirely within the file's valid size that memset() does not run, so the gap retains whatever was in the just-vmapped pages. All pages are then marked uptodate and returned to userspace, disclosing uninitialized (recently-freed) kernel page memory. A crafted compressed file whose stream decompresses to only a few bytes leaks the remainder of every frame on a plain read(2), which is enough to recover kernel pointers and defeat KASLR. Zero the [unc_size, frame_size) tail immediately after a successful LZNT decompress so the remainder reads back as zero.

Linux

In the Linux kernel, the following vulnerability has been resolved: fs/ntfs3: bound page_lcns[] index by the log record The copy_lcns loop and the redo shorten loop index page_lcns[] at j + i, where i runs up to the log record's lcns_follow. That count is checked only against the record's own length, not the target entry, so check_dp_table() (which validates the entry's lcns_follow) does not cover it: the copy_lcns entry may even be freshly allocated after that check, and find_dp() bounds j but not i. A crafted record thus overflows page_lcns[] of an otherwise valid entry. Add dp_range_ok() and reject, before each loop, any record whose run does not fit the entry. These are the only two page_lcns[] accesses indexed by the record rather than the entry, so together with the entry validation every access is now bounded. [almaz.alexandrovich@paragon-software.com: original patch contained changes to the problem already handled, applied partly]

Linux

In the Linux kernel, the following vulnerability has been resolved: ecryptfs: hold msg ctx list lock when cleaning daemon queue ecryptfs_exorcise_daemon() drops queued messages from a dying daemon without holding ecryptfs_msg_ctx_lists_mux, but ecryptfs_msg_ctx_alloc_to_free() requires that lock. Take the list lock while moving the queued contexts back to the free list to avoid racing with other global msg ctx list users.

Linux

In the Linux kernel, the following vulnerability has been resolved: ecryptfs: pass packet set buffer size to parser ecryptfs_parse_packet_set() receives a pointer into the file header, but it calculates the remaining packet buffer size from PAGE_SIZE - 8. For version 1 headers the packet set starts later in the header, so this can overstate the available buffer. Pass the actual packet set buffer length from the caller and calculate per-packet limits from the remaining bytes in that buffer. Recompute the remaining length after consuming a tag 3 packet before parsing the following tag 11 packet.

Linux

In the Linux kernel, the following vulnerability has been resolved: ecryptfs: reject oversized encrypted_key_size in parse_tag_3_packet parse_tag_3_packet() set encrypted_key_size from the Tag 3 packet body without bounding it against ECRYPTFS_MAX_KEY_BYTES (64). When encrypted_key_size > 64, decrypt_passphrase_encrypted_session_key() sets decrypted_key_size = encrypted_key_size and performs two out-of-bounds writes: 1. crypto_skcipher_decrypt() writes encrypted_key_size bytes into decrypted_key[64] via scatterlist, overflowing into the parent ecryptfs_auth_tok struct. 2. memcpy(crypt_stat->key, decrypted_key, decrypted_key_size) writes into crypt_stat->key[64], corrupting root_iv, keysig_list, and mutexes in ecryptfs_crypt_stat. Only AES-192 (cipher code 0x08) enables this because it sets crypt_stat->key_size = 24 independently of encrypted_key_size, allowing crypto_skcipher_setkey() to succeed while encrypted_key_size exceeds ECRYPTFS_MAX_KEY_BYTES. The PKI decryption path (parse_tag_65_packet) already validates decrypted_key_size <= ECRYPTFS_MAX_KEY_BYTES; the passphrase path omits this check. Bound encrypted_key_size against ECRYPTFS_MAX_KEY_BYTES (64) rather than ECRYPTFS_MAX_ENCRYPTED_KEY_BYTES (512). The 64-byte limit also protects the 512-byte encrypted_key[] buffer, so the former 512-byte check is removed as redundant. [tyhicks: Adjust the code comment to refer to macros representing the buffer sizes rather than mentioning the buffer size values since they may change in the future]

Linux

In the Linux kernel, the following vulnerability has been resolved: ecryptfs: reject too-small tag 70 packets ecryptfs_parse_tag_70_packet() subtracts fixed metadata fields from the parsed packet body size to derive the encrypted filename size. A malformed packet with a body smaller than those fixed fields can underflow that size calculation. Reject tag 70 packets before the subtraction unless the body contains the signature, cipher code, and at least one byte of encrypted filename data.

Linux

In the Linux kernel, the following vulnerability has been resolved: ecryptfs: release message context on send failure ecryptfs_send_message_locked() moves a message context from the free list to the allocated list before sending the request to the userspace daemon. If ecryptfs_send_miscdev() fails, the context is left on the allocated list and cannot be reused. Move it back to the free list on failure and clear the caller's pointer.

Linux

In the Linux kernel, the following vulnerability has been resolved: erofs: skip sufficiently large global buffers when resizing z_erofs_gbuf_nrpages is advanced only after every global buffer has been grown. If a resize fails after some buffers were enlarged, a retry revisits those enlarged buffers. Retrying the same size then returns -ENOMEM because alloc_pages_bulk() has no pages to add and the unchanged return value is treated as a failure. Retrying an intermediate size allocates a temporary pointer array smaller than gbuf->nrpages and copies more existing pointers than the array can hold. Skip buffers that already satisfy the request. Once all remaining buffers have caught up, advancing z_erofs_gbuf_nrpages again describes the guaranteed minimum size across the pool.

Linux

In the Linux kernel, the following vulnerability has been resolved: ext2: Fix lost inode updates for IS_SYNC inodes ext2_setsize() and ext2_xattr_set2() had a construct like: if (IS_SYNC(inode)) { sync_inode_metadata(inode, 1); } else { mark_inode_dirty(inode); } which leads to lost inode updates for IS_SYNC inodes because sync_inode_metadata() does anything only if the inode is already dirty and hence inode updates may be simply lost. Fix the problem by unconditionally marking the inode dirty and *then* call sync_inode_metadata().

Linux

In the Linux kernel, the following vulnerability has been resolved: fanotify: fix use-after-free of file range info fsnotify_pre_content() builds its file_range on the triggering task's stack. fanotify_alloc_perm_event() saves a pointer to range.pos in the heap-allocated permission event so copy_range_info_to_user() can report the offset later. The event reader can set the event state to FAN_EVENT_REPORTED and then sleep while preparing the file descriptor. If a signal interrupts the triggering task at that point, fanotify_get_response() changes the state to FAN_EVENT_CANCELED and returns. This unwinds the file_range stack frame while the reader still owns the event. The reader then dereferences pevent->ppos and copies the stale stack value to userspace. KASAN reported: BUG: KASAN: use-after-free in fanotify_read+0x293e/0x2970 Read of size 8 at addr ffff88811434fc50 by task fanotify_inotif/95 Call Trace: fanotify_read+0x293e/0x2970 vfs_read+0x177/0xa20 ksys_read+0xf7/0x1c0 do_syscall_64+0xf9/0x540 entry_SYSCALL_64_after_hwframe+0x77/0x7f Store the range position directly in the permission event and use FANOTIFY_NO_RANGE when range information is unavailable. The event remains alive until the reader finishes, so the reported offset no longer depends on the triggering task's stack.

Linux

In the Linux kernel, the following vulnerability has been resolved: fbdev: omapfb: panel-dsi-cm: initialize lock before registering display dsicm_probe() registers the display before initializing ddata->lock. Once omapdss_register_display() publishes the display, another consumer can reach a dsicm callback that takes this mutex while it is still uninitialized. Initialize the mutex before registering the display so the published callbacks always see a valid lock.

Linux

In the Linux kernel, the following vulnerability has been resolved: fbdev: uvesafb: unregister connector callback on init failure uvesafb_init() registers the v86d connector callback before registering the platform driver. If platform_driver_register() fails, the function returns the error directly and leaves the connector callback registered. The later platform-device failure path already unregisters the callback. Add the same cleanup before the final return when platform-driver registration fails. This issue was identified during our ongoing static-analysis research while reviewing kernel code.

Linux

In the Linux kernel, the following vulnerability has been resolved: forcedeth: fix off-by-one when saving/restoring non-PCI config space nv_suspend() and nv_resume() walk the non-PCI configuration space with for (i = 0; i <= np->register_size/sizeof(u32); i++) which runs one iteration too many. saved_config_space is declared as u32 saved_config_space[NV_PCI_REGSZ_MAX/4]; and NV_PCI_REGSZ_VER3 is equal to NV_PCI_REGSZ_MAX (0x604), so on a VER3 device register_size/sizeof(u32) is exactly the array length and the last iteration addresses one element past the end. The element it lands on is np->name_rx[0..3]: saved_config_space[] is followed immediately by char name_rx[IFNAMSIZ + 3], and char needs no padding. Nothing observable is corrupted by that, because nv_request_irq() rewrites name_rx with sprintf() before it is ever passed to request_irq(). The bug is the out-of-bounds access itself, which UBSAN reports and which CONFIG_UBSAN_TRAP=y turns into a trap that aborts the running kernel code, plus an MMIO read and, on resume, an MMIO writel() to base + 0x604, one dword past the range the driver mapped: np->base = ioremap(addr, np->register_size); VER1 and VER2 devices stay inside the array, but they too get the stray read and the stray write one dword past their own window. Caught by UBSAN on an Apple Macmini3,1 (MCP79) during a deep S3 cycle. The splat below is trimmed: the build path in the file name, the CPU and taint lines, the Workqueue line, the "?" hint frames, and the frames below device_suspend are all cut. The kernel was tainted, with an out-of-tree nouveau and CPU_OUT_OF_SPEC; forcedeth itself was the stock module. UBSAN: array-index-out-of-bounds in drivers/net/ethernet/nvidia/forcedeth.c:6225:25 index 385 is out of range for type 'u32 [385]' Call Trace: dump_stack_lvl+0x5d/0x80 ubsan_epilogue+0x5/0x2b __ubsan_handle_out_of_bounds.cold+0x54/0x59 __this_module+0xe398c/0xe9010 [forcedeth] pci_pm_suspend+0x80/0x170 dpm_run_callback+0x51/0x160 device_suspend+0x1a2/0x4a0 ... Both loops are hit. UBSAN reports each source location only once per module load (__ubsan_handle_out_of_bounds() calls suppress_report(), which does test_and_set_bit(REPORTED_BIT, ...) on the struct source_location), so the two splats land in the first S3 cycle after the module is loaded and later cycles are silent even though the access still runs off the end every time. In that first cycle line 6225 is reported from pci_pm_suspend and line 6240 from pci_pm_resume. The same off-by-one was fixed in nv_get_regs() by commit ba9aa134287f ("forcedeth: fix buffer overflow") in 2012; these two loops were missed. The suspend and resume side was reported on LKML in September 2013 by Marc Weber, with the same analysis and the same one-character fix, but the patch was attached rather than sent inline and the thread ended there. Use < instead of <=, which saves and restores exactly register_size bytes.

Linux

In the Linux kernel, the following vulnerability has been resolved: hsi: omap_ssi_core: fix missing DMA mask setup for SSI controller device The OMAP SSI driver uses a synthetic HSI controller device allocated via hsi_alloc_controller(), which does not go through the normal OF/platform device initialization path. As a result, the embedded struct device does not have a DMA mask initialized by default. After recent DMA API hardening changes, dma_map_sg() and related helpers now require a valid dma_mask to be present, otherwise the driver may crash or trigger warnings when attempting DMA mapping operations. Fix this by explicitly initializing the DMA mask for the SSI controller device and setting a 32-bit DMA mask, which matches the hardware capabilities.

Linux

In the Linux kernel, the following vulnerability has been resolved: hugetlb: only adjust reservation during unmapping if mapcount is 0 Since df7a6d1f6405, __unmap_hugepage_range can adjust reservations. In the case of folio mapped in both a parent and a child, if the parent unmaps the range first, the reservation adjustment will result in an underflow of the reserved count. Once the child unmaps the range, the count is restored. Change __unmap_hugepage_range() to check the mapcount before adjusting the reservation.

Linux

In the Linux kernel, the following vulnerability has been resolved: ACPI: APEI: GHES: fix ARM section length accounting after header In ghes_handle_arm_hw_error(), after skipping the cper_sec_proc_arm header with (err + 1), the remaining length was reduced by sizeof(err) (pointer size) instead of sizeof(*err) (structure size). That overestimates the bytes left for cper_arm_err_info records and can let the parser read past the CPER section when err_info_num is large enough relative to error_data_length. Use sizeof(*err) so the length accounting matches the pointer advance and the earlier sizeof(*err) size check.

Linux

In the Linux kernel, the following vulnerability has been resolved: ACPI: pfr_update: fix stack buffer overflow in query_capability() query_capability() copies four ACPI buffer objects returned by the firmware _DSM into fixed-size u8[16] fields in struct pfru_update_cap_info using memcpy with the firmware-supplied length: memcpy(&cap_hdr->code_type, elements[CAP_CODE_TYPE_IDX].buffer.pointer, elements[CAP_CODE_TYPE_IDX].buffer.length); The same pattern repeats for drv_type, platform_id, and oem_id. If the firmware returns buffer.length > 16 for any of these fields, memcpy writes past the destination array. struct pfru_update_cap_info is stack-allocated in pfru_ioctl(). Confirmed with KASAN on 7.2-rc6: three stack-out-of-bounds reports are generated when a DSM returns 64-byte buffers, with writes reaching 44 bytes past the end of cap_hdr's [64, 156) frame window into adjacent stack redzones. Introduce a helper pointer to out_obj->package.elements and use it to validate each buffer length against its destination field size before copying, returning -EINVAL if the firmware supplies an oversized buffer.

Linux

In the Linux kernel, the following vulnerability has been resolved: ata: libata-scsi: fix DSM TRIM for sector sizes larger than 2048 bytes ata_scsi_write_same_xlat() translates a SCSI WRITE SAME command with the UNMAP bit set into an ATA DATA SET MANAGEMENT TRIM command. The TRIM descriptor is built by ata_format_dsm_trim_descr() into the 2048-byte ata_scsi_rbuf staging buffer, and the number of bytes copied is compared against the logical sector size by the caller: size = ata_format_dsm_trim_descr(scmd, trmax, block, n_block); if (size != len) /* len == sdp->sector_size */ goto invalid_param_len; ata_format_dsm_trim_descr() clamps the copy length to ATA_SCSI_RBUF_SIZE (2048). On a device whose logical sector size exceeds that (e.g. a 4Kn device, where sector_size == 4096) the function can never return more than 2048, while the caller expects it to return sector_size. The comparison therefore always fails, so every TRIM is rejected with "Parameter list length error" and WARN_ON() splats on each attempt. TRIM / discard is thus completely broken on such devices. The descriptor was incorrectly sized from the logical sector size. A DSM TRIM payload is a list of 512-byte pages, each holding up to ATA_MAX_TRIM_RNUM (64) LBA Range Entries, and is independent of the logical sector size. The Block Limits VPD page already advertises a single such page as the maximum WRITE SAME length (65535 * ATA_MAX_TRIM_RNUM logical blocks), so the block layer never sends a request that needs more than one page. Emit exactly one 512-byte page, independent of the logical sector size, and transfer only that page (COUNT == 1). For a 512-byte-sector device this is unchanged; devices with larger logical sectors now work instead of failing every TRIM.

Linux

In the Linux kernel, the following vulnerability has been resolved: auxdisplay: charlcd: cancel backlight work on registration failure With CONFIG_CHARLCD_BL_FLASH, charlcd_init() schedules bl_work before charlcd_register() calls misc_register(). If registration fails, the caller frees the charlcd object while delayed work still contains its address. Add charlcd_deinit() to cancel the delayed work and turn the backlight off. Use it for both registration rollback and normal unregistration.

Linux

In the Linux kernel, the following vulnerability has been resolved: block: validate user space vectors during extraction The bio-based drivers don't necessarily check the alignment split, and stacking block drivers don't always handle a misalignment detected after submitting the bio. Validate user vectors against the device's dma_alignment as the bio is built from the iov_iter, rejecting misaligned early with -EINVAL.

Linux

In the Linux kernel, the following vulnerability has been resolved: Bluetooth: eir: Fix OOB read in eir_get_service_data() eir_get_service_data() walks the advertising data for a Service Data field with a matching UUID. On a mismatch it advances: eir += dlen; eir_len -= dlen; eir_get_data() reports dlen as the field's data length, but the field spans dlen + 2 bytes once its length and type bytes count, and more when non-Service-Data fields were skipped to reach it. The pointer lands correctly on the next field. eir_len does not, and the shortfall compounds across fields until eir_get_data() reads the length and type bytes of a "field" past the end of the buffer. For an ISO broadcast sink that buffer is hcon->le_per_adv_data[], filled from the periodic advertising reports of a remote broadcaster. A PA payload packed with mismatching Service Data fields walks off the array into the rest of struct hci_conn. A drifted field that matches the BAA UUID puts those bytes in iso_pi(sk)->base, where user space reads them back with getsockopt(BT_ISO_BASE). Recompute eir_len from the end of the buffer each iteration.

Linux

In the Linux kernel, the following vulnerability has been resolved: bpf, x86: Fix per-CPU address resolution into an extended register The destination of the per-CPU address MOV is encoded in ModRM.reg, which is extended by REX.R, but the REX prefix is built with add_1mod(), which sets REX.B. REX.B extends ModRM.rm and SIB.base, and this instruction addresses memory as disp32 with no base, so the bit has no effect at all and the high register bit is simply lost. Every is_ereg() destination therefore resolves to the wrong register, picking whichever one shares the low three bits: R5 -> RAX R7 -> RBP R8 -> RSI R9 -> RDI With BPF_REG_5, whose reg2hex is 0, the emitted 65 49 03 04 25 <off> add %gs:<off>,%rax adds the per-CPU offset to RAX rather than R8. The destination keeps the unadjusted address and RAX is clobbered, so the program goes on to dereference a pointer that was never made per-CPU: BUG: unable to handle page fault for address: 0000607e386a8894 RIP: bpf_prog_707837aafd2aa9ae_update_percpu_data+0x93/0xc9 Call Trace: __bpf_prog_test_run_raw_tp+0x2dc/0x7d0 __flush_smp_call_function_queue+0x1e9/0xc80 Kernel panic - not syncing: Fatal exception in interrupt R5 is the mildest of the four, aliasing a scratch register and faulting at the store. R7 aliases RBP and would corrupt the frame pointer, R8 and R9 alias the argument registers. Use add_2mod() so the register goes through REX.R, matching how add_2reg() places it in ModRM.reg and how emit_priv_frame_ptr() hardcodes 0x4c for the same instruction with R9. Encodings for the non-extended registers are unchanged. Problem showed up when trying to resurrect BPF_GCC CI (selftests built with BPF_GCC). This has gone unnoticed because clang reloads the address into R1 before each per-CPU access, so the destination is never an extended register. GCC keeps several per-CPU addresses live at once, and test_progs-bpf_gcc panics the kernel in global_percpu_data/init, where the address of a .percpu variable ends up in R5.

Linux

In the Linux kernel, the following vulnerability has been resolved: bpf: Disable preemption in __bpf_get_stack get_perf_callchain() returns a per-CPU perf_callchain_entry buffer and releases its recursion slot via put_callchain_entry() before returning, so nothing keeps the entry reserved while __bpf_get_stack() consumes it below. A preemptible BPF program (e.g. a non-sleepable raw tracepoint program on a PREEMPT kernel, which runs under migrate_disable() but not preempt_disable()) can be scheduled out between obtaining the entry and the copy. Another task scheduled on the same CPU then reuses the same per-CPU buffer and overwrites trace->nr with a larger value. copy_len is then computed from the inflated trace->nr and can exceed the caller's buffer, causing an out-of-bounds write in the memcpy() and in the build_id path. The rcu_read_lock() taken here alone does not prevent this. It is only taken on the may_fault path, and under CONFIG_PREEMPT_RCU it does not disable preemption; it merely keeps perf's callchain buffer array alive (freed via call_rcu()) and does nothing to stop another task from reusing the entry. Disable preemption around obtaining the callchain entry and copying it into the caller's buffer, so the entry cannot be reused underneath us and trace->nr stays bounded by max_depth. Build ID resolution may fault and is therefore deferred until after preemption is re-enabled; by then the instruction pointers have already been copied into buf, so it operates only on that private copy. Note, preempt_disable() also subsumes the buffer-lifetime guarantee the rcu_read_lock() provided, since a preempt-disabled section is an RCU read-side critical section for the callchain buffers' call_rcu() reclaim. [ changed Fixes: commit ]

Linux

In the Linux kernel, the following vulnerability has been resolved: bpf: Harden bloom filter sizing and indexing on 32-bit kernels bloom_map_alloc() has two 32-bit-specific problems when the computed bitmap reaches the U32_MAX fallback case. First, BITS_TO_BYTES(U32_MAX) is evaluated with 32-bit arithmetic. The addition performed by DIV_ROUND_UP wraps, so the map allocates only the fixed-size bloom filter object while keeping bitset_mask == U32_MAX. Subsequent updates can then write past the allocated object. Second, fixing only the allocation size is not sufficient. The bloom hash is a u32, but set_bit() takes a signed long bit number and x86 test_bit() eventually feeds the index to variable_test_bit(long, ...). On 32-bit kernels, hashes in [0x80000000, U32_MAX] therefore become negative bit offsets. x86 bt/bts with a memory operand interpret those offsets relative to the supplied base, so a map with bitset_mask == U32_MAX can read or write before bloom->bitset even after allocating the full 512 MiB bitmap. Keep the U32_MAX fallback, but split each hash into a word pointer and an in-word bit number before calling test_bit() or set_bit(). The bitops argument is then always in [0, BITS_PER_LONG - 1], while BIT_WORD(h) still selects the intended word in the full bitmap. Compute the bitset size from (u64)bitset_mask + 1 before passing the final size to bpf_map_area_alloc(). This fixes the original under-allocation and keeps the allocated storage consistent with the addressable bitset. Exploitation note: local privilege escalation is possible on a 32-bit x86 kernel using the under-allocation bug from a binary with CAP_BPF.

Linux

In the Linux kernel, the following vulnerability has been resolved: dm array: validate array block headers on read array_block_check() validates blocknr and csum and nothing else, while node_check(), next to it, has bounded the structural fields since both were written. dm_array_cursor_next() takes its loop bound from the on-disk nr_entries and element_at() is unguarded pointer arithmetic, so a count larger than the block holds keeps the cursor in one block while the index grows past it and the read walks off the dm-bufio buffer -- dm_cache_load_mappings() drives it once per cache block at activation. Check the header against itself: reject a zero value_size, require max_entries to equal calc_max_entries() for that value_size and block size, and require nr_entries to fit. Equality rather than an upper bound, since a count below the real capacity trips BUG_ON() in fill_ablock() and trim_ablock(). Metadata dm-array writes satisfies all three.

Linux

In the Linux kernel, the following vulnerability has been resolved: dm array: reject an array block whose value size is not the caller's array_block_check() can only compare the header against itself, so a block with value_size 4 and max_entries 1018 is internally consistent and passes. dm-cache keeps two arrays -- mappings at 8 bytes and hints at 4 -- and the roots for both live in the superblock. Point the mappings root at a hint block and __load_mappings() walks it through an info whose value size is 8, so element_at() strides 8 bytes over 4-byte entries and reaches offset 8160 of a 4096-byte block. get_ablock() and __shadow_ablock() are the two places that hold the block and the caller at once. Reject there when the two value sizes disagree. Arrays only ever read their own blocks, so this fires on crafted metadata only.

Linux

In the Linux kernel, the following vulnerability has been resolved: cxl/features: bound fwctl command payload to the input buffer fwctl_cmd_rpc() copies cmd->in_len bytes into inbuf = kvzalloc(cmd->in_len) and passes inbuf and in_len to ->fw_rpc(). The CXL callback cxlctl_fw_rpc() ignores in_len and never checks the user-controlled op_size against it. cxlctl_set_feature() bounds op_size only from below (op_size <= sizeof(feat_in->hdr)) and then reads op_size - sizeof(hdr) bytes from feat_in->feat_data via cxl_set_feature(). With a small in_len and a large op_size the first memcpy() already reads past the kvzalloc(in_len) buffer; the out-of-bounds bytes are placed in the mailbox payload and sent to the device, and a large enough op_size can walk into unmapped memory and oops the kernel. The Get paths pin op_size to a fixed size but likewise read the input struct without checking in_len. Reject, at the single dispatch point, any request whose fixed header plus op_size does not fit in the copied-in buffer. The lower-bound test guards the subtraction and ensures op_size was copied in before it is read.

Linux

In the Linux kernel, the following vulnerability has been resolved: cxl/mce: Make the MCE notifier per-region Flavien Solt reported lifetime issues with the CXL MCE notifier, which can lead to NULL dereferences and use-after-free in the MCE handler. The notifier was registered per memory device and stored in 'struct cxl_memdev_state', even though it only needs the region state (the region's SPA range and its extended linear cache size). Instead of keeping the memory device and endpoint alive, the correct fix is to move the notifier into 'struct cxl_region' and register it from cxl_region_probe() as it should be a per-region notifier. Setup the registration to only happen for regions that have an extended linear cache as that is the only current usage. Remove cxl_port_get_spa_cache_alias() as it is now dead code. [ dj: Update dev_warn() when notifier fails due to kconfig. (Ben) ]

Linux

In the Linux kernel, the following vulnerability has been resolved: Bluetooth: RFCOMM: serialize security confirmation handling rfcomm_security_cfm() looks up a session on session_list and then walks its DLC list without holding rfcomm_mutex. Since RFCOMM session teardown uses rfcomm_mutex, krfcommd can close and free the same session and DLCs concurrently: hci_rx_work krfcommd ----------- --------- rfcomm_session_get() rfcomm_lock() rfcomm_session_close() rfcomm_dlc_unlink() rfcomm_session_del() kfree(s) rfcomm_unlock() walk s->dlcs The callback can then read a freed session list head and touch freed DLCs while updating their flags or timers. Serialize the session lookup and DLC traversal in rfcomm_security_cfm() with rfcomm_mutex. This matches the existing RFCOMM session lifetime rules and prevents concurrent rfcomm_session_del() / rfcomm_dlc_unlink() from tearing the objects down while the callback is using them. KASAN reported: BUG: KASAN: slab-use-after-free in rfcomm_security_cfm+0x41c/0x440 Read of size 8 at addr ffff888111fb3960 by task kworker/u17:1/89 Workqueue: hci0 hci_rx_work Call Trace: rfcomm_security_cfm+0x41c/0x440 hci_encrypt_cfm+0x139/0x590 hci_encrypt_change_evt+0x37b/0xc40 hci_event_packet+0x71b/0xb20 hci_rx_work+0x293/0x730 Allocated by task 69: rfcomm_session_add+0x9e/0x2f0 rfcomm_run+0x44b/0x41e0 Freed by task 69: kfree+0x131/0x3c0 rfcomm_session_del+0x188/0x220 rfcomm_run+0x1985/0x41e0

Linux

In the Linux kernel, the following vulnerability has been resolved: ip: orphan prefetched skbs before multicast forwarding IPv4 and IPv6 input preserve an skb->sk association installed by bpf_sk_assign() so that local delivery can use the selected socket under RCU. Both address families can also prefetch a socket in UDP early demux. In both paths (BPF and UDP early demux) a reference is not guaranteed to be held on the socket. When a multicast packet is not locally deliverable, IPv6 hands the original skb to ip6_mr_input(). IPv4's ip_mr_input() similarly keeps the original skb when local delivery is not needed. Either path can put the skb on an unresolved multicast route queue or forward it after the receive-side RCU section ends. After the prefetched socket is destroyed, a later skb free invokes sock_pfree() and dereferences the stale skb->sk. Orphan the skb before each non-local multicast forwarding path. Local delivery retains the original skb; the existing skb_clone() calls provide multicast forwarding with a socket-free clone.

Linux

In the Linux kernel, the following vulnerability has been resolved: ip6_tunnel: use skb_cow_head() in ip6_tnl_xmit() ip6_tnl_xmit() may need to expand headroom before it can push the outer IPv6 and optional encap headers. It currently does that with skb_realloc_headroom(), copies skb->sk ownership, consumes the original skb, and then continues processing with the replacement skb kept only in its local variable. That is safe only if the helper cannot fail afterwards. But this helper still has post-reallocation error exits. collect_md tunnels reject non-NONE encap after the replacement, and ip6_tnl_encap() can also fail later. In those cases the helper returns an error to its callers while the caller still only has the original skb pointer. Both ip6_tnl_start_xmit() and the IPv6 GRE paths free the caller skb on error, so they can end up freeing an skb that ip6_tnl_xmit() already consumed. Use skb_cow_head() instead. It provides the required headroom and writability without privately replacing the caller-owned skb, so later error returns cannot leave callers with a stale pointer. The Ethernet users, ip6gretap and ip6erspan, clear IFF_TX_SKB_SHARING and already call skb_cow_head() before entering ip6_tnl_xmit(). They do not rely on the removed skb_shared() reallocation. This also makes the IPv6 tunnel path consistent with ip_tunnel_xmit().

Linux

In the Linux kernel, the following vulnerability has been resolved: ip6_gre: fix hardware header length for NBMA tunnels ip6gre_tnl_link_config_route() accumulates the lower device's hardware header length into dev->hard_header_len whenever header_ops is set. This is incorrect for both users of header_ops. ip6gretap and ip6erspan have a fixed Ethernet hardware header length. For an NBMA ip6gre tunnel, ip6gre_header() creates only the GRE header, the optional FOU or GUE header, and the outer IPv6 header. The lower device header is headroom needed later, not part of the tunnel device's hardware header. Keep the lower device header in needed_headroom. Set hard_header_len to the tunnel header length only for ARPHRD_IP6GRE devices with header_ops, and leave the fixed Ethernet header length unchanged for tap and erspan devices.

Linux

In the Linux kernel, the following vulnerability has been resolved: ipv6: rpl: fix NULL dereference of idev in ipv6_rpl_srh_rcv() ipv6_rpl_srh_rcv() dereferences idev from __in6_dev_get() without a NULL check when reading idev->cnf.rpl_seg_enabled. When the device's MTU drops below IPV6_MIN_MTU, addrconf_ifdown() clears dev->ip6_ptr through RCU_INIT_POINTER(). A packet that passed the idev check in ip6_rcv_core() can then reach ipv6_rpl_srh_rcv() with dev->ip6_ptr already NULL. Reproduced by flooding the receiving interface with ping6 traffic while flapping its MTU between 1500 and 1200: BUG: KASAN: null-ptr-deref in ipv6_rpl_srh_rcv+0xb3/0x1070 Read of size 4 at addr 00000000000006b4 by task ping6/394 CPU: 2 UID: 0 PID: 394 Comm: ping6 Not tainted 7.2.0-rc7-micro-vm-dev-00095-g24ef02f934ee #240 PREEMPT(full) Call Trace: <IRQ> kasan_report+0xc6/0x100 ipv6_rpl_srh_rcv+0xb3/0x1070 ip6_protocol_deliver_rcu+0x759/0x9a0 ip6_input_finish+0xa8/0x1b0 ip6_input+0xe1/0x490 ipv6_rcv+0x33d/0x460 __netif_receive_skb_one_core+0xd6/0x130 process_backlog+0x2cc/0xa00 __napi_poll.constprop.0+0x56/0x270 net_rx_action+0x327/0x730 handle_softirqs+0x11e/0x630 do_softirq+0xb3/0xf0 </IRQ> Both ipv6_rpl_srh_rcv() and ipv6_srh_rcv() are called only from ipv6_rthdr_rcv(), which already has an idev lookup. Fix the NULL dereference on the RPL path by checking idev in ipv6_rthdr_rcv(), before it calls either function. The callees take idev as an argument and no longer call __in6_dev_get(), so the packet is now dropped in one place, with SKB_DROP_REASON_IPV6DISABLED on both paths.

Linux

In the Linux kernel, the following vulnerability has been resolved: landlock: Require LANDLOCK_ACCESS_FS_MAKE_REG for whiteout creation Whiteout objects are used in the upper layer of an OverlayFS to indicate that the file with this name does not exist in the unified view, even if it is present in one of the lower layer file systems. For the userspace implementations of OverlayFS (fuse-overlayfs), whiteout objects can be created from userspace as well: * mknod(2) with S_IFCHR and makedev(0, 0) * renameat2(2) with RENAME_WHITEOUT, creating the whiteout in the old place of the moved file. This commit guards whiteout creation in both of these cases with LANDLOCK_ACCESS_FS_MAKE_REG. Whiteout objects are *not* considered character devices and are not bound to a driver. LANDLOCK_ACCESS_FS_MAKE_REG describes the same permission class as a whiteout object: creating one is the only S_IFCHR creation that the VFS exempts from CAP_MKNOD, so it is as unprivileged as creating a regular file, while LANDLOCK_ACCESS_FS_MAKE_CHAR and LANDLOCK_ACCESS_FS_MAKE_BLOCK keep meaning the creation of devices that expose a kernel interface [1]. For the mknod(2) case, introduce a Landlock erratum. The creation of whiteout objects through mknod(2) was previously guarded using LANDLOCK_ACCESS_FS_MAKE_CHAR, and it is now guarded using LANDLOCK_ACCESS_FS_MAKE_REG. For the renameat2(2) case, fix a bug: Before this commit, renameat2(2) with RENAME_WHITEOUT would create a directory entry even when all LANDLOCK_ACCESS_FS_MAKE_* rights were denied. This does not affect normal renames within layered OverlayFS mounts: When doing a regular rename() on a mounted fuse-overlayfs, it is the fuse-overlayfs daemon that exercises renameat2() with RENAME_WHITEOUT, and only the Landlock domain of that daemon is checked there. Depends-on: 49c9e09d9610 ("landlock: Fix handling of disconnected directories") Depends-on: fe72ce6710cb ("landlock: Add errata documentation section") [mic: Record why LANDLOCK_ACCESS_FS_MAKE_REG is the matching right, and add link(2) to the user doc]

Linux

In the Linux kernel, the following vulnerability has been resolved: libnvdimm/labels: Prevent integer overflow in __nd_label_validate() The on-media namespace index field nslot is a u32 read from the DIMM label storage area. __nd_label_validate() bounds it against the config area size, but sizeof_namespace_label() returns unsigned, so the product nslot * label_size is evaluated in 32-bit and wraps modulo 2^32 before the comparison. A crafted nslot passes the bound and is then used as the loop trip count in nd_label_data_init(), whose memset() walks off the end of the config_size buffer: an out-of-bounds write. The field is not trusted -- it comes from the medium, or from userspace via ND_CMD_SET_CONFIG_DATA. Evaluate the product in 64-bit so the bound check is exact; conforming labels are unaffected. The check was safe when introduced by commit 4a826c83db4e ("libnvdimm: namespace indices: read and validate"): it multiplied by sizeof(struct nd_namespace_label), a size_t, so on a 64-bit build the product did not wrap. Commit 564e871aa66f ("libnvdimm, label: add v1.2 nvdimm label definitions") narrowed it to 32 bits when the label size became a runtime value read via sizeof_namespace_label().

Linux

In the Linux kernel, the following vulnerability has been resolved: md: do overflow check for sb->bblog_shift in super_1_load() In super_1_load(), sb->bblog_shift is an __u8 type value loaded from on- disk superblock. It is used for badblocks API badblocks_set() by the following sequence, 1930 rdev->badblocks.shift = sb->bblog_shift; 1931 for (i = 0 ; i < (sectors << (9-3)) ; i++, bbp++) { 1932 u64 bb = le64_to_cpu(*bbp); 1933 int count = bb & (0x3ff); 1934 u64 sector = bb >> 10; 1935 sector <<= sb->bblog_shift; 1936 count <<= sb->bblog_shift; 1937 if (bb + 1 == 0) 1938 break; 1939 if (!badblocks_set(&rdev->badblocks, sector, count, 1)) 1940 return -EINVAL; 1941 } bb->bblog_shit is in range of 0-255, variable sector is 64bit width, for an invalid bb->bblog_shit, it is possible to make sector be overflowed by the following calculation, 1935 sector <<= sb->bblog_shift; Then in turn when call badblocks_set() at line 1939 with the invalid rdev->badblocks.shift set at line 1930, may result an overflow inside _badblocks_clear() in block/badblocks.c. Although there are many places to call badblocks APIs, the non-zero shift value is only used in super_1_load(), other places always use 0 as the shift value. Therefore it is unnecessary to do a general shift value overflow check inside badblock API, and just check here as the caller. This may avoid unnecessary check, make the badblocks API code more simple and elegant.

Linux

In the Linux kernel, the following vulnerability has been resolved: mptcp: fix uninitialized local_id in syncookie MP_JOIN reconstruction mptcp_token_join_cookie_init_state() restores remote_nonce, local_nonce, backup, join_id, token and msk from the saved cookie entry when rebuilding the request socket for a MP_JOIN 4th-ACK handled under SYN cookies, but it does not restore local_id, even though the SYN path saved it. subflow_ulp_clone() then reads that uninitialized field and stores it as the joined subflow's address-ID. Because the request-sock slab is SLAB_TYPESAFE_BY_RCU and not zeroed on allocation, the value is the stale byte of a previously freed request socket, which an off-path peer can influence by sending concurrent MP_JOIN SYNs. This corrupts the path manager's id-based subflow bookkeeping for the connection. Restore subflow_req->local_id from the cookie entry, as done for the other fields.

Linux

In the Linux kernel, the following vulnerability has been resolved: nouveau/gem: reserve the bo in the info ioctl around the vma lookup In the non-uvmm path, there could be a race between the info lookup finding the vma, and the gem close path closing the vma leading to a use-after-free. Spotted with the help of Opus 4.6.

Linux

In the Linux kernel, the following vulnerability has been resolved: sunrpc: route to a populated pool in svc_pool_for_cpu() svc_set_num_threads() spreads the requested threads evenly across the service's pools (base = nrservs / sv_nrpools). When a service runs fewer threads than it has pools -- e.g. an nfsd configured with fewer threads than the host has NUMA nodes while running in "pernode" or "percpu" mode -- the trailing pools are left with no threads at all. svc_xprt_enqueue() selects a pool from the CPU servicing the transport, queues the transport on that pool's sp_xprts, and only wakes a thread from the same pool. Each thread services exclusively its own pool, so a transport that lands on a threadless pool is enqueued on sp_xprts and never picked up: the connection hangs indefinitely. Have svc_pool_for_cpu() skip pools that currently have no threads, falling back to the next populated pool. This trades NUMA locality for a guarantee that the work is actually serviced. sp_nrthreads is only updated under the service mutex; the lockless read here is a best-effort routing hint, so annotate it with data_race().

Linux

In the Linux kernel, the following vulnerability has been resolved: SUNRPC: always drain cache_cleaner before destroying a cache_detail sunrpc_destroy_cache_detail() only cancels the global cache_cleaner delayed_work when cache_list is empty. During per-netns teardown cache_list is never empty because init_net's caches remain registered, so the cancel never fires. After unlink, the caller proceeds to cache_destroy_net() which kfrees the cache_detail while cache_clean() may still hold a dangling pointer to it. The result is a use-after-free: cache_dequeue() takes cd->queue_lock on freed memory, and cache_put() dereferences cd->cache_put as a function pointer from freed slab. Drop the list_empty guard so that cancel_delayed_work_sync() always runs, ensuring any in-flight cache_clean() completes before the cache_detail is freed. Re-arm the cleaner afterwards if other caches are still registered.

Linux

In the Linux kernel, the following vulnerability has been resolved: SUNRPC: Check svc pool percpu counter allocation __svc_create() initializes three per-pool percpu_counter stats and ignores every return value. On SMP, percpu_counter_init() fails when __alloc_percpu_gfp() cannot satisfy the allocation, leaving the failed counter with fbc->counters == NULL and its embedded raw_spinlock_t, list_head, and count never initialized. __svc_create() returns the half-constructed svc_serv to nfsd, lockd, or the NFS callback service anyway. Once that service is live, the hot-path increments in svc_xprt_enqueue(), svc_handle_xprt(), and svc_pool_wake_idle_thread() reach a counter whose backing pointer is NULL. The pointer is a per-cpu offset, so the access does not fault: it resolves to offset zero of the current CPU's per-cpu area and silently corrupts whatever variable lives there. A /proc/fs/nfsd/pool_stats read walks the same NULL per-cpu storage and returns garbage, and on CONFIG_DEBUG_SPINLOCK or lockdep it splats on the never-initialized lock. Creating the broken service requires a percpu allocation failure during RPC server startup, so it is reachable only by a local administrator under memory pressure or fault injection; a remote peer cannot induce the bad state on its own. Check each percpu_counter_init() return value in __svc_create() and fail when an allocation fails, unwinding the counters already set up in the current pool and in every pool initialized before it. A discrete percpu_counter_destroy() per counter at teardown frees each per-cpu allocation exactly once.

Linux

In the Linux kernel, the following vulnerability has been resolved: sunrpc: defer rq_argp and rq_resp free until after RCU grace period svc_rqst_free() frees rqstp->rq_argp and rqstp->rq_resp synchronously via kfree(), but defers the rqstp struct free via kfree_rcu(). After svc_exit_thread() calls list_del_rcu() and svc_rqst_free(), there is a window where RCU readers that started before list_del_rcu() can still traverse the thread list and find the rqstp. These readers (e.g. nfsd_nl_rpc_status_get_dumpit()) dereference rqstp->rq_argp, which has already been freed — a use-after-free. Fix this by moving the kfree of rq_argp and rq_resp into an explicit call_rcu() callback alongside the struct free. Resources not accessed by RCU readers (bvec, buffer pages, scratch folio, auth_data) remain synchronously freed.

Linux

In the Linux kernel, the following vulnerability has been resolved: SUNRPC: fix gssx_dec_option_array error path bugs Four coupled defects in the gssx XDR option-array decoder make the error paths unsafe: a NULL deref in the caller, a refcount leak on the decoded group_info, and a latent use-after-free that the leak fix would otherwise expose. gssx_dec_option_array() sets oa->count = 1 before allocating oa->data. If that allocation fails, -ENOMEM is returned with oa->count == 1 and oa->data == NULL. All other error paths jump to free_oa: which frees oa->data and NULLs it but also leaves oa->count == 1. The caller trusts the count: gssp_accept_sec_context_upcall() gssx_dec_accept_sec_context() gssx_dec_option_array() /* fails, count=1 data=NULL */ data = res.options.data[0].value /* NULL deref */ Independently, free_creds: releases the partially decoded svc_cred with a bare kfree(creds). gssx_dec_linux_creds() installs a groups_alloc() result into creds->cr_group_info; that object is kvmalloc-backed and refcounted, and only put_group_info() reaches kvfree(). A plain kfree(creds) drops the wrapper and leaks the group_info allocation. The natural fix for the leak is to call free_svc_cred(creds) before kfree(creds), but free_svc_cred() invokes put_group_info() on creds->cr_group_info unconditionally when non-NULL. The existing out_free_groups: path in gssx_dec_linux_creds() already called groups_free() on that pointer without clearing it, so once free_svc_cred() is wired in, the subsequent put_group_info() would touch freed memory. Fix all four together: - Move the oa->count = 1 assignment below the oa->data allocation so it is never set when oa->data is NULL. - Reset oa->count to 0 at free_oa: so count and data stay coherent and the caller sees an empty option array. - Call free_svc_cred(creds) before kfree(creds) at free_creds: so the refcounted cr_group_info is released. free_svc_cred() either NULL-guards each field explicitly (cr_group_info has an if() check) or delegates to a helper that is NULL-safe itself (kfree for the string fields, gss_mech_put() which guards with if(gm) at gss_mech_switch.c:342), so it is safe to call on a partially decoded svc_cred where only cr_uid/cr_gid/cr_group_info have been written and everything else is zero from kzalloc. - In gssx_dec_linux_creds()'s out_free_groups: path, release cr_group_info with put_group_info() rather than groups_free() so the teardown matches free_svc_cred()'s refcount-aware path, and clear the pointer so a later free_svc_cred() on the same creds does not release it a second time.

Linux

In the Linux kernel, the following vulnerability has been resolved: sunrpc: init gssp_lock before publishing proc entry create_use_gss_proxy_proc_entry() publishes /proc/net/rpc/use-gss-proxy via proc_create_data() before init_gssp_clnt() runs mutex_init() on sn->gssp_lock. Once the dentry is linked under proc_subdir_lock it is immediately reachable from userspace, so a write that lands in the window drives set_gssp_clnt() into mutex_lock() on a zero-initialized struct mutex. create_use_gss_proxy_proc_entry(net) proc_create_data("use-gss-proxy", ...) /* dentry live */ init_gssp_clnt(sn) mutex_init(&sn->gssp_lock) /* too late */ write_gssp() set_gssp_clnt(net) mutex_lock(&sn->gssp_lock) /* uninitialized */ gssp_rpc_create(...) sn->gssp_clnt = clnt mutex_unlock(&sn->gssp_lock) The window spans only the two statements between proc_create_data() returning and init_gssp_clnt(), so a writer reaches it only if the registering thread is preempted there while another task is already opening the freshly published file. register_pernet_subsys() runs in preemptible context under pernet_ops_rwsem, so that preemption is possible, and the window widens on auth_rpcgss module load, when the proc entry is created for every live net namespace whose tasks are already running. A writer that wins the race locks a zero-filled struct mutex. On CONFIG_DEBUG_MUTEXES the missing magic value trips a "lock used without init" splat; on a production kernel the fast path acquires the lock via CMPXCHG(owner, 0, current). In the latter case a second writer that arrives before init_gssp_clnt() re-zeroes owner can enter set_gssp_clnt() concurrently, shut down the first writer's clnt while it is still in use, and leak the loser's clnt. Fix by initializing sn->gssp_lock in sunrpc_init_net() so its lifetime matches the sunrpc_net it lives in. sn->gssp_clnt is already NULL from the kzalloc that backs net_generic storage, so the lazy helper is no longer needed; drop init_gssp_clnt(), its prototype, and the call from create_use_gss_proxy_proc_entry(). sunrpc.ko is a build-time dependency of auth_rpcgss.ko, so sunrpc_init_net() has always run on every netns before any auth_gss pernet init can publish the proc entry.

Linux

In the Linux kernel, the following vulnerability has been resolved: svcrdma: Reorder rpcrdma_rn_unregister before rdma_destroy_id svc_rdma_free() caches rdma->sc_cm_id->device before teardown, then calls rdma_destroy_id(sc_cm_id) which frees the cm_id. rpcrdma_rn_unregister() follows, but between those two calls the transport's sc_rn entry is still installed in the device's rd_xa. A concurrent ib_unregister_device walk can dispatch svc_rdma_xprt_done() against the now-freed sc_cm_id. Move rpcrdma_rn_unregister() before rdma_destroy_id() so the transport's notification entry is removed from the xarray before the cm_id it references is destroyed. Also guard the sc_cm_id dereference with a NULL check: the following patches introduce paths that reach svc_rdma_free() with sc_cm_id == NULL (listener create failure, ADDR_CHANGE replacement failure).

Linux

In the Linux kernel, the following vulnerability has been resolved: svcrdma: Clear sc_cm_id when ADDR_CHANGE replacement fails When svc_rdma_listen_handler() handles RDMA_CM_EVENT_ADDR_CHANGE, it creates a replacement listener cm_id and returns 1, telling the CM core to destroy the old one. If the replacement allocation fails, sc_cm_id still points at the old cm_id that the CM core is about to destroy. Any subsequent dereference of sc_cm_id -- such as svc_rdma_detach()'s rdma_disconnect() call -- is a use-after-free. NULL sc_cm_id on the failure path and guard svc_rdma_detach()'s rdma_disconnect() call against NULL so that the listener can be torn down safely when the server shuts down.

Linux

In the Linux kernel, the following vulnerability has been resolved: svcrdma: Reject Read lists that exceed the page budget Individual Read segment lengths are validated at decode time, but nothing prevents a requester from sending multiple segments whose cumulative length exceeds the rq_pages array budget. When one segment fills the page array exactly, the runtime guard in svc_rdma_build_read_segment() is bypassed because len reaches zero. A subsequent segment then accesses the NULL sentinel slot at rq_pages[rq_maxpages], resulting in a NULL pointer dereference during DMA mapping. Accumulate pages across all Read segments and reject the message at decode time when the total would overflow the page budget.

Linux

In the Linux kernel, the following vulnerability has been resolved: wifi: ath6kl: clamp assoc request/response lengths before subtracting IE offsets ath6kl_cfg80211_connect_event() subtracts fixed IE offsets from assoc_req_len (-= 4) and assoc_resp_len (-= 6), both u8, with no lower bound. The aggregate check recently added to ath6kl_wmi_connect_event_rx() bounds the declared lengths from above (their sum must fit the received event), but an assoc request/response shorter than its fixed offset still underflows here: the u8 wraps to ~250, and cfg80211_connect_result() / cfg80211_roamed() then treat that wrapped value as the IE length and copy that many bytes out of the small assoc_info buffer to user space via nl80211, disclosing adjacent slab memory. Clamp both lengths to their offsets before subtracting. Found by 0sec (https://0sec.ai) using automated source analysis; the missing lower bound is evident from source. Compile-tested.

Linux

In the Linux kernel, the following vulnerability has been resolved: wifi: mt76: mt7925: cancel pending mlo_pm_work If the device is reset, suspended or unregistered within that window, the pending work can still run and access vif/bss data that may already be freed, or send MCU commands while the firmware is not available. Add cancel_delayed_work_sync(&dev->mlo_pm_work) in all relevant teardown and suspend paths: - mt7925_mac_reset_work() (chip reset recovery) - mt7925e_unregister_device() (PCIe unbind) - mt7925_pci_suspend() (PCIe bus suspend) - mt7925_suspend() (mac80211 suspend) - mt7925u_suspend() (USB bus / runtime suspend) This ensures the work is stopped before the device state becomes invalid.

Linux

In the Linux kernel, the following vulnerability has been resolved: media: staging/ipu7: fix async notifier UAF on probe error path isys_register_devices() registers the V4L2 async notifier via isys_notifier_init(). If a subsequent probe step such as isys_fw_log_init() fails, isys_probe() jumps to the out_cleanup label which only calls isys_unregister_devices(). That helper tears down the video devices, subdevices, V4L2 device and media device, but never unregisters or cleans up the async notifier. As a result the notifier stays chained in the global notifier_list while the enclosing struct ipu7_isys is freed by devres, leading to list corruption and a use-after-free the next time the list is walked. The remove path already does the right thing by calling isys_notifier_cleanup() before isys_unregister_devices(). Mirror that on the probe error path so the notifier is unregistered and cleaned up before the device is torn down.

Linux

In the Linux kernel, the following vulnerability has been resolved: sched/core: Handle pick_task() releasing the rq lock Core scheduling's pick_next_task() breaks when a ->pick_task() implementation can release the rq lock. The selection state derived on entry is only valid while the lock is held continuously. Once a pick can drop the lock, an interleaving selection can invalidate all of it: the single-CPU fast path can commit an uncookied pick although the core went cookied during the release, and forceidle committed by the interleaving selection skews the restarted pass's accounting. Fix it by restarting the whole selection when a pick returns RETRY_TASK after releasing the lock: a single restart point above the state derivation replaces the per-loop restart labels, so a retry picks up state committed by interleaving selections and accounts and resets forceidle like a fresh selection would. need_sync and fi_before latch across retries. Clock validity can't be re-derived - there is no program-ordered way to tell whether the own and core rq clocks are still updated after the lock was released, as other lockers' pin cycles may or may not have invalidated them. When restarting, clear core_clock_updated so that the sibling loop re-updates the core rq, and update the own rq clock if invalidated.

Linux

In the Linux kernel, the following vulnerability has been resolved: sched/core: Make core-sched flips wait for in-flight selections Core scheduling's pick_next_task() operates on all sibling rqs under one acquisition of the shared core-wide lock. A ->pick_task() that releases the rq lock leaves every sibling __lock momentarily free, letting __sched_core_flip(false) complete mid-selection and rebind rq_lockp() under it. The selection resumes on the split locks, touching sibling state it no longer protects, and __schedule() finally releases a lock that was never taken while leaking the one that was. Count in-flight core-wide selections in the leader's rq->core_pick_in_flight and make __sched_core_flip() wait for the count to drain. The count only changes under the shared lock, which the flip holds while sampling, so no other ordering is needed. The wait can repeat while selections overlap, but the flip backs off between samples and flips are rare cookie-lifetime events. sched_core_cpu_deactivate() moves the count to the new leader - a stale copy left behind would bias it forever if that CPU later returns as its own leader.

Linux

In the Linux kernel, the following vulnerability has been resolved: RISC-V: KVM: Fix PMU event info array size overflow SBI PMU EVENT_GET_INFO stores guest-controlled num_events * sizeof(*einfo) in a 32-bit integer. On RV64, num_events = 0x10000001 makes 0x100000010 truncate to 16. KVM then allocates one entry but loops over the original num_events, causing out-of-bounds reads and writes. A nested guest triggered: BUG: KASAN: slab-out-of-bounds in kvm_riscv_vcpu_pmu_event_info+0xa4/0x142 Read of size 4 at addr ff600000074d46b0 by task init/1 Call Trace: [<ffffffff8006471c>] kvm_riscv_vcpu_pmu_event_info+0xa4/0x142 [<ffffffff800690c0>] kvm_sbi_ext_pmu_handler+0xca/0x268 [<ffffffff8006779e>] kvm_riscv_vcpu_sbi_ecall+0xec/0x1e6 [<ffffffff8006008c>] kvm_riscv_vcpu_exit+0x48c/0x540 [<ffffffff8005ea0a>] kvm_arch_vcpu_ioctl_run+0x37e/0xc80 Allocated by task 1: __kmalloc_noprof+0x19e/0x4b0 kvm_riscv_vcpu_pmu_event_info+0x72/0x142 kvm_sbi_ext_pmu_handler+0xca/0x268 kvm_riscv_vcpu_sbi_ecall+0xec/0x1e6 kvm_riscv_vcpu_exit+0x48c/0x540 kvm_arch_vcpu_ioctl_run+0x37e/0xc80 The buggy address is located 0 bytes to the right of allocated 16-byte region [ff600000074d46a0, ff600000074d46b0) Store the shared-memory size in size_t and reject multiplication overflow. Allocate the guest-driven array with GFP_KERNEL_ACCOUNT so it is charged to kmemcg, and use __GFP_NOWARN to suppress allocation failure warnings. Use kvcalloc() to allow vmalloc fallback and an unsigned long loop index to match num_events.

Linux

In the Linux kernel, the following vulnerability has been resolved: qede: Fix NULL pointer dereference in TPA fragment processing Under memory pressure, the qede driver encounters NULL pointer dereferences when processing TPA continuation fragments. Commit 8a8633978b84 ("qede: Add build_skb() support.") accidentally dropped the assignment of tpa_info->buffer.data in qede_tpa_start(). When memory pressure causes an SKB allocation failure in qede_tpa_start(), the driver sets tpa_start_fail = true and attempts to recycle the physical page later in qede_tpa_end() via qede_reuse_page(). However, because buffer.data was left uninitialized (NULL), qede_reuse_page() pushes a "ghost" BD (valid DMA mapping but NULL data pointer) back into the active Rx ring. The next time the hardware uses this ring slot, it passes a NULL page to qede_fill_frag_skb(), causing a kernel panic. Example crash from production system: BUG: unable to handle kernel NULL pointer dereference at 0x8 RIP: qede_fill_frag_skb+0x96/0x430 [qede] Call Trace: qede_rx_int+0xb06/0x1de0 qede_poll+0x2f4/0x6c0 __napi_poll+0x2d/0x130 Fix the root cause by restoring the tpa_info->buffer.data assignment in qede_tpa_start(), ensuring valid pages are correctly tracked and recycled. Additionally, update the stale comment for struct qede_agg_info::buffer to reflect its current usage.

Linux

In the Linux kernel, the following vulnerability has been resolved: RDMA/cxgb4: Cancel reg_work before freeing device on remove c4iw_uld_state_change() queues reg_work to register the RDMA device. c4iw_remove() can free ctx->dev while this work is pending or running, leaving c4iw_register_device() accessing the freed device. Cancel reg_work before removing the device. The registration work can tear down ctx->dev when registration fails, so do not unregister or deallocate it again in that case. This issue was found by an in-house static analysis tool.

Linux

In the Linux kernel, the following vulnerability has been resolved: RDMA/ucma: Lock the handler in ucma_set_ib_path() ucma_set_ib_path() calls ucma_event_handler() straight from the write() path, without the handler lock that keeps ctx->file stable while a uevent is queued. The handler re-reads ctx->file for every dereference: mutex_lock(&ctx->file->mut); /* file A */ list_add_tail(&uevent->list, &ctx->file->event_list); /* file B */ mutex_unlock(&ctx->file->mut); /* file B */ wake_up_interruptible(&ctx->file->poll_wait); /* file B */ A concurrent ucma_migrate_id() reassigns ctx->file while the SET_OPTION caller sleeps in mutex_lock(), so the list_add_tail() lands on file B's event_list while only file A's mutex is held, racing every other user of that list: BUG: KASAN: slab-use-after-free in __list_add_valid_or_report+0x1aa/0x1c0 Read of size 8 at addr ffff888153c6a418 by task poc_corr/486 Call Trace: __list_add_valid_or_report+0x1aa/0x1c0 ucma_event_handler+0x1be/0xc00 ucma_set_ib_path+0x45e/0x710 ucma_set_option+0x32e/0x590 ucma_write+0x1f9/0x330 Allocated by task 505: ucma_write_cm_event+0x1a1/0x660 Freed by task 505: kfree+0x1da/0x4c0 ucma_get_event+0x5d5/0x7e0 The freed object is a ucma_event that another thread dequeued from file B's list under file B's mutex. File A's mut is left held on top of that, wedging its next writer in uninterruptible sleep. This path needs a bound and address-resolved cm_id, so it requires an RDMA device to be present. Take the handler lock around the call.

Linux

In the Linux kernel, the following vulnerability has been resolved: RDMA/ucma: Lock the handler in ucma_write_cm_event() ctx->file may only be changed under the handler lock and the xa_lock, which is what stops uevents being queued for a ctx while ucma_migrate_id() moves it to another file. The CM core takes that lock before invoking ucma_event_handler(), but the write() paths that queue uevents themselves do not. ucma_write_cm_event() re-reads ctx->file for each of its four dereferences, so ucma_migrate_id() can swap it mid-sequence: mutex_lock(&ctx->file->mut); /* file A */ list_add_tail(&uevent->list, &ctx->file->event_list); /* file B */ mutex_unlock(&ctx->file->mut); /* file B */ wake_up_interruptible(&ctx->file->poll_wait); /* file B */ The window is the mutex_lock() itself: the writer sleeps in it while the migration reassigns ctx->file. The list_add_tail() then runs on file B's event_list holding only file A's mutex: list_add corruption. prev->next should be next (ffff888101320f30), but was ffff88814a08c418. (prev=ffff88814a075c18). kernel BUG at lib/list_debug.c:32! Call Trace: ucma_write_cm_event+0x36e/0x5e0 and file A's mut is left held forever, wedging its next writer in D state. The uevent is also stranded on a list ucma_cleanup_ctx_events() will not walk, so it outlives its context. /dev/infiniband/rdma_cm is 0666 and no RDMA device is involved, so an unprivileged user reaches all of this. Take the handler lock, as ucma_cleanup_mc_events() does; ctx->cm_id is pinned by the ucma_get_ctx() reference.

Linux

In the Linux kernel, the following vulnerability has been resolved: regulator: as3722_get_regulator_dt_data: fix premature of_node_put leaving dangling of_node pointer In as3722_get_regulator_dt_data(), of_get_child_by_name() acquires a reference on np, which is then assigned to pdev->dev.of_node. The function immediately calls of_node_put(np), releasing the reference and leaving pdev->dev.of_node as a dangling pointer. Remove the of_node_put(np) call to let the device hold the reference.

Linux

In the Linux kernel, the following vulnerability has been resolved: ring-buffer: Fix subbuf resize race with ring_buffer_alloc_read_page() ring_buffer_alloc_read_page() is racy with ring_buffer_subbuf_order_set, it can allocate a reader page with an outdated order. This isn't a big issue, the user can still re-allocate a new reader page and try again. However, what is more problematic is if the value of subbuf_order changes in the middle of ring_buffer_alloc_read_page(). In that case, bpage->order might not match the actual allocated memory. Use bpage->order for the allocation to prevent this race.

Linux

In the Linux kernel, the following vulnerability has been resolved: ring-buffer: Hold cpu_buffer::lock when resizing a subbuf Because, ring_buffer_subbuf_order_set() can clear cpu_buffer->free_page, hold cpu_buffer->lock to prevent races with ring_buffer_alloc_read_page() and ring_buffer_free_read_page().

Linux

In the Linux kernel, the following vulnerability has been resolved: ring-buffer: Make cpu_buffer::free_page a buffer_data_read_page Discarding a cached reader page after a concurrent ring buffer resize uses the new global subbuf_order for the free_pages() call. This mismatched order may crashes the kernel or leaks memory because the cached page was allocated under the old size. Save the actual free_page order alongside the page address to ensure we always refer to the correct value and do not rely on the potentially stalled cpu_buffer->subbuf_order value. The simplest is to make free_page a buffer_data_read_page which already covers exactly what we need: a page address and a page order.

Linux

In the Linux kernel, the following vulnerability has been resolved: ring-buffer: Stop remote reader update when page swap fails The remote swap_reader_page callback can return -EBUSY when the writer moves the head before the remote catches it, particularly during an event storm on a small buffer. __rb_get_reader_page_from_remote() currently warns about that failure but continues with the unchanged reader ID and rearranges the local page list as though the swap succeeded. Handle the callback failure as a recoverable error. Report it with pr_warn_ratelimited() and return NULL. Callers already handle a NULL reader page as a failed attempt. This avoids splicing the same page as both the previous and new reader without flooding the log under contention.

Linux

In the Linux kernel, the following vulnerability has been resolved: orangefs: skip leading spaces before parsing client debug masks orangefs_prepare_cdm_array() sizes each client debug keyword buffer with strcspn(cds_head, " "), but then parses the keyword with %s. The %s conversion skips leading whitespace, while strcspn() does not. If a client debug entry starts with a space, the allocation can be sized for an empty keyword while sscanf() copies the following non-empty token. This can write past the end of the allocated keyword buffer. Skip leading spaces before computing the keyword length so the allocation matches the string parsed by sscanf().

Linux

In the Linux kernel, the following vulnerability has been resolved: ocfs2: validate rl_used against rl_count in refcount block validator ocfs2_find_refcount_rec_in_rl() walks the on-disk refcount record array with: for (; i < le16_to_cpu(rb->rf_records.rl_used); i++) { rec = &rb->rf_records.rl_recs[i]; ... rl_recs[] lives in a single metadata block (4096 bytes on the common configuration), so its real capacity is fixed by ocfs2_refcount_recs_per_rb(sb) (247 records for a 4K block with the 16-byte ocfs2_refcount_rec). rl_used and rl_count are both read directly off disk by ocfs2_validate_refcount_block() and are never checked against that capacity, nor against each other, before any refcount/reflink/CoW operation walks the array. A crafted (or corrupted) refcount block with rl_used == 0xffff makes the loop above walk far past the end of the block, dereferencing rl_recs[i] for i up to 65534. The resulting index is then handed to the sibling ocfs2_insert_refcount_rec(), whose insert-shift does: if (index < le16_to_cpu(rf_list->rl_used)) memmove(&rf_list->rl_recs[index + 1], &rf_list->rl_recs[index], (le16_to_cpu(rf_list->rl_used) - index) * sizeof(struct ocfs2_refcount_rec)); i.e. a memmove() of up to (0xffff - index) * 16 bytes (~1 MiB) from an offset already past the block. This is reachable from an ordinary reflink (FICLONE) against a crafted/corrupted ocfs2 image: attaching an extent whose cpos sorts past every real record in the leaf forces the lookup to run off the end instead of returning early on a match. The attacker model is local: CAP_SYS_ADMIN mounting a crafted or corrupted ocfs2 image, or a raw write to the block device backing an already-mounted ocfs2 filesystem. ocfs2_validate_refcount_block() already validates the block's ECC, signature, rf_blkno and rf_fs_generation, but never rl_count/rl_used against the block's actual on-disk capacity. This is the same class of gap that ocfs2_validate_extent_block() (fs/ocfs2/alloc.c) already closes for the sibling extent-list header, which checks both the record capacity and the "used" bound before any code walks h_list.l_recs[]: if (le16_to_cpu(eb->h_list.l_count) != ocfs2_extent_recs_per_eb(sb)) { rc = ocfs2_error(...); goto bail; } if (le16_to_cpu(eb->h_list.l_next_free_rec) > le16_to_cpu(eb->h_list.l_count)) { rc = ocfs2_error(...); goto bail; } Add the equivalent pair of checks to ocfs2_validate_refcount_block(): reject a refcount block whose rl_count does not match the fixed per-block capacity returned by ocfs2_refcount_recs_per_rb(), and reject rl_used > rl_count. Both checks are skipped when OCFS2_REFCOUNT_TREE_FL is set, because in that case the same union bytes hold an ocfs2_extent_list (rf_list), not the refcount record list (rf_records) -- that layout is already validated separately by ocfs2_validate_extent_block() when the referenced extent block is read. This mirrors the existing "!(rb->rf_flags & OCFS2_REFCOUNT_TREE_FL)" guard used elsewhere in this file (e.g. ocfs2_get_refcount_rec()) to decide whether rf_records or rf_list is the live member of the union. With this in place, a forged rl_used/rl_count is caught at block validation time (ocfs2_error()), consistent with every other corruption check in this function, instead of driving an out-of-bounds read in ocfs2_find_refcount_rec_in_rl() and a subsequent out-of-bounds memmove() in ocfs2_insert_refcount_rec(). Verified against a crafted image on a v6.19 KASAN (KASAN_GENERIC) build: replaying the same reflink (FICLONE) reliably hit a KASAN report in __ocfs2_increase_refcount()/ocfs2_insert_refcount_rec() before this patch, and triggers no report once ocfs2_validate_refcount_block() rejects the forged rl_used/rl_count.

Linux

In the Linux kernel, the following vulnerability has been resolved: openrisc: fix arbitrary kernel memory access via or1k_atomic syscall sys_or1k_atomic() (syscall 244 in the "or1k" ABI) takes two user pointers, v1 and v2, and swaps the words they point to in hand-written assembly. l.lwz r29,0(r4) l.lwz r27,0(r5) l.sw 0(r4),r27 l.sw 0(r5),r29 The pointers are not checked with access_ok(). The four memory accesses also have no exception table entries. A caller passes a kernel address as either pointer, and the syscall reads from and writes to it directly. This gives an unprivileged process a kernel read/write primitive. It overwrites kernel data such as the sys_call_table, gaining code execution in kernel context. Check both pointers before entering the critical section. Add fixups for the four memory accesses so faults on valid but unmapped user addresses return -EFAULT. [shorne@gmail.com: fix comment style]

Linux

In the Linux kernel, the following vulnerability has been resolved: openvswitch: Fix CT limit teardown use-after-free Packet processing uses CT limit state under RCU, while netns teardown frees that state under ovs_mutex. The CT limit pointer was neither removed from readers nor protected by a grace period, allowing packet processing to dereference the freed state. An unprivileged user can trigger this bug from a user and network namespace, causing a slab-use-after-free in ovs_ct_execute() when the netns is torn down. Publish the CT limit pointer through RCU, remove it before teardown, and wait for readers before freeing its contents. Keep ovs_mutex around individual CT limit updates, and use the RCU read-side lock while GET traverses the RCU-protected limit lists. Netns teardown detaches the RCU-protected CT limit state in the pernet .pre_exit callback while holding ovs_mutex. The pernet core guarantees an RCU grace period between the .pre_exit and .exit callbacks, so the .exit callback completes the teardown without adding any extra synchronization. The netlink command handlers do not need NULL checks because the userspace netlink socket holds an active reference to its network namespace while a request is processed. The per-netns exit path therefore cannot run concurrently with SET, DEL, or GET for that socket's namespace.

Linux

In the Linux kernel, the following vulnerability has been resolved: openvswitch: only skb_tx_error() a packet we are about to drop queue_userspace_packet() borrows the packet skb -- it only copies it into a private netlink message (user_skb) and does not own it; on return do_execute_actions() keeps forwarding it through the flow's remaining actions. Its error path nevertheless calls skb_tx_error(skb), which via skb_zcopy_clear() does skb_shinfo(skb)->flags &= ~SKBFL_ALL_ZEROCOPY, stripping SKBFL_SHARED_FRAG from that live skb (skb_tx_error()'s kerneldoc says "skb must be freed afterwards"). For a MSG_ZEROCOPY skb carrying page-cache frags, SKBFL_SHARED_FRAG is what makes esp_input() skb_cow_data() before in-place AEAD; once it is stripped a later local ESP-in-UDP delivery decrypts in place over pages the sender does not own -- an unprivileged page-cache write (the "Fragnesia" primitive). do_execute_actions() ignores output_userspace()'s return value, so any action after a failed USERSPACE upcall inherits the stripped skb. Move the skb_tx_error() to the flow-miss drop path - the "default" branch of ovs_dp_process_packet()'s switch(error), before kfree_skb(). The call has been here since commit 36d5fe6a0007 ("core, nfqueue, openvswitch: Orphan frags in skb_zerocopy and handle errors") but was harmless until esp_input() began relying on SKBFL_SHARED_FRAG to gate in-place decrypt; only then did stripping it on a still-forwarded skb become a page-cache write primitive.

Linux

In the Linux kernel, the following vulnerability has been resolved: ipmi: Fix use-after-free of cmd_rcvr in _ipmi_destroy_user() Commit 9e91f8a6c868 ("ipmi:msghandler: Remove srcu for the ipmi_interfaces list") dropped the synchronize_rcu() between unlinking the command receivers from intf->cmd_rcvrs and freeing them, updating only the comment that explains why the barrier is needed. The cmd_rcvrs list is still traversed under plain RCU: find_cmd_rcvr() walks it inside rcu_read_lock(), and handle_ipmb_get_msg_cmd() borrows rcvr->user from that lookup within the same read-side section. Without the grace period, _ipmi_destroy_user() can kfree() a cmd_rcvr while a reader still holds a pointer to it, causing a use-after-free. The rework only made srcu unnecessary for the interfaces list; the cmd_rcvrs list still relies on plain RCU. Restore the synchronize_rcu() before freeing the receivers.

Linux

In the Linux kernel, the following vulnerability has been resolved: nvme: zero the discard fallback page nvme_setup_discard() always maps sizeof(struct nvme_dsm_range) * NVME_DSM_MAX_RANGES = 4096 bytes as the DSM payload however many ranges the command declares, because some devices ignore the 'Number of Ranges' field - the Fixes: commit records two that read past the declared ranges. A single-range discard fills only the first 16 bytes. Normally the buffer comes from kzalloc() and the other 4080 bytes are zero. When that allocation fails the code falls back to the per-controller ctrl->discard_page, which nvme_init_ctrl() obtains with alloc_page(GFP_KERNEL) and nothing ever zeroes, so those 4080 bytes are whatever the page last held and are handed to the controller. Reaching it requires the kzalloc(GFP_ATOMIC | __GFP_NOWARN) to fail, that is memory pressure; it is not remotely triggerable. Failing the allocation under KMSAN reproduces it, with the leaked tail full of vmemmap struct page pointers. The extent in the report is a partial transfer of the payload, not the whole 4096 bytes; the 16-byte boundary in it is the one declared range: [ 11.991601] BUG: KMSAN: uninit-value in dma_map_phys+0x14c8/0x1900 [ 11.991969] dma_map_phys+0x14c8/0x1900 [ 11.992220] dma_map_page_attrs+0xcf/0x130 [ 11.992485] e1000_xmit_frame+0x4099/0x6d10 [ 11.992768] dev_hard_start_xmit+0x22f/0xa80 [ 11.993068] sch_direct_xmit+0x35c/0xcb0 [ 11.993315] __dev_queue_xmit+0x1ee5/0x5eb0 [ 11.993608] ip_finish_output2+0x1903/0x1c30 [ 11.993881] ip_finish_output+0x288/0x870 [ 11.994125] ip_output+0x15e/0x400 [ 11.994365] __ip_queue_xmit+0x1e85/0x1fb0 [ 11.994639] ip_queue_xmit+0x60/0x80 [ 11.994899] __tcp_transmit_skb+0x4e71/0x5fa0 [ 11.995210] tcp_write_xmit+0x3a36/0x9160 [ 11.995533] __tcp_push_pending_frames+0xc5/0x3c0 [ 11.995854] tcp_push+0x7dc/0x840 [ 11.996076] tcp_sendmsg_locked+0x766c/0x8400 [ 11.996371] tcp_sendmsg+0x4b/0x90 [ 11.996572] inet_sendmsg+0x134/0x2a0 [ 11.996823] __sock_sendmsg+0x265/0x360 [ 11.997076] sock_sendmsg+0x100/0x1e0 [ 11.997293] nvme_tcp_try_send+0x196f/0x6370 [ 11.997605] nvme_tcp_queue_rq+0x1d54/0x20b0 [ 11.997882] blk_mq_dispatch_rq_list+0x5ee/0x2e50 [ 11.998175] __blk_mq_sched_dispatch_requests+0x16dc/0x24a0 [ 11.998539] blk_mq_sched_dispatch_requests+0x11b/0x2c0 [ 11.998865] blk_mq_run_work_fn+0x13b/0x280 [ 11.999146] process_scheduled_works+0x966/0x1ad0 [ 11.999465] worker_thread+0xe44/0x1480 [ 11.999709] kthread+0x53b/0x600 [ 11.999927] ret_from_fork+0x29f/0x7c0 [ 12.000191] ret_from_fork_asm+0x1a/0x30 [ 12.000460] [ 12.000558] Uninit was created at: [ 12.000788] __alloc_frozen_pages_noprof+0x8bf/0xd30 [ 12.001096] alloc_pages_mpol+0x1d0/0x5f0 [ 12.001326] alloc_pages_noprof+0x102/0x290 [ 12.001627] nvme_init_ctrl+0x5a3/0x9f0 [ 12.001891] nvme_tcp_create_ctrl+0xd75/0x19b0 [ 12.002170] nvmf_dev_write+0x4c68/0x4fd0 [ 12.002426] vfs_write+0x587/0x1a10 [ 12.002636] __x64_sys_write+0x207/0x4f0 [ 12.002874] x64_sys_call+0x2ff0/0x3ea0 [ 12.003123] do_syscall_64+0x147/0x3b0 [ 12.003400] entry_SYSCALL_64_after_hwframe+0x77/0x7f [ 12.003680] [ 12.003777] Bytes 16-2843 of 2844 are uninitialized [ 12.004068] Memory access of size 2844 starts at ffff888109f82000 [ 12.004412] [ 12.004530] CPU: 0 UID: 0 PID: 101 Comm: kworker/0:1H Not tainted 7.2.0-rc5-NVMECTL-gf5098b6bae76 #1 PREEMPT(lazy) [ 12.005127] Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch_caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 [ 12.005762] Workqueue: kblockd blk_mq_run_work_fn [ 12.006073] ===================================================== Allocate the page with __GFP_ZERO. The single allocation site covers every use of it: bytes no discard has written stay zero, and bytes one did write hold that controller's own range list, which it has already been sent.

Linux

In the Linux kernel, the following vulnerability has been resolved: nvme-tcp: fix host memory disclosure on R2T for a read command nvme_tcp_handle_r2t() does not check the direction of the request the R2T refers to. A malicious controller can send an R2T for a READ and the host will answer it: nvme_tcp_setup_h2c_data_pdu() builds the H2CData header and nvme_tcp_try_send_data() sends the request's data buffer. That buffer is the READ destination, so its contents go to the controller. The command then completes normally and nothing is logged. Against a test controller that answers every READ with an R2T, a 4096 byte buffered read returned all 4096 bytes, split over two R2Ts. The pages contained stale kernel data, including an array of struct page pointers. Reject an R2T for a request that is not a write.

Linux

In the Linux kernel, the following vulnerability has been resolved: nvme-tcp: reject a read that transferred too few bytes nvme_tcp_recv_data() completes a request once the current C2HData PDU has been consumed. Nothing compares the total bytes received against the length the command asked for: struct nvme_tcp_request has no receive-side counter, queue->data_remaining is per queue, and blk_mq_end_request() completes for blk_rq_bytes(rq) unconditionally with no residual concept anywhere above. A controller can therefore answer a 4096-byte read with 512 bytes and have it reported as a complete read; user space then gets 4096 bytes of which 3584 are whatever was already in the page. I reproduced that with a test target. Count the bytes received and refuse to complete a successful read whose count does not match, at the two NVME_TCP_F_DATA_SUCCESS paths and in nvme_tcp_process_nvme_cqe(). The success test shifts req->status right by one, because the driver keeps the wire value there and shifts it on completion, so the check must see what the completion path will see. Only REQ_OP_READ is checked, because there the length comes from the sectors the request covers; a passthrough command is built by its submitter, which picks both command and buffer, so the kernel has nothing to compare against.

Linux

In the Linux kernel, the following vulnerability has been resolved: sctp: fix NULL deref on untransmitted RECONF completion sctp_process_strreset_outreq(), sctp_process_strreset_addstrm_out() and sctp_process_strreset_resp() complete a pending stream reconfiguration request by stopping the reconf timer on the transport it was sent on: t = asoc->strreset_chunk->transport; if (timer_delete(&t->reconf_timer)) sctp_transport_put(t); chunk->transport is assigned by __sctp_packet_append_chunk() when the chunk is appended to an outbound packet, and sctp_outq_flush_ctrl() arms the reconf timer at that same point. A request already published in asoc->strreset_chunk but not yet transmitted has neither, so completing it dereferences NULL. Two ways to get there. sctp_send_asconf_del_ip() sets asoc->src_out_of_asoc_ok without sending anything when the address being removed is the association's last one, and sctp_outq_flush_ctrl() then leaves every non-ASCONF control chunk queued; as only sctp_process_asconf_ack() clears that flag, it persists. An unprivileged process that removes such an address and then asks for a stream reset panics the kernel from softirq. A peer needs neither ASCONF nor local help: sctp_cmd_interpreter() uncorks the outqueue only once the whole packet has been processed, so a reply built while walking a RECONF chunk stays untransmitted for the rest of that walk, and one RECONF chunk carrying [Incoming SSN Reset Request, Outgoing SSN Reset Request, Response] -- or two RECONF chunks in one packet -- reaches the same dereference. KASAN: null-ptr-deref in range [0x00000000000001e8-0x00000000000001ef] RIP: 0010:timer_delete+0x67/0x110 Call Trace: <IRQ> sctp_process_strreset_addstrm_out (net/sctp/stream.c:832) sctp_sf_do_reconf (net/sctp/sm_statefuns.c:4212) sctp_do_sm (net/sctp/sm_sideeffect.c:1172) sctp_assoc_bh_rcv (net/sctp/associola.c:1044) sctp_rcv (net/sctp/input.c:243) ip_local_deliver (net/ipv4/ip_input.c:262) process_backlog (net/core/dev.c:6680) </IRQ> A response can only acknowledge a request that was actually sent, so do not match asoc->strreset_chunk while chunk->transport is NULL. Guarding the lookup covers all three completion sites.

Linux

In the Linux kernel, the following vulnerability has been resolved: sctp: fix stream->outcnt underflow on duplicate RECONF responses A cached RECONF chunk may contain more than one request parameter. A duplicate response can therefore find and process the same ADD_OUT request again while another parameter is still outstanding, rolling back outcnt twice and possibly underflowing it. Track outstanding request types as bits and clear each bit after its first response. Later responses for the same request are then ignored.