Vulnerabilities

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ACPICA: Fix NULL pointer dereference in acpi_ns_custom_package()<br /> <br /> acpi_ns_custom_package() unconditionally dereferences the first element<br /> of the package to read the _BIX version number, without checking for<br /> NULL:<br /> <br /> if ((*Elements)-&gt;Common.Type != ACPI_TYPE_INTEGER)<br /> <br /> When firmware returns a _BIX package whose first element is an<br /> unresolvable reference, ACPICA evaluates that entry to NULL.<br /> acpi_ns_remove_null_elements() does not strip NULL entries for<br /> ACPI_PTYPE_CUSTOM packages (fixed-position format would break if<br /> elements were shifted), so acpi_ns_custom_package() sees the NULL<br /> and causes a crash.<br /> <br /> Add a NULL check for the first element (version field) before<br /> dereferencing it. The caller then receives AE_AML_OPERAND_TYPE<br /> instead of crashing.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ACPICA: Enhance OEM ID and Table ID validation in acpi_ex_load_table_op()<br /> <br /> Enhance OEM ID and Table ID validation in acpi_ex_load_table_op() to<br /> prevent buffer overflows.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> xhci: Prevent queuing new commands if xhci is inaccessible<br /> <br /> Refuse to queue a new command on the command ring if xHC is marked<br /> inaccessible with the HCD_FLAG_HW_ACCESSIBLE.<br /> <br /> HCD_FLAG_HW_ACCESSIBLE is set and cleared in suspend and resume.<br /> <br /> Also print a warning if xhci is being suspended with commands<br /> still pending on the command ring.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amd/display: Avoid DPMS-on for phantom stream<br /> <br /> [Why &amp; How]<br /> Calling dc_update_planes_and_stream separately for stream and its<br /> phantom stream causes a NULL pointer dereference, since the phantom is<br /> destroyed on the first call.<br /> <br /> Skip the call for phantom streams.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> wifi: iwlwifi: mvm: fix P2P-Device binding handling<br /> <br /> Our binding handling for P2P-Device can run into the following<br /> scenario, as observed by our testing:<br /> <br /> - a station interface is connected on some channel<br /> - the P2P-Device does a remain-on-channel (ROC) on that channel<br /> - the ROC ends, and the P2P-Device is removed from the binding,<br /> but the phy_ctxt pointer is left around as a PHY cache so we<br /> don&amp;#39;t need to recalibrate to the channel again and again in<br /> case it&amp;#39;s not shared<br /> - a binding update by the station interface, even a removal,<br /> will re-add the P2P-Device to the binding<br /> - the P2P-Device is removed, which removes the PHY context, but<br /> it&amp;#39;s still in the binding so the firmware crashes<br /> <br /> Since the P2P device is removed from the binding and only re-<br /> added by unrelated code, but we want to keep the phy_ctxt around<br /> as a cache for future ROC usage, fix it by adding a boolean that<br /> indicates whether or not the P2P-Device should be added to the<br /> binding, and handle that in the binding iterator. That way, the<br /> station interface cannot re-add the P2P-Device to the binding<br /> when that isn&amp;#39;t active.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> nvme: validate FDP configuration descriptor sizes<br /> <br /> Validate descriptor sizes while walking the FDP configurations log so<br /> dsze == 0 or a descriptor past the log end cannot cause unbounded<br /> iteration or reads past the buffer.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> dpaa2-switch: fix handling of NAPI on the remove path<br /> <br /> All the NAPI instances for a DPSW device are attached to the first<br /> switch port&amp;#39;s net_device but shared by all ports. The NAPI instances get<br /> disabled only once the last port goes down.<br /> <br /> This causes an issue on the .remove() path where each port is<br /> unregistered and freed one at a time, causing the NAPI instances to be<br /> deleted even though they are not disabled.<br /> <br /> In order to avoid this, split up the unregister_netdev() calls from the<br /> free_netdev() so that we make sure all ports go down before we attempt<br /> a deletion of NAPI instances. Also, make the netif_napi_del() explicit<br /> as it is on the .probe() path.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: dsa: sja1105: flower: reject cross-chip redirect<br /> <br /> dsa_port_from_netdev() may return a valid port from a different switch<br /> chip. Programming another chip&amp;#39;s port index into the local hardware<br /> causes redirection to the wrong port, or an out-of-bounds access if the<br /> index exceeds the local chip&amp;#39;s port count.<br /> <br /> Apply a minimal fix that adds a check to catch this case and adjusts the<br /> extack message. When cls-&gt;common.skip_sw is not set, the operation could<br /> instead redirect to the upstream port and let the software or upstream<br /> switch(es) handle the forward, but that is not addressed here.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> dpaa2-switch: rework FDB management on the bridge leave path<br /> <br /> On bridge leave, the dpaa2_switch_port_set_fdb() function always<br /> allocates a new FDB for the port which is becoming standalone. In case<br /> no FDB is found, then the port leaving a bridge will continue to use the<br /> current one.<br /> <br /> The above logic does not cover the case in which there are multiple<br /> bridges which have ports from the same DPSW instance. In this case, when<br /> the last port leaves bridge #1, it finds an unused FDB to switch to, but<br /> the old FDB is not marked as unused. Since the number of FDBs is equal<br /> to the number of DPSW interfaces, this will eventually lead to multiple<br /> ports sharing the same FDB.<br /> <br /> Fix this by changing how we are managing the FDBs on the leave path.<br /> Instead of directly allocating a new FDB, first verify if the current<br /> port is the last one to leave a bridge. If this is the case, then<br /> continue to use the current FDB and only allocate another FDB if there<br /> are other ports remaining in the bridge.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ntfs3: fix out-of-bounds read in ntfs_dir_emit() and hdr_find_e()<br /> <br /> The bounds check in ntfs_dir_emit() compares fname-&gt;name_len (a<br /> character count) against e-&gt;size (a byte count) without accounting<br /> for the 2-byte-per-character UTF-16LE encoding or the ATTR_FILE_NAME<br /> header size:<br /> <br /> if (fname-&gt;name_len + sizeof(struct NTFS_DE) &gt; le16_to_cpu(e-&gt;size))<br /> <br /> This computes: name_len + 16 &gt; e_size<br /> <br /> The correct check must account for the ATTR_FILE_NAME header (66 bytes<br /> before the name) and the UTF-16LE character size (2 bytes each):<br /> <br /> sizeof(NTFS_DE) + offsetof(ATTR_FILE_NAME, name) +<br /> name_len * sizeof(short) &gt; e_size<br /> <br /> Which computes: 16 + 66 + name_len * 2 &gt; e_size<br /> <br /> The correct calculation already exists as fname_full_size() in ntfs.h<br /> and is used in cmp_fnames(), namei.c, and fslog.c, but was not used<br /> in the readdir path.<br /> <br /> A crafted NTFS image with an index entry containing a small e-&gt;size<br /> but large fname-&gt;name_len bypasses the current check, causing<br /> ntfs_utf16_to_nls() to read past the entry boundary.<br /> <br /> Additionally, add a key_size validation in hdr_find_e() to ensure the<br /> declared key_size does not exceed the available entry data, preventing<br /> comparison functions from reading past entry boundaries on the lookup<br /> path.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> fs/ntfs3: validate index entry key bounds<br /> <br /> [BUG]<br /> A malformed NTFS directory index entry can advertise a key_size larger<br /> than the bytes actually present in its NTFS_DE payload. Directory lookup<br /> then passes that malformed key to cmp_fnames(), which can read past the<br /> end of the kmalloc&amp;#39;ed index buffer.<br /> <br /> BUG: KASAN: slab-out-of-bounds in fname_full_size fs/ntfs3/ntfs.h:590 [inline]<br /> BUG: KASAN: slab-out-of-bounds in cmp_fnames+0x1ea/0x230 fs/ntfs3/index.c:46<br /> Read of size 1 at addr ffff88801c313018 by task syz.6.3365/9279<br /> <br /> Call Trace:<br /> __dump_stack lib/dump_stack.c:94 [inline]<br /> dump_stack_lvl+0xbe/0x130 lib/dump_stack.c:120<br /> print_address_description mm/kasan/report.c:378 [inline]<br /> print_report+0xd1/0x650 mm/kasan/report.c:482<br /> kasan_report+0xfb/0x140 mm/kasan/report.c:595<br /> __asan_report_load1_noabort+0x14/0x30 mm/kasan/report_generic.c:378<br /> fname_full_size fs/ntfs3/ntfs.h:590 [inline]<br /> cmp_fnames+0x1ea/0x230 fs/ntfs3/index.c:46<br /> hdr_find_e.isra.0+0x3ed/0x670 fs/ntfs3/index.c:762<br /> indx_find+0x4b5/0x900 fs/ntfs3/index.c:1186<br /> dir_search_u+0x2c0/0x460 fs/ntfs3/dir.c:254<br /> ntfs_lookup+0x1cc/0x2a0 fs/ntfs3/namei.c:85<br /> __lookup_slow+0x241/0x450 fs/namei.c:1816<br /> lookup_slow fs/namei.c:1833 [inline]<br /> walk_component+0x31c/0x570 fs/namei.c:2151<br /> link_path_walk+0x592/0xd60 fs/namei.c:2519<br /> path_lookupat+0x138/0x660 fs/namei.c:2675<br /> filename_lookup+0x1f3/0x560 fs/namei.c:2705<br /> filename_setxattr+0xad/0x1c0 fs/xattr.c:660<br /> path_setxattrat+0x1d8/0x280 fs/xattr.c:713<br /> __do_sys_lsetxattr fs/xattr.c:754 [inline]<br /> __se_sys_lsetxattr fs/xattr.c:750 [inline]<br /> __x64_sys_lsetxattr+0xd0/0x150 fs/xattr.c:750<br /> ...<br /> <br /> Allocated by task 9279:<br /> kasan_save_stack+0x39/0x70 mm/kasan/common.c:56<br /> kasan_save_track+0x14/0x40 mm/kasan/common.c:77<br /> kasan_save_alloc_info+0x37/0x60 mm/kasan/generic.c:573<br /> poison_kmalloc_redzone mm/kasan/common.c:400 [inline]<br /> __kasan_kmalloc+0xc3/0xd0 mm/kasan/common.c:417<br /> kasan_kmalloc include/linux/kasan.h:262 [inline]<br /> __do_kmalloc_node mm/slub.c:5650 [inline]<br /> __kmalloc_noprof+0x2bd/0x900 mm/slub.c:5662<br /> kmalloc_noprof include/linux/slab.h:961 [inline]<br /> indx_read+0x41d/0xad0 fs/ntfs3/index.c:1059<br /> indx_find+0x447/0x900 fs/ntfs3/index.c:1179<br /> dir_search_u+0x2c0/0x460 fs/ntfs3/dir.c:254<br /> ntfs_lookup+0x1cc/0x2a0 fs/ntfs3/namei.c:85<br /> __lookup_slow+0x241/0x450 fs/namei.c:1816<br /> lookup_slow fs/namei.c:1833 [inline]<br /> walk_component+0x31c/0x570 fs/namei.c:2151<br /> link_path_walk+0x592/0xd60 fs/namei.c:2519<br /> path_lookupat+0x138/0x660 fs/namei.c:2675<br /> filename_lookup+0x1f3/0x560 fs/namei.c:2705<br /> filename_setxattr+0xad/0x1c0 fs/xattr.c:660<br /> path_setxattrat+0x1d8/0x280 fs/xattr.c:713<br /> __do_sys_lsetxattr fs/xattr.c:754 [inline]<br /> __se_sys_lsetxattr fs/xattr.c:750 [inline]<br /> __x64_sys_lsetxattr+0xd0/0x150 fs/xattr.c:750<br /> ...<br /> <br /> [CAUSE]<br /> The index-header validators only validated INDEX_HDR-level geometry.<br /> They did not walk each NTFS_DE to verify entry alignment, subnode<br /> layout, or that key_size fit inside the entry payload. They also<br /> allowed a last sentinel entry to carry a non-zero key_size.<br /> <br /> [FIX]<br /> Walk every NTFS_DE in ntfs3&amp;#39;s index-header validators and reject<br /> entries with invalid layout, mismatched subnode state, oversized<br /> key_size, or non-zero sentinel keys before lookup or log replay can<br /> consume them.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdkfd: fix SMI event cross-process information leak<br /> <br /> kfd_smi_ev_enabled() skips the suser privilege check when pid=0.<br /> PROCESS_START, PROCESS_END, and VMFAULT events are emitted with<br /> pid=0 while carrying another process&amp;#39;s PID and command name, so any<br /> /dev/kfd user in the render group can monitor all GPU workloads.<br /> <br /> Pass the target process PID into kfd_smi_event_add() for these events<br /> so the existing per-client filter restricts delivery to the owning<br /> process or CAP_SYS_ADMIN subscribers.