Vulnerabilities

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ipv4: Handle attempt to delete multipath route when fib_info contains an nh reference<br /> <br /> Gwangun Jung reported a slab-out-of-bounds access in fib_nh_match:<br /> fib_nh_match+0xf98/0x1130 linux-6.0-rc7/net/ipv4/fib_semantics.c:961<br /> fib_table_delete+0x5f3/0xa40 linux-6.0-rc7/net/ipv4/fib_trie.c:1753<br /> inet_rtm_delroute+0x2b3/0x380 linux-6.0-rc7/net/ipv4/fib_frontend.c:874<br /> <br /> Separate nexthop objects are mutually exclusive with the legacy<br /> multipath spec. Fix fib_nh_match to return if the config for the<br /> to be deleted route contains a multipath spec while the fib_info<br /> is using a nexthop object.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> iommu/vt-d: Fix PCI device refcount leak in has_external_pci()<br /> <br /> for_each_pci_dev() is implemented by pci_get_device(). The comment of<br /> pci_get_device() says that it will increase the reference count for the<br /> returned pci_dev and also decrease the reference count for the input<br /> pci_dev @from if it is not NULL.<br /> <br /> If we break for_each_pci_dev() loop with pdev not NULL, we need to call<br /> pci_dev_put() to decrease the reference count. Add the missing<br /> pci_dev_put() before &amp;#39;return true&amp;#39; to avoid reference count leak.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> riscv: fix race when vmap stack overflow<br /> <br /> Currently, when detecting vmap stack overflow, riscv firstly switches<br /> to the so called shadow stack, then use this shadow stack to call the<br /> get_overflow_stack() to get the overflow stack. However, there&amp;#39;s<br /> a race here if two or more harts use the same shadow stack at the same<br /> time.<br /> <br /> To solve this race, we introduce spin_shadow_stack atomic var, which<br /> will be swap between its own address and 0 in atomic way, when the<br /> var is set, it means the shadow_stack is being used; when the var<br /> is cleared, it means the shadow_stack isn&amp;#39;t being used.<br /> <br /> [Palmer: Add AQ to the swap, and also some comments.]

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> iommu/vt-d: Fix PCI device refcount leak in dmar_dev_scope_init()<br /> <br /> for_each_pci_dev() is implemented by pci_get_device(). The comment of<br /> pci_get_device() says that it will increase the reference count for the<br /> returned pci_dev and also decrease the reference count for the input<br /> pci_dev @from if it is not NULL.<br /> <br /> If we break for_each_pci_dev() loop with pdev not NULL, we need to call<br /> pci_dev_put() to decrease the reference count. Add the missing<br /> pci_dev_put() for the error path to avoid reference count leak.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> nvme: fix SRCU protection of nvme_ns_head list<br /> <br /> Walking the nvme_ns_head siblings list is protected by the head&amp;#39;s srcu<br /> in nvme_ns_head_submit_bio() but not nvme_mpath_revalidate_paths().<br /> Removing namespaces from the list also fails to synchronize the srcu.<br /> Concurrent scan work can therefore cause use-after-frees.<br /> <br /> Hold the head&amp;#39;s srcu lock in nvme_mpath_revalidate_paths() and<br /> synchronize with the srcu, not the global RCU, in nvme_ns_remove().<br /> <br /> Observed the following panic when making NVMe/RDMA connections<br /> with native multipath on the Rocky Linux 8.6 kernel<br /> (it seems the upstream kernel has the same race condition).<br /> Disassembly shows the faulting instruction is cmp 0x50(%rdx),%rcx;<br /> computing capacity != get_capacity(ns-&gt;disk).<br /> Address 0x50 is dereferenced because ns-&gt;disk is NULL.<br /> The NULL disk appears to be the result of concurrent scan work<br /> freeing the namespace (note the log line in the middle of the panic).<br /> <br /> [37314.206036] BUG: unable to handle kernel NULL pointer dereference at 0000000000000050<br /> [37314.206036] nvme0n3: detected capacity change from 0 to 11811160064<br /> [37314.299753] PGD 0 P4D 0<br /> [37314.299756] Oops: 0000 [#1] SMP PTI<br /> [37314.299759] CPU: 29 PID: 322046 Comm: kworker/u98:3 Kdump: loaded Tainted: G W X --------- - - 4.18.0-372.32.1.el8test86.x86_64 #1<br /> [37314.299762] Hardware name: Dell Inc. PowerEdge R720/0JP31P, BIOS 2.7.0 05/23/2018<br /> [37314.299763] Workqueue: nvme-wq nvme_scan_work [nvme_core]<br /> [37314.299783] RIP: 0010:nvme_mpath_revalidate_paths+0x26/0xb0 [nvme_core]<br /> [37314.299790] Code: 1f 44 00 00 66 66 66 66 90 55 53 48 8b 5f 50 48 8b 83 c8 c9 00 00 48 8b 13 48 8b 48 50 48 39 d3 74 20 48 8d 42 d0 48 8b 50 20 3b 4a 50 74 05 f0 80 60 70 ef 48 8b 50 30 48 8d 42 d0 48 39 d3<br /> [37315.058803] RSP: 0018:ffffabe28f913d10 EFLAGS: 00010202<br /> [37315.121316] RAX: ffff927a077da800 RBX: ffff92991dd70000 RCX: 0000000001600000<br /> [37315.206704] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff92991b719800<br /> [37315.292106] RBP: ffff929a6b70c000 R08: 000000010234cd4a R09: c0000000ffff7fff<br /> [37315.377501] R10: 0000000000000001 R11: ffffabe28f913a30 R12: 0000000000000000<br /> [37315.462889] R13: ffff92992716600c R14: ffff929964e6e030 R15: ffff92991dd70000<br /> [37315.548286] FS: 0000000000000000(0000) GS:ffff92b87fb80000(0000) knlGS:0000000000000000<br /> [37315.645111] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> [37315.713871] CR2: 0000000000000050 CR3: 0000002208810006 CR4: 00000000000606e0<br /> [37315.799267] Call Trace:<br /> [37315.828515] nvme_update_ns_info+0x1ac/0x250 [nvme_core]<br /> [37315.892075] nvme_validate_or_alloc_ns+0x2ff/0xa00 [nvme_core]<br /> [37315.961871] ? __blk_mq_free_request+0x6b/0x90<br /> [37316.015021] nvme_scan_work+0x151/0x240 [nvme_core]<br /> [37316.073371] process_one_work+0x1a7/0x360<br /> [37316.121318] ? create_worker+0x1a0/0x1a0<br /> [37316.168227] worker_thread+0x30/0x390<br /> [37316.212024] ? create_worker+0x1a0/0x1a0<br /> [37316.258939] kthread+0x10a/0x120<br /> [37316.297557] ? set_kthread_struct+0x50/0x50<br /> [37316.347590] ret_from_fork+0x35/0x40<br /> [37316.390360] Modules linked in: nvme_rdma nvme_tcp(X) nvme_fabrics nvme_core netconsole iscsi_tcp libiscsi_tcp dm_queue_length dm_service_time nf_conntrack_netlink br_netfilter bridge stp llc overlay nft_chain_nat ipt_MASQUERADE nf_nat xt_addrtype xt_CT nft_counter xt_state xt_conntrack nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 xt_comment xt_multiport nft_compat nf_tables libcrc32c nfnetlink dm_multipath tg3 rpcrdma sunrpc rdma_ucm ib_srpt ib_isert iscsi_target_mod target_core_mod ib_iser libiscsi scsi_transport_iscsi ib_umad rdma_cm ib_ipoib iw_cm ib_cm intel_rapl_msr iTCO_wdt iTCO_vendor_support dcdbas intel_rapl_common sb_edac x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel ipmi_ssif kvm irqbypass crct10dif_pclmul crc32_pclmul mlx5_ib ghash_clmulni_intel ib_uverbs rapl intel_cstate intel_uncore ib_core ipmi_si joydev mei_me pcspkr ipmi_devintf mei lpc_ich wmi ipmi_msghandler acpi_power_meter ex<br /> ---truncated---

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> riscv: Sync efi page table&amp;#39;s kernel mappings before switching<br /> <br /> The EFI page table is initially created as a copy of the kernel page table.<br /> With VMAP_STACK enabled, kernel stacks are allocated in the vmalloc area:<br /> if the stack is allocated in a new PGD (one that was not present at the<br /> moment of the efi page table creation or not synced in a previous vmalloc<br /> fault), the kernel will take a trap when switching to the efi page table<br /> when the vmalloc kernel stack is accessed, resulting in a kernel panic.<br /> <br /> Fix that by updating the efi kernel mappings before switching to the efi<br /> page table.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: dsa: sja1105: avoid out of bounds access in sja1105_init_l2_policing()<br /> <br /> The SJA1105 family has 45 L2 policing table entries<br /> (SJA1105_MAX_L2_POLICING_COUNT) and SJA1110 has 110<br /> (SJA1110_MAX_L2_POLICING_COUNT). Keeping the table structure but<br /> accounting for the difference in port count (5 in SJA1105 vs 10 in<br /> SJA1110) does not fully explain the difference. Rather, the SJA1110 also<br /> has L2 ingress policers for multicast traffic. If a packet is classified<br /> as multicast, it will be processed by the policer index 99 + SRCPORT.<br /> <br /> The sja1105_init_l2_policing() function initializes all L2 policers such<br /> that they don&amp;#39;t interfere with normal packet reception by default. To have<br /> a common code between SJA1105 and SJA1110, the index of the multicast<br /> policer for the port is calculated because it&amp;#39;s an index that is out of<br /> bounds for SJA1105 but in bounds for SJA1110, and a bounds check is<br /> performed.<br /> <br /> The code fails to do the proper thing when determining what to do with the<br /> multicast policer of port 0 on SJA1105 (ds-&gt;num_ports = 5). The "mcast"<br /> index will be equal to 45, which is also equal to<br /> table-&gt;ops-&gt;max_entry_count (SJA1105_MAX_L2_POLICING_COUNT). So it passes<br /> through the check. But at the same time, SJA1105 doesn&amp;#39;t have multicast<br /> policers. So the code programs the SHARINDX field of an out-of-bounds<br /> element in the L2 Policing table of the static config.<br /> <br /> The comparison between index 45 and 45 entries should have determined the<br /> code to not access this policer index on SJA1105, since its memory wasn&amp;#39;t<br /> even allocated.<br /> <br /> With enough bad luck, the out-of-bounds write could even overwrite other<br /> valid kernel data, but in this case, the issue was detected using KASAN.<br /> <br /> Kernel log:<br /> <br /> sja1105 spi5.0: Probed switch chip: SJA1105Q<br /> ==================================================================<br /> BUG: KASAN: slab-out-of-bounds in sja1105_setup+0x1cbc/0x2340<br /> Write of size 8 at addr ffffff880bd57708 by task kworker/u8:0/8<br /> ...<br /> Workqueue: events_unbound deferred_probe_work_func<br /> Call trace:<br /> ...<br /> sja1105_setup+0x1cbc/0x2340<br /> dsa_register_switch+0x1284/0x18d0<br /> sja1105_probe+0x748/0x840<br /> ...<br /> Allocated by task 8:<br /> ...<br /> sja1105_setup+0x1bcc/0x2340<br /> dsa_register_switch+0x1284/0x18d0<br /> sja1105_probe+0x748/0x840<br /> ...

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/shmem-helper: Remove errant put in error path<br /> <br /> drm_gem_shmem_mmap() doesn&amp;#39;t own this reference, resulting in the GEM<br /> object getting prematurely freed leading to a later use-after-free.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> Bluetooth: Fix crash when replugging CSR fake controllers<br /> <br /> It seems fake CSR 5.0 clones can cause the suspend notifier to be<br /> registered twice causing the following kernel panic:<br /> <br /> [ 71.986122] Call Trace:<br /> [ 71.986124] <br /> [ 71.986125] blocking_notifier_chain_register+0x33/0x60<br /> [ 71.986130] hci_register_dev+0x316/0x3d0 [bluetooth 99b5497ea3d09708fa1366c1dc03288bf3cca8da]<br /> [ 71.986154] btusb_probe+0x979/0xd85 [btusb e1e0605a4f4c01984a4b9c8ac58c3666ae287477]<br /> [ 71.986159] ? __pm_runtime_set_status+0x1a9/0x300<br /> [ 71.986162] ? ktime_get_mono_fast_ns+0x3e/0x90<br /> [ 71.986167] usb_probe_interface+0xe3/0x2b0<br /> [ 71.986171] really_probe+0xdb/0x380<br /> [ 71.986174] ? pm_runtime_barrier+0x54/0x90<br /> [ 71.986177] __driver_probe_device+0x78/0x170<br /> [ 71.986180] driver_probe_device+0x1f/0x90<br /> [ 71.986183] __device_attach_driver+0x89/0x110<br /> [ 71.986186] ? driver_allows_async_probing+0x70/0x70<br /> [ 71.986189] bus_for_each_drv+0x8c/0xe0<br /> [ 71.986192] __device_attach+0xb2/0x1e0<br /> [ 71.986195] bus_probe_device+0x92/0xb0<br /> [ 71.986198] device_add+0x422/0x9a0<br /> [ 71.986201] ? sysfs_merge_group+0xd4/0x110<br /> [ 71.986205] usb_set_configuration+0x57a/0x820<br /> [ 71.986208] usb_generic_driver_probe+0x4f/0x70<br /> [ 71.986211] usb_probe_device+0x3a/0x110<br /> [ 71.986213] really_probe+0xdb/0x380<br /> [ 71.986216] ? pm_runtime_barrier+0x54/0x90<br /> [ 71.986219] __driver_probe_device+0x78/0x170<br /> [ 71.986221] driver_probe_device+0x1f/0x90<br /> [ 71.986224] __device_attach_driver+0x89/0x110<br /> [ 71.986227] ? driver_allows_async_probing+0x70/0x70<br /> [ 71.986230] bus_for_each_drv+0x8c/0xe0<br /> [ 71.986232] __device_attach+0xb2/0x1e0<br /> [ 71.986235] bus_probe_device+0x92/0xb0<br /> [ 71.986237] device_add+0x422/0x9a0<br /> [ 71.986239] ? _dev_info+0x7d/0x98<br /> [ 71.986242] ? blake2s_update+0x4c/0xc0<br /> [ 71.986246] usb_new_device.cold+0x148/0x36d<br /> [ 71.986250] hub_event+0xa8a/0x1910<br /> [ 71.986255] process_one_work+0x1c4/0x380<br /> [ 71.986259] worker_thread+0x51/0x390<br /> [ 71.986262] ? rescuer_thread+0x3b0/0x3b0<br /> [ 71.986264] kthread+0xdb/0x110<br /> [ 71.986266] ? kthread_complete_and_exit+0x20/0x20<br /> [ 71.986268] ret_from_fork+0x1f/0x30<br /> [ 71.986273] <br /> [ 71.986274] ---[ end trace 0000000000000000 ]---<br /> [ 71.986284] btusb: probe of 2-1.6:1.0 failed with error -17

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> io_uring: Fix a null-ptr-deref in io_tctx_exit_cb()<br /> <br /> Syzkaller reports a NULL deref bug as follows:<br /> <br /> BUG: KASAN: null-ptr-deref in io_tctx_exit_cb+0x53/0xd3<br /> Read of size 4 at addr 0000000000000138 by task file1/1955<br /> <br /> CPU: 1 PID: 1955 Comm: file1 Not tainted 6.1.0-rc7-00103-gef4d3ea40565 #75<br /> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014<br /> Call Trace:<br /> <br /> dump_stack_lvl+0xcd/0x134<br /> ? io_tctx_exit_cb+0x53/0xd3<br /> kasan_report+0xbb/0x1f0<br /> ? io_tctx_exit_cb+0x53/0xd3<br /> kasan_check_range+0x140/0x190<br /> io_tctx_exit_cb+0x53/0xd3<br /> task_work_run+0x164/0x250<br /> ? task_work_cancel+0x30/0x30<br /> get_signal+0x1c3/0x2440<br /> ? lock_downgrade+0x6e0/0x6e0<br /> ? lock_downgrade+0x6e0/0x6e0<br /> ? exit_signals+0x8b0/0x8b0<br /> ? do_raw_read_unlock+0x3b/0x70<br /> ? do_raw_spin_unlock+0x50/0x230<br /> arch_do_signal_or_restart+0x82/0x2470<br /> ? kmem_cache_free+0x260/0x4b0<br /> ? putname+0xfe/0x140<br /> ? get_sigframe_size+0x10/0x10<br /> ? do_execveat_common.isra.0+0x226/0x710<br /> ? lockdep_hardirqs_on+0x79/0x100<br /> ? putname+0xfe/0x140<br /> ? do_execveat_common.isra.0+0x238/0x710<br /> exit_to_user_mode_prepare+0x15f/0x250<br /> syscall_exit_to_user_mode+0x19/0x50<br /> do_syscall_64+0x42/0xb0<br /> entry_SYSCALL_64_after_hwframe+0x63/0xcd<br /> RIP: 0023:0x0<br /> Code: Unable to access opcode bytes at 0xffffffffffffffd6.<br /> RSP: 002b:00000000fffb7790 EFLAGS: 00000200 ORIG_RAX: 000000000000000b<br /> RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000<br /> RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000<br /> RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000<br /> R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000<br /> R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000<br /> <br /> Kernel panic - not syncing: panic_on_warn set ...<br /> <br /> This happens because the adding of task_work from io_ring_exit_work()<br /> isn&amp;#39;t synchronized with canceling all work items from eg exec. The<br /> execution of the two are ordered in that they are both run by the task<br /> itself, but if io_tctx_exit_cb() is queued while we&amp;#39;re canceling all<br /> work items off exec AND gets executed when the task exits to userspace<br /> rather than in the main loop in io_uring_cancel_generic(), then we can<br /> find current-&gt;io_uring == NULL and hit the above crash.<br /> <br /> It&amp;#39;s safe to add this NULL check here, because the execution of the two<br /> paths are done by the task itself.<br /> <br /> [axboe: add code comment and also put an explanation in the commit msg]

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> can: slcan: fix freed work crash<br /> <br /> The LTP test pty03 is causing a crash in slcan:<br /> BUG: kernel NULL pointer dereference, address: 0000000000000008<br /> #PF: supervisor read access in kernel mode<br /> #PF: error_code(0x0000) - not-present page<br /> PGD 0 P4D 0<br /> Oops: 0000 [#1] PREEMPT SMP NOPTI<br /> CPU: 0 PID: 348 Comm: kworker/0:3 Not tainted 6.0.8-1-default #1 openSUSE Tumbleweed 9d20364b934f5aab0a9bdf84e8f45cfdfae39dab<br /> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.15.0-0-g2dd4b9b-rebuilt.opensuse.org 04/01/2014<br /> Workqueue: 0x0 (events)<br /> RIP: 0010:process_one_work (/home/rich/kernel/linux/kernel/workqueue.c:706 /home/rich/kernel/linux/kernel/workqueue.c:2185)<br /> Code: 49 89 ff 41 56 41 55 41 54 55 53 48 89 f3 48 83 ec 10 48 8b 06 48 8b 6f 48 49 89 c4 45 30 e4 a8 04 b8 00 00 00 00 4c 0f 44 e0 8b 44 24 08 44 8b a8 00 01 00 00 41 83 e5 20 f6 45 10 04 75 0e<br /> RSP: 0018:ffffaf7b40f47e98 EFLAGS: 00010046<br /> RAX: 0000000000000000 RBX: ffff9d644e1b8b48 RCX: ffff9d649e439968<br /> RDX: 00000000ffff8455 RSI: ffff9d644e1b8b48 RDI: ffff9d64764aa6c0<br /> RBP: ffff9d649e4335c0 R08: 0000000000000c00 R09: ffff9d64764aa734<br /> R10: 0000000000000007 R11: 0000000000000001 R12: 0000000000000000<br /> R13: ffff9d649e4335e8 R14: ffff9d64490da780 R15: ffff9d64764aa6c0<br /> FS: 0000000000000000(0000) GS:ffff9d649e400000(0000) knlGS:0000000000000000<br /> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> CR2: 0000000000000008 CR3: 0000000036424000 CR4: 00000000000006f0<br /> Call Trace:<br /> <br /> worker_thread (/home/rich/kernel/linux/kernel/workqueue.c:2436)<br /> kthread (/home/rich/kernel/linux/kernel/kthread.c:376)<br /> ret_from_fork (/home/rich/kernel/linux/arch/x86/entry/entry_64.S:312)<br /> <br /> Apparently, the slcan&amp;#39;s tx_work is freed while being scheduled. While<br /> slcan_netdev_close() (netdev side) calls flush_work(&amp;sl-&gt;tx_work),<br /> slcan_close() (tty side) does not. So when the netdev is never set UP,<br /> but the tty is stuffed with bytes and forced to wakeup write, the work<br /> is scheduled, but never flushed.<br /> <br /> So add an additional flush_work() to slcan_close() to be sure the work<br /> is flushed under all circumstances.<br /> <br /> The Fixes commit below moved flush_work() from slcan_close() to<br /> slcan_netdev_close(). What was the rationale behind it? Maybe we can<br /> drop the one in slcan_netdev_close()?<br /> <br /> I see the same pattern in can327. So it perhaps needs the very same fix.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: mana: Fix race on per-CQ variable napi work_done<br /> <br /> After calling napi_complete_done(), the NAPIF_STATE_SCHED bit may be<br /> cleared, and another CPU can start napi thread and access per-CQ variable,<br /> cq-&gt;work_done. If the other thread (for example, from busy_poll) sets<br /> it to a value &gt;= budget, this thread will continue to run when it should<br /> stop, and cause memory corruption and panic.<br /> <br /> To fix this issue, save the per-CQ work_done variable in a local variable<br /> before napi_complete_done(), so it won&amp;#39;t be corrupted by a possible<br /> concurrent thread after napi_complete_done().<br /> <br /> Also, add a flag bit to advertise to the NIC firmware: the NAPI work_done<br /> variable race is fixed, so the driver is able to reliably support features<br /> like busy_poll.