CVE-2024-26645

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> tracing: Ensure visibility when inserting an element into tracing_map<br /> <br /> Running the following two commands in parallel on a multi-processor<br /> AArch64 machine can sporadically produce an unexpected warning about<br /> duplicate histogram entries:<br /> <br /> $ while true; do<br /> echo hist:key=id.syscall:val=hitcount &gt; \<br /> /sys/kernel/debug/tracing/events/raw_syscalls/sys_enter/trigger<br /> cat /sys/kernel/debug/tracing/events/raw_syscalls/sys_enter/hist<br /> sleep 0.001<br /> done<br /> $ stress-ng --sysbadaddr $(nproc)<br /> <br /> The warning looks as follows:<br /> <br /> [ 2911.172474] ------------[ cut here ]------------<br /> [ 2911.173111] Duplicates detected: 1<br /> [ 2911.173574] WARNING: CPU: 2 PID: 12247 at kernel/trace/tracing_map.c:983 tracing_map_sort_entries+0x3e0/0x408<br /> [ 2911.174702] Modules linked in: iscsi_ibft(E) iscsi_boot_sysfs(E) rfkill(E) af_packet(E) nls_iso8859_1(E) nls_cp437(E) vfat(E) fat(E) ena(E) tiny_power_button(E) qemu_fw_cfg(E) button(E) fuse(E) efi_pstore(E) ip_tables(E) x_tables(E) xfs(E) libcrc32c(E) aes_ce_blk(E) aes_ce_cipher(E) crct10dif_ce(E) polyval_ce(E) polyval_generic(E) ghash_ce(E) gf128mul(E) sm4_ce_gcm(E) sm4_ce_ccm(E) sm4_ce(E) sm4_ce_cipher(E) sm4(E) sm3_ce(E) sm3(E) sha3_ce(E) sha512_ce(E) sha512_arm64(E) sha2_ce(E) sha256_arm64(E) nvme(E) sha1_ce(E) nvme_core(E) nvme_auth(E) t10_pi(E) sg(E) scsi_mod(E) scsi_common(E) efivarfs(E)<br /> [ 2911.174738] Unloaded tainted modules: cppc_cpufreq(E):1<br /> [ 2911.180985] CPU: 2 PID: 12247 Comm: cat Kdump: loaded Tainted: G E 6.7.0-default #2 1b58bbb22c97e4399dc09f92d309344f69c44a01<br /> [ 2911.182398] Hardware name: Amazon EC2 c7g.8xlarge/, BIOS 1.0 11/1/2018<br /> [ 2911.183208] pstate: 61400005 (nZCv daif +PAN -UAO -TCO +DIT -SSBS BTYPE=--)<br /> [ 2911.184038] pc : tracing_map_sort_entries+0x3e0/0x408<br /> [ 2911.184667] lr : tracing_map_sort_entries+0x3e0/0x408<br /> [ 2911.185310] sp : ffff8000a1513900<br /> [ 2911.185750] x29: ffff8000a1513900 x28: ffff0003f272fe80 x27: 0000000000000001<br /> [ 2911.186600] x26: ffff0003f272fe80 x25: 0000000000000030 x24: 0000000000000008<br /> [ 2911.187458] x23: ffff0003c5788000 x22: ffff0003c16710c8 x21: ffff80008017f180<br /> [ 2911.188310] x20: ffff80008017f000 x19: ffff80008017f180 x18: ffffffffffffffff<br /> [ 2911.189160] x17: 0000000000000000 x16: 0000000000000000 x15: ffff8000a15134b8<br /> [ 2911.190015] x14: 0000000000000000 x13: 205d373432323154 x12: 5b5d313131333731<br /> [ 2911.190844] x11: 00000000fffeffff x10: 00000000fffeffff x9 : ffffd1b78274a13c<br /> [ 2911.191716] x8 : 000000000017ffe8 x7 : c0000000fffeffff x6 : 000000000057ffa8<br /> [ 2911.192554] x5 : ffff0012f6c24ec0 x4 : 0000000000000000 x3 : ffff2e5b72b5d000<br /> [ 2911.193404] x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffff0003ff254480<br /> [ 2911.194259] Call trace:<br /> [ 2911.194626] tracing_map_sort_entries+0x3e0/0x408<br /> [ 2911.195220] hist_show+0x124/0x800<br /> [ 2911.195692] seq_read_iter+0x1d4/0x4e8<br /> [ 2911.196193] seq_read+0xe8/0x138<br /> [ 2911.196638] vfs_read+0xc8/0x300<br /> [ 2911.197078] ksys_read+0x70/0x108<br /> [ 2911.197534] __arm64_sys_read+0x24/0x38<br /> [ 2911.198046] invoke_syscall+0x78/0x108<br /> [ 2911.198553] el0_svc_common.constprop.0+0xd0/0xf8<br /> [ 2911.199157] do_el0_svc+0x28/0x40<br /> [ 2911.199613] el0_svc+0x40/0x178<br /> [ 2911.200048] el0t_64_sync_handler+0x13c/0x158<br /> [ 2911.200621] el0t_64_sync+0x1a8/0x1b0<br /> [ 2911.201115] ---[ end trace 0000000000000000 ]---<br /> <br /> The problem appears to be caused by CPU reordering of writes issued from<br /> __tracing_map_insert().<br /> <br /> The check for the presence of an element with a given key in this<br /> function is:<br /> <br /> val = READ_ONCE(entry-&gt;val);<br /> if (val &amp;&amp; keys_match(key, val-&gt;key, map-&gt;key_size)) ...<br /> <br /> The write of a new entry is:<br /> <br /> elt = get_free_elt(map);<br /> memcpy(elt-&gt;key, key, map-&gt;key_size);<br /> entry-&gt;val = elt;<br /> <br /> The "memcpy(elt-&gt;key, key, map-&gt;key_size);" and "entry-&gt;val = elt;"<br /> stores may become visible in the reversed order on another CPU. This<br /> second CPU might then incorrectly determine that a new key doesn&amp;#39;t match<br /> an already present val-&gt;key and subse<br /> ---truncated---