From 40dddd4b8bd08a69471efd96107a4e1c73fabefc Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 9 Oct 2024 18:58:02 +0000 Subject: [PATCH 01/16] ppp: fix ppp_async_encode() illegal access syzbot reported an issue in ppp_async_encode() [1] In this case, pppoe_sendmsg() is called with a zero size. Then ppp_async_encode() is called with an empty skb. BUG: KMSAN: uninit-value in ppp_async_encode drivers/net/ppp/ppp_async.c:545 [inline] BUG: KMSAN: uninit-value in ppp_async_push+0xb4f/0x2660 drivers/net/ppp/ppp_async.c:675 ppp_async_encode drivers/net/ppp/ppp_async.c:545 [inline] ppp_async_push+0xb4f/0x2660 drivers/net/ppp/ppp_async.c:675 ppp_async_send+0x130/0x1b0 drivers/net/ppp/ppp_async.c:634 ppp_channel_bridge_input drivers/net/ppp/ppp_generic.c:2280 [inline] ppp_input+0x1f1/0xe60 drivers/net/ppp/ppp_generic.c:2304 pppoe_rcv_core+0x1d3/0x720 drivers/net/ppp/pppoe.c:379 sk_backlog_rcv+0x13b/0x420 include/net/sock.h:1113 __release_sock+0x1da/0x330 net/core/sock.c:3072 release_sock+0x6b/0x250 net/core/sock.c:3626 pppoe_sendmsg+0x2b8/0xb90 drivers/net/ppp/pppoe.c:903 sock_sendmsg_nosec net/socket.c:729 [inline] __sock_sendmsg+0x30f/0x380 net/socket.c:744 ____sys_sendmsg+0x903/0xb60 net/socket.c:2602 ___sys_sendmsg+0x28d/0x3c0 net/socket.c:2656 __sys_sendmmsg+0x3c1/0x960 net/socket.c:2742 __do_sys_sendmmsg net/socket.c:2771 [inline] __se_sys_sendmmsg net/socket.c:2768 [inline] __x64_sys_sendmmsg+0xbc/0x120 net/socket.c:2768 x64_sys_call+0xb6e/0x3ba0 arch/x86/include/generated/asm/syscalls_64.h:308 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xcd/0x1e0 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f Uninit was created at: slab_post_alloc_hook mm/slub.c:4092 [inline] slab_alloc_node mm/slub.c:4135 [inline] kmem_cache_alloc_node_noprof+0x6bf/0xb80 mm/slub.c:4187 kmalloc_reserve+0x13d/0x4a0 net/core/skbuff.c:587 __alloc_skb+0x363/0x7b0 net/core/skbuff.c:678 alloc_skb include/linux/skbuff.h:1322 [inline] sock_wmalloc+0xfe/0x1a0 net/core/sock.c:2732 pppoe_sendmsg+0x3a7/0xb90 drivers/net/ppp/pppoe.c:867 sock_sendmsg_nosec net/socket.c:729 [inline] __sock_sendmsg+0x30f/0x380 net/socket.c:744 ____sys_sendmsg+0x903/0xb60 net/socket.c:2602 ___sys_sendmsg+0x28d/0x3c0 net/socket.c:2656 __sys_sendmmsg+0x3c1/0x960 net/socket.c:2742 __do_sys_sendmmsg net/socket.c:2771 [inline] __se_sys_sendmmsg net/socket.c:2768 [inline] __x64_sys_sendmmsg+0xbc/0x120 net/socket.c:2768 x64_sys_call+0xb6e/0x3ba0 arch/x86/include/generated/asm/syscalls_64.h:308 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xcd/0x1e0 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f CPU: 1 UID: 0 PID: 5411 Comm: syz.1.14 Not tainted 6.12.0-rc1-syzkaller-00165-g360c1f1f24c6 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: syzbot+1d121645899e7692f92a@syzkaller.appspotmail.com Signed-off-by: Eric Dumazet Reviewed-by: Simon Horman Link: https://patch.msgid.link/20241009185802.3763282-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- drivers/net/ppp/ppp_async.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/ppp/ppp_async.c b/drivers/net/ppp/ppp_async.c index a940b9a67107..c97406c6004d 100644 --- a/drivers/net/ppp/ppp_async.c +++ b/drivers/net/ppp/ppp_async.c @@ -542,7 +542,7 @@ ppp_async_encode(struct asyncppp *ap) * and 7 (code-reject) must be sent as though no options * had been negotiated. */ - islcp = proto == PPP_LCP && 1 <= data[2] && data[2] <= 7; + islcp = proto == PPP_LCP && count >= 3 && 1 <= data[2] && data[2] <= 7; if (i == 0) { if (islcp) -- 2.51.0 From 6fd27ea183c208e478129a85e11d880fc70040f2 Mon Sep 17 00:00:00 2001 From: "D. Wythe" Date: Wed, 9 Oct 2024 14:55:16 +0800 Subject: [PATCH 02/16] net/smc: fix lacks of icsk_syn_mss with IPPROTO_SMC Eric report a panic on IPPROTO_SMC, and give the facts that when INET_PROTOSW_ICSK was set, icsk->icsk_sync_mss must be set too. Bug: Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000 Mem abort info: ESR = 0x0000000086000005 EC = 0x21: IABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 FSC = 0x05: level 1 translation fault user pgtable: 4k pages, 48-bit VAs, pgdp=00000001195d1000 [0000000000000000] pgd=0800000109c46003, p4d=0800000109c46003, pud=0000000000000000 Internal error: Oops: 0000000086000005 [#1] PREEMPT SMP Modules linked in: CPU: 1 UID: 0 PID: 8037 Comm: syz.3.265 Not tainted 6.11.0-rc7-syzkaller-g5f5673607153 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/06/2024 pstate: 80400005 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : 0x0 lr : cipso_v4_sock_setattr+0x2a8/0x3c0 net/ipv4/cipso_ipv4.c:1910 sp : ffff80009b887a90 x29: ffff80009b887aa0 x28: ffff80008db94050 x27: 0000000000000000 x26: 1fffe0001aa6f5b3 x25: dfff800000000000 x24: ffff0000db75da00 x23: 0000000000000000 x22: ffff0000d8b78518 x21: 0000000000000000 x20: ffff0000d537ad80 x19: ffff0000d8b78000 x18: 1fffe000366d79ee x17: ffff8000800614a8 x16: ffff800080569b84 x15: 0000000000000001 x14: 000000008b336894 x13: 00000000cd96feaa x12: 0000000000000003 x11: 0000000000040000 x10: 00000000000020a3 x9 : 1fffe0001b16f0f1 x8 : 0000000000000000 x7 : 0000000000000000 x6 : 000000000000003f x5 : 0000000000000040 x4 : 0000000000000001 x3 : 0000000000000000 x2 : 0000000000000002 x1 : 0000000000000000 x0 : ffff0000d8b78000 Call trace: 0x0 netlbl_sock_setattr+0x2e4/0x338 net/netlabel/netlabel_kapi.c:1000 smack_netlbl_add+0xa4/0x154 security/smack/smack_lsm.c:2593 smack_socket_post_create+0xa8/0x14c security/smack/smack_lsm.c:2973 security_socket_post_create+0x94/0xd4 security/security.c:4425 __sock_create+0x4c8/0x884 net/socket.c:1587 sock_create net/socket.c:1622 [inline] __sys_socket_create net/socket.c:1659 [inline] __sys_socket+0x134/0x340 net/socket.c:1706 __do_sys_socket net/socket.c:1720 [inline] __se_sys_socket net/socket.c:1718 [inline] __arm64_sys_socket+0x7c/0x94 net/socket.c:1718 __invoke_syscall arch/arm64/kernel/syscall.c:35 [inline] invoke_syscall+0x98/0x2b8 arch/arm64/kernel/syscall.c:49 el0_svc_common+0x130/0x23c arch/arm64/kernel/syscall.c:132 do_el0_svc+0x48/0x58 arch/arm64/kernel/syscall.c:151 el0_svc+0x54/0x168 arch/arm64/kernel/entry-common.c:712 el0t_64_sync_handler+0x84/0xfc arch/arm64/kernel/entry-common.c:730 el0t_64_sync+0x190/0x194 arch/arm64/kernel/entry.S:598 Code: ???????? ???????? ???????? ???????? (????????) ---[ end trace 0000000000000000 ]--- This patch add a toy implementation that performs a simple return to prevent such panic. This is because MSS can be set in sock_create_kern or smc_setsockopt, similar to how it's done in AF_SMC. However, for AF_SMC, there is currently no way to synchronize MSS within __sys_connect_file. This toy implementation lays the groundwork for us to support such feature for IPPROTO_SMC in the future. Fixes: d25a92ccae6b ("net/smc: Introduce IPPROTO_SMC") Reported-by: Eric Dumazet Signed-off-by: D. Wythe Reviewed-by: Eric Dumazet Reviewed-by: Wenjia Zhang Link: https://patch.msgid.link/1728456916-67035-1-git-send-email-alibuda@linux.alibaba.com Signed-off-by: Jakub Kicinski --- net/smc/smc_inet.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/net/smc/smc_inet.c b/net/smc/smc_inet.c index a5b2041600f9..a944e7dcb8b9 100644 --- a/net/smc/smc_inet.c +++ b/net/smc/smc_inet.c @@ -108,12 +108,23 @@ static struct inet_protosw smc_inet6_protosw = { }; #endif /* CONFIG_IPV6 */ +static unsigned int smc_sync_mss(struct sock *sk, u32 pmtu) +{ + /* No need pass it through to clcsock, mss can always be set by + * sock_create_kern or smc_setsockopt. + */ + return 0; +} + static int smc_inet_init_sock(struct sock *sk) { struct net *net = sock_net(sk); /* init common smc sock */ smc_sk_init(net, sk, IPPROTO_SMC); + + inet_csk(sk)->icsk_sync_mss = smc_sync_mss; + /* create clcsock */ return smc_create_clcsk(net, sk, sk->sk_family); } -- 2.51.0 From 7d3fce8cbe3a70a1c7c06c9b53696be5d5d8dd5c Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 9 Oct 2024 09:11:32 +0000 Subject: [PATCH 03/16] slip: make slhc_remember() more robust against malicious packets syzbot found that slhc_remember() was missing checks against malicious packets [1]. slhc_remember() only checked the size of the packet was at least 20, which is not good enough. We need to make sure the packet includes the IPv4 and TCP header that are supposed to be carried. Add iph and th pointers to make the code more readable. [1] BUG: KMSAN: uninit-value in slhc_remember+0x2e8/0x7b0 drivers/net/slip/slhc.c:666 slhc_remember+0x2e8/0x7b0 drivers/net/slip/slhc.c:666 ppp_receive_nonmp_frame+0xe45/0x35e0 drivers/net/ppp/ppp_generic.c:2455 ppp_receive_frame drivers/net/ppp/ppp_generic.c:2372 [inline] ppp_do_recv+0x65f/0x40d0 drivers/net/ppp/ppp_generic.c:2212 ppp_input+0x7dc/0xe60 drivers/net/ppp/ppp_generic.c:2327 pppoe_rcv_core+0x1d3/0x720 drivers/net/ppp/pppoe.c:379 sk_backlog_rcv+0x13b/0x420 include/net/sock.h:1113 __release_sock+0x1da/0x330 net/core/sock.c:3072 release_sock+0x6b/0x250 net/core/sock.c:3626 pppoe_sendmsg+0x2b8/0xb90 drivers/net/ppp/pppoe.c:903 sock_sendmsg_nosec net/socket.c:729 [inline] __sock_sendmsg+0x30f/0x380 net/socket.c:744 ____sys_sendmsg+0x903/0xb60 net/socket.c:2602 ___sys_sendmsg+0x28d/0x3c0 net/socket.c:2656 __sys_sendmmsg+0x3c1/0x960 net/socket.c:2742 __do_sys_sendmmsg net/socket.c:2771 [inline] __se_sys_sendmmsg net/socket.c:2768 [inline] __x64_sys_sendmmsg+0xbc/0x120 net/socket.c:2768 x64_sys_call+0xb6e/0x3ba0 arch/x86/include/generated/asm/syscalls_64.h:308 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xcd/0x1e0 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f Uninit was created at: slab_post_alloc_hook mm/slub.c:4091 [inline] slab_alloc_node mm/slub.c:4134 [inline] kmem_cache_alloc_node_noprof+0x6bf/0xb80 mm/slub.c:4186 kmalloc_reserve+0x13d/0x4a0 net/core/skbuff.c:587 __alloc_skb+0x363/0x7b0 net/core/skbuff.c:678 alloc_skb include/linux/skbuff.h:1322 [inline] sock_wmalloc+0xfe/0x1a0 net/core/sock.c:2732 pppoe_sendmsg+0x3a7/0xb90 drivers/net/ppp/pppoe.c:867 sock_sendmsg_nosec net/socket.c:729 [inline] __sock_sendmsg+0x30f/0x380 net/socket.c:744 ____sys_sendmsg+0x903/0xb60 net/socket.c:2602 ___sys_sendmsg+0x28d/0x3c0 net/socket.c:2656 __sys_sendmmsg+0x3c1/0x960 net/socket.c:2742 __do_sys_sendmmsg net/socket.c:2771 [inline] __se_sys_sendmmsg net/socket.c:2768 [inline] __x64_sys_sendmmsg+0xbc/0x120 net/socket.c:2768 x64_sys_call+0xb6e/0x3ba0 arch/x86/include/generated/asm/syscalls_64.h:308 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xcd/0x1e0 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f CPU: 0 UID: 0 PID: 5460 Comm: syz.2.33 Not tainted 6.12.0-rc2-syzkaller-00006-g87d6aab2389e #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024 Fixes: b5451d783ade ("slip: Move the SLIP drivers") Reported-by: syzbot+2ada1bc857496353be5a@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/670646db.050a0220.3f80e.0027.GAE@google.com/T/#u Signed-off-by: Eric Dumazet Link: https://patch.msgid.link/20241009091132.2136321-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- drivers/net/slip/slhc.c | 57 ++++++++++++++++++++++++----------------- 1 file changed, 34 insertions(+), 23 deletions(-) diff --git a/drivers/net/slip/slhc.c b/drivers/net/slip/slhc.c index 252cd757d3a2..ee9fd3a94b96 100644 --- a/drivers/net/slip/slhc.c +++ b/drivers/net/slip/slhc.c @@ -643,46 +643,57 @@ bad: int slhc_remember(struct slcompress *comp, unsigned char *icp, int isize) { - struct cstate *cs; - unsigned ihl; - + const struct tcphdr *th; unsigned char index; + struct iphdr *iph; + struct cstate *cs; + unsigned int ihl; - if(isize < 20) { - /* The packet is shorter than a legal IP header */ + /* The packet is shorter than a legal IP header. + * Also make sure isize is positive. + */ + if (isize < (int)sizeof(struct iphdr)) { +runt: comp->sls_i_runt++; - return slhc_toss( comp ); + return slhc_toss(comp); } + iph = (struct iphdr *)icp; /* Peek at the IP header's IHL field to find its length */ - ihl = icp[0] & 0xf; - if(ihl < 20 / 4){ - /* The IP header length field is too small */ - comp->sls_i_runt++; - return slhc_toss( comp ); - } - index = icp[9]; - icp[9] = IPPROTO_TCP; + ihl = iph->ihl; + /* The IP header length field is too small, + * or packet is shorter than the IP header followed + * by minimal tcp header. + */ + if (ihl < 5 || isize < ihl * 4 + sizeof(struct tcphdr)) + goto runt; + + index = iph->protocol; + iph->protocol = IPPROTO_TCP; if (ip_fast_csum(icp, ihl)) { /* Bad IP header checksum; discard */ comp->sls_i_badcheck++; - return slhc_toss( comp ); + return slhc_toss(comp); } - if(index > comp->rslot_limit) { + if (index > comp->rslot_limit) { comp->sls_i_error++; return slhc_toss(comp); } - + th = (struct tcphdr *)(icp + ihl * 4); + if (th->doff < sizeof(struct tcphdr) / 4) + goto runt; + if (isize < ihl * 4 + th->doff * 4) + goto runt; /* Update local state */ cs = &comp->rstate[comp->recv_current = index]; comp->flags &=~ SLF_TOSS; - memcpy(&cs->cs_ip,icp,20); - memcpy(&cs->cs_tcp,icp + ihl*4,20); + memcpy(&cs->cs_ip, iph, sizeof(*iph)); + memcpy(&cs->cs_tcp, th, sizeof(*th)); if (ihl > 5) - memcpy(cs->cs_ipopt, icp + sizeof(struct iphdr), (ihl - 5) * 4); - if (cs->cs_tcp.doff > 5) - memcpy(cs->cs_tcpopt, icp + ihl*4 + sizeof(struct tcphdr), (cs->cs_tcp.doff - 5) * 4); - cs->cs_hsize = ihl*2 + cs->cs_tcp.doff*2; + memcpy(cs->cs_ipopt, &iph[1], (ihl - 5) * 4); + if (th->doff > 5) + memcpy(cs->cs_tcpopt, &th[1], (th->doff - 5) * 4); + cs->cs_hsize = ihl*2 + th->doff*2; cs->initialized = true; /* Put headers back on packet * Neither header checksum is recalculated -- 2.51.0 From 9937aae39bc09645cd67d53e0320926cd91570de Mon Sep 17 00:00:00 2001 From: Simon Horman Date: Wed, 9 Oct 2024 09:47:22 +0100 Subject: [PATCH 04/16] MAINTAINERS: consistently exclude wireless files from NETWORKING [GENERAL] We already exclude wireless drivers from the netdev@ traffic, to delegate it to linux-wireless@, and avoid overwhelming netdev@. Many of the following wireless-related sections MAINTAINERS are already not included in the NETWORKING [GENERAL] section. For consistency, exclude those that are. * 802.11 (including CFG80211/NL80211) * MAC80211 * RFKILL Acked-by: Johannes Berg Signed-off-by: Simon Horman Link: https://patch.msgid.link/20241009-maint-net-hdrs-v2-1-f2c86e7309c8@kernel.org Signed-off-by: Jakub Kicinski --- MAINTAINERS | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/MAINTAINERS b/MAINTAINERS index 58380aeafbf0..89789bddfb6a 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -16197,8 +16197,19 @@ F: lib/random32.c F: net/ F: tools/net/ F: tools/testing/selftests/net/ +X: Documentation/networking/mac80211-injection.rst +X: Documentation/networking/mac80211_hwsim/ +X: Documentation/networking/regulatory.rst +X: include/net/cfg80211.h +X: include/net/ieee80211_radiotap.h +X: include/net/iw_handler.h +X: include/net/mac80211.h +X: include/net/wext.h X: net/9p/ X: net/bluetooth/ +X: net/mac80211/ +X: net/rfkill/ +X: net/wireless/ NETWORKING [IPSEC] M: Steffen Klassert -- 2.51.0 From 5404b5a2fea9831a1f5be4ab9a94de07d976b177 Mon Sep 17 00:00:00 2001 From: Simon Horman Date: Wed, 9 Oct 2024 09:47:23 +0100 Subject: [PATCH 05/16] MAINTAINERS: Add headers and mailing list to UDP section Add netdev mailing list and some more udp.h headers to the UDP section. This is now more consistent with the TCP section. Acked-by: Willem de Bruijn Signed-off-by: Simon Horman Link: https://patch.msgid.link/20241009-maint-net-hdrs-v2-2-f2c86e7309c8@kernel.org Signed-off-by: Jakub Kicinski --- MAINTAINERS | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/MAINTAINERS b/MAINTAINERS index 89789bddfb6a..baf2ba9dc070 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -24184,8 +24184,12 @@ F: drivers/usb/host/xhci* USER DATAGRAM PROTOCOL (UDP) M: Willem de Bruijn +L: netdev@vger.kernel.org S: Maintained F: include/linux/udp.h +F: include/net/udp.h +F: include/trace/events/udp.h +F: include/uapi/linux/udp.h F: net/ipv4/udp.c F: net/ipv6/udp.c -- 2.51.0 From f7345ccc62a4b880cf76458db5f320725f28e400 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 10 Oct 2024 18:36:09 +0200 Subject: [PATCH 06/16] rcu/nocb: Fix rcuog wake-up from offline softirq After a CPU has set itself offline and before it eventually calls rcutree_report_cpu_dead(), there are still opportunities for callbacks to be enqueued, for example from a softirq. When that happens on NOCB, the rcuog wake-up is deferred through an IPI to an online CPU in order not to call into the scheduler and risk arming the RT-bandwidth after hrtimers have been migrated out and disabled. But performing a synchronized IPI from a softirq is buggy as reported in the following scenario: WARNING: CPU: 1 PID: 26 at kernel/smp.c:633 smp_call_function_single Modules linked in: rcutorture torture CPU: 1 UID: 0 PID: 26 Comm: migration/1 Not tainted 6.11.0-rc1-00012-g9139f93209d1 #1 Stopper: multi_cpu_stop+0x0/0x320 <- __stop_cpus+0xd0/0x120 RIP: 0010:smp_call_function_single swake_up_one_online __call_rcu_nocb_wake __call_rcu_common ? rcu_torture_one_read call_timer_fn __run_timers run_timer_softirq handle_softirqs irq_exit_rcu ? tick_handle_periodic sysvec_apic_timer_interrupt Fix this with forcing deferred rcuog wake up through the NOCB timer when the CPU is offline. The actual wake up will happen from rcutree_report_cpu_dead(). Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-lkp/202409231644.4c55582d-lkp@intel.com Fixes: 9139f93209d1 ("rcu/nocb: Fix RT throttling hrtimer armed from offline CPU") Reviewed-by: "Joel Fernandes (Google)" Signed-off-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree_nocb.h | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h index 97b99cd06923..16865475120b 100644 --- a/kernel/rcu/tree_nocb.h +++ b/kernel/rcu/tree_nocb.h @@ -554,13 +554,19 @@ static void __call_rcu_nocb_wake(struct rcu_data *rdp, bool was_alldone, rcu_nocb_unlock(rdp); wake_nocb_gp_defer(rdp, RCU_NOCB_WAKE_LAZY, TPS("WakeLazy")); - } else if (!irqs_disabled_flags(flags)) { + } else if (!irqs_disabled_flags(flags) && cpu_online(rdp->cpu)) { /* ... if queue was empty ... */ rcu_nocb_unlock(rdp); wake_nocb_gp(rdp, false); trace_rcu_nocb_wake(rcu_state.name, rdp->cpu, TPS("WakeEmpty")); } else { + /* + * Don't do the wake-up upfront on fragile paths. + * Also offline CPUs can't call swake_up_one_online() from + * (soft-)IRQs. Rely on the final deferred wake-up from + * rcutree_report_cpu_dead() + */ rcu_nocb_unlock(rdp); wake_nocb_gp_defer(rdp, RCU_NOCB_WAKE, TPS("WakeEmptyIsDeferred")); -- 2.51.0 From 6e0391e48cf9fb8b1b5e27c0cbbaf2e4639f2c33 Mon Sep 17 00:00:00 2001 From: Stephen Boyd Date: Wed, 9 Oct 2024 13:41:31 -0700 Subject: [PATCH 07/16] of: Skip kunit tests when arm64+ACPI doesn't populate root node A root node is required to apply DT overlays. A root node is usually present after commit 7b937cc243e5 ("of: Create of_root if no dtb provided by firmware"), except for on arm64 systems booted with ACPI tables. In that case, the root node is intentionally not populated because it would "allow DT devices to be instantiated atop an ACPI base system"[1]. Introduce an OF function that skips the kunit test if the root node isn't populated. Limit the test to when both CONFIG_ARM64 and CONFIG_ACPI are set, because otherwise the lack of a root node is a bug. Make the function private and take a kunit test parameter so that it can't be abused to test for the presence of the root node in non-test code. Use this function to skip tests that require the root node. Currently that's the DT tests and any tests that apply overlays. Reported-by: Guenter Roeck Closes: https://lore.kernel.org/r/6cd337fb-38f0-41cb-b942-5844b84433db@roeck-us.net Link: https://lore.kernel.org/r/Zd4dQpHO7em1ji67@FVFF77S0Q05N.cambridge.arm.com [1] Fixes: 893ecc6d2d61 ("of: Add KUnit test to confirm DTB is loaded") Signed-off-by: Stephen Boyd Tested-by: Guenter Roeck Acked-by: Mark Rutland Link: https://lore.kernel.org/r/20241009204133.1169931-1-sboyd@kernel.org Signed-off-by: Rob Herring (Arm) --- drivers/of/of_kunit_helpers.c | 15 +++++++++++++++ drivers/of/of_private.h | 3 +++ drivers/of/of_test.c | 3 +++ drivers/of/overlay_test.c | 3 +++ 4 files changed, 24 insertions(+) diff --git a/drivers/of/of_kunit_helpers.c b/drivers/of/of_kunit_helpers.c index 287d6c91bb37..7b3ed5a382aa 100644 --- a/drivers/of/of_kunit_helpers.c +++ b/drivers/of/of_kunit_helpers.c @@ -10,6 +10,19 @@ #include #include +#include "of_private.h" + +/** + * of_root_kunit_skip() - Skip test if the root node isn't populated + * @test: test to skip if the root node isn't populated + */ +void of_root_kunit_skip(struct kunit *test) +{ + if (IS_ENABLED(CONFIG_ARM64) && IS_ENABLED(CONFIG_ACPI) && !of_root) + kunit_skip(test, "arm64+acpi doesn't populate a root node"); +} +EXPORT_SYMBOL_GPL(of_root_kunit_skip); + #if defined(CONFIG_OF_OVERLAY) && defined(CONFIG_OF_EARLY_FLATTREE) static void of_overlay_fdt_apply_kunit_exit(void *ovcs_id) @@ -36,6 +49,8 @@ int of_overlay_fdt_apply_kunit(struct kunit *test, void *overlay_fdt, int ret; int *copy_id; + of_root_kunit_skip(test); + copy_id = kunit_kmalloc(test, sizeof(*copy_id), GFP_KERNEL); if (!copy_id) return -ENOMEM; diff --git a/drivers/of/of_private.h b/drivers/of/of_private.h index 04aa2a91f851..c235d6c909a1 100644 --- a/drivers/of/of_private.h +++ b/drivers/of/of_private.h @@ -42,6 +42,9 @@ extern raw_spinlock_t devtree_lock; extern struct list_head aliases_lookup; extern struct kset *of_kset; +struct kunit; +extern void of_root_kunit_skip(struct kunit *test); + #if defined(CONFIG_OF_DYNAMIC) extern int of_property_notify(int action, struct device_node *np, struct property *prop, struct property *old_prop); diff --git a/drivers/of/of_test.c b/drivers/of/of_test.c index c85a258bc6ae..b0557ded838f 100644 --- a/drivers/of/of_test.c +++ b/drivers/of/of_test.c @@ -7,6 +7,8 @@ #include +#include "of_private.h" + /* * Test that the root node "/" can be found by path. */ @@ -36,6 +38,7 @@ static struct kunit_case of_dtb_test_cases[] = { static int of_dtb_test_init(struct kunit *test) { + of_root_kunit_skip(test); if (!IS_ENABLED(CONFIG_OF_EARLY_FLATTREE)) kunit_skip(test, "requires CONFIG_OF_EARLY_FLATTREE"); diff --git a/drivers/of/overlay_test.c b/drivers/of/overlay_test.c index 19695bdf77be..1f76d50fb16a 100644 --- a/drivers/of/overlay_test.c +++ b/drivers/of/overlay_test.c @@ -11,6 +11,8 @@ #include #include +#include "of_private.h" + static const char * const kunit_node_name = "kunit-test"; static const char * const kunit_compatible = "test,empty"; @@ -62,6 +64,7 @@ static void of_overlay_apply_kunit_cleanup(struct kunit *test) struct device *dev; struct device_node *np; + of_root_kunit_skip(test); if (!IS_ENABLED(CONFIG_OF_EARLY_FLATTREE)) kunit_skip(test, "requires CONFIG_OF_EARLY_FLATTREE for root node"); -- 2.51.0 From 8956c582ac6b1693a351230179f898979dd00bdf Mon Sep 17 00:00:00 2001 From: Christophe Leroy Date: Sat, 5 Oct 2024 10:53:29 +0200 Subject: [PATCH 08/16] powerpc/8xx: Fix kernel DTLB miss on dcbz Following OOPS is encountered while loading test_bpf module on powerpc 8xx: [ 218.835567] BUG: Unable to handle kernel data access on write at 0xcb000000 [ 218.842473] Faulting instruction address: 0xc0017a80 [ 218.847451] Oops: Kernel access of bad area, sig: 11 [#1] [ 218.852854] BE PAGE_SIZE=16K PREEMPT CMPC885 [ 218.857207] SAF3000 DIE NOTIFICATION [ 218.860713] Modules linked in: test_bpf(+) test_module [ 218.865867] CPU: 0 UID: 0 PID: 527 Comm: insmod Not tainted 6.11.0-s3k-dev-09856-g3de3d71ae2e6-dirty #1280 [ 218.875546] Hardware name: MIAE 8xx 0x500000 CMPC885 [ 218.880521] NIP: c0017a80 LR: beab859c CTR: 000101d4 [ 218.885584] REGS: cac2bc90 TRAP: 0300 Not tainted (6.11.0-s3k-dev-09856-g3de3d71ae2e6-dirty) [ 218.894308] MSR: 00009032 CR: 55005555 XER: a0007100 [ 218.901290] DAR: cb000000 DSISR: c2000000 [ 218.901290] GPR00: 000185d1 cac2bd50 c21b9580 caf7c030 c3883fcc 00000008 cafffffc 00000000 [ 218.901290] GPR08: 00040000 18300000 20000000 00000004 99005555 100d815e ca669d08 00000369 [ 218.901290] GPR16: ca730000 00000000 ca2c004c 00000000 00000000 0000035d 00000311 00000369 [ 218.901290] GPR24: ca732240 00000001 00030ba3 c3800000 00000000 00185d48 caf7c000 ca2c004c [ 218.941087] NIP [c0017a80] memcpy+0x88/0xec [ 218.945277] LR [beab859c] test_bpf_init+0x22c/0x3c90 [test_bpf] [ 218.951476] Call Trace: [ 218.953916] [cac2bd50] [beab8570] test_bpf_init+0x200/0x3c90 [test_bpf] (unreliable) [ 218.962034] [cac2bde0] [c0004c04] do_one_initcall+0x4c/0x1fc [ 218.967706] [cac2be40] [c00a2ec4] do_init_module+0x68/0x360 [ 218.973292] [cac2be60] [c00a5194] init_module_from_file+0x8c/0xc0 [ 218.979401] [cac2bed0] [c00a5568] sys_finit_module+0x250/0x3f0 [ 218.985248] [cac2bf20] [c000e390] system_call_exception+0x8c/0x15c [ 218.991444] [cac2bf30] [c00120a8] ret_from_syscall+0x0/0x28 This happens in the main loop of memcpy() ==> c0017a80: 7c 0b 37 ec dcbz r11,r6 c0017a84: 80 e4 00 04 lwz r7,4(r4) c0017a88: 81 04 00 08 lwz r8,8(r4) c0017a8c: 81 24 00 0c lwz r9,12(r4) c0017a90: 85 44 00 10 lwzu r10,16(r4) c0017a94: 90 e6 00 04 stw r7,4(r6) c0017a98: 91 06 00 08 stw r8,8(r6) c0017a9c: 91 26 00 0c stw r9,12(r6) c0017aa0: 95 46 00 10 stwu r10,16(r6) c0017aa4: 42 00 ff dc bdnz c0017a80 Commit ac9f97ff8b32 ("powerpc/8xx: Inconditionally use task PGDIR in DTLB misses") relies on re-reading DAR register to know if an error is due to a missing copy of a PMD entry in task's PGDIR, allthough DAR was already read in the exception prolog and copied into thread struct. This is because is it done very early in the exception and there are not enough registers available to keep a pointer to thread struct. However, dcbz instruction is buggy and doesn't update DAR register on fault. That is detected and generates a call to FixupDAR workaround which updates DAR copy in thread struct but doesn't fix DAR register. Let's fix DAR in addition to the update of DAR copy in thread struct. Fixes: ac9f97ff8b32 ("powerpc/8xx: Inconditionally use task PGDIR in DTLB misses") Signed-off-by: Christophe Leroy Signed-off-by: Michael Ellerman Link: https://msgid.link/2b851399bd87e81c6ccb87ea3a7a6b32c7aa04d7.1728118396.git.christophe.leroy@csgroup.eu --- arch/powerpc/kernel/head_8xx.S | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/powerpc/kernel/head_8xx.S b/arch/powerpc/kernel/head_8xx.S index 811a7130505c..56c5ebe21b99 100644 --- a/arch/powerpc/kernel/head_8xx.S +++ b/arch/powerpc/kernel/head_8xx.S @@ -494,6 +494,7 @@ FixupDAR:/* Entry point for dcbx workaround. */ bctr /* jump into table */ 152: mfdar r11 + mtdar r10 mtctr r11 /* restore ctr reg from DAR */ mfspr r11, SPRN_SPRG_THREAD stw r10, DAR(r11) -- 2.51.0 From a0cc649353bb726d4aa0db60dce467432197b746 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Tue, 8 Oct 2024 21:28:01 -0400 Subject: [PATCH 09/16] selftests/rseq: Fix mm_cid test failure Adapt the rseq.c/rseq.h code to follow GNU C library changes introduced by: glibc commit 2e456ccf0c34 ("Linux: Make __rseq_size useful for feature detection (bug 31965)") Without this fix, rseq selftests for mm_cid fail: ./run_param_test.sh Default parameters Running test spinlock Running compare-twice test spinlock Running mm_cid test spinlock Error: cpu id getter unavailable Fixes: 18c2355838e7 ("selftests/rseq: Implement rseq mm_cid field support") Signed-off-by: Mathieu Desnoyers Cc: Peter Zijlstra CC: Boqun Feng CC: "Paul E. McKenney" Cc: Shuah Khan CC: Carlos O'Donell CC: Florian Weimer CC: linux-kselftest@vger.kernel.org CC: stable@vger.kernel.org Signed-off-by: Shuah Khan --- tools/testing/selftests/rseq/rseq.c | 110 +++++++++++++++++++--------- tools/testing/selftests/rseq/rseq.h | 10 +-- 2 files changed, 77 insertions(+), 43 deletions(-) diff --git a/tools/testing/selftests/rseq/rseq.c b/tools/testing/selftests/rseq/rseq.c index 96e812bdf8a4..5b9772cdf265 100644 --- a/tools/testing/selftests/rseq/rseq.c +++ b/tools/testing/selftests/rseq/rseq.c @@ -60,12 +60,6 @@ unsigned int rseq_size = -1U; /* Flags used during rseq registration. */ unsigned int rseq_flags; -/* - * rseq feature size supported by the kernel. 0 if the registration was - * unsuccessful. - */ -unsigned int rseq_feature_size = -1U; - static int rseq_ownership; static int rseq_reg_success; /* At least one rseq registration has succeded. */ @@ -111,6 +105,43 @@ int rseq_available(void) } } +/* The rseq areas need to be at least 32 bytes. */ +static +unsigned int get_rseq_min_alloc_size(void) +{ + unsigned int alloc_size = rseq_size; + + if (alloc_size < ORIG_RSEQ_ALLOC_SIZE) + alloc_size = ORIG_RSEQ_ALLOC_SIZE; + return alloc_size; +} + +/* + * Return the feature size supported by the kernel. + * + * Depending on the value returned by getauxval(AT_RSEQ_FEATURE_SIZE): + * + * 0: Return ORIG_RSEQ_FEATURE_SIZE (20) + * > 0: Return the value from getauxval(AT_RSEQ_FEATURE_SIZE). + * + * It should never return a value below ORIG_RSEQ_FEATURE_SIZE. + */ +static +unsigned int get_rseq_kernel_feature_size(void) +{ + unsigned long auxv_rseq_feature_size, auxv_rseq_align; + + auxv_rseq_align = getauxval(AT_RSEQ_ALIGN); + assert(!auxv_rseq_align || auxv_rseq_align <= RSEQ_THREAD_AREA_ALLOC_SIZE); + + auxv_rseq_feature_size = getauxval(AT_RSEQ_FEATURE_SIZE); + assert(!auxv_rseq_feature_size || auxv_rseq_feature_size <= RSEQ_THREAD_AREA_ALLOC_SIZE); + if (auxv_rseq_feature_size) + return auxv_rseq_feature_size; + else + return ORIG_RSEQ_FEATURE_SIZE; +} + int rseq_register_current_thread(void) { int rc; @@ -119,7 +150,7 @@ int rseq_register_current_thread(void) /* Treat libc's ownership as a successful registration. */ return 0; } - rc = sys_rseq(&__rseq_abi, rseq_size, 0, RSEQ_SIG); + rc = sys_rseq(&__rseq_abi, get_rseq_min_alloc_size(), 0, RSEQ_SIG); if (rc) { if (RSEQ_READ_ONCE(rseq_reg_success)) { /* Incoherent success/failure within process. */ @@ -140,28 +171,12 @@ int rseq_unregister_current_thread(void) /* Treat libc's ownership as a successful unregistration. */ return 0; } - rc = sys_rseq(&__rseq_abi, rseq_size, RSEQ_ABI_FLAG_UNREGISTER, RSEQ_SIG); + rc = sys_rseq(&__rseq_abi, get_rseq_min_alloc_size(), RSEQ_ABI_FLAG_UNREGISTER, RSEQ_SIG); if (rc) return -1; return 0; } -static -unsigned int get_rseq_feature_size(void) -{ - unsigned long auxv_rseq_feature_size, auxv_rseq_align; - - auxv_rseq_align = getauxval(AT_RSEQ_ALIGN); - assert(!auxv_rseq_align || auxv_rseq_align <= RSEQ_THREAD_AREA_ALLOC_SIZE); - - auxv_rseq_feature_size = getauxval(AT_RSEQ_FEATURE_SIZE); - assert(!auxv_rseq_feature_size || auxv_rseq_feature_size <= RSEQ_THREAD_AREA_ALLOC_SIZE); - if (auxv_rseq_feature_size) - return auxv_rseq_feature_size; - else - return ORIG_RSEQ_FEATURE_SIZE; -} - static __attribute__((constructor)) void rseq_init(void) { @@ -178,28 +193,54 @@ void rseq_init(void) } if (libc_rseq_size_p && libc_rseq_offset_p && libc_rseq_flags_p && *libc_rseq_size_p != 0) { + unsigned int libc_rseq_size; + /* rseq registration owned by glibc */ rseq_offset = *libc_rseq_offset_p; - rseq_size = *libc_rseq_size_p; + libc_rseq_size = *libc_rseq_size_p; rseq_flags = *libc_rseq_flags_p; - rseq_feature_size = get_rseq_feature_size(); - if (rseq_feature_size > rseq_size) - rseq_feature_size = rseq_size; + + /* + * Previous versions of glibc expose the value + * 32 even though the kernel only supported 20 + * bytes initially. Therefore treat 32 as a + * special-case. glibc 2.40 exposes a 20 bytes + * __rseq_size without using getauxval(3) to + * query the supported size, while still allocating a 32 + * bytes area. Also treat 20 as a special-case. + * + * Special-cases are handled by using the following + * value as active feature set size: + * + * rseq_size = min(32, get_rseq_kernel_feature_size()) + */ + switch (libc_rseq_size) { + case ORIG_RSEQ_FEATURE_SIZE: + fallthrough; + case ORIG_RSEQ_ALLOC_SIZE: + { + unsigned int rseq_kernel_feature_size = get_rseq_kernel_feature_size(); + + if (rseq_kernel_feature_size < ORIG_RSEQ_ALLOC_SIZE) + rseq_size = rseq_kernel_feature_size; + else + rseq_size = ORIG_RSEQ_ALLOC_SIZE; + break; + } + default: + /* Otherwise just use the __rseq_size from libc as rseq_size. */ + rseq_size = libc_rseq_size; + break; + } return; } rseq_ownership = 1; if (!rseq_available()) { rseq_size = 0; - rseq_feature_size = 0; return; } rseq_offset = (void *)&__rseq_abi - rseq_thread_pointer(); rseq_flags = 0; - rseq_feature_size = get_rseq_feature_size(); - if (rseq_feature_size == ORIG_RSEQ_FEATURE_SIZE) - rseq_size = ORIG_RSEQ_ALLOC_SIZE; - else - rseq_size = RSEQ_THREAD_AREA_ALLOC_SIZE; } static __attribute__((destructor)) @@ -209,7 +250,6 @@ void rseq_exit(void) return; rseq_offset = 0; rseq_size = -1U; - rseq_feature_size = -1U; rseq_ownership = 0; } diff --git a/tools/testing/selftests/rseq/rseq.h b/tools/testing/selftests/rseq/rseq.h index d7364ea4d201..4e217b620e0c 100644 --- a/tools/testing/selftests/rseq/rseq.h +++ b/tools/testing/selftests/rseq/rseq.h @@ -68,12 +68,6 @@ extern unsigned int rseq_size; /* Flags used during rseq registration. */ extern unsigned int rseq_flags; -/* - * rseq feature size supported by the kernel. 0 if the registration was - * unsuccessful. - */ -extern unsigned int rseq_feature_size; - enum rseq_mo { RSEQ_MO_RELAXED = 0, RSEQ_MO_CONSUME = 1, /* Unused */ @@ -193,7 +187,7 @@ static inline uint32_t rseq_current_cpu(void) static inline bool rseq_node_id_available(void) { - return (int) rseq_feature_size >= rseq_offsetofend(struct rseq_abi, node_id); + return (int) rseq_size >= rseq_offsetofend(struct rseq_abi, node_id); } /* @@ -207,7 +201,7 @@ static inline uint32_t rseq_current_node_id(void) static inline bool rseq_mm_cid_available(void) { - return (int) rseq_feature_size >= rseq_offsetofend(struct rseq_abi, mm_cid); + return (int) rseq_size >= rseq_offsetofend(struct rseq_abi, mm_cid); } static inline uint32_t rseq_current_mm_cid(void) -- 2.51.0 From 4ee5ca9a29384fcf3f18232fdf8474166dea8dca Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Oct 2024 16:52:35 -0400 Subject: [PATCH 10/16] ftrace/selftest: Test combination of function_graph tracer and function profiler Masami reported a bug when running function graph tracing then the function profiler. The following commands would cause a kernel crash: # cd /sys/kernel/tracing/ # echo function_graph > current_tracer # echo 1 > function_profile_enabled In that order. Create a test to test this two to make sure this does not come back as a regression. Link: https://lore.kernel.org/172398528350.293426.8347220120333730248.stgit@devnote2 Link: https://lore.kernel.org/all/20241010165235.35122877@gandalf.local.home/ Acked-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) Signed-off-by: Shuah Khan --- .../ftrace/test.d/ftrace/fgraph-profiler.tc | 31 +++++++++++++++++++ 1 file changed, 31 insertions(+) create mode 100644 tools/testing/selftests/ftrace/test.d/ftrace/fgraph-profiler.tc diff --git a/tools/testing/selftests/ftrace/test.d/ftrace/fgraph-profiler.tc b/tools/testing/selftests/ftrace/test.d/ftrace/fgraph-profiler.tc new file mode 100644 index 000000000000..ffff8646733c --- /dev/null +++ b/tools/testing/selftests/ftrace/test.d/ftrace/fgraph-profiler.tc @@ -0,0 +1,31 @@ +#!/bin/sh +# SPDX-License-Identifier: GPL-2.0 +# description: ftrace - function profiler with function graph tracing +# requires: function_profile_enabled set_ftrace_filter function_graph:tracer + +# The function graph tracer can now be run along side of the function +# profiler. But there was a bug that caused the combination of the two +# to crash. It also required the function graph tracer to be started +# first. +# +# This test triggers that bug +# +# We need both function_graph and profiling to run this test + +fail() { # mesg + echo $1 + exit_fail +} + +echo "Enabling function graph tracer:" +echo function_graph > current_tracer +echo "enable profiler" + +# Older kernels do not allow function_profile to be enabled with +# function graph tracer. If the below fails, mark it as unsupported +echo 1 > function_profile_enabled || exit_unsupported + +# Let it run for a bit to make sure nothing explodes +sleep 1 + +exit 0 -- 2.51.0 From 8e929cb546ee42c9a61d24fae60605e9e3192354 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 13 Oct 2024 14:33:32 -0700 Subject: [PATCH 11/16] Linux 6.12-rc3 --- Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Makefile b/Makefile index c5493c0c0ca1..8cf3cf528892 100644 --- a/Makefile +++ b/Makefile @@ -2,7 +2,7 @@ VERSION = 6 PATCHLEVEL = 12 SUBLEVEL = 0 -EXTRAVERSION = -rc2 +EXTRAVERSION = -rc3 NAME = Baby Opossum Posse # *DOCUMENTATION* -- 2.51.0 From ed1cb76ebdeb88cf0603b9cb543f43f09ab704a1 Mon Sep 17 00:00:00 2001 From: Josh Poimboeuf Date: Thu, 3 Oct 2024 17:31:10 -0700 Subject: [PATCH 12/16] objtool: Detect non-relocated text references When kernel IBT is enabled, objtool detects all text references in order to determine which functions can be indirectly branched to. In text, such references look like one of the following: mov $0x0,%rax R_X86_64_32S .init.text+0x7e0a0 lea 0x0(%rip),%rax R_X86_64_PC32 autoremove_wake_function-0x4 Either way the function pointer is denoted by a relocation, so objtool just reads that. However there are some "lea xxx(%rip)" cases which don't use relocations because they're referencing code in the same translation unit. Objtool doesn't have visibility to those. The only currently known instances of that are a few hand-coded asm text references which don't actually need ENDBR. So it's not actually a problem at the moment. However if we enable -fpie, the compiler would start generating them and there would definitely be bugs in the IBT sealing. Detect non-relocated text references and handle them appropriately. [ Note: I removed the manual static_call_tramp check -- that should already be handled by the noendbr check. ] Reported-by: Ard Biesheuvel Tested-by: Ard Biesheuvel Acked-by: Peter Zijlstra (Intel) Signed-off-by: Josh Poimboeuf --- arch/x86/kernel/acpi/wakeup_64.S | 1 + arch/x86/kernel/head_64.S | 1 + tools/objtool/arch/x86/decode.c | 15 ++-- tools/objtool/check.c | 112 +++++++++++++++------------ tools/objtool/include/objtool/arch.h | 1 + 5 files changed, 77 insertions(+), 53 deletions(-) diff --git a/arch/x86/kernel/acpi/wakeup_64.S b/arch/x86/kernel/acpi/wakeup_64.S index 94ff83f3d3fe..b200a193beeb 100644 --- a/arch/x86/kernel/acpi/wakeup_64.S +++ b/arch/x86/kernel/acpi/wakeup_64.S @@ -87,6 +87,7 @@ SYM_FUNC_START(do_suspend_lowlevel) .align 4 .Lresume_point: + ANNOTATE_NOENDBR /* We don't restore %rax, it must be 0 anyway */ movq $saved_context, %rax movq saved_context_cr4(%rax), %rbx diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S index 16752b8dfa89..56163e2124cf 100644 --- a/arch/x86/kernel/head_64.S +++ b/arch/x86/kernel/head_64.S @@ -77,6 +77,7 @@ SYM_CODE_START_NOALIGN(startup_64) lretq .Lon_kernel_cs: + ANNOTATE_NOENDBR UNWIND_HINT_END_OF_STACK #ifdef CONFIG_AMD_MEM_ENCRYPT diff --git a/tools/objtool/arch/x86/decode.c b/tools/objtool/arch/x86/decode.c index ed6bff0e01dc..fe1362c34564 100644 --- a/tools/objtool/arch/x86/decode.c +++ b/tools/objtool/arch/x86/decode.c @@ -456,10 +456,6 @@ int arch_decode_instruction(struct objtool_file *file, const struct section *sec if (!rex_w) break; - /* skip RIP relative displacement */ - if (is_RIP()) - break; - /* skip nontrivial SIB */ if (have_SIB()) { modrm_rm = sib_base; @@ -467,6 +463,12 @@ int arch_decode_instruction(struct objtool_file *file, const struct section *sec break; } + /* lea disp(%rip), %dst */ + if (is_RIP()) { + insn->type = INSN_LEA_RIP; + break; + } + /* lea disp(%src), %dst */ ADD_OP(op) { op->src.offset = ins.displacement.value; @@ -737,7 +739,10 @@ int arch_decode_instruction(struct objtool_file *file, const struct section *sec break; } - insn->immediate = ins.immediate.nbytes ? ins.immediate.value : 0; + if (ins.immediate.nbytes) + insn->immediate = ins.immediate.value; + else if (ins.displacement.nbytes) + insn->immediate = ins.displacement.value; return 0; } diff --git a/tools/objtool/check.c b/tools/objtool/check.c index 6604f5d038aa..7fc96c30b79c 100644 --- a/tools/objtool/check.c +++ b/tools/objtool/check.c @@ -4392,6 +4392,51 @@ static bool noendbr_range(struct objtool_file *file, struct instruction *insn) return insn->offset == sym->offset + sym->len; } +static int __validate_ibt_insn(struct objtool_file *file, struct instruction *insn, + struct instruction *dest) +{ + if (dest->type == INSN_ENDBR) { + mark_endbr_used(dest); + return 0; + } + + if (insn_func(dest) && insn_func(insn) && + insn_func(dest)->pfunc == insn_func(insn)->pfunc) { + /* + * Anything from->to self is either _THIS_IP_ or + * IRET-to-self. + * + * There is no sane way to annotate _THIS_IP_ since the + * compiler treats the relocation as a constant and is + * happy to fold in offsets, skewing any annotation we + * do, leading to vast amounts of false-positives. + * + * There's also compiler generated _THIS_IP_ through + * KCOV and such which we have no hope of annotating. + * + * As such, blanket accept self-references without + * issue. + */ + return 0; + } + + /* + * Accept anything ANNOTATE_NOENDBR. + */ + if (dest->noendbr) + return 0; + + /* + * Accept if this is the instruction after a symbol + * that is (no)endbr -- typical code-range usage. + */ + if (noendbr_range(file, dest)) + return 0; + + WARN_INSN(insn, "relocation to !ENDBR: %s", offstr(dest->sec, dest->offset)); + return 1; +} + static int validate_ibt_insn(struct objtool_file *file, struct instruction *insn) { struct instruction *dest; @@ -4404,6 +4449,7 @@ static int validate_ibt_insn(struct objtool_file *file, struct instruction *insn * direct/indirect branches: */ switch (insn->type) { + case INSN_CALL: case INSN_CALL_DYNAMIC: case INSN_JUMP_CONDITIONAL: @@ -4413,6 +4459,23 @@ static int validate_ibt_insn(struct objtool_file *file, struct instruction *insn case INSN_RETURN: case INSN_NOP: return 0; + + case INSN_LEA_RIP: + if (!insn_reloc(file, insn)) { + /* local function pointer reference without reloc */ + + off = arch_jump_destination(insn); + + dest = find_insn(file, insn->sec, off); + if (!dest) { + WARN_INSN(insn, "corrupt function pointer reference"); + return 1; + } + + return __validate_ibt_insn(file, insn, dest); + } + break; + default: break; } @@ -4423,13 +4486,6 @@ static int validate_ibt_insn(struct objtool_file *file, struct instruction *insn reloc_offset(reloc) + 1, (insn->offset + insn->len) - (reloc_offset(reloc) + 1))) { - /* - * static_call_update() references the trampoline, which - * doesn't have (or need) ENDBR. Skip warning in that case. - */ - if (reloc->sym->static_call_tramp) - continue; - off = reloc->sym->offset; if (reloc_type(reloc) == R_X86_64_PC32 || reloc_type(reloc) == R_X86_64_PLT32) @@ -4441,47 +4497,7 @@ static int validate_ibt_insn(struct objtool_file *file, struct instruction *insn if (!dest) continue; - if (dest->type == INSN_ENDBR) { - mark_endbr_used(dest); - continue; - } - - if (insn_func(dest) && insn_func(insn) && - insn_func(dest)->pfunc == insn_func(insn)->pfunc) { - /* - * Anything from->to self is either _THIS_IP_ or - * IRET-to-self. - * - * There is no sane way to annotate _THIS_IP_ since the - * compiler treats the relocation as a constant and is - * happy to fold in offsets, skewing any annotation we - * do, leading to vast amounts of false-positives. - * - * There's also compiler generated _THIS_IP_ through - * KCOV and such which we have no hope of annotating. - * - * As such, blanket accept self-references without - * issue. - */ - continue; - } - - /* - * Accept anything ANNOTATE_NOENDBR. - */ - if (dest->noendbr) - continue; - - /* - * Accept if this is the instruction after a symbol - * that is (no)endbr -- typical code-range usage. - */ - if (noendbr_range(file, dest)) - continue; - - WARN_INSN(insn, "relocation to !ENDBR: %s", offstr(dest->sec, dest->offset)); - - warnings++; + warnings += __validate_ibt_insn(file, insn, dest); } return warnings; diff --git a/tools/objtool/include/objtool/arch.h b/tools/objtool/include/objtool/arch.h index 0b303eba660e..d63b46a19f39 100644 --- a/tools/objtool/include/objtool/arch.h +++ b/tools/objtool/include/objtool/arch.h @@ -28,6 +28,7 @@ enum insn_type { INSN_CLD, INSN_TRAP, INSN_ENDBR, + INSN_LEA_RIP, INSN_OTHER, }; -- 2.51.0 From 32b504854bd96f707a03c6ddecb0af9d7fbc4775 Mon Sep 17 00:00:00 2001 From: HONG Yifan Date: Tue, 8 Oct 2024 23:47:17 +0000 Subject: [PATCH 13/16] objtool: Also include tools/include/uapi When building objtool against a sysroot that contains a stripped down version of the UAPI headers, the following error happens: In file included from arch/x86/decode.c:10: In file included from .../tools/arch/x86/include/asm/insn.h:10: In file included from /include/asm/byteorder.h:9: In file included from /include/linux/byteorder/little_endian.h:15: In file included from /include/linux/stddef.h:9: In file included from .../tools/include/linux/compiler_types.h:36: .../tools/include/linux/compiler-gcc.h:3:2: error: "Please don't include directly, include instead." 3 | #error "Please don't include directly, include instead." | ^ 1 error generated. As hinted by the error, this is because /include/linux/stddef.h (a stripped-down version of uapi/include/linux/stddef.h) includes linux/compiler_types.h directly. However, this gets resolved to tools/include/linux/compiler_types.h, which is not expected to be included directly. To resolve this, I added tools/include/uapi to the include paths when building objtool. With this trick, linux/stddef.h is resolved to tools/include/uapi/linux/stddef.h, which doesn't include linux/compiler_types.h. Signed-off-by: HONG Yifan Signed-off-by: Josh Poimboeuf --- tools/objtool/Makefile | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/objtool/Makefile b/tools/objtool/Makefile index bf7f7f84ac62..f56e27727534 100644 --- a/tools/objtool/Makefile +++ b/tools/objtool/Makefile @@ -24,6 +24,7 @@ LIBELF_LIBS := $(shell $(HOSTPKG_CONFIG) libelf --libs 2>/dev/null || echo -lel all: $(OBJTOOL) INCLUDES := -I$(srctree)/tools/include \ + -I$(srctree)/tools/include/uapi \ -I$(srctree)/tools/arch/$(HOSTARCH)/include/uapi \ -I$(srctree)/tools/arch/$(SRCARCH)/include \ -I$(srctree)/tools/objtool/include \ -- 2.51.0 From 3bf19a0fb690022ec22ce87a5afeb1030cbcb56c Mon Sep 17 00:00:00 2001 From: Zheng Yejian Date: Fri, 13 Sep 2024 10:45:01 +0800 Subject: [PATCH 14/16] x86/unwind/orc: Fix unwind for newly forked tasks When arch_stack_walk_reliable() is called to unwind for newly forked tasks, the return value is negative which means the call stack is unreliable. This obviously does not meet expectations. The root cause is that after commit 3aec4ecb3d1f ("x86: Rewrite ret_from_fork() in C"), the 'ret_addr' of newly forked task is changed to 'ret_from_fork_asm' (see copy_thread()), then at the start of the unwind, it is incorrectly interprets not as a "signal" one because 'ret_from_fork' is still used to determine the initial "signal" (see __unwind_start()). Then the address gets incorrectly decremented in the call to orc_find() (see unwind_next_frame()) and resulting in the incorrect ORC data. To fix it, check 'ret_from_fork_asm' rather than 'ret_from_fork' in __unwind_start(). Fixes: 3aec4ecb3d1f ("x86: Rewrite ret_from_fork() in C") Signed-off-by: Zheng Yejian Signed-off-by: Josh Poimboeuf --- arch/x86/kernel/unwind_orc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/kernel/unwind_orc.c b/arch/x86/kernel/unwind_orc.c index d00c28aaa5be..d4705a348a80 100644 --- a/arch/x86/kernel/unwind_orc.c +++ b/arch/x86/kernel/unwind_orc.c @@ -723,7 +723,7 @@ void __unwind_start(struct unwind_state *state, struct task_struct *task, state->sp = task->thread.sp + sizeof(*frame); state->bp = READ_ONCE_NOCHECK(frame->bp); state->ip = READ_ONCE_NOCHECK(frame->ret_addr); - state->signal = (void *)state->ip == ret_from_fork; + state->signal = (void *)state->ip == ret_from_fork_asm; } if (get_stack_info((unsigned long *)state->sp, state->task, -- 2.51.0 From 56ac7bd2c58a4e93d19f0ccb181035d075b315d3 Mon Sep 17 00:00:00 2001 From: Carlos Llamas Date: Mon, 12 Aug 2024 23:01:20 +0000 Subject: [PATCH 15/16] Revert "scripts/faddr2line: Check only two symbols when calculating symbol size" This reverts commit c02904f05ff805d6c0631634d5751ebd338f75ec. Such commit assumed that only two symbols are relevant for the symbol size calculation. However, this can lead to an incorrect symbol size calculation when there are mapping symbols emitted by readelf. For instance, when feeding 'update_irq_load_avg+0x1c/0x1c4', faddr2line might need to process the following readelf lines: 784284: ffffffc0081cca30 428 FUNC GLOBAL DEFAULT 2 update_irq_load_avg 87319: ffffffc0081ccb0c 0 NOTYPE LOCAL DEFAULT 2 $x.62522 87321: ffffffc0081ccbdc 0 NOTYPE LOCAL DEFAULT 2 $x.62524 87323: ffffffc0081ccbe0 0 NOTYPE LOCAL DEFAULT 2 $x.62526 87325: ffffffc0081ccbe4 0 NOTYPE LOCAL DEFAULT 2 $x.62528 87327: ffffffc0081ccbe8 0 NOTYPE LOCAL DEFAULT 2 $x.62530 87329: ffffffc0081ccbec 0 NOTYPE LOCAL DEFAULT 2 $x.62532 87331: ffffffc0081ccbf0 0 NOTYPE LOCAL DEFAULT 2 $x.62534 87332: ffffffc0081ccbf4 0 NOTYPE LOCAL DEFAULT 2 $x.62535 783403: ffffffc0081ccbf4 424 FUNC GLOBAL DEFAULT 2 sched_pelt_multiplier The symbol size of 'update_irq_load_avg' should be calculated with the address of 'sched_pelt_multiplier', after skipping the mapping symbols seen in between. However, the offending commit cuts the list short and faddr2line incorrectly assumes 'update_irq_load_avg' is the last symbol in the section, resulting in: $ scripts/faddr2line vmlinux update_irq_load_avg+0x1c/0x1c4 skipping update_irq_load_avg address at 0xffffffc0081cca4c due to size mismatch (0x1c4 != 0x3ff9a59988) no match for update_irq_load_avg+0x1c/0x1c4 After reverting the commit the issue is resolved: $ scripts/faddr2line vmlinux update_irq_load_avg+0x1c/0x1c4 update_irq_load_avg+0x1c/0x1c4: cpu_of at kernel/sched/sched.h:1109 (inlined by) update_irq_load_avg at kernel/sched/pelt.c:481 Fixes: c02904f05ff8 ("scripts/faddr2line: Check only two symbols when calculating symbol size") Signed-off-by: Carlos Llamas Acked-by: Will Deacon Acked-by: Brian Johannesmeyer Signed-off-by: Josh Poimboeuf --- scripts/faddr2line | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/faddr2line b/scripts/faddr2line index fe0cc45f03be..1fa6beef9f97 100755 --- a/scripts/faddr2line +++ b/scripts/faddr2line @@ -252,7 +252,7 @@ __faddr2line() { found=2 break fi - done < <(echo "${ELF_SYMS}" | sed 's/\[.*\]//' | ${AWK} -v sec=$sym_sec '$7 == sec' | sort --key=2 | ${GREP} -A1 --no-group-separator " ${sym_name}$") + done < <(echo "${ELF_SYMS}" | sed 's/\[.*\]//' | ${AWK} -v sec=$sym_sec '$7 == sec' | sort --key=2) if [[ $found = 0 ]]; then warn "can't find symbol: sym_name: $sym_name sym_sec: $sym_sec sym_addr: $sym_addr sym_elf_size: $sym_elf_size" -- 2.51.0 From d5173f7537505315557d8580e3a648f07f17deda Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 8 Nov 2024 10:32:02 +0100 Subject: [PATCH 16/16] objtool: Exclude __tracepoints data from ENDBR checks For some, as of yet unexplained reason, Clang-19, but not GCC, generates and endless stream of: drivers/iio/imu/bno055/bno055_ser.o: warning: objtool: __tracepoint_send_chunk+0x20: data relocation to !ENDBR: __SCT__tp_func_send_chunk+0x0 drivers/iio/imu/bno055/bno055_ser.o: warning: objtool: __tracepoint_cmd_retry+0x20: data relocation to !ENDBR: __SCT__tp_func_cmd_retry+0x0 drivers/iio/imu/bno055/bno055_ser.o: warning: objtool: __tracepoint_write_reg+0x20: data relocation to !ENDBR: __SCT__tp_func_write_reg+0x0 drivers/iio/imu/bno055/bno055_ser.o: warning: objtool: __tracepoint_read_reg+0x20: data relocation to !ENDBR: __SCT__tp_func_read_reg+0x0 drivers/iio/imu/bno055/bno055_ser.o: warning: objtool: __tracepoint_recv+0x20: data relocation to !ENDBR: __SCT__tp_func_recv+0x0 Which is entirely correct, but harmless. Add the __tracepoints section to the exclusion list. Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20241108184618.GG38786@noisy.programming.kicks-ass.net --- tools/objtool/check.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/objtool/check.c b/tools/objtool/check.c index 7fc96c30b79c..f7586f82b967 100644 --- a/tools/objtool/check.c +++ b/tools/objtool/check.c @@ -4573,6 +4573,7 @@ static int validate_ibt(struct objtool_file *file) !strcmp(sec->name, "__jump_table") || !strcmp(sec->name, "__mcount_loc") || !strcmp(sec->name, ".kcfi_traps") || + !strcmp(sec->name, "__tracepoints") || strstr(sec->name, "__patchable_function_entries")) continue; -- 2.51.0