lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <388bbc4c805ce029bbd08010fd30405494f998a9.camel@web.de>
Date: Sat, 17 May 2025 13:34:11 +0200
From: Bert Karwatzki <spasswolf@....de>
To: Johannes Berg <johannes@...solutions.net>,
 "linux-kernel@...r.kernel.org"	 <linux-kernel@...r.kernel.org>
Cc: "linux-next@...r.kernel.org" <linux-next@...r.kernel.org>, 
 "llvm@...ts.linux.dev"
	 <llvm@...ts.linux.dev>, Thomas Gleixner <tglx@...utronix.de>, 
	linux-wireless@...r.kernel.org, spasswolf@....de
Subject: Re: lockup and kernel panic in linux-next-202505{09,12} when
 compiled with clang

Am Freitag, dem 16.05.2025 um 20:19 +0200 schrieb Bert Karwatzki:
> I've added a debugging statement:
> 
> diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
> index 3bd5ee0995fe..853493eca4f5 100644
> --- a/net/mac80211/tx.c
> +++ b/net/mac80211/tx.c
> @@ -4586,7 +4586,11 @@ static noinline void ieee80211_8023_xmit_clang_debug_helper(struct sk_buff *skb,
>                                                             struct ieee80211_local *local,
>                                                             struct ieee80211_tx_info *info)
>  {
> -       if (unlikely(skb->sk && sock_flag(skb->sk, SOCK_WIFI_STATUS))) {
> +       if (unlikely(skb->sk && ((skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) ||
> +                               sock_flag(skb->sk, SOCK_WIFI_STATUS)))) {
> +               if ((skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) ^ sock_flag(skb->sk, SOCK_WIFI_STATUS))
> +                       printk(KERN_INFO "%s: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = %u sock_flag(skb->sk,
> SOCK_WIFI_STATUS) = %u\n",
> +                                       __func__, (skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS), sock_flag(skb->sk,
> SOCK_WIFI_STATUS));
>                 info->status_data = ieee80211_store_ack_skb(local, skb,
>                                                             &info->flags, NULL);
>                 if (info->status_data)
> 
> This gives the following logoutput (and a lockup), indicating that sock_flag(skb->sk, SOCK_WIFI_STATUS) and
> (skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) are actually NOT equivalent (when compiled with clang and
> PREEMPT_RT=y)

I've added more debugging output:

diff --git a/include/net/sock.h b/include/net/sock.h
index e223102337c7..e13560b5b7a8 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -2735,8 +2735,10 @@ static inline void _sock_tx_timestamp(struct sock *sk,
 				*tskey = atomic_inc_return(&sk->sk_tskey) - 1;
 		}
 	}
-	if (unlikely(sock_flag(sk, SOCK_WIFI_STATUS)))
+	if (unlikely(sock_flag(sk, SOCK_WIFI_STATUS))) {
+		printk(KERN_INFO "%s: setting SKBTX_WIFI_STATUS for sk = %px\n", __func__, sk);
 		*tx_flags |= SKBTX_WIFI_STATUS;
+	}
 }
 
 static inline void sock_tx_timestamp(struct sock *sk,
diff --git a/net/core/sock.c b/net/core/sock.c
index e02a78538e3e..f6589ad5ba36 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -1548,6 +1548,7 @@ int sk_setsockopt(struct sock *sk, int level, int optname,
 		break;
 
 	case SO_WIFI_STATUS:
+		printk(KERN_INFO "%s: setting SOCK_WIFI_STATUS to %u for sk = %px\n", __func__, valbool, sk);
 		sock_valbool_flag(sk, SOCK_WIFI_STATUS, valbool);
 		break;
 
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index 853493eca4f5..eee2f80949c6 100644
--- a/net/mac80211/tx.c
+++ b/net/mac80211/tx.c
@@ -4588,9 +4588,12 @@ static noinline void ieee80211_8023_xmit_clang_debug_helper(struct sk_buff *skb,
 {
 	if (unlikely(skb->sk && ((skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) ||
 				sock_flag(skb->sk, SOCK_WIFI_STATUS)))) {
-		if ((skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) ^ sock_flag(skb->sk, SOCK_WIFI_STATUS))
+		if ((skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) ^ sock_flag(skb->sk, SOCK_WIFI_STATUS)) {
 			printk(KERN_INFO "%s: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = %u sock_flag(skb->sk, SOCK_WIFI_STATUS) = %u\n",
 					__func__, (skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS), sock_flag(skb->sk, SOCK_WIFI_STATUS));
+			printk(KERN_INFO "%s: skb->sk = %px skb->sk->sk_flags = 0x%lx\n", __func__, skb->sk, skb->sk->sk_flags);
+			return; // This should make this case non-fatal.
+		}
 		info->status_data = ieee80211_store_ack_skb(local, skb,
 							    &info->flags, NULL);
 		if (info->status_data)



This gives after ~15min uptime

[  189.337797] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  189.337803] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1b798c4e00 skb->sk->sk_flags = 0xffffffffb4efe640
[  191.325256] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  191.325259] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1b798c5a00 skb->sk->sk_flags = 0xffffffffb4efe640
[  257.591831] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  257.591844] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1baf3bca00 skb->sk->sk_flags = 0xffffffffb4efe640
[  301.786963] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  301.786967] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1c1bc40100 skb->sk->sk_flags = 0xffffffffb4efe640
[  302.780881] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  302.780884] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1a44cf6000 skb->sk->sk_flags = 0xffffffffb4efe640
[  482.792298] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  482.792304] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1da0f4de00 skb->sk->sk_flags = 0xffffffffb4efe640
[  482.806144] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  482.806148] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1da0f4c500 skb->sk->sk_flags = 0xffffffffb4efe640
[  482.817280] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  482.817284] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1da0f4df00 skb->sk->sk_flags = 0xffffffffb4efe640
[  552.327291] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  552.327295] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1da0f4de00 skb->sk->sk_flags = 0xffffffffb4efe640
[  916.971599] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS = 0 sock_flag(skb->sk, SOCK_WIFI_STATUS) = 1
[  916.971607] [    T576] ieee80211_8023_xmit_clang_debug_helper: skb->sk = ffff8c1a62834000 skb->sk->sk_flags = 0xffffffffb4efe640

The printk()s in sk_set_sockopt() and _sock_tx_timestamp() are not called at all so the flagĀ 
SOCK_WIFI_STATUS is actually nevers set! What is printed when printing skb->sk->sk_flags looks
suspiciously like a pointer, and as sk_flags is actually a member of a union in struct sock_common
it seems clang is using sk_flags for one of the other union members here

struct sock_common {
[...]
	union {
		unsigned long	skc_flags;
		struct sock	*skc_listener; /* request_sock */
		struct inet_timewait_death_row *skc_tw_dr; /* inet_timewait_sock */
	};
[...]
}

Bert Karwatzki

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ