linux-kernel - [RFC patch 15/19] bpf: Use migrate_disable() in sys

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: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20200214161504.431560798@linutronix.de>
Date:   Fri, 14 Feb 2020 14:39:32 +0100
From:   Thomas Gleixner <tglx@...utronix.de>
To:     LKML <linux-kernel@...r.kernel.org>
Cc:     David Miller <davem@...emloft.net>, bpf@...r.kernel.org,
        netdev@...r.kernel.org, Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Sebastian Sewior <bigeasy@...utronix.de>,
        Peter Zijlstra <peterz@...radead.org>,
        Clark Williams <williams@...hat.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Juri Lelli <juri.lelli@...hat.com>,
        Ingo Molnar <mingo@...nel.org>
Subject: [RFC patch 15/19] bpf: Use migrate_disable() in sys_bpf()

The required protection is that the caller cannot be migrated to a
different CPU as these functions end up in places which take either a hash
bucket lock or might trigger a kprobe inside the memory allocator. Both
scenarios can lead to deadlocks. The deadlock prevention is per CPU by
incrementing a per CPU variable which temporarily blocks the invocation of
BPF programs from perf and kprobes.

Replace the preempt_disable/enable() pairs with migrate_disable/enable()
pairs to prepare BPF to work on PREEMPT_RT enabled kernels. On a non-RT
kernel this maps to preempt_disable/enable(), i.e. no functional change.

Signed-off-by: Thomas Gleixner <tglx@...utronix.de>
---
 kernel/bpf/syscall.c |   16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -174,7 +174,7 @@ static int bpf_map_update_value(struct b
 	/* must increment bpf_prog_active to avoid kprobe+bpf triggering from
 	 * inside bpf map update or delete otherwise deadlocks are possible
 	 */
-	preempt_disable();
+	migrate_disable();
 	__this_cpu_inc(bpf_prog_active);
 	if (map->map_type == BPF_MAP_TYPE_PERCPU_HASH ||
 	    map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH) {
@@ -207,7 +207,7 @@ static int bpf_map_update_value(struct b
 		rcu_read_unlock();
 	}
 	__this_cpu_dec(bpf_prog_active);
-	preempt_enable();
+	migrate_enable();
 	maybe_wait_bpf_programs(map);
 
 	return err;
@@ -222,7 +222,7 @@ static int bpf_map_copy_value(struct bpf
 	if (bpf_map_is_dev_bound(map))
 		return bpf_map_offload_lookup_elem(map, key, value);
 
-	preempt_disable();
+	migrate_disable();
 	this_cpu_inc(bpf_prog_active);
 	if (map->map_type == BPF_MAP_TYPE_PERCPU_HASH ||
 	    map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH) {
@@ -269,7 +269,7 @@ static int bpf_map_copy_value(struct bpf
 	}
 
 	this_cpu_dec(bpf_prog_active);
-	preempt_enable();
+	migrate_enable();
 	maybe_wait_bpf_programs(map);
 
 	return err;
@@ -1136,13 +1136,13 @@ static int map_delete_elem(union bpf_att
 		goto out;
 	}
 
-	preempt_disable();
+	migrate_disable();
 	__this_cpu_inc(bpf_prog_active);
 	rcu_read_lock();
 	err = map->ops->map_delete_elem(map, key);
 	rcu_read_unlock();
 	__this_cpu_dec(bpf_prog_active);
-	preempt_enable();
+	migrate_enable();
 	maybe_wait_bpf_programs(map);
 out:
 	kfree(key);
@@ -1254,13 +1254,13 @@ int generic_map_delete_batch(struct bpf_
 			break;
 		}
 
-		preempt_disable();
+		migrate_disable();
 		__this_cpu_inc(bpf_prog_active);
 		rcu_read_lock();
 		err = map->ops->map_delete_elem(map, key);
 		rcu_read_unlock();
 		__this_cpu_dec(bpf_prog_active);
-		preempt_enable();
+		migrate_enable();
 		maybe_wait_bpf_programs(map);
 		if (err)
 			break;