bpf: fix stackmap overflow check in __bpf_get_stackid()

Syzkaller reported a KASAN slab-out-of-bounds write in __bpf_get_stackid()
when copying stack trace data. The issue occurs when the perf trace
 contains more stack entries than the stack map bucket can hold,
 leading to an out-of-bounds write in the bucket's data array.

Changes in v2:
 - Fixed max_depth names across get stack id

Changes in v4:
 - Removed unnecessary empty line in __bpf_get_stackid

Link to v4: https://lore.kernel.org/all/[email protected]/

Reported-by: [email protected]
Closes: https://syzkaller.appspot.com/bug?extid=c9b724fbb41cf2538b7b
Signed-off-by: Arnaud Lecomte <[email protected]>
Acked-by: Yonghong Song <[email protected]>

Author: ArnaudLcm

kernel/bpf/stackmap.c

@@ -247,7 +247,7 @@ get_callchain_entry_for_task(struct task_struct *task, u32 max_depth)
 }
 
 static long __bpf_get_stackid(struct bpf_map *map,
-			      struct perf_callchain_entry *trace, u64 flags)
+			      struct perf_callchain_entry *trace, u64 flags, u32 max_depth)
 {
 	struct bpf_stack_map *smap = container_of(map, struct bpf_stack_map, map);
 	struct stack_map_bucket *bucket, *new_bucket, *old_bucket;
@@ -263,6 +263,8 @@ static long __bpf_get_stackid(struct bpf_map *map,
 
 	trace_nr = trace->nr - skip;
 	trace_len = trace_nr * sizeof(u64);
+	trace_nr = min(trace_nr, max_depth - skip);
+
 	ips = trace->ip + skip;
 	hash = jhash2((u32 *)ips, trace_len / sizeof(u32), 0);
 	id = hash & (smap->n_buckets - 1);
@@ -322,19 +324,17 @@ static long __bpf_get_stackid(struct bpf_map *map,
 BPF_CALL_3(bpf_get_stackid, struct pt_regs *, regs, struct bpf_map *, map,
 	   u64, flags)
 {
-	u32 max_depth = map->value_size / stack_map_data_size(map);
-	u32 skip = flags & BPF_F_SKIP_FIELD_MASK;
+	u32 elem_size = stack_map_data_size(map);
 	bool user = flags & BPF_F_USER_STACK;
 	struct perf_callchain_entry *trace;
 	bool kernel = !user;
+	u32 max_depth;
 
 	if (unlikely(flags & ~(BPF_F_SKIP_FIELD_MASK | BPF_F_USER_STACK |
 			       BPF_F_FAST_STACK_CMP | BPF_F_REUSE_STACKID)))
 		return -EINVAL;
 
-	max_depth += skip;
-	if (max_depth > sysctl_perf_event_max_stack)
-		max_depth = sysctl_perf_event_max_stack;
+	max_depth = stack_map_calculate_max_depth(map->value_size, elem_size, flags);
 
 	trace = get_perf_callchain(regs, 0, kernel, user, max_depth,
 				   false, false);
@@ -343,7 +343,7 @@ BPF_CALL_3(bpf_get_stackid, struct pt_regs *, regs, struct bpf_map *, map,
 		/* couldn't fetch the stack trace */
 		return -EFAULT;
 
-	return __bpf_get_stackid(map, trace, flags);
+	return __bpf_get_stackid(map, trace, flags, max_depth);
 }
 
 const struct bpf_func_proto bpf_get_stackid_proto = {
@@ -375,6 +375,7 @@ BPF_CALL_3(bpf_get_stackid_pe, struct bpf_perf_event_data_kern *, ctx,
 	bool kernel, user;
 	__u64 nr_kernel;
 	int ret;
+	u32 elem_size, max_depth;
 
 	/* perf_sample_data doesn't have callchain, use bpf_get_stackid */
 	if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
@@ -393,12 +394,13 @@ BPF_CALL_3(bpf_get_stackid_pe, struct bpf_perf_event_data_kern *, ctx,
 		return -EFAULT;
 
 	nr_kernel = count_kernel_ip(trace);
-
+	elem_size = stack_map_data_size(map);
 	if (kernel) {
 		__u64 nr = trace->nr;
 
 		trace->nr = nr_kernel;
-		ret = __bpf_get_stackid(map, trace, flags);
+		max_depth = stack_map_calculate_max_depth(map->value_size, elem_size, flags);
+		ret = __bpf_get_stackid(map, trace, flags, max_depth);
 
 		/* restore nr */
 		trace->nr = nr;
@@ -410,7 +412,8 @@ BPF_CALL_3(bpf_get_stackid_pe, struct bpf_perf_event_data_kern *, ctx,
 			return -EFAULT;
 
 		flags = (flags & ~BPF_F_SKIP_FIELD_MASK) | skip;
-		ret = __bpf_get_stackid(map, trace, flags);
+		max_depth = stack_map_calculate_max_depth(map->value_size, elem_size, flags);
+		ret = __bpf_get_stackid(map, trace, flags, max_depth);
 	}
 	return ret;
 }