bpf: Use kvmalloc for map values in syscall

commit f0dce1d upstream

ANBZ:torvalds#198

[BackportNotes]
Modifications for bpf new interfaces(such as generic_map_update_batch,
generic_map_lookup_batch) is not applied, since they are not
implemented.

Use kvmalloc/kvfree for temporary value when manipulating a map via
syscall. kmalloc might not be sufficient for percpu maps where the value
is big (and further multiplied by hundreds of CPUs).

Can be reproduced with netcnt test on qemu with "-smp 255".

Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210818235216.1159202-1-sdf@google.com
Signed-off-by: Qiao Ma <mqaio@linux.alibaba.com>
Acked-by: Tony Lu <tonylu@linux.alibaba.com>

kernel/bpf/syscall.c

@@ -798,7 +798,7 @@ static int map_lookup_elem(union bpf_attr *attr)
 		value_size = map->value_size;
 
 	err = -ENOMEM;
-	value = kmalloc(value_size, GFP_USER | __GFP_NOWARN);
+	value = kvmalloc(value_size, GFP_USER | __GFP_NOWARN);
 	if (!value)
 		goto free_key;
 
@@ -847,7 +847,7 @@ static int map_lookup_elem(union bpf_attr *attr)
 	err = 0;
 
 free_value:
-	kfree(value);
+	kvfree(value);
 free_key:
 	kfree(key);
 err_put:
@@ -906,7 +906,7 @@ static int map_update_elem(union bpf_attr *attr)
 		value_size = map->value_size;
 
 	err = -ENOMEM;
-	value = kmalloc(value_size, GFP_USER | __GFP_NOWARN);
+	value = kvmalloc(value_size, GFP_USER | __GFP_NOWARN);
 	if (!value)
 		goto free_key;
 
@@ -959,7 +959,7 @@ static int map_update_elem(union bpf_attr *attr)
 	maybe_wait_bpf_programs(map);
 out:
 free_value:
-	kfree(value);
+	kvfree(value);
 free_key:
 	kfree(key);
 err_put: