Re: [PATCH bpf-next v9 3/4] selftests: bpf: crypto skcipher algo selftests

[Martin KaFai Lau <martin.lau@linux.dev>]

On 4/16/24 1:40 PM, Vadim Fedorenko wrote:
> +void test_crypto_sanity(void)
> +{
> +	struct crypto_syscall_args sargs = {
> +		.key_len = 16,
> +	};
> +	LIBBPF_OPTS(bpf_tc_hook, qdisc_hook, .attach_point = BPF_TC_EGRESS);
> +	LIBBPF_OPTS(bpf_tc_opts, tc_attach_enc);
> +	LIBBPF_OPTS(bpf_tc_opts, tc_attach_dec);
> +	LIBBPF_OPTS(bpf_test_run_opts, opts,
> +		    .ctx_in = &sargs,
> +		    .ctx_size_in = sizeof(sargs),
> +	);
> +	struct nstoken *nstoken = NULL;
> +	struct crypto_sanity *skel;
> +	char afalg_plain[16] = {0};
> +	char afalg_dst[16] = {0};
> +	struct sockaddr_in6 addr;
> +	int sockfd, err, pfd;
> +	socklen_t addrlen;
> +
> +	SYS(fail, "ip netns add %s", NS_TEST);
> +	SYS(fail, "ip -net %s -6 addr add %s/128 dev lo nodad", NS_TEST, IPV6_IFACE_ADDR);
> +	SYS(fail, "ip -net %s link set dev lo up", NS_TEST);
> +
> +	nstoken = open_netns(NS_TEST);
> +	if (!ASSERT_OK_PTR(nstoken, "open_netns"))
> +		goto fail;

skel is not initialized. The "fail:" case needs it.

> +
> +	err = init_afalg();
> +	if (!ASSERT_OK(err, "AF_ALG init fail"))
> +		goto fail;
> +
> +	qdisc_hook.ifindex = if_nametoindex("lo");
> +	if (!ASSERT_GT(qdisc_hook.ifindex, 0, "if_nametoindex lo"))
> +		goto fail;
> +
> +	skel = crypto_sanity__open_and_load();
> +	if (!ASSERT_OK_PTR(skel, "skel open"))
> +		return;

The netns "crypto_sanity_ns" is not deleted.

> +
> +	memcpy(skel->bss->key, crypto_key, sizeof(crypto_key));
> +	snprintf(skel->bss->algo, 128, "%s", algo);
> +	pfd = bpf_program__fd(skel->progs.skb_crypto_setup);
> +	if (!ASSERT_GT(pfd, 0, "skb_crypto_setup fd"))
> +		goto fail;
> +
> +	err = bpf_prog_test_run_opts(pfd, &opts);
> +	if (!ASSERT_OK(err, "skb_crypto_setup") ||
> +	    !ASSERT_OK(opts.retval, "skb_crypto_setup retval"))
> +		goto fail;
> +
> +	if (!ASSERT_OK(skel->bss->status, "skb_crypto_setup status"))
> +		goto fail;
> +
> +	err = crypto_sanity__attach(skel);

This attach is a left over from previous revision?

> +	if (!ASSERT_OK(err, "crypto_sanity__attach"))
> +		goto fail;
> +
> +	err = bpf_tc_hook_create(&qdisc_hook);
> +	if (!ASSERT_OK(err, "create qdisc hook"))
> +		goto fail;
> +
> +	addrlen = sizeof(addr);
> +	err = make_sockaddr(AF_INET6, IPV6_IFACE_ADDR, UDP_TEST_PORT,
> +			    (void *)&addr, &addrlen);
> +	if (!ASSERT_OK(err, "make_sockaddr"))
> +		goto fail;
> +
> +	tc_attach_enc.prog_fd = bpf_program__fd(skel->progs.encrypt_sanity);
> +	err = bpf_tc_attach(&qdisc_hook, &tc_attach_enc);
> +	if (!ASSERT_OK(err, "attach encrypt filter"))
> +		goto fail;
> +
> +	sockfd = socket(AF_INET6, SOCK_DGRAM, 0);
> +	if (!ASSERT_NEQ(sockfd, -1, "encrypt socket"))
> +		goto fail;
> +	err = sendto(sockfd, plain_text, sizeof(plain_text), 0, (void *)&addr, addrlen);
> +	close(sockfd);
> +	if (!ASSERT_EQ(err, sizeof(plain_text), "encrypt send"))
> +		goto fail;
> +
> +	do_crypt_afalg(plain_text, afalg_dst, sizeof(afalg_dst), true);
> +
> +	bpf_tc_detach(&qdisc_hook, &tc_attach_enc);

Check error.

I suspect this detach should have failed because at least the 
tc_attach_enc.prog_fd is not 0.

The following attach (&tc_attach_"dec") may just happen to have a higher 
priority such that the left over here does not matter. It is still better to get 
it right.

> +	if (!ASSERT_OK(skel->bss->status, "encrypt status"))
> +		goto fail;
> +	if (!ASSERT_STRNEQ(skel->bss->dst, afalg_dst, sizeof(afalg_dst), "encrypt AF_ALG"))
> +		goto fail;
> +
> +	tc_attach_dec.prog_fd = bpf_program__fd(skel->progs.decrypt_sanity);
> +	err = bpf_tc_attach(&qdisc_hook, &tc_attach_dec);
> +	if (!ASSERT_OK(err, "attach decrypt filter"))
> +		goto fail;
> +
> +	sockfd = socket(AF_INET6, SOCK_DGRAM, 0);
> +	if (!ASSERT_NEQ(sockfd, -1, "decrypt socket"))
> +		goto fail;
> +	err = sendto(sockfd, afalg_dst, sizeof(afalg_dst), 0, (void *)&addr, addrlen);
> +	close(sockfd);
> +	if (!ASSERT_EQ(err, sizeof(afalg_dst), "decrypt send"))
> +		goto fail;
> +
> +	do_crypt_afalg(afalg_dst, afalg_plain, sizeof(afalg_plain), false);
> +
> +	bpf_tc_detach(&qdisc_hook, &tc_attach_dec);
> +	if (!ASSERT_OK(skel->bss->status, "decrypt status"))
> +		goto fail;
> +	if (!ASSERT_STRNEQ(skel->bss->dst, afalg_plain, sizeof(afalg_plain), "decrypt AF_ALG"))
> +		goto fail;
> +
> +fail:
> +	if (nstoken) {

No need to check NULL. close_netns() can handle it.

> +		bpf_tc_hook_destroy(&qdisc_hook);

This also does not destroy the clsact qdisc. Although the function name feels 
like it would, from a quick look at bpf_tc_hook_destroy, it depends on both 
BPF_TC_INGRESS and BPF_TC_EGRESS are set in the qdisc_hook.attach_point.

I would skip the whole bpf_tc_hook_destroy. It will go away together with the netns.

[ ... ]

> diff --git a/tools/testing/selftests/bpf/progs/crypto_sanity.c b/tools/testing/selftests/bpf/progs/crypto_sanity.c
> new file mode 100644
> index 000000000000..57df5776bcaf
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/progs/crypto_sanity.c
> @@ -0,0 +1,161 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright (c) 2024 Meta Platforms, Inc. and affiliates. */
> +
> +#include "vmlinux.h"
> +#include "bpf_tracing_net.h"
> +#include <bpf/bpf_helpers.h>
> +#include <bpf/bpf_endian.h>
> +#include <bpf/bpf_tracing.h>
> +#include "bpf_misc.h"
> +#include "bpf_kfuncs.h"
> +#include "crypto_common.h"
> +
> +unsigned char key[256] = {};
> +char algo[128] = {};
> +char dst[16] = {};
> +int status;
> +
> +static int skb_dynptr_validate(struct __sk_buff *skb, struct bpf_dynptr *psrc)
> +{
> +	struct ipv6hdr ip6h;
> +	struct udphdr udph;
> +	u32 offset;
> +
> +	if (skb->protocol != __bpf_constant_htons(ETH_P_IPV6))
> +		return -1;
> +
> +	if (bpf_skb_load_bytes(skb, ETH_HLEN, &ip6h, sizeof(ip6h)))
> +		return -1;
> +
> +	if (ip6h.nexthdr != IPPROTO_UDP)
> +		return -1;
> +
> +	if (bpf_skb_load_bytes(skb, ETH_HLEN + sizeof(ip6h), &udph, sizeof(udph)))
> +		return -1;
> +
> +	if (udph.dest != __bpf_constant_htons(UDP_TEST_PORT))
> +		return -1;
> +
> +	offset = ETH_HLEN + sizeof(ip6h) + sizeof(udph);
> +	if (skb->len < offset + 16)
> +		return -1;
> +
> +	/* let's make sure that 16 bytes of payload are in the linear part of skb */
> +	bpf_skb_pull_data(skb, offset + 16);
> +	bpf_dynptr_from_skb(skb, 0, psrc);
> +	bpf_dynptr_adjust(psrc, offset, offset + 16);
> +
> +	return 0;
> +}
> +
> +SEC("syscall")
> +int skb_crypto_setup(struct crypto_syscall_args *ctx)
> +{
> +	struct bpf_crypto_params params = {
> +		.type = "skcipher",
> +		.key_len = ctx->key_len,
> +		.authsize = ctx->authsize,
> +	};
> +	struct bpf_crypto_ctx *cctx;
> +	int err = 0;
> +
> +	status = 0;
> +
> +	if (ctx->key_len > 255) {

key_len == 256 won't work ?

> +		status = -EINVAL;
> +		return 0;
> +	}
> +
> +	__builtin_memcpy(&params.algo, algo, sizeof(algo));
> +	__builtin_memcpy(&params.key, key, sizeof(key));

It will be useful to comment here what problem was hit such that the key cannot 
be passed in the "struct crypto_syscall_args" and need to go back to the global 
variable.

Instead of "key_len" in the crypto_syscall_args and the actual "key" in global, 
how about skip using the "struct crypto_syscall_args" altogether and put key_len 
(and authsize) in the global?

Put UDP_TEST_PORT as a global variable for config/filter usage also and the 
"crypto_share.h" can go away.

> +	cctx = bpf_crypto_ctx_create(&params, &err);
> +
> +	if (!cctx) {
> +		status = err;
> +		return 0;
> +	}
> +
> +	err = crypto_ctx_insert(cctx);
> +	if (err && err != -EEXIST)
> +		status = err;
> +
> +	return 0;
> +}
> +
> +SEC("tc")
> +int decrypt_sanity(struct __sk_buff *skb)
> +{
> +	struct __crypto_ctx_value *v;
> +	struct bpf_crypto_ctx *ctx;
> +	struct bpf_dynptr psrc, pdst, iv;
> +	int err;
> +
> +	err = skb_dynptr_validate(skb, &psrc);
> +	if (err < 0) {
> +		status = err;
> +		return TC_ACT_SHOT;
> +	}
> +
> +	v = crypto_ctx_value_lookup();
> +	if (!v) {
> +		status = -ENOENT;
> +		return TC_ACT_SHOT;
> +	}
> +
> +	ctx = v->ctx;
> +	if (!ctx) {
> +		status = -ENOENT;
> +		return TC_ACT_SHOT;
> +	}
> +
> +	bpf_dynptr_from_mem(dst, sizeof(dst), 0, &pdst);

dst is now a global which makes it easier to test the result. A comment here to 
note this point for people referencing this test for production use case and 
suggest a percpu map could be used.

It will be useful to have dynptr working with stack memory in the future.

> +	/* iv dynptr has to be initialized with 0 size, but proper memory region
> +	 * has to be provided anyway
> +	 */
> +	bpf_dynptr_from_mem(dst, 0, 0, &iv);
> +
> +	status = bpf_crypto_decrypt(ctx, &psrc, &pdst, &iv);
> +
> +	return TC_ACT_SHOT;
> +}
> +
> +SEC("tc")
> +int encrypt_sanity(struct __sk_buff *skb)
> +{
> +	struct __crypto_ctx_value *v;
> +	struct bpf_crypto_ctx *ctx;
> +	struct bpf_dynptr psrc, pdst, iv;
> +	int err;
> +
> +	status = 0;
> +
> +	err = skb_dynptr_validate(skb, &psrc);
> +	if (err < 0) {
> +		status = err;
> +		return TC_ACT_SHOT;
> +	}
> +
> +	v = crypto_ctx_value_lookup();
> +	if (!v) {
> +		status = -ENOENT;
> +		return TC_ACT_SHOT;
> +	}
> +
> +	ctx = v->ctx;
> +	if (!ctx) {
> +		status = -ENOENT;
> +		return TC_ACT_SHOT;
> +	}
> +
> +	bpf_dynptr_from_mem(dst, sizeof(dst), 0, &pdst);
> +	/* iv dynptr has to be initialized with 0 size, but proper memory region
> +	 * has to be provided anyway
> +	 */
> +	bpf_dynptr_from_mem(dst, 0, 0, &iv);
> +
> +	status = bpf_crypto_encrypt(ctx, &psrc, &pdst, &iv);
> +
> +	return TC_ACT_SHOT;
> +}
> +
> +char __license[] SEC("license") = "GPL";
> diff --git a/tools/testing/selftests/bpf/progs/crypto_share.h b/tools/testing/selftests/bpf/progs/crypto_share.h
> new file mode 100644
> index 000000000000..c5a6ef65156d
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/progs/crypto_share.h
> @@ -0,0 +1,10 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/* Copyright (c) 2024 Meta Platforms, Inc. and affiliates. */
> +
> +#define UDP_TEST_PORT 7777
> +
> +struct crypto_syscall_args {
> +	u32 key_len;
> +	u32 authsize;
> +};
> +