x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

When using eBPF to probe CS CALL instructions within a function,
a crash can be triggered.

The eBPF tool probes offset 257 of the __hrtimer_run_queues()
function:

<__hrtimer_run_queues+249>:  nopl   0x0(%rax,%rax,1)
<__hrtimer_run_queues+254>:  mov    %r14,%rdi
<__hrtimer_run_queues+257>:  cs call <__x86_indirect_thunk_r12>
<__hrtimer_run_queues+263>:  mov    %eax,%r12d
<__hrtimer_run_queues+266>:  xchg   %ax,%ax
<__hrtimer_run_queues+268>:  mov    %r13,%rdi

Which triggers this crash:

  BUG: unable to handle page fault for address: 00000000000f41c9
  #PF: supervisor write access in kernel mode
  #PF: error_code(0x0002) - not-present page
  PGD 0 P4D 0
  Oops: 0002 [#1] SMP NOPTI
  CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P
  RIP: 0010:__hrtimer_run_queues+0x106/0x230

Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is
at the 6th byte of the above CS CALL instruction. Since the CS CALL
instruction occupies 6 bytes, the exception occurred in the middle of that
call instruction.

The root cause is that when using eBPF tools to probe in the middle of a
function, a kprobe with INT3 is used as the underlying implementation.

During single-step emulation of the original CALL instruction,
int3_emulate_call() assumes that the probed CALL instruction is 5 bytes
long. However, the actual CS-prefixed CALL instruction occupies 6 bytes,
so it constructs an incorrect exception return address. When the CPU
returns from the kprobe handler, the next instruction to be executed is at
the address of the last byte of that CS CALL instruction. Coincidentally,
starting from that address, the CPU fetches and decodes a completely
different instruction, which ultimately triggers a kernel crash.

Fix the issue by using the actual instruction length obtained from
the instruction decoder when constructing the exception return
address, rather than relying on the hardcoded CALL_INSN_SIZE macro.

[ mingo: Refined the changelog ]

Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step")
Suggested-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 262 insertions, 0 deletions

diff --git a/crypto/tea.c b/crypto/tea.c
new file mode 100644
index 000000000..7c66efcb5
--- /dev/null
+++ b/crypto/tea.c
@@ -0,0 +1,262 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* 
+ * Cryptographic API.
+ *
+ * TEA, XTEA, and XETA crypto algorithms
+ *
+ * The TEA and Xtended TEA algorithms were developed by David Wheeler 
+ * and Roger Needham at the Computer Laboratory of Cambridge University.
+ *
+ * Due to the order of evaluation in XTEA many people have incorrectly
+ * implemented it.  XETA (XTEA in the wrong order), exists for
+ * compatibility with these implementations.
+ *
+ * Copyright (c) 2004 Aaron Grothe ajgrothe@yahoo.com
+ */
+
+#include <crypto/algapi.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/unaligned.h>
+#include <linux/types.h>
+
+#define TEA_KEY_SIZE		16
+#define TEA_BLOCK_SIZE		8
+#define TEA_ROUNDS		32
+#define TEA_DELTA		0x9e3779b9
+
+#define XTEA_KEY_SIZE		16
+#define XTEA_BLOCK_SIZE		8
+#define XTEA_ROUNDS		32
+#define XTEA_DELTA		0x9e3779b9
+
+struct tea_ctx {
+	u32 KEY[4];
+};
+
+struct xtea_ctx {
+	u32 KEY[4];
+};
+
+static int tea_setkey(struct crypto_tfm *tfm, const u8 *in_key,
+		      unsigned int key_len)
+{
+	struct tea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	ctx->KEY[0] = get_unaligned_le32(&in_key[0]);
+	ctx->KEY[1] = get_unaligned_le32(&in_key[4]);
+	ctx->KEY[2] = get_unaligned_le32(&in_key[8]);
+	ctx->KEY[3] = get_unaligned_le32(&in_key[12]);
+
+	return 0; 
+
+}
+
+static void tea_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	u32 y, z, n, sum = 0;
+	u32 k0, k1, k2, k3;
+	struct tea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	y = get_unaligned_le32(&src[0]);
+	z = get_unaligned_le32(&src[4]);
+
+	k0 = ctx->KEY[0];
+	k1 = ctx->KEY[1];
+	k2 = ctx->KEY[2];
+	k3 = ctx->KEY[3];
+
+	n = TEA_ROUNDS;
+
+	while (n-- > 0) {
+		sum += TEA_DELTA;
+		y += ((z << 4) + k0) ^ (z + sum) ^ ((z >> 5) + k1);
+		z += ((y << 4) + k2) ^ (y + sum) ^ ((y >> 5) + k3);
+	}
+	
+	put_unaligned_le32(y, &dst[0]);
+	put_unaligned_le32(z, &dst[4]);
+}
+
+static void tea_decrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	u32 y, z, n, sum;
+	u32 k0, k1, k2, k3;
+	struct tea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	y = get_unaligned_le32(&src[0]);
+	z = get_unaligned_le32(&src[4]);
+
+	k0 = ctx->KEY[0];
+	k1 = ctx->KEY[1];
+	k2 = ctx->KEY[2];
+	k3 = ctx->KEY[3];
+
+	sum = TEA_DELTA << 5;
+
+	n = TEA_ROUNDS;
+
+	while (n-- > 0) {
+		z -= ((y << 4) + k2) ^ (y + sum) ^ ((y >> 5) + k3);
+		y -= ((z << 4) + k0) ^ (z + sum) ^ ((z >> 5) + k1);
+		sum -= TEA_DELTA;
+	}
+	
+	put_unaligned_le32(y, &dst[0]);
+	put_unaligned_le32(z, &dst[4]);
+}
+
+static int xtea_setkey(struct crypto_tfm *tfm, const u8 *in_key,
+		       unsigned int key_len)
+{
+	struct xtea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	ctx->KEY[0] = get_unaligned_le32(&in_key[0]);
+	ctx->KEY[1] = get_unaligned_le32(&in_key[4]);
+	ctx->KEY[2] = get_unaligned_le32(&in_key[8]);
+	ctx->KEY[3] = get_unaligned_le32(&in_key[12]);
+
+	return 0; 
+
+}
+
+static void xtea_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	u32 y, z, sum = 0;
+	u32 limit = XTEA_DELTA * XTEA_ROUNDS;
+	struct xtea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	y = get_unaligned_le32(&src[0]);
+	z = get_unaligned_le32(&src[4]);
+
+	while (sum != limit) {
+		y += ((z << 4 ^ z >> 5) + z) ^ (sum + ctx->KEY[sum&3]); 
+		sum += XTEA_DELTA;
+		z += ((y << 4 ^ y >> 5) + y) ^ (sum + ctx->KEY[sum>>11 &3]); 
+	}
+	
+	put_unaligned_le32(y, &dst[0]);
+	put_unaligned_le32(z, &dst[4]);
+}
+
+static void xtea_decrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	u32 y, z, sum;
+	struct tea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	y = get_unaligned_le32(&src[0]);
+	z = get_unaligned_le32(&src[4]);
+
+	sum = XTEA_DELTA * XTEA_ROUNDS;
+
+	while (sum) {
+		z -= ((y << 4 ^ y >> 5) + y) ^ (sum + ctx->KEY[sum>>11 & 3]);
+		sum -= XTEA_DELTA;
+		y -= ((z << 4 ^ z >> 5) + z) ^ (sum + ctx->KEY[sum & 3]);
+	}
+	
+	put_unaligned_le32(y, &dst[0]);
+	put_unaligned_le32(z, &dst[4]);
+}
+
+
+static void xeta_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	u32 y, z, sum = 0;
+	u32 limit = XTEA_DELTA * XTEA_ROUNDS;
+	struct xtea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	y = get_unaligned_le32(&src[0]);
+	z = get_unaligned_le32(&src[4]);
+
+	while (sum != limit) {
+		y += (z << 4 ^ z >> 5) + (z ^ sum) + ctx->KEY[sum&3];
+		sum += XTEA_DELTA;
+		z += (y << 4 ^ y >> 5) + (y ^ sum) + ctx->KEY[sum>>11 &3];
+	}
+	
+	put_unaligned_le32(y, &dst[0]);
+	put_unaligned_le32(z, &dst[4]);
+}
+
+static void xeta_decrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+	u32 y, z, sum;
+	struct tea_ctx *ctx = crypto_tfm_ctx(tfm);
+
+	y = get_unaligned_le32(&src[0]);
+	z = get_unaligned_le32(&src[4]);
+
+	sum = XTEA_DELTA * XTEA_ROUNDS;
+
+	while (sum) {
+		z -= (y << 4 ^ y >> 5) + (y ^ sum) + ctx->KEY[sum>>11 & 3];
+		sum -= XTEA_DELTA;
+		y -= (z << 4 ^ z >> 5) + (z ^ sum) + ctx->KEY[sum & 3];
+	}
+	
+	put_unaligned_le32(y, &dst[0]);
+	put_unaligned_le32(z, &dst[4]);
+}
+
+static struct crypto_alg tea_algs[3] = { {
+	.cra_name		=	"tea",
+	.cra_driver_name	=	"tea-generic",
+	.cra_flags		=	CRYPTO_ALG_TYPE_CIPHER,
+	.cra_blocksize		=	TEA_BLOCK_SIZE,
+	.cra_ctxsize		=	sizeof (struct tea_ctx),
+	.cra_module		=	THIS_MODULE,
+	.cra_u			=	{ .cipher = {
+	.cia_min_keysize	=	TEA_KEY_SIZE,
+	.cia_max_keysize	=	TEA_KEY_SIZE,
+	.cia_setkey		= 	tea_setkey,
+	.cia_encrypt		=	tea_encrypt,
+	.cia_decrypt		=	tea_decrypt } }
+}, {
+	.cra_name		=	"xtea",
+	.cra_driver_name	=	"xtea-generic",
+	.cra_flags		=	CRYPTO_ALG_TYPE_CIPHER,
+	.cra_blocksize		=	XTEA_BLOCK_SIZE,
+	.cra_ctxsize		=	sizeof (struct xtea_ctx),
+	.cra_module		=	THIS_MODULE,
+	.cra_u			=	{ .cipher = {
+	.cia_min_keysize	=	XTEA_KEY_SIZE,
+	.cia_max_keysize	=	XTEA_KEY_SIZE,
+	.cia_setkey		= 	xtea_setkey,
+	.cia_encrypt		=	xtea_encrypt,
+	.cia_decrypt		=	xtea_decrypt } }
+}, {
+	.cra_name		=	"xeta",
+	.cra_driver_name	=	"xeta-generic",
+	.cra_flags		=	CRYPTO_ALG_TYPE_CIPHER,
+	.cra_blocksize		=	XTEA_BLOCK_SIZE,
+	.cra_ctxsize		=	sizeof (struct xtea_ctx),
+	.cra_module		=	THIS_MODULE,
+	.cra_u			=	{ .cipher = {
+	.cia_min_keysize	=	XTEA_KEY_SIZE,
+	.cia_max_keysize	=	XTEA_KEY_SIZE,
+	.cia_setkey		= 	xtea_setkey,
+	.cia_encrypt		=	xeta_encrypt,
+	.cia_decrypt		=	xeta_decrypt } }
+} };
+
+static int __init tea_mod_init(void)
+{
+	return crypto_register_algs(tea_algs, ARRAY_SIZE(tea_algs));
+}
+
+static void __exit tea_mod_fini(void)
+{
+	crypto_unregister_algs(tea_algs, ARRAY_SIZE(tea_algs));
+}
+
+MODULE_ALIAS_CRYPTO("tea");
+MODULE_ALIAS_CRYPTO("xtea");
+MODULE_ALIAS_CRYPTO("xeta");
+
+module_init(tea_mod_init);
+module_exit(tea_mod_fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("TEA, XTEA & XETA Cryptographic Algorithms");