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, 351 insertions, 0 deletions

diff --git a/drivers/net/netdevsim/macsec.c b/drivers/net/netdevsim/macsec.c
new file mode 100644
index 000000000..bdc8020d5
--- /dev/null
+++ b/drivers/net/netdevsim/macsec.c
@@ -0,0 +1,351 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <net/macsec.h>
+#include "netdevsim.h"
+
+static int nsim_macsec_find_secy(struct netdevsim *ns, sci_t sci)
+{
+	int i;
+
+	for (i = 0; i < NSIM_MACSEC_MAX_SECY_COUNT; i++) {
+		if (ns->macsec.nsim_secy[i].sci == sci)
+			return i;
+	}
+
+	return -1;
+}
+
+static int nsim_macsec_find_rxsc(struct nsim_secy *ns_secy, sci_t sci)
+{
+	int i;
+
+	for (i = 0; i < NSIM_MACSEC_MAX_RXSC_COUNT; i++) {
+		if (ns_secy->nsim_rxsc[i].sci == sci)
+			return i;
+	}
+
+	return -1;
+}
+
+static int nsim_macsec_add_secy(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	int idx;
+
+	if (ns->macsec.nsim_secy_count == NSIM_MACSEC_MAX_SECY_COUNT)
+		return -ENOSPC;
+
+	for (idx = 0; idx < NSIM_MACSEC_MAX_SECY_COUNT; idx++) {
+		if (!ns->macsec.nsim_secy[idx].used)
+			break;
+	}
+
+	if (idx == NSIM_MACSEC_MAX_SECY_COUNT) {
+		netdev_err(ctx->netdev, "%s: nsim_secy_count not full but all SecYs used\n",
+			   __func__);
+		return -ENOSPC;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: adding new secy with sci %016llx at index %d\n",
+		   __func__, sci_to_cpu(ctx->secy->sci), idx);
+	ns->macsec.nsim_secy[idx].used = true;
+	ns->macsec.nsim_secy[idx].nsim_rxsc_count = 0;
+	ns->macsec.nsim_secy[idx].sci = ctx->secy->sci;
+	ns->macsec.nsim_secy_count++;
+
+	return 0;
+}
+
+static int nsim_macsec_upd_secy(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: updating secy with sci %016llx at index %d\n",
+		   __func__, sci_to_cpu(ctx->secy->sci), idx);
+
+	return 0;
+}
+
+static int nsim_macsec_del_secy(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: removing SecY with SCI %016llx at index %d\n",
+		   __func__, sci_to_cpu(ctx->secy->sci), idx);
+
+	ns->macsec.nsim_secy[idx].used = false;
+	memset(&ns->macsec.nsim_secy[idx], 0, sizeof(ns->macsec.nsim_secy[idx]));
+	ns->macsec.nsim_secy_count--;
+
+	return 0;
+}
+
+static int nsim_macsec_add_rxsc(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	struct nsim_secy *secy;
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+	secy = &ns->macsec.nsim_secy[idx];
+
+	if (secy->nsim_rxsc_count == NSIM_MACSEC_MAX_RXSC_COUNT)
+		return -ENOSPC;
+
+	for (idx = 0; idx < NSIM_MACSEC_MAX_RXSC_COUNT; idx++) {
+		if (!secy->nsim_rxsc[idx].used)
+			break;
+	}
+
+	if (idx == NSIM_MACSEC_MAX_RXSC_COUNT)
+		netdev_err(ctx->netdev, "%s: nsim_rxsc_count not full but all RXSCs used\n",
+			   __func__);
+
+	netdev_dbg(ctx->netdev, "%s: adding new rxsc with sci %016llx at index %d\n",
+		   __func__, sci_to_cpu(ctx->rx_sc->sci), idx);
+	secy->nsim_rxsc[idx].used = true;
+	secy->nsim_rxsc[idx].sci = ctx->rx_sc->sci;
+	secy->nsim_rxsc_count++;
+
+	return 0;
+}
+
+static int nsim_macsec_upd_rxsc(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	struct nsim_secy *secy;
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+	secy = &ns->macsec.nsim_secy[idx];
+
+	idx = nsim_macsec_find_rxsc(secy, ctx->rx_sc->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in RXSC table\n",
+			   __func__, sci_to_cpu(ctx->rx_sc->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: updating RXSC with sci %016llx at index %d\n",
+		   __func__, sci_to_cpu(ctx->rx_sc->sci), idx);
+
+	return 0;
+}
+
+static int nsim_macsec_del_rxsc(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	struct nsim_secy *secy;
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+	secy = &ns->macsec.nsim_secy[idx];
+
+	idx = nsim_macsec_find_rxsc(secy, ctx->rx_sc->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in RXSC table\n",
+			   __func__, sci_to_cpu(ctx->rx_sc->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: removing RXSC with sci %016llx at index %d\n",
+		   __func__, sci_to_cpu(ctx->rx_sc->sci), idx);
+
+	secy->nsim_rxsc[idx].used = false;
+	memset(&secy->nsim_rxsc[idx], 0, sizeof(secy->nsim_rxsc[idx]));
+	secy->nsim_rxsc_count--;
+
+	return 0;
+}
+
+static int nsim_macsec_add_rxsa(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	struct nsim_secy *secy;
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+	secy = &ns->macsec.nsim_secy[idx];
+
+	idx = nsim_macsec_find_rxsc(secy, ctx->sa.rx_sa->sc->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in RXSC table\n",
+			   __func__, sci_to_cpu(ctx->sa.rx_sa->sc->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: RXSC with sci %016llx, AN %u\n",
+		   __func__, sci_to_cpu(ctx->sa.rx_sa->sc->sci), ctx->sa.assoc_num);
+
+	return 0;
+}
+
+static int nsim_macsec_upd_rxsa(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	struct nsim_secy *secy;
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+	secy = &ns->macsec.nsim_secy[idx];
+
+	idx = nsim_macsec_find_rxsc(secy, ctx->sa.rx_sa->sc->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in RXSC table\n",
+			   __func__, sci_to_cpu(ctx->sa.rx_sa->sc->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: RXSC with sci %016llx, AN %u\n",
+		   __func__, sci_to_cpu(ctx->sa.rx_sa->sc->sci), ctx->sa.assoc_num);
+
+	return 0;
+}
+
+static int nsim_macsec_del_rxsa(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	struct nsim_secy *secy;
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+	secy = &ns->macsec.nsim_secy[idx];
+
+	idx = nsim_macsec_find_rxsc(secy, ctx->sa.rx_sa->sc->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in RXSC table\n",
+			   __func__, sci_to_cpu(ctx->sa.rx_sa->sc->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: RXSC with sci %016llx, AN %u\n",
+		   __func__, sci_to_cpu(ctx->sa.rx_sa->sc->sci), ctx->sa.assoc_num);
+
+	return 0;
+}
+
+static int nsim_macsec_add_txsa(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: SECY with sci %016llx, AN %u\n",
+		   __func__, sci_to_cpu(ctx->secy->sci), ctx->sa.assoc_num);
+
+	return 0;
+}
+
+static int nsim_macsec_upd_txsa(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: SECY with sci %016llx, AN %u\n",
+		   __func__, sci_to_cpu(ctx->secy->sci), ctx->sa.assoc_num);
+
+	return 0;
+}
+
+static int nsim_macsec_del_txsa(struct macsec_context *ctx)
+{
+	struct netdevsim *ns = netdev_priv(ctx->netdev);
+	int idx;
+
+	idx = nsim_macsec_find_secy(ns, ctx->secy->sci);
+	if (idx < 0) {
+		netdev_err(ctx->netdev, "%s: sci %016llx not found in secy table\n",
+			   __func__, sci_to_cpu(ctx->secy->sci));
+		return -ENOENT;
+	}
+
+	netdev_dbg(ctx->netdev, "%s: SECY with sci %016llx, AN %u\n",
+		   __func__, sci_to_cpu(ctx->secy->sci), ctx->sa.assoc_num);
+
+	return 0;
+}
+
+static const struct macsec_ops nsim_macsec_ops = {
+	.mdo_add_secy = nsim_macsec_add_secy,
+	.mdo_upd_secy = nsim_macsec_upd_secy,
+	.mdo_del_secy = nsim_macsec_del_secy,
+	.mdo_add_rxsc = nsim_macsec_add_rxsc,
+	.mdo_upd_rxsc = nsim_macsec_upd_rxsc,
+	.mdo_del_rxsc = nsim_macsec_del_rxsc,
+	.mdo_add_rxsa = nsim_macsec_add_rxsa,
+	.mdo_upd_rxsa = nsim_macsec_upd_rxsa,
+	.mdo_del_rxsa = nsim_macsec_del_rxsa,
+	.mdo_add_txsa = nsim_macsec_add_txsa,
+	.mdo_upd_txsa = nsim_macsec_upd_txsa,
+	.mdo_del_txsa = nsim_macsec_del_txsa,
+};
+
+void nsim_macsec_init(struct netdevsim *ns)
+{
+	ns->netdev->macsec_ops = &nsim_macsec_ops;
+	ns->netdev->features |= NETIF_F_HW_MACSEC;
+	memset(&ns->macsec, 0, sizeof(ns->macsec));
+}
+
+void nsim_macsec_teardown(struct netdevsim *ns)
+{
+}