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, 130 insertions, 0 deletions

diff --git a/fs/nfsd/flexfilelayoutxdr.c b/fs/nfsd/flexfilelayoutxdr.c
new file mode 100644
index 000000000..374e52d30
--- /dev/null
+++ b/fs/nfsd/flexfilelayoutxdr.c
@@ -0,0 +1,130 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2016 Tom Haynes <loghyr@primarydata.com>
+ */
+#include <linux/sunrpc/svc.h>
+#include <linux/nfs4.h>
+
+#include "nfsd.h"
+#include "flexfilelayoutxdr.h"
+
+#define NFSDDBG_FACILITY	NFSDDBG_PNFS
+
+struct ff_idmap {
+	char buf[11];
+	int len;
+};
+
+__be32
+nfsd4_ff_encode_layoutget(struct xdr_stream *xdr,
+		const struct nfsd4_layoutget *lgp)
+{
+	const struct pnfs_ff_layout *fl = lgp->lg_content;
+	int len, mirror_len, ds_len, fh_len;
+	__be32 *p;
+
+	/*
+	 * Unlike nfsd4_encode_user, we know these will
+	 * always be stringified.
+	 */
+	struct ff_idmap uid;
+	struct ff_idmap gid;
+
+	fh_len = 4 + xdr_align_size(fl->fh.size);
+
+	uid.len = sprintf(uid.buf, "%u", from_kuid(&init_user_ns, fl->uid));
+	gid.len = sprintf(gid.buf, "%u", from_kgid(&init_user_ns, fl->gid));
+
+	/* data server entry: deviceid + efficiency + stateid + fh list +
+	 * user + group + flags + stats_collect_hint
+	 */
+	ds_len = 16 + 4 + 4 + sizeof(stateid_opaque_t) + 4 + fh_len +
+		 4 + xdr_align_size(uid.len) +
+		 4 + xdr_align_size(gid.len) +
+		 4 + 4;
+
+	/* mirror: ds_count + ds */
+	mirror_len = 4 + ds_len;
+
+	/* stripe_unit + mirror_count + mirror */
+	len = 12 + mirror_len;
+
+	p = xdr_reserve_space(xdr, sizeof(__be32) + len);
+	if (!p)
+		return nfserr_toosmall;
+
+	*p++ = cpu_to_be32(len);
+	p = xdr_encode_hyper(p, 0);		/* stripe unit of 1 */
+
+	*p++ = cpu_to_be32(1);			/* single mirror */
+	*p++ = cpu_to_be32(1);			/* single data server */
+
+	p = svcxdr_encode_deviceid4(p, &fl->deviceid);
+
+	*p++ = cpu_to_be32(1);			/* efficiency */
+
+	*p++ = cpu_to_be32(fl->stateid.si_generation);
+	p = xdr_encode_opaque_fixed(p, &fl->stateid.si_opaque,
+				    sizeof(stateid_opaque_t));
+
+	*p++ = cpu_to_be32(1);			/* single file handle */
+	p = xdr_encode_opaque(p, fl->fh.data, fl->fh.size);
+
+	p = xdr_encode_opaque(p, uid.buf, uid.len);
+	p = xdr_encode_opaque(p, gid.buf, gid.len);
+
+	*p++ = cpu_to_be32(fl->flags);
+	*p++ = cpu_to_be32(0);			/* No stats collect hint */
+
+	return 0;
+}
+
+__be32
+nfsd4_ff_encode_getdeviceinfo(struct xdr_stream *xdr,
+		const struct nfsd4_getdeviceinfo *gdp)
+{
+	struct pnfs_ff_device_addr *da = gdp->gd_device;
+	int len;
+	int ver_len;
+	int addr_len;
+	__be32 *p;
+
+	/*
+	 * See paragraph 5 of RFC 8881 S18.40.3.
+	 */
+	if (!gdp->gd_maxcount) {
+		if (xdr_stream_encode_u32(xdr, 0) != XDR_UNIT)
+			return nfserr_resource;
+		return nfs_ok;
+	}
+
+	/* len + padding for two strings */
+	addr_len = 8 + xdr_align_size(da->netaddr.netid_len) +
+		   xdr_align_size(da->netaddr.addr_len);
+	ver_len = 20;
+
+	len = 4 + ver_len + 4 + addr_len;
+
+	p = xdr_reserve_space(xdr, len + sizeof(__be32));
+	if (!p)
+		return nfserr_resource;
+
+	/*
+	 * Fill in the overall length and number of volumes at the beginning
+	 * of the layout.
+	 */
+	*p++ = cpu_to_be32(len);
+	*p++ = cpu_to_be32(1);			/* 1 netaddr */
+	p = xdr_encode_opaque(p, da->netaddr.netid, da->netaddr.netid_len);
+	p = xdr_encode_opaque(p, da->netaddr.addr, da->netaddr.addr_len);
+
+	*p++ = cpu_to_be32(1);			/* 1 versions */
+
+	*p++ = cpu_to_be32(da->version);
+	*p++ = cpu_to_be32(da->minor_version);
+	*p++ = cpu_to_be32(da->rsize);
+	*p++ = cpu_to_be32(da->wsize);
+	*p++ = cpu_to_be32(da->tightly_coupled);
+
+	return 0;
+}