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, 139 insertions, 0 deletions

diff --git a/fs/orangefs/dcache.c b/fs/orangefs/dcache.c
new file mode 100644
index 000000000..a19d1ad70
--- /dev/null
+++ b/fs/orangefs/dcache.c
@@ -0,0 +1,139 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * (C) 2001 Clemson University and The University of Chicago
+ *
+ * See COPYING in top-level directory.
+ */
+
+/*
+ *  Implementation of dentry (directory cache) functions.
+ */
+
+#include "protocol.h"
+#include "orangefs-kernel.h"
+
+/* Returns 1 if dentry can still be trusted, else 0. */
+static int orangefs_revalidate_lookup(struct inode *parent_inode, const struct qstr *name,
+				      struct dentry *dentry)
+{
+	struct orangefs_inode_s *parent = ORANGEFS_I(parent_inode);
+	struct inode *inode = dentry->d_inode;
+	struct orangefs_kernel_op_s *new_op;
+	int ret = 0;
+	int err = 0;
+
+	gossip_debug(GOSSIP_DCACHE_DEBUG, "%s: attempting lookup.\n", __func__);
+
+	new_op = op_alloc(ORANGEFS_VFS_OP_LOOKUP);
+	if (!new_op)
+		return -ENOMEM;
+
+	new_op->upcall.req.lookup.sym_follow = ORANGEFS_LOOKUP_LINK_NO_FOLLOW;
+	new_op->upcall.req.lookup.parent_refn = parent->refn;
+	/* op_alloc() leaves ->upcall zeroed */
+	memcpy(new_op->upcall.req.lookup.d_name, name->name,
+			min(name->len, ORANGEFS_NAME_MAX - 1));
+
+	gossip_debug(GOSSIP_DCACHE_DEBUG,
+		     "%s:%s:%d interrupt flag [%d]\n",
+		     __FILE__,
+		     __func__,
+		     __LINE__,
+		     get_interruptible_flag(parent_inode));
+
+	err = service_operation(new_op, "orangefs_lookup",
+			get_interruptible_flag(parent_inode));
+
+	/* Positive dentry: reject if error or not the same inode. */
+	if (inode) {
+		if (err) {
+			gossip_debug(GOSSIP_DCACHE_DEBUG,
+			    "%s:%s:%d lookup failure.\n",
+			    __FILE__, __func__, __LINE__);
+			goto out_drop;
+		}
+		if (!match_handle(new_op->downcall.resp.lookup.refn.khandle,
+		    inode)) {
+			gossip_debug(GOSSIP_DCACHE_DEBUG,
+			    "%s:%s:%d no match.\n",
+			    __FILE__, __func__, __LINE__);
+			goto out_drop;
+		}
+
+	/* Negative dentry: reject if success or error other than ENOENT. */
+	} else {
+		gossip_debug(GOSSIP_DCACHE_DEBUG, "%s: negative dentry.\n",
+		    __func__);
+		if (!err || err != -ENOENT) {
+			if (new_op->downcall.status != 0)
+				gossip_debug(GOSSIP_DCACHE_DEBUG,
+				    "%s:%s:%d lookup failure.\n",
+				    __FILE__, __func__, __LINE__);
+			goto out_drop;
+		}
+	}
+
+	orangefs_set_timeout(dentry);
+	ret = 1;
+out_release_op:
+	op_release(new_op);
+	return ret;
+out_drop:
+	gossip_debug(GOSSIP_DCACHE_DEBUG, "%s:%s:%d revalidate failed\n",
+	    __FILE__, __func__, __LINE__);
+	goto out_release_op;
+}
+
+/*
+ * Verify that dentry is valid.
+ *
+ * Should return 1 if dentry can still be trusted, else 0.
+ */
+static int orangefs_d_revalidate(struct inode *dir, const struct qstr *name,
+				 struct dentry *dentry, unsigned int flags)
+{
+	int ret;
+	unsigned long time = (unsigned long) dentry->d_fsdata;
+
+	if (time_before(jiffies, time))
+		return 1;
+
+	if (flags & LOOKUP_RCU)
+		return -ECHILD;
+
+	gossip_debug(GOSSIP_DCACHE_DEBUG, "%s: called on dentry %p.\n",
+		     __func__, dentry);
+
+	/* skip root handle lookups. */
+	if (dentry->d_inode && is_root_handle(dentry->d_inode))
+		return 1;
+
+	/*
+	 * If this passes, the positive dentry still exists or the negative
+	 * dentry still does not exist.
+	 */
+	if (!orangefs_revalidate_lookup(dir, name, dentry))
+		return 0;
+
+	/* We do not need to continue with negative dentries. */
+	if (!dentry->d_inode) {
+		gossip_debug(GOSSIP_DCACHE_DEBUG,
+		    "%s: negative dentry or positive dentry and inode valid.\n",
+		    __func__);
+		return 1;
+	}
+
+	/* Now we must perform a getattr to validate the inode contents. */
+
+	ret = orangefs_inode_check_changed(dentry->d_inode);
+	if (ret < 0) {
+		gossip_debug(GOSSIP_DCACHE_DEBUG, "%s:%s:%d getattr failure.\n",
+		    __FILE__, __func__, __LINE__);
+		return 0;
+	}
+	return !ret;
+}
+
+const struct dentry_operations orangefs_dentry_operations = {
+	.d_revalidate = orangefs_d_revalidate,
+};