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, 158 insertions, 0 deletions

diff --git a/fs/nfs/cache_lib.c b/fs/nfs/cache_lib.c
new file mode 100644
index 000000000..9738a1ae9
--- /dev/null
+++ b/fs/nfs/cache_lib.c
@@ -0,0 +1,158 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * linux/fs/nfs/cache_lib.c
+ *
+ * Helper routines for the NFS client caches
+ *
+ * Copyright (c) 2009 Trond Myklebust <Trond.Myklebust@netapp.com>
+ */
+#include <linux/kmod.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/mount.h>
+#include <linux/namei.h>
+#include <linux/slab.h>
+#include <linux/sunrpc/cache.h>
+#include <linux/sunrpc/rpc_pipe_fs.h>
+#include <net/net_namespace.h>
+
+#include "cache_lib.h"
+
+#define NFS_CACHE_UPCALL_PATHLEN 256
+#define NFS_CACHE_UPCALL_TIMEOUT 15
+
+static char nfs_cache_getent_prog[NFS_CACHE_UPCALL_PATHLEN] =
+				"/sbin/nfs_cache_getent";
+static unsigned long nfs_cache_getent_timeout = NFS_CACHE_UPCALL_TIMEOUT;
+
+module_param_string(cache_getent, nfs_cache_getent_prog,
+		sizeof(nfs_cache_getent_prog), 0600);
+MODULE_PARM_DESC(cache_getent, "Path to the client cache upcall program");
+module_param_named(cache_getent_timeout, nfs_cache_getent_timeout, ulong, 0600);
+MODULE_PARM_DESC(cache_getent_timeout, "Timeout (in seconds) after which "
+		"the cache upcall is assumed to have failed");
+
+int nfs_cache_upcall(struct cache_detail *cd, char *entry_name)
+{
+	static char *envp[] = { "HOME=/",
+		"TERM=linux",
+		"PATH=/sbin:/usr/sbin:/bin:/usr/bin",
+		NULL
+	};
+	char *argv[] = {
+		nfs_cache_getent_prog,
+		cd->name,
+		entry_name,
+		NULL
+	};
+	int ret = -EACCES;
+
+	if (nfs_cache_getent_prog[0] == '\0')
+		goto out;
+	ret = call_usermodehelper(argv[0], argv, envp, UMH_WAIT_EXEC);
+	/*
+	 * Disable the upcall mechanism if we're getting an ENOENT or
+	 * EACCES error. The admin can re-enable it on the fly by using
+	 * sysfs to set the 'cache_getent' parameter once the problem
+	 * has been fixed.
+	 */
+	if (ret == -ENOENT || ret == -EACCES)
+		nfs_cache_getent_prog[0] = '\0';
+out:
+	return ret > 0 ? 0 : ret;
+}
+
+/*
+ * Deferred request handling
+ */
+void nfs_cache_defer_req_put(struct nfs_cache_defer_req *dreq)
+{
+	if (refcount_dec_and_test(&dreq->count))
+		kfree(dreq);
+}
+
+static void nfs_dns_cache_revisit(struct cache_deferred_req *d, int toomany)
+{
+	struct nfs_cache_defer_req *dreq;
+
+	dreq = container_of(d, struct nfs_cache_defer_req, deferred_req);
+
+	complete(&dreq->completion);
+	nfs_cache_defer_req_put(dreq);
+}
+
+static struct cache_deferred_req *nfs_dns_cache_defer(struct cache_req *req)
+{
+	struct nfs_cache_defer_req *dreq;
+
+	dreq = container_of(req, struct nfs_cache_defer_req, req);
+	dreq->deferred_req.revisit = nfs_dns_cache_revisit;
+	refcount_inc(&dreq->count);
+
+	return &dreq->deferred_req;
+}
+
+struct nfs_cache_defer_req *nfs_cache_defer_req_alloc(void)
+{
+	struct nfs_cache_defer_req *dreq;
+
+	dreq = kzalloc_obj(*dreq);
+	if (dreq) {
+		init_completion(&dreq->completion);
+		refcount_set(&dreq->count, 1);
+		dreq->req.defer = nfs_dns_cache_defer;
+	}
+	return dreq;
+}
+
+int nfs_cache_wait_for_upcall(struct nfs_cache_defer_req *dreq)
+{
+	if (wait_for_completion_timeout(&dreq->completion,
+			nfs_cache_getent_timeout * HZ) == 0)
+		return -ETIMEDOUT;
+	return 0;
+}
+
+int nfs_cache_register_sb(struct super_block *sb, struct cache_detail *cd)
+{
+	int ret;
+	struct dentry *dir;
+
+	dir = rpc_d_lookup_sb(sb, "cache");
+	ret = sunrpc_cache_register_pipefs(dir, cd->name, 0600, cd);
+	dput(dir);
+	return ret;
+}
+
+int nfs_cache_register_net(struct net *net, struct cache_detail *cd)
+{
+	struct super_block *pipefs_sb;
+	int ret = 0;
+
+	sunrpc_init_cache_detail(cd);
+	pipefs_sb = rpc_get_sb_net(net);
+	if (pipefs_sb) {
+		ret = nfs_cache_register_sb(pipefs_sb, cd);
+		rpc_put_sb_net(net);
+		if (ret)
+			sunrpc_destroy_cache_detail(cd);
+	}
+	return ret;
+}
+
+void nfs_cache_unregister_sb(struct super_block *sb, struct cache_detail *cd)
+{
+	sunrpc_cache_unregister_pipefs(cd);
+}
+
+void nfs_cache_unregister_net(struct net *net, struct cache_detail *cd)
+{
+	struct super_block *pipefs_sb;
+
+	pipefs_sb = rpc_get_sb_net(net);
+	if (pipefs_sb) {
+		nfs_cache_unregister_sb(pipefs_sb, cd);
+		rpc_put_sb_net(net);
+	}
+	sunrpc_destroy_cache_detail(cd);
+}