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, 242 insertions, 0 deletions

diff --git a/drivers/md/dm-thin-metadata.h b/drivers/md/dm-thin-metadata.h
new file mode 100644
index 000000000..2f64f48b5
--- /dev/null
+++ b/drivers/md/dm-thin-metadata.h
@@ -0,0 +1,242 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2010-2011 Red Hat, Inc.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef DM_THIN_METADATA_H
+#define DM_THIN_METADATA_H
+
+#include "persistent-data/dm-block-manager.h"
+#include "persistent-data/dm-space-map.h"
+#include "persistent-data/dm-space-map-metadata.h"
+
+#define THIN_METADATA_BLOCK_SIZE DM_SM_METADATA_BLOCK_SIZE
+
+/*
+ * The metadata device is currently limited in size.
+ */
+#define THIN_METADATA_MAX_SECTORS DM_SM_METADATA_MAX_SECTORS
+
+/*
+ * A metadata device larger than 16GB triggers a warning.
+ */
+#define THIN_METADATA_MAX_SECTORS_WARNING (16 * (1024 * 1024 * 1024 >> SECTOR_SHIFT))
+
+/*----------------------------------------------------------------*/
+
+/*
+ * Thin metadata superblock flags.
+ */
+#define THIN_METADATA_NEEDS_CHECK_FLAG (1 << 0)
+
+struct dm_pool_metadata;
+struct dm_thin_device;
+
+/*
+ * Device identifier
+ */
+typedef uint64_t dm_thin_id;
+
+/*
+ * Reopens or creates a new, empty metadata volume.
+ */
+struct dm_pool_metadata *dm_pool_metadata_open(struct block_device *bdev,
+					       sector_t data_block_size,
+					       bool format_device);
+
+int dm_pool_metadata_close(struct dm_pool_metadata *pmd);
+
+/*
+ * Compat feature flags.  Any incompat flags beyond the ones
+ * specified below will prevent use of the thin metadata.
+ */
+#define THIN_FEATURE_COMPAT_SUPP	  0UL
+#define THIN_FEATURE_COMPAT_RO_SUPP	  0UL
+#define THIN_FEATURE_INCOMPAT_SUPP	  0UL
+
+/*
+ * Device creation/deletion.
+ */
+int dm_pool_create_thin(struct dm_pool_metadata *pmd, dm_thin_id dev);
+
+/*
+ * An internal snapshot.
+ *
+ * You can only snapshot a quiesced origin i.e. one that is either
+ * suspended or not instanced at all.
+ */
+int dm_pool_create_snap(struct dm_pool_metadata *pmd, dm_thin_id dev,
+			dm_thin_id origin);
+
+/*
+ * Deletes a virtual device from the metadata.  It _is_ safe to call this
+ * when that device is open.  Operations on that device will just start
+ * failing.  You still need to call close() on the device.
+ */
+int dm_pool_delete_thin_device(struct dm_pool_metadata *pmd,
+			       dm_thin_id dev);
+
+/*
+ * Commits _all_ metadata changes: device creation, deletion, mapping
+ * updates.
+ */
+int dm_pool_commit_metadata(struct dm_pool_metadata *pmd);
+
+/*
+ * Discards all uncommitted changes.  Rereads the superblock, rolling back
+ * to the last good transaction.  Thin devices remain open.
+ * dm_thin_aborted_changes() tells you if they had uncommitted changes.
+ *
+ * If this call fails it's only useful to call dm_pool_metadata_close().
+ * All other methods will fail with -EINVAL.
+ */
+int dm_pool_abort_metadata(struct dm_pool_metadata *pmd);
+
+/*
+ * Set/get userspace transaction id.
+ */
+int dm_pool_set_metadata_transaction_id(struct dm_pool_metadata *pmd,
+					uint64_t current_id,
+					uint64_t new_id);
+
+int dm_pool_get_metadata_transaction_id(struct dm_pool_metadata *pmd,
+					uint64_t *result);
+
+/*
+ * Hold/get root for userspace transaction.
+ *
+ * The metadata snapshot is a copy of the current superblock (minus the
+ * space maps).  Userland can access the data structures for READ
+ * operations only.  A small performance hit is incurred by providing this
+ * copy of the metadata to userland due to extra copy-on-write operations
+ * on the metadata nodes.  Release this as soon as you finish with it.
+ */
+int dm_pool_reserve_metadata_snap(struct dm_pool_metadata *pmd);
+int dm_pool_release_metadata_snap(struct dm_pool_metadata *pmd);
+
+int dm_pool_get_metadata_snap(struct dm_pool_metadata *pmd,
+			      dm_block_t *result);
+
+/*
+ * Actions on a single virtual device.
+ */
+
+/*
+ * Opening the same device more than once will fail with -EBUSY.
+ */
+int dm_pool_open_thin_device(struct dm_pool_metadata *pmd, dm_thin_id dev,
+			     struct dm_thin_device **td);
+
+int dm_pool_close_thin_device(struct dm_thin_device *td);
+
+dm_thin_id dm_thin_dev_id(struct dm_thin_device *td);
+
+struct dm_thin_lookup_result {
+	dm_block_t block;
+	bool shared:1;
+};
+
+/*
+ * Returns:
+ *   -EWOULDBLOCK iff @can_issue_io is set and would issue IO
+ *   -ENODATA iff that mapping is not present.
+ *   0 success
+ */
+int dm_thin_find_block(struct dm_thin_device *td, dm_block_t block,
+		       int can_issue_io, struct dm_thin_lookup_result *result);
+
+/*
+ * Retrieve the next run of contiguously mapped blocks.  Useful for working
+ * out where to break up IO.  Returns 0 on success, < 0 on error.
+ */
+int dm_thin_find_mapped_range(struct dm_thin_device *td,
+			      dm_block_t begin, dm_block_t end,
+			      dm_block_t *thin_begin, dm_block_t *thin_end,
+			      dm_block_t *pool_begin, bool *maybe_shared);
+
+/*
+ * Obtain an unused block.
+ */
+int dm_pool_alloc_data_block(struct dm_pool_metadata *pmd, dm_block_t *result);
+
+/*
+ * Insert or remove block.
+ */
+int dm_thin_insert_block(struct dm_thin_device *td, dm_block_t block,
+			 dm_block_t data_block);
+
+int dm_thin_remove_range(struct dm_thin_device *td,
+			 dm_block_t begin, dm_block_t end);
+
+/*
+ * Queries.
+ */
+bool dm_thin_changed_this_transaction(struct dm_thin_device *td);
+
+bool dm_pool_changed_this_transaction(struct dm_pool_metadata *pmd);
+
+bool dm_thin_aborted_changes(struct dm_thin_device *td);
+
+int dm_thin_get_highest_mapped_block(struct dm_thin_device *td,
+				     dm_block_t *highest_mapped);
+
+int dm_thin_get_mapped_count(struct dm_thin_device *td, dm_block_t *result);
+
+int dm_pool_get_free_block_count(struct dm_pool_metadata *pmd,
+				 dm_block_t *result);
+
+int dm_pool_get_free_metadata_block_count(struct dm_pool_metadata *pmd,
+					  dm_block_t *result);
+
+int dm_pool_get_metadata_dev_size(struct dm_pool_metadata *pmd,
+				  dm_block_t *result);
+
+int dm_pool_get_data_dev_size(struct dm_pool_metadata *pmd, dm_block_t *result);
+
+int dm_pool_block_is_shared(struct dm_pool_metadata *pmd, dm_block_t b, bool *result);
+
+int dm_pool_inc_data_range(struct dm_pool_metadata *pmd, dm_block_t b, dm_block_t e);
+int dm_pool_dec_data_range(struct dm_pool_metadata *pmd, dm_block_t b, dm_block_t e);
+
+/*
+ * Returns -ENOSPC if the new size is too small and already allocated
+ * blocks would be lost.
+ */
+int dm_pool_resize_data_dev(struct dm_pool_metadata *pmd, dm_block_t new_size);
+int dm_pool_resize_metadata_dev(struct dm_pool_metadata *pmd, dm_block_t new_size);
+
+/*
+ * Flicks the underlying block manager into read only mode, so you know
+ * that nothing is changing.
+ */
+void dm_pool_metadata_read_only(struct dm_pool_metadata *pmd);
+void dm_pool_metadata_read_write(struct dm_pool_metadata *pmd);
+
+int dm_pool_register_metadata_threshold(struct dm_pool_metadata *pmd,
+					dm_block_t threshold,
+					dm_sm_threshold_fn fn,
+					void *context);
+
+/*
+ * Updates the superblock immediately.
+ */
+int dm_pool_metadata_set_needs_check(struct dm_pool_metadata *pmd);
+bool dm_pool_metadata_needs_check(struct dm_pool_metadata *pmd);
+
+/*
+ * Issue any prefetches that may be useful.
+ */
+void dm_pool_issue_prefetches(struct dm_pool_metadata *pmd);
+
+/* Pre-commit callback */
+typedef int (*dm_pool_pre_commit_fn)(void *context);
+
+void dm_pool_register_pre_commit_callback(struct dm_pool_metadata *pmd,
+					  dm_pool_pre_commit_fn fn,
+					  void *context);
+
+/*----------------------------------------------------------------*/
+
+#endif