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, 77 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-class-firmware b/Documentation/ABI/testing/sysfs-class-firmware
new file mode 100644
index 000000000..44ca1b78a
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-class-firmware
@@ -0,0 +1,77 @@
+What: 		/sys/class/firmware/.../data
+Date:		July 2022
+KernelVersion:	5.19
+Contact:	driver-core@lists.linux.dev
+Description:	The data sysfs file is used for firmware-fallback and for
+		firmware uploads. Cat a firmware image to this sysfs file
+		after you echo 1 to the loading sysfs file. When the firmware
+		image write is complete, echo 0 to the loading sysfs file. This
+		sequence will signal the completion of the firmware write and
+		signal the lower-level driver that the firmware data is
+		available.
+
+What: 		/sys/class/firmware/.../cancel
+Date:		July 2022
+KernelVersion:	5.19
+Contact:	driver-core@lists.linux.dev
+Description:	Write-only. For firmware uploads, write a "1" to this file to
+		request that the transfer of firmware data to the lower-level
+		device be canceled. This request will be rejected (EBUSY) if
+		the update cannot be canceled (e.g. a FLASH write is in
+		progress) or (ENODEV) if there is no firmware update in progress.
+
+What: 		/sys/class/firmware/.../error
+Date:		July 2022
+KernelVersion:	5.19
+Contact:	driver-core@lists.linux.dev
+Description:	Read-only. Returns a string describing a failed firmware
+		upload. This string will be in the form of <STATUS>:<ERROR>,
+		where <STATUS> will be one of the status strings described
+		for the status sysfs file and <ERROR> will be one of the
+		following: "hw-error", "timeout", "user-abort", "device-busy",
+		"invalid-file-size", "read-write-error", "flash-wearout". The
+		error sysfs file is only meaningful when the current firmware
+		upload status is "idle". If this file is read while a firmware
+		transfer is in progress, then the read will fail with EBUSY.
+
+What: 		/sys/class/firmware/.../loading
+Date:		July 2022
+KernelVersion:	5.19
+Contact:	driver-core@lists.linux.dev
+Description:	The loading sysfs file is used for both firmware-fallback and
+		for firmware uploads. Echo 1 onto the loading file to indicate
+		you are writing a firmware file to the data sysfs node. Echo
+		-1 onto this file to abort the data write or echo 0 onto this
+		file to indicate that the write is complete. For firmware
+		uploads, the zero value also triggers the transfer of the
+		firmware data to the lower-level device driver.
+
+What: 		/sys/class/firmware/.../remaining_size
+Date:		July 2022
+KernelVersion:	5.19
+Contact:	driver-core@lists.linux.dev
+Description:	Read-only. For firmware upload, this file contains the size
+		of the firmware data that remains to be transferred to the
+		lower-level device driver. The size value is initialized to
+		the full size of the firmware image that was previously
+		written to the data sysfs file. This value is periodically
+		updated during the "transferring" phase of the firmware
+		upload.
+		Format: "%u".
+
+What: 		/sys/class/firmware/.../status
+Date:		July 2022
+KernelVersion:	5.19
+Contact:	driver-core@lists.linux.dev
+Description:	Read-only. Returns a string describing the current status of
+		a firmware upload. The string will be one of the following:
+		idle, "receiving", "preparing", "transferring", "programming".
+
+What: 		/sys/class/firmware/.../timeout
+Date:		July 2022
+KernelVersion:	5.19
+Contact:	driver-core@lists.linux.dev
+Description:	This file supports the timeout mechanism for firmware
+		fallback.  This file has no affect on firmware uploads. For
+		more information on timeouts please see the documentation
+		for firmware fallback.