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, 104 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-class-remoteproc b/Documentation/ABI/testing/sysfs-class-remoteproc
new file mode 100644
index 000000000..b2b8e2db2
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-class-remoteproc
@@ -0,0 +1,104 @@
+What:		/sys/class/remoteproc/.../firmware
+Date:		October 2016
+Contact:	Matt Redfearn <matt.redfearn@mips.com>
+Description:	Remote processor firmware
+
+		Reports the name of the firmware currently loaded to the
+		remote processor.
+
+		To change the running firmware, ensure the remote processor is
+		stopped (using /sys/class/remoteproc/.../state) and write a new filename.
+
+What:		/sys/class/remoteproc/.../state
+Date:		October 2016
+Contact:	Matt Redfearn <matt.redfearn@mips.com>
+Description:	Remote processor state
+
+		Reports the state of the remote processor, which will be one of:
+
+		- "offline"
+		- "suspended"
+		- "running"
+		- "crashed"
+		- "invalid"
+
+		"offline" means the remote processor is powered off.
+
+		"suspended" means that the remote processor is suspended and
+		must be woken to receive messages.
+
+		"running" is the normal state of an available remote processor
+
+		"crashed" indicates that a problem/crash has been detected on
+		the remote processor.
+
+		"invalid" is returned if the remote processor is in an
+		unknown state.
+
+		Writing this file controls the state of the remote processor.
+		The following states can be written:
+
+		- "start"
+		- "stop"
+
+		Writing "start" will attempt to start the processor running the
+		firmware indicated by, or written to,
+		/sys/class/remoteproc/.../firmware. The remote processor should
+		transition to "running" state.
+
+		Writing "stop" will attempt to halt the remote processor and
+		return it to the "offline" state.
+
+What:		/sys/class/remoteproc/.../name
+Date:		August 2019
+KernelVersion:	5.4
+Contact:	Suman Anna <s-anna@ti.com>
+Description:	Remote processor name
+
+		Reports the name of the remote processor. This can be used by
+		userspace in exactly identifying a remote processor and ease
+		up the usage in modifying the 'firmware' or 'state' files.
+
+What:		/sys/class/remoteproc/.../coredump
+Date:		July 2020
+Contact:	Bjorn Andersson <bjorn.andersson@linaro.org>, Ohad Ben-Cohen <ohad@wizery.com>
+Description:	Remote processor coredump configuration
+
+		Reports the coredump configuration of the remote processor,
+		which will be one of:
+
+		"disabled"
+		"enabled"
+		"inline"
+
+		"disabled" means no dump will be collected.
+
+		"enabled" means when the remote processor's coredump is
+		collected it will be copied to a separate buffer and that
+		buffer is exposed to userspace.
+
+		"inline" means when the remote processor's coredump is
+		collected userspace will directly read from the remote
+		processor's device memory. Extra buffer will not be used to
+		copy the dump. Also recovery process will not proceed until
+		all data is read by userspace.
+
+What:		/sys/class/remoteproc/.../recovery
+Date:		July 2020
+Contact:	Bjorn Andersson <bjorn.andersson@linaro.org>, Ohad Ben-Cohen <ohad@wizery.com>
+Description:	Remote processor recovery mechanism
+
+		Reports the recovery mechanism of the remote processor,
+		which will be one of:
+
+		"enabled"
+		"disabled"
+
+		"enabled" means, the remote processor will be automatically
+		recovered whenever it crashes. Moreover, if the remote
+		processor crashes while recovery is disabled, it will
+		be automatically recovered too as soon as recovery is enabled.
+
+		"disabled" means, a remote processor will remain in a crashed
+		state if it crashes. This is useful for debugging purposes;
+		without it, debugging a crash is substantially harder.