From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions 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) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- Documentation/ABI/testing/sysfs-class-mei | 105 ++++++++++++++++++++++++++++++ 1 file changed, 105 insertions(+) create mode 100644 Documentation/ABI/testing/sysfs-class-mei (limited to 'Documentation/ABI/testing/sysfs-class-mei') diff --git a/Documentation/ABI/testing/sysfs-class-mei b/Documentation/ABI/testing/sysfs-class-mei new file mode 100644 index 000000000..1db36ddf8 --- /dev/null +++ b/Documentation/ABI/testing/sysfs-class-mei @@ -0,0 +1,105 @@ +What: /sys/class/mei/ +Date: May 2014 +KernelVersion: 3.17 +Contact: Tomas Winkler +Description: + The mei/ class sub-directory belongs to mei device class + + +What: /sys/class/mei/mei/ +Date: May 2014 +KernelVersion: 3.17 +Contact: Tomas Winkler +Description: + The /sys/class/mei/meiN directory is created for + each probed mei device + +What: /sys/class/mei/mei/fw_status +Date: Nov 2014 +KernelVersion: 3.19 +Contact: Tomas Winkler +Description: Display fw status registers content + + The ME FW writes its status information into fw status + registers for BIOS and OS to monitor fw health. + + The register contains running state, power management + state, error codes, and others. The way the registers + are decoded depends on PCH or SoC generation. + Also number of registers varies between 1 and 6 + depending on generation. + +What: /sys/class/mei/mei/hbm_ver +Date: Aug 2016 +KernelVersion: 4.9 +Contact: Tomas Winkler +Description: Display the negotiated HBM protocol version. + + The HBM protocol version negotiated + between the driver and the device. + +What: /sys/class/mei/mei/hbm_ver_drv +Date: Aug 2016 +KernelVersion: 4.9 +Contact: Tomas Winkler +Description: Display the driver HBM protocol version. + + The HBM protocol version supported by the driver. + +What: /sys/class/mei/mei/tx_queue_limit +Date: Jan 2018 +KernelVersion: 4.16 +Contact: Tomas Winkler +Description: Configure tx queue limit + + Set maximal number of pending writes + per opened session. + +What: /sys/class/mei/mei/fw_ver +Date: May 2018 +KernelVersion: 4.18 +Contact: Tomas Winkler +Description: Display the ME firmware version. + + The version of the platform ME firmware is in format: + :.... + There can be up to three such blocks for different + FW components. + +What: /sys/class/mei/mei/dev_state +Date: Mar 2019 +KernelVersion: 5.1 +Contact: Tomas Winkler +Description: Display the ME device state. + + The device state can have following values: + INITIALIZING + INIT_CLIENTS + ENABLED + RESETTING + DISABLED + POWER_DOWN + POWER_UP + +What: /sys/class/mei/mei/trc +Date: Nov 2019 +KernelVersion: 5.5 +Contact: Tomas Winkler +Description: Display trc status register content + + The ME FW writes Glitch Detection HW (TRC) + status information into trc status register + for BIOS and OS to monitor fw health. + +What: /sys/class/mei/mei/kind +Date: Jul 2020 +KernelVersion: 5.8 +Contact: Tomas Winkler +Description: Display kind of the device + + Generic devices are marked as "mei" + while special purpose have their own + names. + Available options: + - mei: generic mei device. + - itouch: itouch (ipts) mei device. -- cgit v1.3.1