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/arch/parisc/debugging.rst | 46 +++++++++++++++++++++++++++++++++ 1 file changed, 46 insertions(+) create mode 100644 Documentation/arch/parisc/debugging.rst (limited to 'Documentation/arch/parisc/debugging.rst') diff --git a/Documentation/arch/parisc/debugging.rst b/Documentation/arch/parisc/debugging.rst new file mode 100644 index 000000000..de1b60402 --- /dev/null +++ b/Documentation/arch/parisc/debugging.rst @@ -0,0 +1,46 @@ +================= +PA-RISC Debugging +================= + +okay, here are some hints for debugging the lower-level parts of +linux/parisc. + + +1. Absolute addresses +===================== + +A lot of the assembly code currently runs in real mode, which means +absolute addresses are used instead of virtual addresses as in the +rest of the kernel. To translate an absolute address to a virtual +address you can lookup in System.map, add __PAGE_OFFSET (0x10000000 +currently). + + +2. HPMCs +======== + +When real-mode code tries to access non-existent memory, you'll get +an HPMC instead of a kernel oops. To debug an HPMC, try to find +the System Responder/Requestor addresses. The System Requestor +address should match (one of the) processor HPAs (high addresses in +the I/O range); the System Responder address is the address real-mode +code tried to access. + +Typical values for the System Responder address are addresses larger +than __PAGE_OFFSET (0x10000000) which mean a virtual address didn't +get translated to a physical address before real-mode code tried to +access it. + + +3. Q bit fun +============ + +Certain, very critical code has to clear the Q bit in the PSW. What +happens when the Q bit is cleared is the CPU does not update the +registers interruption handlers read to find out where the machine +was interrupted - so if you get an interruption between the instruction +that clears the Q bit and the RFI that sets it again you don't know +where exactly it happened. If you're lucky the IAOQ will point to the +instruction that cleared the Q bit, if you're not it points anywhere +at all. Usually Q bit problems will show themselves in unexplainable +system hangs or running off the end of physical memory. -- cgit v1.3.1