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 --- drivers/gpu/host1x/syncpt.h | 132 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 132 insertions(+) create mode 100644 drivers/gpu/host1x/syncpt.h (limited to 'drivers/gpu/host1x/syncpt.h') diff --git a/drivers/gpu/host1x/syncpt.h b/drivers/gpu/host1x/syncpt.h new file mode 100644 index 000000000..9eff42efc --- /dev/null +++ b/drivers/gpu/host1x/syncpt.h @@ -0,0 +1,132 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * Tegra host1x Syncpoints + * + * Copyright (c) 2010-2013, NVIDIA Corporation. + */ + +#ifndef __HOST1X_SYNCPT_H +#define __HOST1X_SYNCPT_H + +#include +#include +#include +#include +#include +#include + +#include "fence.h" +#include "intr.h" + +struct host1x; + +/* Reserved for replacing an expired wait with a NOP */ +#define HOST1X_SYNCPT_RESERVED 0 + +struct host1x_syncpt_base { + unsigned int id; + bool requested; +}; + +struct host1x_syncpt { + struct kref ref; + + unsigned int id; + atomic_t min_val; + atomic_t max_val; + u32 base_val; + const char *name; + bool client_managed; + struct host1x *host; + struct host1x_syncpt_base *base; + + /* interrupt data */ + struct host1x_fence_list fences; + + /* + * If a submission incrementing this syncpoint fails, lock it so that + * further submission cannot be made until application has handled the + * failure. + */ + bool locked; +}; + +/* Initialize sync point array */ +int host1x_syncpt_init(struct host1x *host); + +/* Free sync point array */ +void host1x_syncpt_deinit(struct host1x *host); + +/* Return number of sync point supported. */ +unsigned int host1x_syncpt_nb_pts(struct host1x *host); + +/* Return number of wait bases supported. */ +unsigned int host1x_syncpt_nb_bases(struct host1x *host); + +/* Return number of mlocks supported. */ +unsigned int host1x_syncpt_nb_mlocks(struct host1x *host); + +/* + * Check sync point sanity. If max is larger than min, there have too many + * sync point increments. + * + * Client managed sync point are not tracked. + * */ +static inline bool host1x_syncpt_check_max(struct host1x_syncpt *sp, u32 real) +{ + u32 max; + if (sp->client_managed) + return true; + max = host1x_syncpt_read_max(sp); + return (s32)wrapping_sub(u32, max, real) >= 0; +} + +/* Return true if sync point is client managed. */ +static inline bool host1x_syncpt_client_managed(struct host1x_syncpt *sp) +{ + return sp->client_managed; +} + +/* + * Returns true if syncpoint min == max, which means that there are no + * outstanding operations. + */ +static inline bool host1x_syncpt_idle(struct host1x_syncpt *sp) +{ + int min, max; + smp_rmb(); + min = atomic_read(&sp->min_val); + max = atomic_read(&sp->max_val); + return (min == max); +} + +/* Load current value from hardware to the shadow register. */ +u32 host1x_syncpt_load(struct host1x_syncpt *sp); + +/* Check if the given syncpoint value has already passed */ +bool host1x_syncpt_is_expired(struct host1x_syncpt *sp, u32 thresh); + +/* Save host1x sync point state into shadow registers. */ +void host1x_syncpt_save(struct host1x *host); + +/* Reset host1x sync point state from shadow registers. */ +void host1x_syncpt_restore(struct host1x *host); + +/* Read current wait base value into shadow register and return it. */ +u32 host1x_syncpt_load_wait_base(struct host1x_syncpt *sp); + +/* Indicate future operations by incrementing the sync point max. */ +u32 host1x_syncpt_incr_max(struct host1x_syncpt *sp, u32 incrs); + +/* Check if sync point id is valid. */ +static inline int host1x_syncpt_is_valid(struct host1x_syncpt *sp) +{ + return sp->id < host1x_syncpt_nb_pts(sp->host); +} + +static inline void host1x_syncpt_set_locked(struct host1x_syncpt *sp) +{ + sp->locked = true; +} + +#endif -- cgit v1.3.1