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/scsi/arm/queue.c | 319 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 319 insertions(+) create mode 100644 drivers/scsi/arm/queue.c (limited to 'drivers/scsi/arm/queue.c') diff --git a/drivers/scsi/arm/queue.c b/drivers/scsi/arm/queue.c new file mode 100644 index 000000000..797cf4971 --- /dev/null +++ b/drivers/scsi/arm/queue.c @@ -0,0 +1,319 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * linux/drivers/acorn/scsi/queue.c: queue handling primitives + * + * Copyright (C) 1997-2000 Russell King + * + * Changelog: + * 15-Sep-1997 RMK Created. + * 11-Oct-1997 RMK Corrected problem with queue_remove_exclude + * not updating internal linked list properly + * (was causing commands to go missing). + * 30-Aug-2000 RMK Use Linux list handling and spinlocks + */ +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include + +#define DEBUG + +typedef struct queue_entry { + struct list_head list; + struct scsi_cmnd *SCpnt; +#ifdef DEBUG + unsigned long magic; +#endif +} QE_t; + +#ifdef DEBUG +#define QUEUE_MAGIC_FREE 0xf7e1c9a3 +#define QUEUE_MAGIC_USED 0xf7e1cc33 + +#define SET_MAGIC(q,m) ((q)->magic = (m)) +#define BAD_MAGIC(q,m) ((q)->magic != (m)) +#else +#define SET_MAGIC(q,m) do { } while (0) +#define BAD_MAGIC(q,m) (0) +#endif + +#include "queue.h" + +#define NR_QE 32 + +/* + * Function: void queue_initialise (Queue_t *queue) + * Purpose : initialise a queue + * Params : queue - queue to initialise + */ +int queue_initialise (Queue_t *queue) +{ + unsigned int nqueues = NR_QE; + QE_t *q; + + spin_lock_init(&queue->queue_lock); + INIT_LIST_HEAD(&queue->head); + INIT_LIST_HEAD(&queue->free); + + /* + * If life was easier, then SCpnt would have a + * host-available list head, and we wouldn't + * need to keep free lists or allocate this + * memory. + */ + queue->alloc = q = kmalloc_objs(QE_t, nqueues); + if (q) { + for (; nqueues; q++, nqueues--) { + SET_MAGIC(q, QUEUE_MAGIC_FREE); + q->SCpnt = NULL; + list_add(&q->list, &queue->free); + } + } + + return queue->alloc != NULL; +} + +/* + * Function: void queue_free (Queue_t *queue) + * Purpose : free a queue + * Params : queue - queue to free + */ +void queue_free (Queue_t *queue) +{ + if (!list_empty(&queue->head)) + printk(KERN_WARNING "freeing non-empty queue %p\n", queue); + kfree(queue->alloc); +} + + +/* + * Function: int __queue_add(Queue_t *queue, struct scsi_cmnd *SCpnt, int head) + * Purpose : Add a new command onto a queue, adding REQUEST_SENSE to head. + * Params : queue - destination queue + * SCpnt - command to add + * head - add command to head of queue + * Returns : 0 on error, !0 on success + */ +int __queue_add(Queue_t *queue, struct scsi_cmnd *SCpnt, int head) +{ + unsigned long flags; + struct list_head *l; + QE_t *q; + int ret = 0; + + spin_lock_irqsave(&queue->queue_lock, flags); + if (list_empty(&queue->free)) + goto empty; + + l = queue->free.next; + list_del(l); + + q = list_entry(l, QE_t, list); + BUG_ON(BAD_MAGIC(q, QUEUE_MAGIC_FREE)); + + SET_MAGIC(q, QUEUE_MAGIC_USED); + q->SCpnt = SCpnt; + + if (head) + list_add(l, &queue->head); + else + list_add_tail(l, &queue->head); + + ret = 1; +empty: + spin_unlock_irqrestore(&queue->queue_lock, flags); + return ret; +} + +static struct scsi_cmnd *__queue_remove(Queue_t *queue, struct list_head *ent) +{ + QE_t *q; + + /* + * Move the entry from the "used" list onto the "free" list + */ + list_del(ent); + q = list_entry(ent, QE_t, list); + BUG_ON(BAD_MAGIC(q, QUEUE_MAGIC_USED)); + + SET_MAGIC(q, QUEUE_MAGIC_FREE); + list_add(ent, &queue->free); + + return q->SCpnt; +} + +/* + * Function: struct scsi_cmnd *queue_remove_exclude (queue, exclude) + * Purpose : remove a SCSI command from a queue + * Params : queue - queue to remove command from + * exclude - bit array of target&lun which is busy + * Returns : struct scsi_cmnd if successful (and a reference), or NULL if no command available + */ +struct scsi_cmnd *queue_remove_exclude(Queue_t *queue, unsigned long *exclude) +{ + unsigned long flags; + struct list_head *l; + struct scsi_cmnd *SCpnt = NULL; + + spin_lock_irqsave(&queue->queue_lock, flags); + list_for_each(l, &queue->head) { + QE_t *q = list_entry(l, QE_t, list); + if (!test_bit(q->SCpnt->device->id * 8 + + (u8)(q->SCpnt->device->lun & 0x7), exclude)) { + SCpnt = __queue_remove(queue, l); + break; + } + } + spin_unlock_irqrestore(&queue->queue_lock, flags); + + return SCpnt; +} + +/* + * Function: struct scsi_cmnd *queue_remove (queue) + * Purpose : removes first SCSI command from a queue + * Params : queue - queue to remove command from + * Returns : struct scsi_cmnd if successful (and a reference), or NULL if no command available + */ +struct scsi_cmnd *queue_remove(Queue_t *queue) +{ + unsigned long flags; + struct scsi_cmnd *SCpnt = NULL; + + spin_lock_irqsave(&queue->queue_lock, flags); + if (!list_empty(&queue->head)) + SCpnt = __queue_remove(queue, queue->head.next); + spin_unlock_irqrestore(&queue->queue_lock, flags); + + return SCpnt; +} + +/* + * Function: struct scsi_cmnd *queue_remove_tgtluntag (queue, target, lun, tag) + * Purpose : remove a SCSI command from the queue for a specified target/lun/tag + * Params : queue - queue to remove command from + * target - target that we want + * lun - lun on device + * tag - tag on device + * Returns : struct scsi_cmnd if successful, or NULL if no command satisfies requirements + */ +struct scsi_cmnd *queue_remove_tgtluntag(Queue_t *queue, int target, int lun, + int tag) +{ + unsigned long flags; + struct list_head *l; + struct scsi_cmnd *SCpnt = NULL; + + spin_lock_irqsave(&queue->queue_lock, flags); + list_for_each(l, &queue->head) { + QE_t *q = list_entry(l, QE_t, list); + if (q->SCpnt->device->id == target && q->SCpnt->device->lun == lun && + scsi_cmd_to_rq(q->SCpnt)->tag == tag) { + SCpnt = __queue_remove(queue, l); + break; + } + } + spin_unlock_irqrestore(&queue->queue_lock, flags); + + return SCpnt; +} + +/* + * Function: queue_remove_all_target(queue, target) + * Purpose : remove all SCSI commands from the queue for a specified target + * Params : queue - queue to remove command from + * target - target device id + * Returns : nothing + */ +void queue_remove_all_target(Queue_t *queue, int target) +{ + unsigned long flags; + struct list_head *l; + + spin_lock_irqsave(&queue->queue_lock, flags); + list_for_each(l, &queue->head) { + QE_t *q = list_entry(l, QE_t, list); + if (q->SCpnt->device->id == target) + __queue_remove(queue, l); + } + spin_unlock_irqrestore(&queue->queue_lock, flags); +} + +/* + * Function: int queue_probetgtlun (queue, target, lun) + * Purpose : check to see if we have a command in the queue for the specified + * target/lun. + * Params : queue - queue to look in + * target - target we want to probe + * lun - lun on target + * Returns : 0 if not found, != 0 if found + */ +int queue_probetgtlun (Queue_t *queue, int target, int lun) +{ + unsigned long flags; + struct list_head *l; + int found = 0; + + spin_lock_irqsave(&queue->queue_lock, flags); + list_for_each(l, &queue->head) { + QE_t *q = list_entry(l, QE_t, list); + if (q->SCpnt->device->id == target && q->SCpnt->device->lun == lun) { + found = 1; + break; + } + } + spin_unlock_irqrestore(&queue->queue_lock, flags); + + return found; +} + +/* + * Function: int queue_remove_cmd(Queue_t *queue, struct scsi_cmnd *SCpnt) + * Purpose : remove a specific command from the queues + * Params : queue - queue to look in + * SCpnt - command to find + * Returns : 0 if not found + */ +int queue_remove_cmd(Queue_t *queue, struct scsi_cmnd *SCpnt) +{ + unsigned long flags; + struct list_head *l; + int found = 0; + + spin_lock_irqsave(&queue->queue_lock, flags); + list_for_each(l, &queue->head) { + QE_t *q = list_entry(l, QE_t, list); + if (q->SCpnt == SCpnt) { + __queue_remove(queue, l); + found = 1; + break; + } + } + spin_unlock_irqrestore(&queue->queue_lock, flags); + + return found; +} + +EXPORT_SYMBOL(queue_initialise); +EXPORT_SYMBOL(queue_free); +EXPORT_SYMBOL(__queue_add); +EXPORT_SYMBOL(queue_remove); +EXPORT_SYMBOL(queue_remove_exclude); +EXPORT_SYMBOL(queue_remove_tgtluntag); +EXPORT_SYMBOL(queue_remove_cmd); +EXPORT_SYMBOL(queue_remove_all_target); +EXPORT_SYMBOL(queue_probetgtlun); + +MODULE_AUTHOR("Russell King"); +MODULE_DESCRIPTION("SCSI command queueing"); +MODULE_LICENSE("GPL"); -- cgit v1.3.1