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, 232 insertions, 0 deletions

diff --git a/fs/hpfs/buffer.c b/fs/hpfs/buffer.c
new file mode 100644
index 000000000..d39246865
--- /dev/null
+++ b/fs/hpfs/buffer.c
@@ -0,0 +1,232 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  linux/fs/hpfs/buffer.c
+ *
+ *  Mikulas Patocka (mikulas@artax.karlin.mff.cuni.cz), 1998-1999
+ *
+ *  general buffer i/o
+ */
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/blkdev.h>
+#include "hpfs_fn.h"
+
+secno hpfs_search_hotfix_map(struct super_block *s, secno sec)
+{
+	unsigned i;
+	struct hpfs_sb_info *sbi = hpfs_sb(s);
+	for (i = 0; unlikely(i < sbi->n_hotfixes); i++) {
+		if (sbi->hotfix_from[i] == sec) {
+			return sbi->hotfix_to[i];
+		}
+	}
+	return sec;
+}
+
+unsigned hpfs_search_hotfix_map_for_range(struct super_block *s, secno sec, unsigned n)
+{
+	unsigned i;
+	struct hpfs_sb_info *sbi = hpfs_sb(s);
+	for (i = 0; unlikely(i < sbi->n_hotfixes); i++) {
+		if (sbi->hotfix_from[i] >= sec && sbi->hotfix_from[i] < sec + n) {
+			n = sbi->hotfix_from[i] - sec;
+		}
+	}
+	return n;
+}
+
+void hpfs_prefetch_sectors(struct super_block *s, unsigned secno, int n)
+{
+	struct buffer_head *bh;
+	struct blk_plug plug;
+
+	if (n <= 0 || unlikely(secno >= hpfs_sb(s)->sb_fs_size))
+		return;
+
+	if (unlikely(hpfs_search_hotfix_map_for_range(s, secno, n) != n))
+		return;
+
+	bh = sb_find_get_block(s, secno);
+	if (bh) {
+		if (buffer_uptodate(bh)) {
+			brelse(bh);
+			return;
+		}
+		brelse(bh);
+	}
+
+	blk_start_plug(&plug);
+	while (n > 0) {
+		if (unlikely(secno >= hpfs_sb(s)->sb_fs_size))
+			break;
+		sb_breadahead(s, secno);
+		secno++;
+		n--;
+	}
+	blk_finish_plug(&plug);
+}
+
+/* Map a sector into a buffer and return pointers to it and to the buffer. */
+
+void *hpfs_map_sector(struct super_block *s, unsigned secno, struct buffer_head **bhp,
+		 int ahead)
+{
+	struct buffer_head *bh;
+
+	hpfs_lock_assert(s);
+
+	hpfs_prefetch_sectors(s, secno, ahead);
+
+	cond_resched();
+
+	*bhp = bh = sb_bread(s, hpfs_search_hotfix_map(s, secno));
+	if (bh != NULL)
+		return bh->b_data;
+	else {
+		pr_err("%s(): read error\n", __func__);
+		return NULL;
+	}
+}
+
+/* Like hpfs_map_sector but don't read anything */
+
+void *hpfs_get_sector(struct super_block *s, unsigned secno, struct buffer_head **bhp)
+{
+	struct buffer_head *bh;
+	/*return hpfs_map_sector(s, secno, bhp, 0);*/
+
+	hpfs_lock_assert(s);
+
+	cond_resched();
+
+	if ((*bhp = bh = sb_getblk(s, hpfs_search_hotfix_map(s, secno))) != NULL) {
+		if (!buffer_uptodate(bh)) wait_on_buffer(bh);
+		set_buffer_uptodate(bh);
+		return bh->b_data;
+	} else {
+		pr_err("%s(): getblk failed\n", __func__);
+		return NULL;
+	}
+}
+
+/* Map 4 sectors into a 4buffer and return pointers to it and to the buffer. */
+
+void *hpfs_map_4sectors(struct super_block *s, unsigned secno, struct quad_buffer_head *qbh,
+		   int ahead)
+{
+	char *data;
+
+	hpfs_lock_assert(s);
+
+	cond_resched();
+
+	if (secno & 3) {
+		pr_err("%s(): unaligned read\n", __func__);
+		return NULL;
+	}
+
+	hpfs_prefetch_sectors(s, secno, 4 + ahead);
+
+	if (!hpfs_map_sector(s, secno + 0, &qbh->bh[0], 0)) goto bail0;
+	if (!hpfs_map_sector(s, secno + 1, &qbh->bh[1], 0)) goto bail1;
+	if (!hpfs_map_sector(s, secno + 2, &qbh->bh[2], 0)) goto bail2;
+	if (!hpfs_map_sector(s, secno + 3, &qbh->bh[3], 0)) goto bail3;
+
+	if (likely(qbh->bh[1]->b_data == qbh->bh[0]->b_data + 1 * 512) &&
+	    likely(qbh->bh[2]->b_data == qbh->bh[0]->b_data + 2 * 512) &&
+	    likely(qbh->bh[3]->b_data == qbh->bh[0]->b_data + 3 * 512)) {
+		return qbh->data = qbh->bh[0]->b_data;
+	}
+
+	qbh->data = data = kmalloc(2048, GFP_NOFS);
+	if (!data) {
+		pr_err("%s(): out of memory\n", __func__);
+		goto bail4;
+	}
+
+	memcpy(data + 0 * 512, qbh->bh[0]->b_data, 512);
+	memcpy(data + 1 * 512, qbh->bh[1]->b_data, 512);
+	memcpy(data + 2 * 512, qbh->bh[2]->b_data, 512);
+	memcpy(data + 3 * 512, qbh->bh[3]->b_data, 512);
+
+	return data;
+
+ bail4:
+	brelse(qbh->bh[3]);
+ bail3:
+	brelse(qbh->bh[2]);
+ bail2:
+	brelse(qbh->bh[1]);
+ bail1:
+	brelse(qbh->bh[0]);
+ bail0:
+	return NULL;
+}
+
+/* Don't read sectors */
+
+void *hpfs_get_4sectors(struct super_block *s, unsigned secno,
+                          struct quad_buffer_head *qbh)
+{
+	cond_resched();
+
+	hpfs_lock_assert(s);
+
+	if (secno & 3) {
+		pr_err("%s(): unaligned read\n", __func__);
+		return NULL;
+	}
+
+	if (!hpfs_get_sector(s, secno + 0, &qbh->bh[0])) goto bail0;
+	if (!hpfs_get_sector(s, secno + 1, &qbh->bh[1])) goto bail1;
+	if (!hpfs_get_sector(s, secno + 2, &qbh->bh[2])) goto bail2;
+	if (!hpfs_get_sector(s, secno + 3, &qbh->bh[3])) goto bail3;
+
+	if (likely(qbh->bh[1]->b_data == qbh->bh[0]->b_data + 1 * 512) &&
+	    likely(qbh->bh[2]->b_data == qbh->bh[0]->b_data + 2 * 512) &&
+	    likely(qbh->bh[3]->b_data == qbh->bh[0]->b_data + 3 * 512)) {
+		return qbh->data = qbh->bh[0]->b_data;
+	}
+
+	if (!(qbh->data = kmalloc(2048, GFP_NOFS))) {
+		pr_err("%s(): out of memory\n", __func__);
+		goto bail4;
+	}
+	return qbh->data;
+
+bail4:
+	brelse(qbh->bh[3]);
+bail3:
+	brelse(qbh->bh[2]);
+bail2:
+	brelse(qbh->bh[1]);
+bail1:
+	brelse(qbh->bh[0]);
+bail0:
+	return NULL;
+}
+	
+
+void hpfs_brelse4(struct quad_buffer_head *qbh)
+{
+	if (unlikely(qbh->data != qbh->bh[0]->b_data))
+		kfree(qbh->data);
+	brelse(qbh->bh[0]);
+	brelse(qbh->bh[1]);
+	brelse(qbh->bh[2]);
+	brelse(qbh->bh[3]);
+}	
+
+void hpfs_mark_4buffers_dirty(struct quad_buffer_head *qbh)
+{
+	if (unlikely(qbh->data != qbh->bh[0]->b_data)) {
+		memcpy(qbh->bh[0]->b_data, qbh->data + 0 * 512, 512);
+		memcpy(qbh->bh[1]->b_data, qbh->data + 1 * 512, 512);
+		memcpy(qbh->bh[2]->b_data, qbh->data + 2 * 512, 512);
+		memcpy(qbh->bh[3]->b_data, qbh->data + 3 * 512, 512);
+	}
+	mark_buffer_dirty(qbh->bh[0]);
+	mark_buffer_dirty(qbh->bh[1]);
+	mark_buffer_dirty(qbh->bh[2]);
+	mark_buffer_dirty(qbh->bh[3]);
+}