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, 246 insertions, 0 deletions

diff --git a/fs/hfsplus/bitmap.c b/fs/hfsplus/bitmap.c
new file mode 100644
index 000000000..1b3af8c87
--- /dev/null
+++ b/fs/hfsplus/bitmap.c
@@ -0,0 +1,246 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  linux/fs/hfsplus/bitmap.c
+ *
+ * Copyright (C) 2001
+ * Brad Boyer (flar@allandria.com)
+ * (C) 2003 Ardis Technologies <roman@ardistech.com>
+ *
+ * Handling of allocation file
+ */
+
+#include <linux/pagemap.h>
+
+#include "hfsplus_fs.h"
+#include "hfsplus_raw.h"
+
+#define PAGE_CACHE_BITS	(PAGE_SIZE * 8)
+
+int hfsplus_block_allocate(struct super_block *sb, u32 size,
+		u32 offset, u32 *max)
+{
+	struct hfsplus_sb_info *sbi = HFSPLUS_SB(sb);
+	struct page *page;
+	struct address_space *mapping;
+	__be32 *pptr, *curr, *end;
+	u32 mask, start, len, n;
+	__be32 val;
+	int i;
+
+	len = *max;
+	if (!len)
+		return size;
+
+	hfs_dbg("size %u, offset %u, len %u\n", size, offset, len);
+	mutex_lock(&sbi->alloc_mutex);
+	mapping = sbi->alloc_file->i_mapping;
+	page = read_mapping_page(mapping, offset / PAGE_CACHE_BITS, NULL);
+	if (IS_ERR(page)) {
+		start = size;
+		goto out;
+	}
+	pptr = kmap_local_page(page);
+	curr = pptr + (offset & (PAGE_CACHE_BITS - 1)) / 32;
+	i = offset % 32;
+	offset &= ~(PAGE_CACHE_BITS - 1);
+	if ((size ^ offset) / PAGE_CACHE_BITS)
+		end = pptr + PAGE_CACHE_BITS / 32;
+	else
+		end = pptr + ((size + 31) & (PAGE_CACHE_BITS - 1)) / 32;
+
+	/* scan the first partial u32 for zero bits */
+	val = *curr;
+	if (~val) {
+		n = be32_to_cpu(val);
+		mask = (1U << 31) >> i;
+		for (; i < 32; mask >>= 1, i++) {
+			if (!(n & mask))
+				goto found;
+		}
+	}
+	curr++;
+
+	/* scan complete u32s for the first zero bit */
+	while (1) {
+		while (curr < end) {
+			val = *curr;
+			if (~val) {
+				n = be32_to_cpu(val);
+				mask = 1 << 31;
+				for (i = 0; i < 32; mask >>= 1, i++) {
+					if (!(n & mask))
+						goto found;
+				}
+			}
+			curr++;
+		}
+		kunmap_local(pptr);
+		offset += PAGE_CACHE_BITS;
+		if (offset >= size)
+			break;
+		page = read_mapping_page(mapping, offset / PAGE_CACHE_BITS,
+					 NULL);
+		if (IS_ERR(page)) {
+			start = size;
+			goto out;
+		}
+		curr = pptr = kmap_local_page(page);
+		if ((size ^ offset) / PAGE_CACHE_BITS)
+			end = pptr + PAGE_CACHE_BITS / 32;
+		else
+			end = pptr + ((size + 31) & (PAGE_CACHE_BITS - 1)) / 32;
+	}
+	hfs_dbg("bitmap full\n");
+	start = size;
+	goto out;
+
+found:
+	start = offset + (curr - pptr) * 32 + i;
+	if (start >= size) {
+		hfs_dbg("bitmap full\n");
+		goto out;
+	}
+	/* do any partial u32 at the start */
+	len = min(size - start, len);
+	while (1) {
+		n |= mask;
+		if (++i >= 32)
+			break;
+		mask >>= 1;
+		if (!--len || n & mask)
+			goto done;
+	}
+	if (!--len)
+		goto done;
+	*curr++ = cpu_to_be32(n);
+	/* do full u32s */
+	while (1) {
+		while (curr < end) {
+			n = be32_to_cpu(*curr);
+			if (len < 32)
+				goto last;
+			if (n) {
+				len = 32;
+				goto last;
+			}
+			*curr++ = cpu_to_be32(0xffffffff);
+			len -= 32;
+		}
+		set_page_dirty(page);
+		kunmap_local(pptr);
+		offset += PAGE_CACHE_BITS;
+		page = read_mapping_page(mapping, offset / PAGE_CACHE_BITS,
+					 NULL);
+		if (IS_ERR(page)) {
+			start = size;
+			goto out;
+		}
+		pptr = kmap_local_page(page);
+		curr = pptr;
+		end = pptr + PAGE_CACHE_BITS / 32;
+	}
+last:
+	/* do any partial u32 at end */
+	mask = 1U << 31;
+	for (i = 0; i < len; i++) {
+		if (n & mask)
+			break;
+		n |= mask;
+		mask >>= 1;
+	}
+done:
+	*curr = cpu_to_be32(n);
+	set_page_dirty(page);
+	kunmap_local(pptr);
+	*max = offset + (curr - pptr) * 32 + i - start;
+	sbi->free_blocks -= *max;
+	hfsplus_mark_mdb_dirty(sb);
+	hfs_dbg("start %u, max %u\n", start, *max);
+out:
+	mutex_unlock(&sbi->alloc_mutex);
+	return start;
+}
+
+int hfsplus_block_free(struct super_block *sb, u32 offset, u32 count)
+{
+	struct hfsplus_sb_info *sbi = HFSPLUS_SB(sb);
+	struct page *page;
+	struct address_space *mapping;
+	__be32 *pptr, *curr, *end;
+	u32 mask, len, pnr;
+	int i;
+
+	/* is there any actual work to be done? */
+	if (!count)
+		return 0;
+
+	hfs_dbg("offset %u, count %u\n", offset, count);
+	/* are all of the bits in range? */
+	if ((offset + count) > sbi->total_blocks)
+		return -ENOENT;
+
+	mutex_lock(&sbi->alloc_mutex);
+	mapping = sbi->alloc_file->i_mapping;
+	pnr = offset / PAGE_CACHE_BITS;
+	page = read_mapping_page(mapping, pnr, NULL);
+	if (IS_ERR(page))
+		goto kaboom;
+	pptr = kmap_local_page(page);
+	curr = pptr + (offset & (PAGE_CACHE_BITS - 1)) / 32;
+	end = pptr + PAGE_CACHE_BITS / 32;
+	len = count;
+
+	/* do any partial u32 at the start */
+	i = offset % 32;
+	if (i) {
+		int j = 32 - i;
+		mask = 0xffffffffU << j;
+		if (j > count) {
+			mask |= 0xffffffffU >> (i + count);
+			*curr++ &= cpu_to_be32(mask);
+			goto out;
+		}
+		*curr++ &= cpu_to_be32(mask);
+		count -= j;
+	}
+
+	/* do full u32s */
+	while (1) {
+		while (curr < end) {
+			if (count < 32)
+				goto done;
+			*curr++ = 0;
+			count -= 32;
+		}
+		if (!count)
+			break;
+		set_page_dirty(page);
+		kunmap_local(pptr);
+		page = read_mapping_page(mapping, ++pnr, NULL);
+		if (IS_ERR(page))
+			goto kaboom;
+		pptr = kmap_local_page(page);
+		curr = pptr;
+		end = pptr + PAGE_CACHE_BITS / 32;
+	}
+done:
+	/* do any partial u32 at end */
+	if (count) {
+		mask = 0xffffffffU >> count;
+		*curr &= cpu_to_be32(mask);
+	}
+out:
+	set_page_dirty(page);
+	kunmap_local(pptr);
+	sbi->free_blocks += len;
+	hfsplus_mark_mdb_dirty(sb);
+	mutex_unlock(&sbi->alloc_mutex);
+
+	return 0;
+
+kaboom:
+	pr_crit("unable to mark blocks free: error %ld\n", PTR_ERR(page));
+	mutex_unlock(&sbi->alloc_mutex);
+
+	return -EIO;
+}