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, 534 insertions, 0 deletions

diff --git a/fs/erofs/data.c b/fs/erofs/data.c
new file mode 100644
index 000000000..be63b89f0
--- /dev/null
+++ b/fs/erofs/data.c
@@ -0,0 +1,534 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2017-2018 HUAWEI, Inc.
+ *             https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
+ */
+#include "internal.h"
+#include <linux/filelock.h>
+#include <linux/sched/mm.h>
+#include <trace/events/erofs.h>
+
+void erofs_unmap_metabuf(struct erofs_buf *buf)
+{
+	if (!buf->base)
+		return;
+	kunmap_local(buf->base);
+	buf->base = NULL;
+}
+
+void erofs_put_metabuf(struct erofs_buf *buf)
+{
+	if (!buf->page)
+		return;
+	erofs_unmap_metabuf(buf);
+	folio_put(page_folio(buf->page));
+	buf->page = NULL;
+}
+
+void *erofs_bread(struct erofs_buf *buf, erofs_off_t offset, bool need_kmap)
+{
+	pgoff_t index = (buf->off + offset) >> PAGE_SHIFT;
+	struct folio *folio = NULL;
+
+	if (buf->page) {
+		folio = page_folio(buf->page);
+		if (folio_file_page(folio, index) != buf->page)
+			erofs_unmap_metabuf(buf);
+	}
+	if (!folio || !folio_contains(folio, index)) {
+		erofs_put_metabuf(buf);
+		folio = read_cache_folio(buf->mapping, index,
+				buf->mc ? erofs_read_meta_folio : NULL, NULL);
+		if (IS_ERR(folio))
+			return folio;
+	}
+	buf->page = folio_file_page(folio, index);
+	if (!need_kmap)
+		return NULL;
+	if (!buf->base)
+		buf->base = kmap_local_page(buf->page);
+	return buf->base + ((buf->off + offset) & ~PAGE_MASK);
+}
+
+int erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb,
+		       bool in_metabox)
+{
+	struct erofs_sb_info *sbi = EROFS_SB(sb);
+
+	buf->mc = false;
+	if (in_metabox) {
+		if (unlikely(!sbi->metabox_inode))
+			return -EFSCORRUPTED;
+		buf->mapping = sbi->metabox_inode->i_mapping;
+		return 0;
+	}
+	if (erofs_is_fileio_mode(sbi)) {
+		buf->mapping = sbi->managed_cache->i_mapping;
+		buf->mc = true;
+	} else {
+		buf->off = sbi->dif0.fsoff;
+		buf->mapping = sb->s_bdev->bd_mapping;
+	}
+	return 0;
+}
+
+void *erofs_read_metabuf(struct erofs_buf *buf, struct super_block *sb,
+			 erofs_off_t offset, bool in_metabox)
+{
+	int err;
+
+	err = erofs_init_metabuf(buf, sb, in_metabox);
+	if (err)
+		return ERR_PTR(err);
+	return erofs_bread(buf, offset, true);
+}
+
+static int erofs_map_chunks(struct inode *inode, struct erofs_map_blocks *map)
+{
+	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+	struct super_block *sb = inode->i_sb;
+	struct erofs_inode *vi = EROFS_I(inode);
+	struct erofs_inode_chunk_index *idx;
+	unsigned int unit = vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES ?
+		sizeof(*idx) : EROFS_BLOCK_MAP_ENTRY_SIZE;
+	erofs_blk_t addrmask = (vi->chunkformat & EROFS_CHUNK_FORMAT_48BIT) ?
+		BIT_ULL(48) - 1 : BIT_ULL(32) - 1;
+	u64 nr = map->m_la >> vi->chunkbits, chunksize = 1ULL << vi->chunkbits;
+	erofs_off_t pos = ALIGN(erofs_iloc(inode) + vi->inode_isize +
+				vi->xattr_isize, unit) + unit * nr;
+	/* m_llen will be clamped to EOF in the end */
+	erofs_off_t endpos = round_up(pos + 1, sb->s_blocksize);
+	u64 last, addr;
+
+	idx = erofs_read_metabuf(&buf, sb, pos, erofs_inode_in_metabox(inode));
+	if (IS_ERR(idx))
+		return PTR_ERR(idx);
+
+	map->m_la = nr << vi->chunkbits;
+	map->m_llen = 0;
+	nr = 0;
+	do {
+		if (unit == EROFS_BLOCK_MAP_ENTRY_SIZE) {
+			addr = le32_to_cpu(((__le32 *)idx)[nr]);
+			if (addr == (u32)EROFS_NULL_ADDR)
+				addr = EROFS_NULL_ADDR;
+		} else {
+			addr = (((u64)le16_to_cpu(idx[nr].startblk_hi) << 32) |
+				le32_to_cpu(idx[nr].startblk_lo)) & addrmask;
+			if (addr ^ (EROFS_NULL_ADDR & addrmask))
+				addr |= (u64)(le16_to_cpu(idx[nr].device_id) &
+					EROFS_SB(sb)->device_id_mask) << 48;
+			else
+				addr = EROFS_NULL_ADDR;
+		}
+		if (!nr) {
+			last = addr;
+			continue;
+		}
+		/* expand and account the prior chunk here */
+		map->m_llen += chunksize;
+		if (last != EROFS_NULL_ADDR)
+			last += erofs_blknr(sb, chunksize);
+	} while (addr == last && pos + (++nr) * unit < endpos);
+
+	if (last != EROFS_NULL_ADDR) {
+		map->m_pa = erofs_pos(sb, last & addrmask) - map->m_llen;
+		map->m_deviceid = last >> 48;
+		map->m_flags = EROFS_MAP_MAPPED;
+	}
+	if (addr == last)
+		map->m_llen += chunksize;
+	map->m_llen = min_t(erofs_off_t, map->m_llen,
+			round_up(inode->i_size - map->m_la, sb->s_blocksize));
+	erofs_put_metabuf(&buf);
+	return 0;
+}
+
+int erofs_map_blocks(struct inode *inode, struct erofs_map_blocks *map)
+{
+	struct super_block *sb = inode->i_sb;
+	struct erofs_inode *vi = EROFS_I(inode);
+	bool tailinline = (vi->datalayout == EROFS_INODE_FLAT_INLINE);
+	erofs_off_t pos;
+	int err = 0;
+
+	trace_erofs_map_blocks_enter(inode, map, 0);
+	map->m_deviceid = 0;
+	map->m_flags = 0;
+	if (map->m_la >= inode->i_size)
+		goto out;
+	if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
+		err = erofs_map_chunks(inode, map);
+	} else if (tailinline || vi->startblk != EROFS_NULL_ADDR) {
+		pos = erofs_pos(sb, erofs_iblks(inode) - tailinline);
+		map->m_flags = EROFS_MAP_MAPPED;
+		if (map->m_la < pos) {
+			map->m_pa = erofs_pos(sb, vi->startblk) + map->m_la;
+			map->m_llen = pos - map->m_la;
+		} else {
+			map->m_pa = erofs_iloc(inode) + vi->inode_isize +
+				vi->xattr_isize + erofs_blkoff(sb, map->m_la);
+			map->m_llen = inode->i_size - map->m_la;
+			map->m_flags |= EROFS_MAP_META;
+			if (erofs_blkoff(sb, map->m_pa) + map->m_llen >
+			    sb->s_blocksize) {
+				erofs_err(sb, "inline data across blocks @ nid %llu", vi->nid);
+				return -EFSCORRUPTED;
+			}
+		}
+	}
+out:
+	map->m_plen = err ? 0 : map->m_llen;
+	trace_erofs_map_blocks_exit(inode, map, 0, err);
+	return err;
+}
+
+static void erofs_fill_from_devinfo(struct erofs_map_dev *map,
+		struct super_block *sb, struct erofs_device_info *dif)
+{
+	map->m_sb = sb;
+	map->m_dif = dif;
+	map->m_bdev = NULL;
+	if (dif->file && S_ISBLK(file_inode(dif->file)->i_mode))
+		map->m_bdev = file_bdev(dif->file);
+}
+
+int erofs_map_dev(struct super_block *sb, struct erofs_map_dev *map)
+{
+	struct erofs_dev_context *devs = EROFS_SB(sb)->devs;
+	struct erofs_device_info *dif;
+	erofs_off_t startoff;
+	int id;
+
+	erofs_fill_from_devinfo(map, sb, &EROFS_SB(sb)->dif0);
+	map->m_bdev = sb->s_bdev;	/* use s_bdev for the primary device */
+	if (map->m_deviceid) {
+		down_read(&devs->rwsem);
+		dif = idr_find(&devs->tree, map->m_deviceid - 1);
+		if (!dif) {
+			up_read(&devs->rwsem);
+			return -ENODEV;
+		}
+		if (devs->flatdev) {
+			map->m_pa += erofs_pos(sb, dif->uniaddr);
+			up_read(&devs->rwsem);
+			return 0;
+		}
+		erofs_fill_from_devinfo(map, sb, dif);
+		up_read(&devs->rwsem);
+	} else if (devs->extra_devices && !devs->flatdev) {
+		down_read(&devs->rwsem);
+		idr_for_each_entry(&devs->tree, dif, id) {
+			if (!dif->uniaddr)
+				continue;
+
+			startoff = erofs_pos(sb, dif->uniaddr);
+			if (map->m_pa >= startoff &&
+			    map->m_pa < startoff + erofs_pos(sb, dif->blocks)) {
+				map->m_pa -= startoff;
+				erofs_fill_from_devinfo(map, sb, dif);
+				break;
+			}
+		}
+		up_read(&devs->rwsem);
+	}
+	return 0;
+}
+
+/*
+ * bit 30: I/O error occurred on this folio
+ * bit 29: CPU has dirty data in D-cache (needs aliasing handling);
+ * bit 0 - 29: remaining parts to complete this folio
+ */
+#define EROFS_ONLINEFOLIO_EIO		30
+#define EROFS_ONLINEFOLIO_DIRTY		29
+
+void erofs_onlinefolio_init(struct folio *folio)
+{
+	union {
+		atomic_t o;
+		void *v;
+	} u = { .o = ATOMIC_INIT(1) };
+
+	folio->private = u.v;	/* valid only if file-backed folio is locked */
+}
+
+void erofs_onlinefolio_split(struct folio *folio)
+{
+	atomic_inc((atomic_t *)&folio->private);
+}
+
+void erofs_onlinefolio_end(struct folio *folio, int err, bool dirty)
+{
+	int orig, v;
+
+	do {
+		orig = atomic_read((atomic_t *)&folio->private);
+		DBG_BUGON(orig <= 0);
+		v = dirty << EROFS_ONLINEFOLIO_DIRTY;
+		v |= (orig - 1) | (!!err << EROFS_ONLINEFOLIO_EIO);
+	} while (atomic_cmpxchg((atomic_t *)&folio->private, orig, v) != orig);
+
+	if (v & (BIT(EROFS_ONLINEFOLIO_DIRTY) - 1))
+		return;
+	folio->private = 0;
+	if (v & BIT(EROFS_ONLINEFOLIO_DIRTY))
+		flush_dcache_folio(folio);
+	folio_end_read(folio, !(v & BIT(EROFS_ONLINEFOLIO_EIO)));
+}
+
+struct erofs_iomap_iter_ctx {
+	struct page *page;
+	void *base;
+	struct inode *realinode;
+};
+
+static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
+		unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
+{
+	struct iomap_iter *iter = container_of(iomap, struct iomap_iter, iomap);
+	struct erofs_iomap_iter_ctx *ctx = iter->private;
+	struct inode *realinode = ctx ? ctx->realinode : inode;
+	struct super_block *sb = realinode->i_sb;
+	struct erofs_map_blocks map;
+	struct erofs_map_dev mdev;
+	int ret;
+
+	map.m_la = offset;
+	map.m_llen = length;
+	ret = erofs_map_blocks(realinode, &map);
+	if (ret < 0)
+		return ret;
+
+	iomap->offset = map.m_la;
+	iomap->length = map.m_llen;
+	iomap->flags = 0;
+	iomap->addr = IOMAP_NULL_ADDR;
+	if (!(map.m_flags & EROFS_MAP_MAPPED)) {
+		iomap->type = IOMAP_HOLE;
+		return 0;
+	}
+
+	if (!(map.m_flags & EROFS_MAP_META) || !erofs_inode_in_metabox(realinode)) {
+		mdev = (struct erofs_map_dev) {
+			.m_deviceid = map.m_deviceid,
+			.m_pa = map.m_pa,
+		};
+		ret = erofs_map_dev(sb, &mdev);
+		if (ret)
+			return ret;
+
+		if (flags & IOMAP_DAX)
+			iomap->dax_dev = mdev.m_dif->dax_dev;
+		else
+			iomap->bdev = mdev.m_bdev;
+		iomap->addr = mdev.m_dif->fsoff + mdev.m_pa;
+		if (flags & IOMAP_DAX)
+			iomap->addr += mdev.m_dif->dax_part_off;
+	}
+
+	if (map.m_flags & EROFS_MAP_META) {
+		iomap->type = IOMAP_INLINE;
+		/* read context should read the inlined data */
+		if (ctx) {
+			struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
+			void *ptr;
+
+			ptr = erofs_read_metabuf(&buf, sb, map.m_pa,
+						 erofs_inode_in_metabox(realinode));
+			if (IS_ERR(ptr))
+				return PTR_ERR(ptr);
+			iomap->inline_data = ptr;
+			ctx->page = buf.page;
+			ctx->base = buf.base;
+		}
+	} else {
+		iomap->type = IOMAP_MAPPED;
+	}
+	return 0;
+}
+
+static int erofs_iomap_end(struct inode *inode, loff_t pos, loff_t length,
+		ssize_t written, unsigned int flags, struct iomap *iomap)
+{
+	struct iomap_iter *iter = container_of(iomap, struct iomap_iter, iomap);
+	struct erofs_iomap_iter_ctx *ctx = iter->private;
+
+	if (ctx && ctx->base) {
+		struct erofs_buf buf = {
+			.page = ctx->page,
+			.base = ctx->base,
+		};
+
+		DBG_BUGON(iomap->type != IOMAP_INLINE);
+		erofs_put_metabuf(&buf);
+		ctx->base = NULL;
+	}
+	return written;
+}
+
+static DEFINE_IOMAP_ITER_NEXT_END(erofs_iomap_next, erofs_iomap_begin,
+				  erofs_iomap_end);
+
+static const struct iomap_ops erofs_iomap_ops = {
+	.iomap_next = erofs_iomap_next,
+};
+
+int erofs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
+		 u64 start, u64 len)
+{
+	if (erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout)) {
+		if (!IS_ENABLED(CONFIG_EROFS_FS_ZIP))
+			return -EOPNOTSUPP;
+		return iomap_fiemap(inode, fieinfo, start, len,
+				    &z_erofs_iomap_report_ops);
+	}
+	return iomap_fiemap(inode, fieinfo, start, len, &erofs_iomap_ops);
+}
+
+/*
+ * since we dont have write or truncate flows, so no inode
+ * locking needs to be held at the moment.
+ */
+static int erofs_read_folio(struct file *file, struct folio *folio)
+{
+	struct iomap_read_folio_ctx read_ctx = {
+		.ops		= &iomap_bio_read_ops,
+		.cur_folio	= folio,
+	};
+	bool need_iput;
+	struct erofs_iomap_iter_ctx iter_ctx = {
+		.realinode = erofs_real_inode(folio_inode(folio), &need_iput),
+	};
+
+	trace_erofs_read_folio(iter_ctx.realinode, folio, true);
+	iomap_read_folio(&erofs_iomap_ops, &read_ctx, &iter_ctx);
+	if (need_iput)
+		iput(iter_ctx.realinode);
+	return 0;
+}
+
+static void erofs_readahead(struct readahead_control *rac)
+{
+	struct iomap_read_folio_ctx read_ctx = {
+		.ops		= &iomap_bio_read_ops,
+		.rac		= rac,
+	};
+	bool need_iput;
+	struct erofs_iomap_iter_ctx iter_ctx = {
+		.realinode = erofs_real_inode(rac->mapping->host, &need_iput),
+	};
+
+	trace_erofs_readahead(iter_ctx.realinode, readahead_index(rac),
+			      readahead_count(rac), true);
+	iomap_readahead(&erofs_iomap_ops, &read_ctx, &iter_ctx);
+	if (need_iput)
+		iput(iter_ctx.realinode);
+}
+
+static sector_t erofs_bmap(struct address_space *mapping, sector_t block)
+{
+	return iomap_bmap(mapping, block, &erofs_iomap_ops);
+}
+
+static ssize_t erofs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
+{
+	struct inode *inode = file_inode(iocb->ki_filp);
+
+	/* no need taking (shared) inode lock since it's a ro filesystem */
+	if (!iov_iter_count(to))
+		return 0;
+
+	if (IS_ENABLED(CONFIG_FS_DAX) && IS_DAX(inode))
+		return dax_iomap_rw(iocb, to, &erofs_iomap_ops);
+
+	if ((iocb->ki_flags & IOCB_DIRECT) && inode->i_sb->s_bdev) {
+		struct erofs_iomap_iter_ctx iter_ctx = {
+			.realinode = inode,
+		};
+
+		return iomap_dio_rw(iocb, to, &erofs_iomap_ops,
+				    NULL, 0, &iter_ctx, 0);
+	}
+	return filemap_read(iocb, to, 0);
+}
+
+/* for uncompressed (aligned) files and raw access for other files */
+const struct address_space_operations erofs_aops = {
+	.read_folio = erofs_read_folio,
+	.readahead = erofs_readahead,
+	.bmap = erofs_bmap,
+	.direct_IO = noop_direct_IO,
+	.release_folio = iomap_release_folio,
+	.invalidate_folio = iomap_invalidate_folio,
+};
+
+#ifdef CONFIG_FS_DAX
+static vm_fault_t erofs_dax_huge_fault(struct vm_fault *vmf,
+		unsigned int order)
+{
+	return dax_iomap_fault(vmf, order, NULL, NULL, &erofs_iomap_ops);
+}
+
+static vm_fault_t erofs_dax_fault(struct vm_fault *vmf)
+{
+	return erofs_dax_huge_fault(vmf, 0);
+}
+
+static const struct vm_operations_struct erofs_dax_vm_ops = {
+	.fault		= erofs_dax_fault,
+	.huge_fault	= erofs_dax_huge_fault,
+};
+
+static int erofs_file_mmap_prepare(struct vm_area_desc *desc)
+{
+	if (!IS_DAX(file_inode(desc->file)))
+		return generic_file_readonly_mmap_prepare(desc);
+
+	if (vma_desc_test_all(desc, VMA_SHARED_BIT, VMA_MAYWRITE_BIT))
+		return -EINVAL;
+
+	desc->vm_ops = &erofs_dax_vm_ops;
+	vma_desc_set_flags(desc, VMA_HUGEPAGE_BIT);
+	return 0;
+}
+#else
+#define erofs_file_mmap_prepare	generic_file_readonly_mmap_prepare
+#endif
+
+loff_t erofs_file_llseek(struct file *file, loff_t offset, int whence)
+{
+	struct inode *inode = file->f_mapping->host;
+	const struct iomap_ops *ops = &erofs_iomap_ops;
+
+	if (erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout)) {
+		if (!IS_ENABLED(CONFIG_EROFS_FS_ZIP))
+			return generic_file_llseek(file, offset, whence);
+		ops = &z_erofs_iomap_report_ops;
+	}
+
+	if (whence == SEEK_HOLE)
+		offset = iomap_seek_hole(inode, offset, ops);
+	else if (whence == SEEK_DATA)
+		offset = iomap_seek_data(inode, offset, ops);
+	else
+		return generic_file_llseek(file, offset, whence);
+
+	if (offset < 0)
+		return offset;
+	return vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
+}
+
+const struct file_operations erofs_file_fops = {
+	.llseek		= erofs_file_llseek,
+	.read_iter	= erofs_file_read_iter,
+	.unlocked_ioctl = erofs_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl   = erofs_compat_ioctl,
+#endif
+	.mmap_prepare	= erofs_file_mmap_prepare,
+	.get_unmapped_area = thp_get_unmapped_area,
+	.splice_read	= filemap_splice_read,
+	.setlease	= generic_setlease,
+};