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 --- fs/omfs/bitmap.c | 194 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 194 insertions(+) create mode 100644 fs/omfs/bitmap.c (limited to 'fs/omfs/bitmap.c') diff --git a/fs/omfs/bitmap.c b/fs/omfs/bitmap.c new file mode 100644 index 000000000..7147ba6a6 --- /dev/null +++ b/fs/omfs/bitmap.c @@ -0,0 +1,194 @@ +// SPDX-License-Identifier: GPL-2.0 +#include +#include +#include +#include +#include "omfs.h" + +unsigned long omfs_count_free(struct super_block *sb) +{ + unsigned int i; + unsigned long sum = 0; + struct omfs_sb_info *sbi = OMFS_SB(sb); + int nbits = sb->s_blocksize * 8; + + for (i = 0; i < sbi->s_imap_size; i++) + sum += nbits - bitmap_weight(sbi->s_imap[i], nbits); + + return sum; +} + +/* + * Counts the run of zero bits starting at bit up to max. + * It handles the case where a run might spill over a buffer. + * Called with bitmap lock. + */ +static int count_run(unsigned long **addr, int nbits, + int addrlen, int bit, int max) +{ + int count = 0; + int x; + + for (; addrlen > 0; addrlen--, addr++) { + x = find_next_bit(*addr, nbits, bit); + count += x - bit; + + if (x < nbits || count > max) + return min(count, max); + + bit = 0; + } + return min(count, max); +} + +/* + * Sets or clears the run of count bits starting with bit. + * Called with bitmap lock. + */ +static int set_run(struct super_block *sb, int map, + int nbits, int bit, int count, int set) +{ + int i; + int err; + struct buffer_head *bh; + struct omfs_sb_info *sbi = OMFS_SB(sb); + + err = -ENOMEM; + bh = sb_bread(sb, clus_to_blk(sbi, sbi->s_bitmap_ino) + map); + if (!bh) + goto out; + + for (i = 0; i < count; i++, bit++) { + if (bit >= nbits) { + bit = 0; + map++; + + mark_buffer_dirty(bh); + brelse(bh); + bh = sb_bread(sb, + clus_to_blk(sbi, sbi->s_bitmap_ino) + map); + if (!bh) + goto out; + } + if (set) { + set_bit(bit, sbi->s_imap[map]); + set_bit(bit, (unsigned long *)bh->b_data); + } else { + clear_bit(bit, sbi->s_imap[map]); + clear_bit(bit, (unsigned long *)bh->b_data); + } + } + mark_buffer_dirty(bh); + brelse(bh); + err = 0; +out: + return err; +} + +/* + * Tries to allocate exactly one block. Returns true if successful. + */ +int omfs_allocate_block(struct super_block *sb, u64 block) +{ + struct buffer_head *bh; + struct omfs_sb_info *sbi = OMFS_SB(sb); + int bits_per_entry = 8 * sb->s_blocksize; + unsigned int map, bit; + int ret = 0; + u64 tmp; + + tmp = block; + bit = do_div(tmp, bits_per_entry); + map = tmp; + + mutex_lock(&sbi->s_bitmap_lock); + if (map >= sbi->s_imap_size || test_and_set_bit(bit, sbi->s_imap[map])) + goto out; + + if (sbi->s_bitmap_ino > 0) { + bh = sb_bread(sb, clus_to_blk(sbi, sbi->s_bitmap_ino) + map); + if (!bh) + goto out; + + set_bit(bit, (unsigned long *)bh->b_data); + mark_buffer_dirty(bh); + brelse(bh); + } + ret = 1; +out: + mutex_unlock(&sbi->s_bitmap_lock); + return ret; +} + + +/* + * Tries to allocate a set of blocks. The request size depends on the + * type: for inodes, we must allocate sbi->s_mirrors blocks, and for file + * blocks, we try to allocate sbi->s_clustersize, but can always get away + * with just one block. + */ +int omfs_allocate_range(struct super_block *sb, + int min_request, + int max_request, + u64 *return_block, + int *return_size) +{ + struct omfs_sb_info *sbi = OMFS_SB(sb); + int bits_per_entry = 8 * sb->s_blocksize; + int ret = 0; + int i, run, bit; + + mutex_lock(&sbi->s_bitmap_lock); + for (i = 0; i < sbi->s_imap_size; i++) { + bit = 0; + while (bit < bits_per_entry) { + bit = find_next_zero_bit(sbi->s_imap[i], bits_per_entry, + bit); + + if (bit == bits_per_entry) + break; + + run = count_run(&sbi->s_imap[i], bits_per_entry, + sbi->s_imap_size-i, bit, max_request); + + if (run >= min_request) + goto found; + bit += run; + } + } + ret = -ENOSPC; + goto out; + +found: + *return_block = (u64) i * bits_per_entry + bit; + *return_size = run; + ret = set_run(sb, i, bits_per_entry, bit, run, 1); + +out: + mutex_unlock(&sbi->s_bitmap_lock); + return ret; +} + +/* + * Clears count bits starting at a given block. + */ +int omfs_clear_range(struct super_block *sb, u64 block, int count) +{ + struct omfs_sb_info *sbi = OMFS_SB(sb); + int bits_per_entry = 8 * sb->s_blocksize; + u64 tmp; + unsigned int map, bit; + int ret; + + tmp = block; + bit = do_div(tmp, bits_per_entry); + map = tmp; + + if (map >= sbi->s_imap_size) + return 0; + + mutex_lock(&sbi->s_bitmap_lock); + ret = set_run(sb, map, bits_per_entry, bit, count, 0); + mutex_unlock(&sbi->s_bitmap_lock); + return ret; +} -- cgit v1.3.1