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/jffs2/acl.c | 312 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 312 insertions(+) create mode 100644 fs/jffs2/acl.c (limited to 'fs/jffs2/acl.c') diff --git a/fs/jffs2/acl.c b/fs/jffs2/acl.c new file mode 100644 index 000000000..f0f8a4f57 --- /dev/null +++ b/fs/jffs2/acl.c @@ -0,0 +1,312 @@ +/* + * JFFS2 -- Journalling Flash File System, Version 2. + * + * Copyright © 2006 NEC Corporation + * + * Created by KaiGai Kohei + * + * For licensing information, see the file 'LICENCE' in this directory. + * + */ + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "nodelist.h" + +static size_t jffs2_acl_size(int count) +{ + if (count <= 4) { + return sizeof(struct jffs2_acl_header) + + count * sizeof(struct jffs2_acl_entry_short); + } else { + return sizeof(struct jffs2_acl_header) + + 4 * sizeof(struct jffs2_acl_entry_short) + + (count - 4) * sizeof(struct jffs2_acl_entry); + } +} + +static int jffs2_acl_count(size_t size) +{ + size_t s; + + size -= sizeof(struct jffs2_acl_header); + if (size < 4 * sizeof(struct jffs2_acl_entry_short)) { + if (size % sizeof(struct jffs2_acl_entry_short)) + return -1; + return size / sizeof(struct jffs2_acl_entry_short); + } else { + s = size - 4 * sizeof(struct jffs2_acl_entry_short); + if (s % sizeof(struct jffs2_acl_entry)) + return -1; + return s / sizeof(struct jffs2_acl_entry) + 4; + } +} + +static struct posix_acl *jffs2_acl_from_medium(void *value, size_t size) +{ + void *end = value + size; + struct jffs2_acl_header *header = value; + struct jffs2_acl_entry *entry; + struct posix_acl *acl; + uint32_t ver; + int i, count; + + if (!value) + return NULL; + if (size < sizeof(struct jffs2_acl_header)) + return ERR_PTR(-EINVAL); + ver = je32_to_cpu(header->a_version); + if (ver != JFFS2_ACL_VERSION) { + JFFS2_WARNING("Invalid ACL version. (=%u)\n", ver); + return ERR_PTR(-EINVAL); + } + + value += sizeof(struct jffs2_acl_header); + count = jffs2_acl_count(size); + if (count < 0) + return ERR_PTR(-EINVAL); + if (count == 0) + return NULL; + + acl = posix_acl_alloc(count, GFP_KERNEL); + if (!acl) + return ERR_PTR(-ENOMEM); + + for (i=0; i < count; i++) { + entry = value; + if (value + sizeof(struct jffs2_acl_entry_short) > end) + goto fail; + acl->a_entries[i].e_tag = je16_to_cpu(entry->e_tag); + acl->a_entries[i].e_perm = je16_to_cpu(entry->e_perm); + switch (acl->a_entries[i].e_tag) { + case ACL_USER_OBJ: + case ACL_GROUP_OBJ: + case ACL_MASK: + case ACL_OTHER: + value += sizeof(struct jffs2_acl_entry_short); + break; + + case ACL_USER: + value += sizeof(struct jffs2_acl_entry); + if (value > end) + goto fail; + acl->a_entries[i].e_uid = + make_kuid(&init_user_ns, + je32_to_cpu(entry->e_id)); + break; + case ACL_GROUP: + value += sizeof(struct jffs2_acl_entry); + if (value > end) + goto fail; + acl->a_entries[i].e_gid = + make_kgid(&init_user_ns, + je32_to_cpu(entry->e_id)); + break; + + default: + goto fail; + } + } + if (value != end) + goto fail; + return acl; + fail: + posix_acl_release(acl); + return ERR_PTR(-EINVAL); +} + +static void *jffs2_acl_to_medium(const struct posix_acl *acl, size_t *size) +{ + struct jffs2_acl_header *header; + struct jffs2_acl_entry *entry; + void *e; + size_t i; + + *size = jffs2_acl_size(acl->a_count); + header = kmalloc_flex(*header, a_entries, acl->a_count); + if (!header) + return ERR_PTR(-ENOMEM); + header->a_version = cpu_to_je32(JFFS2_ACL_VERSION); + e = header + 1; + for (i=0; i < acl->a_count; i++) { + const struct posix_acl_entry *acl_e = &acl->a_entries[i]; + entry = e; + entry->e_tag = cpu_to_je16(acl_e->e_tag); + entry->e_perm = cpu_to_je16(acl_e->e_perm); + switch(acl_e->e_tag) { + case ACL_USER: + entry->e_id = cpu_to_je32( + from_kuid(&init_user_ns, acl_e->e_uid)); + e += sizeof(struct jffs2_acl_entry); + break; + case ACL_GROUP: + entry->e_id = cpu_to_je32( + from_kgid(&init_user_ns, acl_e->e_gid)); + e += sizeof(struct jffs2_acl_entry); + break; + + case ACL_USER_OBJ: + case ACL_GROUP_OBJ: + case ACL_MASK: + case ACL_OTHER: + e += sizeof(struct jffs2_acl_entry_short); + break; + + default: + goto fail; + } + } + return header; + fail: + kfree(header); + return ERR_PTR(-EINVAL); +} + +struct posix_acl *jffs2_get_acl(struct inode *inode, int type, bool rcu) +{ + struct posix_acl *acl; + char *value = NULL; + int rc, xprefix; + + if (rcu) + return ERR_PTR(-ECHILD); + + switch (type) { + case ACL_TYPE_ACCESS: + xprefix = JFFS2_XPREFIX_ACL_ACCESS; + break; + case ACL_TYPE_DEFAULT: + xprefix = JFFS2_XPREFIX_ACL_DEFAULT; + break; + default: + BUG(); + } + rc = do_jffs2_getxattr(inode, xprefix, "", NULL, 0); + if (rc > 0) { + value = kmalloc(rc, GFP_KERNEL); + if (!value) + return ERR_PTR(-ENOMEM); + rc = do_jffs2_getxattr(inode, xprefix, "", value, rc); + } + if (rc > 0) { + acl = jffs2_acl_from_medium(value, rc); + } else if (rc == -ENODATA || rc == -ENOSYS) { + acl = NULL; + } else { + acl = ERR_PTR(rc); + } + kfree(value); + return acl; +} + +static int __jffs2_set_acl(struct inode *inode, int xprefix, struct posix_acl *acl) +{ + char *value = NULL; + size_t size = 0; + int rc; + + if (acl) { + value = jffs2_acl_to_medium(acl, &size); + if (IS_ERR(value)) + return PTR_ERR(value); + } + rc = do_jffs2_setxattr(inode, xprefix, "", value, size, 0); + if (!value && rc == -ENODATA) + rc = 0; + kfree(value); + + return rc; +} + +int jffs2_set_acl(struct mnt_idmap *idmap, struct dentry *dentry, + struct posix_acl *acl, int type) +{ + int rc, xprefix; + struct inode *inode = d_inode(dentry); + + switch (type) { + case ACL_TYPE_ACCESS: + xprefix = JFFS2_XPREFIX_ACL_ACCESS; + if (acl) { + umode_t mode; + + rc = posix_acl_update_mode(&nop_mnt_idmap, inode, &mode, + &acl); + if (rc) + return rc; + if (inode->i_mode != mode) { + struct iattr attr; + + attr.ia_valid = ATTR_MODE | ATTR_CTIME; + attr.ia_mode = mode; + attr.ia_ctime = current_time(inode); + rc = jffs2_do_setattr(inode, &attr); + if (rc < 0) + return rc; + } + } + break; + case ACL_TYPE_DEFAULT: + xprefix = JFFS2_XPREFIX_ACL_DEFAULT; + if (!S_ISDIR(inode->i_mode)) + return acl ? -EACCES : 0; + break; + default: + return -EINVAL; + } + rc = __jffs2_set_acl(inode, xprefix, acl); + if (!rc) + set_cached_acl(inode, type, acl); + return rc; +} + +int jffs2_init_acl_pre(struct inode *dir_i, struct inode *inode, umode_t *i_mode) +{ + struct posix_acl *default_acl, *acl; + int rc; + + cache_no_acl(inode); + + rc = posix_acl_create(dir_i, i_mode, &default_acl, &acl); + if (rc) + return rc; + + if (default_acl) { + set_cached_acl(inode, ACL_TYPE_DEFAULT, default_acl); + posix_acl_release(default_acl); + } + if (acl) { + set_cached_acl(inode, ACL_TYPE_ACCESS, acl); + posix_acl_release(acl); + } + return 0; +} + +int jffs2_init_acl_post(struct inode *inode) +{ + int rc; + + if (inode->i_default_acl) { + rc = __jffs2_set_acl(inode, JFFS2_XPREFIX_ACL_DEFAULT, inode->i_default_acl); + if (rc) + return rc; + } + + if (inode->i_acl) { + rc = __jffs2_set_acl(inode, JFFS2_XPREFIX_ACL_ACCESS, inode->i_acl); + if (rc) + return rc; + } + + return 0; +} -- cgit v1.3.1