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 --- net/ethtool/cmis.h | 128 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 128 insertions(+) create mode 100644 net/ethtool/cmis.h (limited to 'net/ethtool/cmis.h') diff --git a/net/ethtool/cmis.h b/net/ethtool/cmis.h new file mode 100644 index 000000000..778783a0f --- /dev/null +++ b/net/ethtool/cmis.h @@ -0,0 +1,128 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ + +#define ETHTOOL_CMIS_CDB_LPL_MAX_PL_LENGTH 120 +#define ETHTOOL_CMIS_CDB_EPL_MAX_PL_LENGTH 2048 +#define ETHTOOL_CMIS_CDB_CMD_PAGE 0x9F +#define ETHTOOL_CMIS_CDB_PAGE_I2C_ADDR 0x50 + +/** + * struct ethtool_cmis_cdb - CDB commands parameters + * @cmis_rev: CMIS revision major. + * @read_write_len_ext: Allowable additional number of byte octets to the LPL + * in a READ or a WRITE CDB commands. + * @max_completion_time: Maximum CDB command completion time in msec. + */ +struct ethtool_cmis_cdb { + u8 cmis_rev; + u8 read_write_len_ext; + u16 max_completion_time; +}; + +enum ethtool_cmis_cdb_cmd_id { + ETHTOOL_CMIS_CDB_CMD_QUERY_STATUS = 0x0000, + ETHTOOL_CMIS_CDB_CMD_MODULE_FEATURES = 0x0040, + ETHTOOL_CMIS_CDB_CMD_FW_MANAGMENT_FEATURES = 0x0041, + ETHTOOL_CMIS_CDB_CMD_START_FW_DOWNLOAD = 0x0101, + ETHTOOL_CMIS_CDB_CMD_WRITE_FW_BLOCK_LPL = 0x0103, + ETHTOOL_CMIS_CDB_CMD_WRITE_FW_BLOCK_EPL = 0x0104, + ETHTOOL_CMIS_CDB_CMD_COMPLETE_FW_DOWNLOAD = 0x0107, + ETHTOOL_CMIS_CDB_CMD_RUN_FW_IMAGE = 0x0109, + ETHTOOL_CMIS_CDB_CMD_COMMIT_FW_IMAGE = 0x010A, +}; + +/** + * struct ethtool_cmis_cdb_request - CDB commands request fields as decribed in + * the CMIS standard + * @id: Command ID. + * @epl_len: EPL memory length. + * @lpl_len: LPL memory length. + * @chk_code: Check code for the previous field and the payload. + * @resv1: Added to match the CMIS standard request continuity. + * @resv2: Added to match the CMIS standard request continuity. + * @payload: Payload for the CDB commands. + * @epl: Extended payload for the CDB commands. + */ +struct ethtool_cmis_cdb_request { + __be16 id; + struct_group(body, + __be16 epl_len; + u8 lpl_len; + u8 chk_code; + u8 resv1; + u8 resv2; + u8 payload[ETHTOOL_CMIS_CDB_LPL_MAX_PL_LENGTH]; + ); + u8 *epl; /* Everything above this field checksummed. */ +}; + +#define CDB_F_COMPLETION_VALID BIT(0) +#define CDB_F_STATUS_VALID BIT(1) +#define CDB_F_MODULE_STATE_VALID BIT(2) + +/** + * struct ethtool_cmis_cdb_cmd_args - CDB commands execution arguments + * @req: CDB command fields as described in the CMIS standard. + * @max_duration: Maximum duration time for command completion in msec. + * @msleep_pre_rpl: Waiting time before checking reply in msec. + * @read_write_len_ext: Allowable additional number of byte octets to the LPL + * in a READ or a WRITE commands. + * @rpl_exp_len: Expected reply length in bytes. + * @flags: Validation flags for CDB commands. + * @err_msg: Error message to be sent to user space. + */ +struct ethtool_cmis_cdb_cmd_args { + struct ethtool_cmis_cdb_request req; + u16 max_duration; + u16 msleep_pre_rpl; + u8 read_write_len_ext; + u8 rpl_exp_len; + u8 flags; + char *err_msg; +}; + +/** + * struct ethtool_cmis_cdb_rpl_hdr - CDB commands reply header arguments + * @rpl_len: Reply length. + * @rpl_chk_code: Reply check code. + */ +struct ethtool_cmis_cdb_rpl_hdr { + u8 rpl_len; + u8 rpl_chk_code; +}; + +/** + * struct ethtool_cmis_cdb_rpl - CDB commands reply arguments + * @hdr: CDB commands reply header arguments. + * @payload: Payload for the CDB commands reply. + */ +struct ethtool_cmis_cdb_rpl { + struct ethtool_cmis_cdb_rpl_hdr hdr; + u8 payload[ETHTOOL_CMIS_CDB_LPL_MAX_PL_LENGTH]; +}; + +u32 ethtool_cmis_get_max_lpl_size(u8 num_of_byte_octs); + +void ethtool_cmis_cdb_compose_args(struct ethtool_cmis_cdb_cmd_args *args, + enum ethtool_cmis_cdb_cmd_id cmd, u8 *lpl, + u8 lpl_len, u8 *epl, u16 epl_len, + u16 max_duration, u8 read_write_len_ext, + u16 msleep_pre_rpl, u8 rpl_exp_len, + u8 flags); + +void ethtool_cmis_cdb_check_completion_flag(u8 cmis_rev, u8 *flags); + +void ethtool_cmis_page_init(struct ethtool_module_eeprom *page_data, + u8 page, u32 offset, u32 length); + +struct ethtool_cmis_cdb * +ethtool_cmis_cdb_init(struct net_device *dev, + const struct ethtool_module_fw_flash_params *params, + struct ethnl_module_fw_flash_ntf_params *ntf_params); +void ethtool_cmis_cdb_fini(struct ethtool_cmis_cdb *cdb); + +int ethtool_cmis_wait_for_cond(struct net_device *dev, u8 flags, u8 flag, + u16 max_duration, u32 offset, + bool (*cond_success)(u8), bool (*cond_fail)(u8), u8 *state); + +int ethtool_cmis_cdb_execute_cmd(struct net_device *dev, + struct ethtool_cmis_cdb_cmd_args *args); -- cgit v1.3.1