Microsoft MVP성태의 닷넷 이야기
Linux: 122. eBPF (bpf2go) - __attribute__((preserve_access_index)) 사용법 [링크 복사], [링크+제목 복사],
조회: 652
글쓴 사람
정성태 (seongtaejeong at gmail.com)
홈페이지
첨부 파일
 

(시리즈 글이 16개 있습니다.)
Linux: 86. Golang + bpf2go를 사용한 eBPF 기본 예제
; https://www.sysnet.pe.kr/2/0/13769

Linux: 94. eBPF - vmlinux.h 헤더 포함하는 방법 (bpf2go에서 사용)
; https://www.sysnet.pe.kr/2/0/13783

Linux: 95. eBPF - kprobe를 이용한 트레이스
; https://www.sysnet.pe.kr/2/0/13784

Linux: 96. eBPF (bpf2go) - fentry, fexit를 이용한 트레이스
; https://www.sysnet.pe.kr/2/0/13788

Linux: 100.  eBPF의 2가지 방식 - libbcc와 libbpf(CO-RE)
; https://www.sysnet.pe.kr/2/0/13801

Linux: 103. eBPF (bpf2go) - Tracepoint를 이용한 트레이스 (BPF_PROG_TYPE_TRACEPOINT)
; https://www.sysnet.pe.kr/2/0/13810

Linux: 105. eBPF - bpf2go에서 전역 변수 설정 방법
; https://www.sysnet.pe.kr/2/0/13815

Linux: 106. eBPF (bpf2go) - (BPF_MAP_TYPE_HASH) Map을 이용한 전역 변수 구현
; https://www.sysnet.pe.kr/2/0/13817

Linux: 107. eBPF - libbpf CO-RE의 CONFIG_DEBUG_INFO_BTF 빌드 여부에 대한 의존성
; https://www.sysnet.pe.kr/2/0/13819

Linux: 109. eBPF (bpf2go) - BPF_PERF_OUTPUT / BPF_MAP_TYPE_PERF_EVENT_ARRAY 사용법
; https://www.sysnet.pe.kr/2/0/13824

Linux: 110. eBPF (bpf2go) - BPF_RINGBUF_OUTPUT / BPF_MAP_TYPE_RINGBUF 사용법
; https://www.sysnet.pe.kr/2/0/13825

Linux: 115. eBPF (bpf2go) - ARRAY / HASH map 기본 사용법
; https://www.sysnet.pe.kr/2/0/13893

Linux: 116. eBPF (bpf2go) - BTF Style Maps 정의 구문과 데이터 정렬 문제
; https://www.sysnet.pe.kr/2/0/13894

Linux: 117. eBPF (bpf2go) - Map에 추가된 요소의 개수를 확인하는 방법
; https://www.sysnet.pe.kr/2/0/13895

Linux: 122. eBPF (bpf2go) - __attribute__((preserve_access_index)) 사용법
; https://www.sysnet.pe.kr/2/0/14016

Linux: 123. eBPF (bpf2go) - BPF_PROG_TYPE_SOCKET_FILTER 예제 - SEC("socket")
; https://www.sysnet.pe.kr/2/0/14017




eBPF (bpf2go) - __attribute__((preserve_access_index)) 사용법

아래의 글에 보면,

eBPF application development: Beyond the basics
; https://developers.redhat.com/articles/2023/10/19/ebpf-application-development-beyond-basics

The BPF program attaches to the net_dev_queue kernel tracepoint to intercept all packets sent or received by any program running on the host. The tracepoint context includes a pointer to the struct sk_buff that holds the packet data. We are only interested in accessing thesk_buff->data and sk_buff->len fields so we can use CO-RE to access them. The BPF program defines its own private version of struct sk_buff that contains only the fields we need. The struct is annotated with the preserve_access_index attribute so that the CO-RE relocation can happen when the BPF program gets loaded.


"Accessing kernel data structures" 절에서 "__attribute__((preserve_access_index)"에 대한 설명이 나옵니다. 정리해 보면, 필요한 커널 구조체가 있을 때 원래의 그 타입이 소유한 모든 필드를 정의할 필요 없이 관심 있는 필드만 정의할 수 있게 해주는 attribute입니다. (어차피 이름이 같은 field의 offset을 런타임 시에 eBPF 로더가 계산해 주므로.)

예를 들어, task_struct를 사용한다고 가정해 볼까요? ^^ 전에도 설명했지만 task_struct는 너무 크고 빌드 시 정의한 전처리 상수에 따라 유동적이므로 그런 구조체의 정의를 BPF 내에서 직접 정의해 사용하는 것은 좀 애매합니다. 바로 그럴 때, __attribute__((preserve_access_index))를 사용해 관심 있는 필드만 정의할 수 있습니다.

struct task_struct {
    unsigned int flags;
    const struct cred *cred;
} __attribute__((preserve_access_index));

한 가지 유의할 점이라면, 저렇게 정의할 수 있는 구조체는 "vmlinux.h"에 정의되지 않은 구조체여야 한다는 점입니다. 왜냐하면 같은 이름의 구조체를 재정의할 수 없기 때문입니다.

즉, 약식 정의한 구조체의 이름을 task_struct라고 정의하면, vmlinux.h에 이미 동일한 이름으로 정의돼 있으므로 컴파일 오류가 발생하는 것입니다. (사실, 굳이 vmlinux.h에 있는 구조체를 재정의할 필요가 없긴 합니다.) 그렇다고 해서, 대신 my_task_struct 같은 이름을 사용하면 어떻게 될까요?

struct my_task_struct {
    unsigned int flags;
    const struct cred *cred;
} __attribute__((preserve_access_index));

SEC("tracepoint/syscalls/sys_enter_execve")
int sys_enter_execve(struct pt_regs *ctx) {
    struct my_task_struct *current_task = (struct my_task_struct *)bpf_get_current_task();

    unsigned int task_flags = BPF_CORE_READ(current_task, flags);

    bpf_printk("task_flags == %d\n", task_flags);

    return 0;
}

저렇게 하면 일단 (bpf2go 등의) 컴파일 단계는 통과하지만, 이후의 실행 단계에서 bpf 프로그램을 로드할 때 다음과 같은 오류가 발생합니다.

program sys_enter_execve: load program: bad CO-RE relocation: invalid func unknown#195896080 (6 line(s) omitted)

왜냐하면 CO-RE 재배치(relocation)를 위해서는 구조체의 이름 정보까지 동일해야 하기 때문입니다.




그렇다면, vmlinux.h 헤더를 포함시키지 않으면 되지 않을까요? 그러기에는 ^^; vmlinux 헤더 파일에 포함된 수많은 정의를 사용할 수 없어 더 큰 불편이 발생합니다.

그래도 아주 불가능한 것은 아닌데요, 이에 대한 좋은 예제를 cilium에서 찾아볼 수 있습니다.

examples/headers
; https://github.com/cilium/ebpf/tree/main/examples/headers

examples/tracepoint_in_c/tracepoint.c
; https://github.com/cilium/ebpf/blob/main/examples/tracepoint_in_c/tracepoint.c

tracepoint.c 예제를 보면, "examples/headers" 경로에 있는 common.h 헤더만 포함시키고 있는데요, 그 파일(및 그것이 포함하는 헤더 파일)은 vmlinux.h의 약식 버전에 해당합니다. 즉, task_struct 같은 커널 구조체 정의는 포함하고 있지 않으므로 다음과 같이 코딩할 수 있습니다.

#include "common.h"
#include <bpf/bpf_core_read.h>

#define TASK_COMM_LEN 16

struct task_struct {
    unsigned int flags;
    const struct cred *cred;
    char comm[TASK_COMM_LEN];
} __attribute__((preserve_access_index));

SEC("tracepoint/syscalls/sys_enter_execve")
int sys_enter_execve(struct pt_regs *ctx) {
    struct task_struct *current_task = (struct task_struct *)bpf_get_current_task();

    unsigned int task_flags = BPF_CORE_READ(current_task, flags);

    bpf_printk("task_flags == %d, %s\n", task_flags, current_task->comm);

    return 0;
}

당연하겠지만, 저 attribute가 적용된 구조체는 BTF가 활성화된 OS에서만 사용 가능합니다. 만약 BTF가 비활성화된 OS에서 저 BPF 프로그램을 로딩하면 다음과 같은 오류가 발생합니다.

program sys_enter_execve: apply CO-RE relocations: load kernel spec: btf: not found




그런데... 굳이 저 옵션이 (vmlinux.h 포함 여부에 상관없이) task_struct와 같은 커널 구조체에 대해 필요한가...입니다. 왜냐하면 vmlinux.h 헤더에 정의된 task_struct도 preserve_access_index 옵션 없이 정의돼 있기 때문입니다. 실제로, (vmlinux.h를 없애고 common.h로 실습한) 위의 예제에서도 task_struct를 preserve_access_index 없이 정의해도 잘 동작합니다. 그러니까, 어차피 BTF 정보가 있는 타입이라면 굳이 저 속성을 붙이지 않아도 된다는 뜻입니다.

그런 의미에서 "eBPF application development: Beyond the basics" 글에서 설명한 저 속성의 용도는 딱히 와닿지 않습니다.

그렇다고 그냥 사용자 코드에서 정의한 구조체에 저 속성을 붙이는 것도,

[BPF] Add preserve_access_index attribute for record definition
; https://reviews.llvm.org/D69759

의미 있어 보이진 않습니다. 혹시 저 옵션에 대한 이력이나 관련 설명을 해주실 분이 계실까요? ^^




[이 글에 대해서 여러분들과 의견을 공유하고 싶습니다. 틀리거나 미흡한 부분 또는 의문 사항이 있으시면 언제든 댓글 남겨주십시오.]







[최초 등록일: ]
[최종 수정일: 9/28/2025]

Creative Commons License
이 저작물은 크리에이티브 커먼즈 코리아 저작자표시-비영리-변경금지 2.0 대한민국 라이센스에 따라 이용하실 수 있습니다.
by SeongTae Jeong, mailto:techsharer at outlook.com

비밀번호

댓글 작성자
 




1  2  3  4  5  6  7  8  9  [10]  11  12  13  14  15  ...
NoWriterDateCnt.TitleFile(s)
13795정성태10/31/20249148Windows: 268. Windows - 리눅스 환경처럼 공백으로 끝나는 프롬프트 만들기
13794정성태10/30/20249631닷넷: 2307. C# - 윈도우에서 한글(및 유니코드)을 포함한 콘솔 프로그램을 컴파일 및 실행하는 방법
13793정성태10/28/20249426C/C++: 183. C++ - 윈도우에서 한글(및 유니코드)을 포함한 콘솔 프로그램을 컴파일 및 실행하는 방법
13792정성태10/27/20248494Linux: 99. Linux - 프로세스의 실행 파일 경로 확인
13791정성태10/27/20248602Windows: 267. Win32 API의 A(ANSI) 버전은 DBCS를 사용할까요?파일 다운로드1
13790정성태10/27/20248225Linux: 98. Ubuntu 22.04 - 리눅스 커널 빌드 및 업그레이드
13789정성태10/27/20249718Linux: 97. menuconfig에 CONFIG_DEBUG_INFO_BTF, CONFIG_DEBUG_INFO_BTF_MODULES 옵션이 없는 경우
13788정성태10/26/20247511Linux: 96. eBPF (bpf2go) - fentry, fexit를 이용한 트레이스
13787정성태10/26/202410253개발 환경 구성: 730. github - Linux 커널 repo를 윈도우 환경에서 git clone하는 방법 [1]
13786정성태10/26/20249233Windows: 266. Windows - 대소문자 구분이 가능한 파일 시스템
13785정성태10/23/202410529C/C++: 182. 윈도우가 운영하는 2개의 Code Page [1]파일 다운로드1
13784정성태10/23/20249058Linux: 95. eBPF - kprobe를 이용한 트레이스
13783정성태10/23/20248745Linux: 94. eBPF - vmlinux.h 헤더 포함하는 방법 (bpf2go에서 사용)
13782정성태10/23/20248093Linux: 93. Ubuntu 22.04 - 커널 이미지로부터 커널 함수 역어셈블
13781정성태10/22/20248801오류 유형: 930. WSL + eBPF: modprobe: FATAL: Module kheaders not found in directory
13780정성태10/22/202410620Linux: 92. WSL 2 - 커널 이미지로부터 커널 함수 역어셈블
13779정성태10/22/20248300개발 환경 구성: 729. WSL 2 - Mariner VM 커널 이미지 업데이트 방법
13778정성태10/21/202410880C/C++: 181. C/C++ - 소스코드 파일의 인코딩, 바이너리 모듈 상태의 인코딩
13777정성태10/20/20249077Windows: 265. Win32 API의 W(유니코드) 버전은 UCS-2일까요? UTF-16 인코딩일까요?
13776정성태10/19/20249819C/C++: 180. C++ - 고수준 FILE I/O 함수에서의 Unicode stream 모드(_O_WTEXT, _O_U16TEXT, _O_U8TEXT)파일 다운로드1
13775정성태10/19/202410529개발 환경 구성: 728. 윈도우 환경의 개발자를 위한 UTF-8 환경 설정
13774정성태10/18/20249500Linux: 91. Container 환경에서 출력하는 eBPF bpf_get_current_pid_tgid의 pid가 존재하지 않는 이유
13773정성태10/18/20248825Linux: 90. pid 네임스페이스 구성으로 본 WSL 2 + docker-desktop
13772정성태10/17/20249609Linux: 89. pid 네임스페이스 구성으로 본 WSL 2 배포본의 계층 관계
13771정성태10/17/20249472Linux: 88. WSL 2 리눅스 배포본 내에서의 pid 네임스페이스 구성
13770정성태10/17/202410223Linux: 87. ps + grep 조합에서 grep 명령어를 사용한 프로세스를 출력에서 제거하는 방법
1  2  3  4  5  6  7  8  9  [10]  11  12  13  14  15  ...