* [PATCH v4 0/3] Fix the incorrect register read for syscalls on x86_64
@ 2022-01-19 13:12 Kenta Tada
2022-01-19 13:12 ` [PATCH v4 1/3] libbpf: Extract syscall wrapper Kenta Tada
` (2 more replies)
0 siblings, 3 replies; 10+ messages in thread
From: Kenta Tada @ 2022-01-19 13:12 UTC (permalink / raw)
To: andrii, bpf
Cc: ast, daniel, kafai, songliubraving, yhs, john.fastabend, kpsingh,
Kenta Tada
Currently, rcx is read as the fourth parameter of syscall on x86_64.
But x86_64 Linux System Call convention uses r10 actually.
This commit adds the wrapper for users who want to access to
syscall params to analyze the user space.
Changelog:
----------
v1 -> v2:
- Rebase to current bpf-next
https://lore.kernel.org/bpf/20211222213924.1869758-1-andrii@kernel.org/
v2 -> v3:
- Modify the definition of SYSCALL macros for only targeted archs.
- Define __BPF_TARGET_MISSING variants for completeness.
- Remove CORE variants. These macros will not be used.
- Add a selftest.
v3 -> v4:
- Modify a selftest not to use serial tests.
- Modify a selftest to use ASSERT_EQ().
- Extract syscall wrapper for all the other tests.
- Add CORE variants.
Kenta Tada (3):
libbpf: Extract syscall wrapper
libbpf: Fix the incorrect register read for syscalls on x86_64
libbpf: Add a test to confirm PT_REGS_PARM4_SYSCALL
tools/lib/bpf/bpf_tracing.h | 34 +++++++++++++
.../bpf/prog_tests/test_bpf_syscall_macro.c | 49 ++++++++++++++++++
tools/testing/selftests/bpf/progs/bpf_misc.h | 19 +++++++
.../selftests/bpf/progs/bpf_syscall_macro.c | 51 +++++++++++++++++++
.../selftests/bpf/progs/test_probe_user.c | 15 +-----
5 files changed, 154 insertions(+), 14 deletions(-)
create mode 100644 tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c
create mode 100644 tools/testing/selftests/bpf/progs/bpf_misc.h
create mode 100644 tools/testing/selftests/bpf/progs/bpf_syscall_macro.c
--
2.32.0
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v4 1/3] libbpf: Extract syscall wrapper
2022-01-19 13:12 [PATCH v4 0/3] Fix the incorrect register read for syscalls on x86_64 Kenta Tada
@ 2022-01-19 13:12 ` Kenta Tada
2022-01-19 13:12 ` [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64 Kenta Tada
2022-01-19 13:12 ` [PATCH v4 3/3] libbpf: Add a test to confirm PT_REGS_PARM4_SYSCALL Kenta Tada
2 siblings, 0 replies; 10+ messages in thread
From: Kenta Tada @ 2022-01-19 13:12 UTC (permalink / raw)
To: andrii, bpf
Cc: ast, daniel, kafai, songliubraving, yhs, john.fastabend, kpsingh,
Kenta Tada
Extract the helper to set up SYS_PREFIX for fentry and kprobe selftests
that use __x86_sys_* attach functions.
Signed-off-by: Kenta Tada <Kenta.Tada@sony.com>
Suggested-by: Andrii Nakryiko <andrii@kernel.org>
---
tools/testing/selftests/bpf/progs/bpf_misc.h | 19 +++++++++++++++++++
.../selftests/bpf/progs/test_probe_user.c | 15 +--------------
2 files changed, 20 insertions(+), 14 deletions(-)
create mode 100644 tools/testing/selftests/bpf/progs/bpf_misc.h
diff --git a/tools/testing/selftests/bpf/progs/bpf_misc.h b/tools/testing/selftests/bpf/progs/bpf_misc.h
new file mode 100644
index 000000000000..0b78bc9b1b4c
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/bpf_misc.h
@@ -0,0 +1,19 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __BPF_MISC_H__
+#define __BPF_MISC_H__
+
+#if defined(__TARGET_ARCH_x86)
+#define SYSCALL_WRAPPER 1
+#define SYS_PREFIX "__x64_"
+#elif defined(__TARGET_ARCH_s390)
+#define SYSCALL_WRAPPER 1
+#define SYS_PREFIX "__s390x_"
+#elif defined(__TARGET_ARCH_arm64)
+#define SYSCALL_WRAPPER 1
+#define SYS_PREFIX "__arm64_"
+#else
+#define SYSCALL_WRAPPER 0
+#define SYS_PREFIX ""
+#endif
+
+#endif
diff --git a/tools/testing/selftests/bpf/progs/test_probe_user.c b/tools/testing/selftests/bpf/progs/test_probe_user.c
index 8812a90da4eb..702578a5e496 100644
--- a/tools/testing/selftests/bpf/progs/test_probe_user.c
+++ b/tools/testing/selftests/bpf/progs/test_probe_user.c
@@ -7,20 +7,7 @@
#include <bpf/bpf_helpers.h>
#include <bpf/bpf_tracing.h>
-
-#if defined(__TARGET_ARCH_x86)
-#define SYSCALL_WRAPPER 1
-#define SYS_PREFIX "__x64_"
-#elif defined(__TARGET_ARCH_s390)
-#define SYSCALL_WRAPPER 1
-#define SYS_PREFIX "__s390x_"
-#elif defined(__TARGET_ARCH_arm64)
-#define SYSCALL_WRAPPER 1
-#define SYS_PREFIX "__arm64_"
-#else
-#define SYSCALL_WRAPPER 0
-#define SYS_PREFIX ""
-#endif
+#include "bpf_misc.h"
static struct sockaddr_in old;
--
2.32.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64
2022-01-19 13:12 [PATCH v4 0/3] Fix the incorrect register read for syscalls on x86_64 Kenta Tada
2022-01-19 13:12 ` [PATCH v4 1/3] libbpf: Extract syscall wrapper Kenta Tada
@ 2022-01-19 13:12 ` Kenta Tada
2022-01-19 18:33 ` Andrii Nakryiko
2022-01-19 13:12 ` [PATCH v4 3/3] libbpf: Add a test to confirm PT_REGS_PARM4_SYSCALL Kenta Tada
2 siblings, 1 reply; 10+ messages in thread
From: Kenta Tada @ 2022-01-19 13:12 UTC (permalink / raw)
To: andrii, bpf
Cc: ast, daniel, kafai, songliubraving, yhs, john.fastabend, kpsingh,
Kenta Tada
Currently, rcx is read as the fourth parameter of syscall on x86_64.
But x86_64 Linux System Call convention uses r10 actually.
This commit adds the wrapper for users who want to access to
syscall params to analyze the user space.
Signed-off-by: Kenta Tada <Kenta.Tada@sony.com>
---
tools/lib/bpf/bpf_tracing.h | 34 ++++++++++++++++++++++++++++++++++
1 file changed, 34 insertions(+)
diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h
index 90f56b0f585f..81673a24973e 100644
--- a/tools/lib/bpf/bpf_tracing.h
+++ b/tools/lib/bpf/bpf_tracing.h
@@ -70,6 +70,7 @@
#define __PT_PARM2_REG si
#define __PT_PARM3_REG dx
#define __PT_PARM4_REG cx
+#define __PT_PARM4_REG_SYSCALL r10 /* syscall uses r10 */
#define __PT_PARM5_REG r8
#define __PT_RET_REG sp
#define __PT_FP_REG bp
@@ -99,6 +100,7 @@
#define __PT_PARM2_REG rsi
#define __PT_PARM3_REG rdx
#define __PT_PARM4_REG rcx
+#define __PT_PARM4_REG_SYSCALL r10 /* syscall uses r10 */
#define __PT_PARM5_REG r8
#define __PT_RET_REG rsp
#define __PT_FP_REG rbp
@@ -263,6 +265,26 @@ struct pt_regs;
#endif
+#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1(x)
+#define PT_REGS_PARM2_SYSCALL(x) PT_REGS_PARM2(x)
+#define PT_REGS_PARM3_SYSCALL(x) PT_REGS_PARM3(x)
+#ifdef __PT_PARM4_REG_SYSCALL
+#define PT_REGS_PARM4_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM4_REG_SYSCALL)
+#else /* __PT_PARM4_REG_SYSCALL */
+#define PT_REGS_PARM4_SYSCALL(x) PT_REGS_PARM4(x)
+#endif
+#define PT_REGS_PARM5_SYSCALL(x) PT_REGS_PARM5(x)
+
+#define PT_REGS_PARM1_CORE_SYSCALL(x) PT_REGS_PARM1_CORE(x)
+#define PT_REGS_PARM2_CORE_SYSCALL(x) PT_REGS_PARM2_CORE(x)
+#define PT_REGS_PARM3_CORE_SYSCALL(x) PT_REGS_PARM3_CORE(x)
+#ifdef __PT_PARM4_REG_SYSCALL
+#define PT_REGS_PARM4_CORE_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM4_REG_SYSCALL)
+#else /* __PT_PARM4_REG_SYSCALL */
+#define PT_REGS_PARM4_CORE_SYSCALL(x) PT_REGS_PARM4_CORE(x)
+#endif
+#define PT_REGS_PARM5_CORE_SYSCALL(x) PT_REGS_PARM5_CORE(x)
+
#else /* defined(bpf_target_defined) */
#define PT_REGS_PARM1(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
@@ -290,6 +312,18 @@ struct pt_regs;
#define BPF_KPROBE_READ_RET_IP(ip, ctx) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
#define BPF_KRETPROBE_READ_RET_IP(ip, ctx) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM1_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM2_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM3_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM4_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM5_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+
+#define PT_REGS_PARM1_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM2_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM3_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM4_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM5_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+
#endif /* defined(bpf_target_defined) */
#ifndef ___bpf_concat
--
2.32.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v4 3/3] libbpf: Add a test to confirm PT_REGS_PARM4_SYSCALL
2022-01-19 13:12 [PATCH v4 0/3] Fix the incorrect register read for syscalls on x86_64 Kenta Tada
2022-01-19 13:12 ` [PATCH v4 1/3] libbpf: Extract syscall wrapper Kenta Tada
2022-01-19 13:12 ` [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64 Kenta Tada
@ 2022-01-19 13:12 ` Kenta Tada
2022-01-19 18:37 ` Andrii Nakryiko
2 siblings, 1 reply; 10+ messages in thread
From: Kenta Tada @ 2022-01-19 13:12 UTC (permalink / raw)
To: andrii, bpf
Cc: ast, daniel, kafai, songliubraving, yhs, john.fastabend, kpsingh,
Kenta Tada
Add a selftest to verify the behavior of PT_REGS_xxx.
Signed-off-by: Kenta Tada <Kenta.Tada@sony.com>
---
.../bpf/prog_tests/test_bpf_syscall_macro.c | 49 ++++++++++++++++++
.../selftests/bpf/progs/bpf_syscall_macro.c | 51 +++++++++++++++++++
2 files changed, 100 insertions(+)
create mode 100644 tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c
create mode 100644 tools/testing/selftests/bpf/progs/bpf_syscall_macro.c
diff --git a/tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c b/tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c
new file mode 100644
index 000000000000..2f725393195b
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c
@@ -0,0 +1,49 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright 2022 Sony Group Corporation */
+#include <sys/prctl.h>
+#include <test_progs.h>
+#include "bpf_syscall_macro.skel.h"
+
+//void serial_bpf_syscall_macro(void)
+void test_bpf_syscall_macro(void)
+{
+ struct bpf_syscall_macro *skel = NULL;
+ int err;
+ int exp_arg1 = 1001;
+ unsigned long exp_arg2 = 12;
+ unsigned long exp_arg3 = 13;
+ unsigned long exp_arg4 = 14;
+ unsigned long exp_arg5 = 15;
+
+ /* check whether it can open program */
+ skel = bpf_syscall_macro__open();
+ if (!ASSERT_OK_PTR(skel, "bpf_syscall_macro__open"))
+ return;
+
+ skel->rodata->filter_pid = getpid();
+
+ /* check whether it can load program */
+ err = bpf_syscall_macro__load(skel);
+ if (!ASSERT_OK(err, "bpf_syscall_macro__load"))
+ goto cleanup;
+
+ /* check whether it can attach kprobe */
+ err = bpf_syscall_macro__attach(skel);
+ if (!ASSERT_OK(err, "bpf_syscall_macro__attach"))
+ goto cleanup;
+
+ /* check whether args of syscall are copied correctly */
+ prctl(exp_arg1, exp_arg2, exp_arg3, exp_arg4, exp_arg5);
+ ASSERT_EQ(skel->bss->arg1, exp_arg1, "syscall_arg1");
+ ASSERT_EQ(skel->bss->arg2, exp_arg2, "syscall_arg2");
+ ASSERT_EQ(skel->bss->arg3, exp_arg3, "syscall_arg3");
+ /* it cannot copy arg4 when uses PT_REGS_PARM4 on x86_64 */
+#ifdef __x86_64__
+ ASSERT_NEQ(skel->bss->arg4_cx, exp_arg4, "syscall_arg4_from_cx");
+#endif
+ ASSERT_EQ(skel->bss->arg4, exp_arg4, "syscall_arg4");
+ ASSERT_EQ(skel->bss->arg5, exp_arg5, "syscall_arg5");
+
+cleanup:
+ bpf_syscall_macro__destroy(skel);
+}
diff --git a/tools/testing/selftests/bpf/progs/bpf_syscall_macro.c b/tools/testing/selftests/bpf/progs/bpf_syscall_macro.c
new file mode 100644
index 000000000000..5a7063de27c3
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/bpf_syscall_macro.c
@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright 2022 Sony Group Corporation */
+#include <linux/bpf.h>
+#include <linux/ptrace.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+#include "bpf_misc.h"
+
+int arg1 = 0;
+unsigned long arg2 = 0;
+unsigned long arg3 = 0;
+unsigned long arg4_cx = 0;
+unsigned long arg4 = 0;
+unsigned long arg5 = 0;
+
+const volatile pid_t filter_pid = 0;
+
+SEC("kprobe/" SYS_PREFIX "sys_prctl")
+int BPF_KPROBE(handle_sys_prctl)
+{
+ struct pt_regs *real_regs;
+ int orig_arg1;
+ unsigned long orig_arg2, orig_arg3, orig_arg4_cx, orig_arg4, orig_arg5;
+ pid_t pid = bpf_get_current_pid_tgid() >> 32;
+
+ if (pid != filter_pid)
+ return 0;
+
+ real_regs = (struct pt_regs *)PT_REGS_PARM1(ctx);
+ bpf_probe_read_kernel(&orig_arg1, sizeof(orig_arg1), &PT_REGS_PARM1_SYSCALL(real_regs));
+ bpf_probe_read_kernel(&orig_arg2, sizeof(orig_arg2), &PT_REGS_PARM2_SYSCALL(real_regs));
+ bpf_probe_read_kernel(&orig_arg3, sizeof(orig_arg3), &PT_REGS_PARM3_SYSCALL(real_regs));
+ bpf_probe_read_kernel(&orig_arg4_cx, sizeof(orig_arg4_cx), &PT_REGS_PARM4(real_regs));
+ bpf_probe_read_kernel(&orig_arg4, sizeof(orig_arg4), &PT_REGS_PARM4_SYSCALL(real_regs));
+ bpf_probe_read_kernel(&orig_arg5, sizeof(orig_arg5), &PT_REGS_PARM5_SYSCALL(real_regs));
+
+ /* copy all actual args and the wrong arg4 on x86_64 */
+ arg1 = orig_arg1;
+ arg2 = orig_arg2;
+ arg3 = orig_arg3;
+ arg4_cx = orig_arg4_cx;
+ arg4 = orig_arg4;
+ arg5 = orig_arg5;
+
+ return 0;
+}
+
+char _license[] SEC("license") = "GPL";
--
2.32.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64
2022-01-19 13:12 ` [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64 Kenta Tada
@ 2022-01-19 18:33 ` Andrii Nakryiko
2022-01-20 22:56 ` Kenta.Tada
0 siblings, 1 reply; 10+ messages in thread
From: Andrii Nakryiko @ 2022-01-19 18:33 UTC (permalink / raw)
To: Kenta Tada
Cc: Andrii Nakryiko, bpf, Alexei Starovoitov, Daniel Borkmann,
Martin Lau, Song Liu, Yonghong Song, john fastabend, KP Singh
On Wed, Jan 19, 2022 at 5:13 AM Kenta Tada <Kenta.Tada@sony.com> wrote:
>
> Currently, rcx is read as the fourth parameter of syscall on x86_64.
> But x86_64 Linux System Call convention uses r10 actually.
> This commit adds the wrapper for users who want to access to
> syscall params to analyze the user space.
>
> Signed-off-by: Kenta Tada <Kenta.Tada@sony.com>
> ---
> tools/lib/bpf/bpf_tracing.h | 34 ++++++++++++++++++++++++++++++++++
> 1 file changed, 34 insertions(+)
>
> diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h
> index 90f56b0f585f..81673a24973e 100644
> --- a/tools/lib/bpf/bpf_tracing.h
> +++ b/tools/lib/bpf/bpf_tracing.h
> @@ -70,6 +70,7 @@
> #define __PT_PARM2_REG si
> #define __PT_PARM3_REG dx
> #define __PT_PARM4_REG cx
> +#define __PT_PARM4_REG_SYSCALL r10 /* syscall uses r10 */
> #define __PT_PARM5_REG r8
> #define __PT_RET_REG sp
> #define __PT_FP_REG bp
> @@ -99,6 +100,7 @@
> #define __PT_PARM2_REG rsi
> #define __PT_PARM3_REG rdx
> #define __PT_PARM4_REG rcx
> +#define __PT_PARM4_REG_SYSCALL r10 /* syscall uses r10 */
> #define __PT_PARM5_REG r8
> #define __PT_RET_REG rsp
> #define __PT_FP_REG rbp
> @@ -263,6 +265,26 @@ struct pt_regs;
>
> #endif
>
> +#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1(x)
> +#define PT_REGS_PARM2_SYSCALL(x) PT_REGS_PARM2(x)
> +#define PT_REGS_PARM3_SYSCALL(x) PT_REGS_PARM3(x)
> +#ifdef __PT_PARM4_REG_SYSCALL
> +#define PT_REGS_PARM4_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM4_REG_SYSCALL)
> +#else /* __PT_PARM4_REG_SYSCALL */
> +#define PT_REGS_PARM4_SYSCALL(x) PT_REGS_PARM4(x)
> +#endif
> +#define PT_REGS_PARM5_SYSCALL(x) PT_REGS_PARM5(x)
> +
> +#define PT_REGS_PARM1_CORE_SYSCALL(x) PT_REGS_PARM1_CORE(x)
> +#define PT_REGS_PARM2_CORE_SYSCALL(x) PT_REGS_PARM2_CORE(x)
> +#define PT_REGS_PARM3_CORE_SYSCALL(x) PT_REGS_PARM3_CORE(x)
> +#ifdef __PT_PARM4_REG_SYSCALL
> +#define PT_REGS_PARM4_CORE_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM4_REG_SYSCALL)
did you check PT_REGS_PARM4_CORE() definition? This should be
BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM4_REG_SYSCALL)
> +#else /* __PT_PARM4_REG_SYSCALL */
> +#define PT_REGS_PARM4_CORE_SYSCALL(x) PT_REGS_PARM4_CORE(x)
> +#endif
> +#define PT_REGS_PARM5_CORE_SYSCALL(x) PT_REGS_PARM5_CORE(x)
> +
> #else /* defined(bpf_target_defined) */
>
> #define PT_REGS_PARM1(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> @@ -290,6 +312,18 @@ struct pt_regs;
> #define BPF_KPROBE_READ_RET_IP(ip, ctx) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> #define BPF_KRETPROBE_READ_RET_IP(ip, ctx) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
>
> +#define PT_REGS_PARM1_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM2_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM3_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM4_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM5_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +
> +#define PT_REGS_PARM1_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM2_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM3_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM4_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +#define PT_REGS_PARM5_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
> +
> #endif /* defined(bpf_target_defined) */
>
> #ifndef ___bpf_concat
> --
> 2.32.0
>
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH v4 3/3] libbpf: Add a test to confirm PT_REGS_PARM4_SYSCALL
2022-01-19 13:12 ` [PATCH v4 3/3] libbpf: Add a test to confirm PT_REGS_PARM4_SYSCALL Kenta Tada
@ 2022-01-19 18:37 ` Andrii Nakryiko
0 siblings, 0 replies; 10+ messages in thread
From: Andrii Nakryiko @ 2022-01-19 18:37 UTC (permalink / raw)
To: Kenta Tada
Cc: Andrii Nakryiko, bpf, Alexei Starovoitov, Daniel Borkmann,
Martin Lau, Song Liu, Yonghong Song, john fastabend, KP Singh
On Wed, Jan 19, 2022 at 5:14 AM Kenta Tada <Kenta.Tada@sony.com> wrote:
>
> Add a selftest to verify the behavior of PT_REGS_xxx.
>
> Signed-off-by: Kenta Tada <Kenta.Tada@sony.com>
> ---
> .../bpf/prog_tests/test_bpf_syscall_macro.c | 49 ++++++++++++++++++
> .../selftests/bpf/progs/bpf_syscall_macro.c | 51 +++++++++++++++++++
> 2 files changed, 100 insertions(+)
> create mode 100644 tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c
> create mode 100644 tools/testing/selftests/bpf/progs/bpf_syscall_macro.c
>
> diff --git a/tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c b/tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c
> new file mode 100644
> index 000000000000..2f725393195b
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/prog_tests/test_bpf_syscall_macro.c
> @@ -0,0 +1,49 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright 2022 Sony Group Corporation */
> +#include <sys/prctl.h>
> +#include <test_progs.h>
> +#include "bpf_syscall_macro.skel.h"
> +
> +//void serial_bpf_syscall_macro(void)
leftover
> +void test_bpf_syscall_macro(void)
> +{
> + struct bpf_syscall_macro *skel = NULL;
> + int err;
> + int exp_arg1 = 1001;
> + unsigned long exp_arg2 = 12;
> + unsigned long exp_arg3 = 13;
> + unsigned long exp_arg4 = 14;
> + unsigned long exp_arg5 = 15;
> +
> + /* check whether it can open program */
> + skel = bpf_syscall_macro__open();
> + if (!ASSERT_OK_PTR(skel, "bpf_syscall_macro__open"))
> + return;
> +
> + skel->rodata->filter_pid = getpid();
> +
> + /* check whether it can load program */
> + err = bpf_syscall_macro__load(skel);
> + if (!ASSERT_OK(err, "bpf_syscall_macro__load"))
> + goto cleanup;
> +
> + /* check whether it can attach kprobe */
> + err = bpf_syscall_macro__attach(skel);
> + if (!ASSERT_OK(err, "bpf_syscall_macro__attach"))
> + goto cleanup;
> +
> + /* check whether args of syscall are copied correctly */
> + prctl(exp_arg1, exp_arg2, exp_arg3, exp_arg4, exp_arg5);
> + ASSERT_EQ(skel->bss->arg1, exp_arg1, "syscall_arg1");
> + ASSERT_EQ(skel->bss->arg2, exp_arg2, "syscall_arg2");
> + ASSERT_EQ(skel->bss->arg3, exp_arg3, "syscall_arg3");
> + /* it cannot copy arg4 when uses PT_REGS_PARM4 on x86_64 */
> +#ifdef __x86_64__
> + ASSERT_NEQ(skel->bss->arg4_cx, exp_arg4, "syscall_arg4_from_cx");
> +#endif
else ASSERT_EQ(arg4_cx and exp_arg4) ?
> + ASSERT_EQ(skel->bss->arg4, exp_arg4, "syscall_arg4");
> + ASSERT_EQ(skel->bss->arg5, exp_arg5, "syscall_arg5");
> +
> +cleanup:
> + bpf_syscall_macro__destroy(skel);
> +}
> diff --git a/tools/testing/selftests/bpf/progs/bpf_syscall_macro.c b/tools/testing/selftests/bpf/progs/bpf_syscall_macro.c
> new file mode 100644
> index 000000000000..5a7063de27c3
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/progs/bpf_syscall_macro.c
> @@ -0,0 +1,51 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/* Copyright 2022 Sony Group Corporation */
> +#include <linux/bpf.h>
> +#include <linux/ptrace.h>
> +#include <sys/types.h>
> +#include <unistd.h>
Use #include <vmlinux.h> instead, please
> +
> +#include <bpf/bpf_helpers.h>
> +#include <bpf/bpf_tracing.h>
> +#include "bpf_misc.h"
> +
> +int arg1 = 0;
> +unsigned long arg2 = 0;
> +unsigned long arg3 = 0;
> +unsigned long arg4_cx = 0;
> +unsigned long arg4 = 0;
> +unsigned long arg5 = 0;
> +
> +const volatile pid_t filter_pid = 0;
> +
> +SEC("kprobe/" SYS_PREFIX "sys_prctl")
> +int BPF_KPROBE(handle_sys_prctl)
> +{
> + struct pt_regs *real_regs;
> + int orig_arg1;
> + unsigned long orig_arg2, orig_arg3, orig_arg4_cx, orig_arg4, orig_arg5;
> + pid_t pid = bpf_get_current_pid_tgid() >> 32;
> +
> + if (pid != filter_pid)
> + return 0;
> +
> + real_regs = (struct pt_regs *)PT_REGS_PARM1(ctx);
> + bpf_probe_read_kernel(&orig_arg1, sizeof(orig_arg1), &PT_REGS_PARM1_SYSCALL(real_regs));
orig_arg1 = PT_REGS_PARM1_CORE_SYSCALL(real_regs);
and same for others (the whole point of _CORE_SYSCALL macros!)
> + bpf_probe_read_kernel(&orig_arg2, sizeof(orig_arg2), &PT_REGS_PARM2_SYSCALL(real_regs));
> + bpf_probe_read_kernel(&orig_arg3, sizeof(orig_arg3), &PT_REGS_PARM3_SYSCALL(real_regs));
> + bpf_probe_read_kernel(&orig_arg4_cx, sizeof(orig_arg4_cx), &PT_REGS_PARM4(real_regs));
> + bpf_probe_read_kernel(&orig_arg4, sizeof(orig_arg4), &PT_REGS_PARM4_SYSCALL(real_regs));
> + bpf_probe_read_kernel(&orig_arg5, sizeof(orig_arg5), &PT_REGS_PARM5_SYSCALL(real_regs));
> +
> + /* copy all actual args and the wrong arg4 on x86_64 */
> + arg1 = orig_arg1;
> + arg2 = orig_arg2;
> + arg3 = orig_arg3;
> + arg4_cx = orig_arg4_cx;
> + arg4 = orig_arg4;
> + arg5 = orig_arg5;
> +
> + return 0;
> +}
> +
> +char _license[] SEC("license") = "GPL";
> --
> 2.32.0
>
^ permalink raw reply [flat|nested] 10+ messages in thread
* RE: [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64
2022-01-19 18:33 ` Andrii Nakryiko
@ 2022-01-20 22:56 ` Kenta.Tada
2022-01-20 23:01 ` Andrii Nakryiko
0 siblings, 1 reply; 10+ messages in thread
From: Kenta.Tada @ 2022-01-20 22:56 UTC (permalink / raw)
To: andrii.nakryiko
Cc: andrii, bpf, ast, daniel, kafai, songliubraving, yhs,
john.fastabend, kpsingh
>did you check PT_REGS_PARM4_CORE() definition? This should be
In my local test, this wrong code can pass the correct arg4 because the test just checks the value.
Anyway I should attach the test for CORE variants at first.
Sorry. I'll fix the issues and add tests for CORE variants this weekend.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64
2022-01-20 22:56 ` Kenta.Tada
@ 2022-01-20 23:01 ` Andrii Nakryiko
2022-01-20 23:34 ` Kenta.Tada
0 siblings, 1 reply; 10+ messages in thread
From: Andrii Nakryiko @ 2022-01-20 23:01 UTC (permalink / raw)
To: Kenta Tada
Cc: Andrii Nakryiko, bpf, Alexei Starovoitov, Daniel Borkmann,
Martin Lau, Song Liu, Yonghong Song, john fastabend, KP Singh
On Thu, Jan 20, 2022 at 2:57 PM <Kenta.Tada@sony.com> wrote:
>
> >did you check PT_REGS_PARM4_CORE() definition? This should be
>
> In my local test, this wrong code can pass the correct arg4 because the test just checks the value.
The biggest problem is the lack of bpf_probe_read_kernel(). Your
definition does direct memory read which won't work if pt_regs is not
an input context to the BPF program. Which is exactly the case for
syscalls.
> Anyway I should attach the test for CORE variants at first.
> Sorry. I'll fix the issues and add tests for CORE variants this weekend.
^ permalink raw reply [flat|nested] 10+ messages in thread
* RE: [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64
2022-01-20 23:01 ` Andrii Nakryiko
@ 2022-01-20 23:34 ` Kenta.Tada
2022-01-20 23:38 ` Kenta.Tada
0 siblings, 1 reply; 10+ messages in thread
From: Kenta.Tada @ 2022-01-20 23:34 UTC (permalink / raw)
To: andrii.nakryiko
Cc: andrii, bpf, ast, daniel, kafai, songliubraving, yhs,
john.fastabend, kpsingh
>>
>> >did you check PT_REGS_PARM4_CORE() definition? This should be
>>
>> In my local test, this wrong code can pass the correct arg4 because the test just checks the value.
>
>The biggest problem is the lack of bpf_probe_read_kernel(). Your definition does direct memory read which won't work if pt_regs is not an input context to the BPF program. Which is exactly the case for syscalls.
Yes.
I'll use BPF_CORE_READ() for PT_REGS_PARM4_CORE() not to read direct memory.
^ permalink raw reply [flat|nested] 10+ messages in thread
* RE: [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64
2022-01-20 23:34 ` Kenta.Tada
@ 2022-01-20 23:38 ` Kenta.Tada
0 siblings, 0 replies; 10+ messages in thread
From: Kenta.Tada @ 2022-01-20 23:38 UTC (permalink / raw)
To: andrii.nakryiko
Cc: andrii, bpf, ast, daniel, kafai, songliubraving, yhs,
john.fastabend, kpsingh
>>>
>>> >did you check PT_REGS_PARM4_CORE() definition? This should be
>>>
>>> In my local test, this wrong code can pass the correct arg4 because the test just checks the value.
>>
>>The biggest problem is the lack of bpf_probe_read_kernel(). Your definition does direct memory read which won't work if pt_regs is not an input context to the BPF program. Which is exactly the case for syscalls.
>
>Yes.
>I'll use BPF_CORE_READ() for PT_REGS_PARM4_CORE() not to read direct memory.
Sorry, Not PT_REGS_PARM4_CORE() but PT_REGS_PARM4_CORE_SYSCALL().
And I cofirmed the definition of PT_REGS_PARM4_CORE() and BPF_CORE_READ().
I'll use BPF_CORE_READ() for PT_REGS_PARM4_CORE_SYSCALL().
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2022-01-20 23:39 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-19 13:12 [PATCH v4 0/3] Fix the incorrect register read for syscalls on x86_64 Kenta Tada
2022-01-19 13:12 ` [PATCH v4 1/3] libbpf: Extract syscall wrapper Kenta Tada
2022-01-19 13:12 ` [PATCH v4 2/3] libbpf: Fix the incorrect register read for syscalls on x86_64 Kenta Tada
2022-01-19 18:33 ` Andrii Nakryiko
2022-01-20 22:56 ` Kenta.Tada
2022-01-20 23:01 ` Andrii Nakryiko
2022-01-20 23:34 ` Kenta.Tada
2022-01-20 23:38 ` Kenta.Tada
2022-01-19 13:12 ` [PATCH v4 3/3] libbpf: Add a test to confirm PT_REGS_PARM4_SYSCALL Kenta Tada
2022-01-19 18:37 ` Andrii Nakryiko
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).