The orig_a0 is missing in struct user_regs_struct of riscv, and there is no way to add it without breaking UAPI. (See Link tag below)
Like NT_ARM_SYSTEM_CALL do, we add a new regset name NT_RISCV_ORIG_A0 to access original a0 register from userspace via ptrace API.
Link: https://lore.kernel.org/all/59505464-c84a-403d-972f-d4b2055eeaac@gmail.com/
Signed-off-by: Celeste Liu uwu@coelacanthus.name --- Changes in v4: - Fix a copy paste error in selftest. (Forget to commit...) - Link to v3: https://lore.kernel.org/r/20241226-riscv-new-regset-v3-0-f5b96465826b@coelac...
Changes in v3: - Use return 0 directly for readability. - Fix test for modify a0. - Add Fixes: tag - Remove useless Cc: stable. - Selftest will check both a0 and orig_a0, but depends on the correctness of PTRACE_GET_SYSCALL_INFO. - Link to v2: https://lore.kernel.org/r/20241203-riscv-new-regset-v2-0-d37da8c0cba6@coelac...
Changes in v2: - Fix integer width. - Add selftest. - Link to v1: https://lore.kernel.org/r/20241201-riscv-new-regset-v1-1-c83c58abcc7b@coelac...
--- Celeste Liu (2): riscv/ptrace: add new regset to access original a0 register riscv: selftests: Add a ptrace test to verify syscall parameter modification
arch/riscv/kernel/ptrace.c | 32 ++++++ include/uapi/linux/elf.h | 1 + tools/testing/selftests/riscv/abi/.gitignore | 1 + tools/testing/selftests/riscv/abi/Makefile | 5 +- tools/testing/selftests/riscv/abi/ptrace.c | 151 +++++++++++++++++++++++++++ 5 files changed, 189 insertions(+), 1 deletion(-) --- base-commit: 0e287d31b62bb53ad81d5e59778384a40f8b6f56 change-id: 20241201-riscv-new-regset-d529b952ad0d
Best regards,
The orig_a0 is missing in struct user_regs_struct of riscv, and there is no way to add it without breaking UAPI. (See Link tag below)
Like NT_ARM_SYSTEM_CALL do, we add a new regset name NT_RISCV_ORIG_A0 to access original a0 register from userspace via ptrace API.
Fixes: e2c0cdfba7f6 ("RISC-V: User-facing API") Link: https://lore.kernel.org/all/59505464-c84a-403d-972f-d4b2055eeaac@gmail.com/ Cc: stable@vger.kernel.org Reviewed-by: Björn Töpel bjorn@rivosinc.com Signed-off-by: Celeste Liu uwu@coelacanthus.name --- arch/riscv/kernel/ptrace.c | 32 ++++++++++++++++++++++++++++++++ include/uapi/linux/elf.h | 1 + 2 files changed, 33 insertions(+)
diff --git a/arch/riscv/kernel/ptrace.c b/arch/riscv/kernel/ptrace.c index ea67e9fb7a583683b922fe2c017ea61f3bc848db..ef9ab74c8575a5c440155973b1c625c06a867c97 100644 --- a/arch/riscv/kernel/ptrace.c +++ b/arch/riscv/kernel/ptrace.c @@ -31,6 +31,7 @@ enum riscv_regset { #ifdef CONFIG_RISCV_ISA_SUPM REGSET_TAGGED_ADDR_CTRL, #endif + REGSET_ORIG_A0, };
static int riscv_gpr_get(struct task_struct *target, @@ -184,6 +185,29 @@ static int tagged_addr_ctrl_set(struct task_struct *target, } #endif
+static int riscv_orig_a0_get(struct task_struct *target, + const struct user_regset *regset, + struct membuf to) +{ + return membuf_store(&to, task_pt_regs(target)->orig_a0); +} + +static int riscv_orig_a0_set(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + unsigned long orig_a0 = task_pt_regs(target)->orig_a0; + int ret; + + ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &orig_a0, 0, -1); + if (ret) + return ret; + + task_pt_regs(target)->orig_a0 = orig_a0; + return 0; +} + static const struct user_regset riscv_user_regset[] = { [REGSET_X] = { .core_note_type = NT_PRSTATUS, @@ -224,6 +248,14 @@ static const struct user_regset riscv_user_regset[] = { .set = tagged_addr_ctrl_set, }, #endif + [REGSET_ORIG_A0] = { + .core_note_type = NT_RISCV_ORIG_A0, + .n = 1, + .size = sizeof(elf_greg_t), + .align = sizeof(elf_greg_t), + .regset_get = riscv_orig_a0_get, + .set = riscv_orig_a0_set, + }, };
static const struct user_regset_view riscv_user_native_view = { diff --git a/include/uapi/linux/elf.h b/include/uapi/linux/elf.h index b44069d29cecc0f9de90ee66bfffd2137f4275a8..390060229601631da2fb27030d9fa2142e676c14 100644 --- a/include/uapi/linux/elf.h +++ b/include/uapi/linux/elf.h @@ -452,6 +452,7 @@ typedef struct elf64_shdr { #define NT_RISCV_CSR 0x900 /* RISC-V Control and Status Registers */ #define NT_RISCV_VECTOR 0x901 /* RISC-V vector registers */ #define NT_RISCV_TAGGED_ADDR_CTRL 0x902 /* RISC-V tagged address control (prctl()) */ +#define NT_RISCV_ORIG_A0 0x903 /* RISC-V original a0 register */ #define NT_LOONGARCH_CPUCFG 0xa00 /* LoongArch CPU config registers */ #define NT_LOONGARCH_CSR 0xa01 /* LoongArch control and status registers */ #define NT_LOONGARCH_LSX 0xa02 /* LoongArch Loongson SIMD Extension registers */
This test checks that orig_a0 allows a syscall argument to be modified, and that changing a0 does not change the syscall argument.
Co-developed-by: Quan Zhou zhouquan@iscas.ac.cn Signed-off-by: Quan Zhou zhouquan@iscas.ac.cn Co-developed-by: Charlie Jenkins charlie@rivosinc.com Signed-off-by: Charlie Jenkins charlie@rivosinc.com Reviewed-by: Björn Töpel bjorn@rivosinc.com Signed-off-by: Celeste Liu uwu@coelacanthus.name --- tools/testing/selftests/riscv/abi/.gitignore | 1 + tools/testing/selftests/riscv/abi/Makefile | 5 +- tools/testing/selftests/riscv/abi/ptrace.c | 151 +++++++++++++++++++++++++++ 3 files changed, 156 insertions(+), 1 deletion(-)
diff --git a/tools/testing/selftests/riscv/abi/.gitignore b/tools/testing/selftests/riscv/abi/.gitignore index b38358f91c4d2240ae64892871d9ca98bda1ae58..378c605919a3b3d58eec2701eb7af430cfe315d6 100644 --- a/tools/testing/selftests/riscv/abi/.gitignore +++ b/tools/testing/selftests/riscv/abi/.gitignore @@ -1 +1,2 @@ pointer_masking +ptrace diff --git a/tools/testing/selftests/riscv/abi/Makefile b/tools/testing/selftests/riscv/abi/Makefile index ed82ff9c664e7eb3f760cbab81fb957ff72579c5..3f74d059dfdcbce4d45d8ff618781ccea1419061 100644 --- a/tools/testing/selftests/riscv/abi/Makefile +++ b/tools/testing/selftests/riscv/abi/Makefile @@ -2,9 +2,12 @@
CFLAGS += -I$(top_srcdir)/tools/include
-TEST_GEN_PROGS := pointer_masking +TEST_GEN_PROGS := pointer_masking ptrace
include ../../lib.mk
$(OUTPUT)/pointer_masking: pointer_masking.c $(CC) -static -o$@ $(CFLAGS) $(LDFLAGS) $^ + +$(OUTPUT)/ptrace: ptrace.c + $(CC) -static -o$@ $(CFLAGS) $(LDFLAGS) $^ diff --git a/tools/testing/selftests/riscv/abi/ptrace.c b/tools/testing/selftests/riscv/abi/ptrace.c new file mode 100644 index 0000000000000000000000000000000000000000..023695352215bb5de3f91c1a6f5ea3b4f9373ff9 --- /dev/null +++ b/tools/testing/selftests/riscv/abi/ptrace.c @@ -0,0 +1,151 @@ +// SPDX-License-Identifier: GPL-2.0-only +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <unistd.h> +#include <fcntl.h> +#include <signal.h> +#include <errno.h> +#include <sys/types.h> +#include <sys/ptrace.h> +#include <sys/stat.h> +#include <sys/user.h> +#include <sys/wait.h> +#include <sys/uio.h> +#include <linux/elf.h> +#include <linux/unistd.h> +#include <linux/ptrace.h> +#include <asm/ptrace.h> + +#include "../../kselftest_harness.h" + +#define ORIG_A0_MODIFY 0x01 +#define A0_MODIFY 0x02 +#define A0_OLD 0x03 +#define A0_NEW 0x04 + +struct a0_regs { + __s64 orig_a0; + __u64 a0; +}; + +#define perr_and_exit(fmt, ...) \ + ({ \ + char buf[256]; \ + snprintf(buf, sizeof(buf), "%s:%d:" fmt ": %m\n", \ + __func__, __LINE__, ##__VA_ARGS__); \ + ksft_exit_fail_perror(buf); \ + }) + +static inline void resume_and_wait_tracee(pid_t pid, int flag) +{ + int status; + + if (ptrace(flag, pid, 0, 0)) + perr_and_exit("failed to resume the tracee %d\n", pid); + + if (waitpid(pid, &status, 0) != pid) + perr_and_exit("failed to wait for the tracee %d\n", pid); +} + +static void ptrace_test(int opt, struct a0_regs *result) +{ + int status; + pid_t pid; + struct user_regs_struct regs; + struct iovec iov = { + .iov_base = ®s, + .iov_len = sizeof(regs), + }; + + unsigned long orig_a0; + struct iovec a0_iov = { + .iov_base = &orig_a0, + .iov_len = sizeof(orig_a0), + }; + struct ptrace_syscall_info syscall_info_entry, syscall_info_exit; + + pid = fork(); + if (pid == 0) { + /* Mark oneself being traced */ + long val = ptrace(PTRACE_TRACEME, 0, 0, 0); + + if (val) + perr_and_exit("failed to request for tracer to trace me: %ld\n", val); + + kill(getpid(), SIGSTOP); + + /* Perform exit syscall that will be intercepted */ + exit(A0_OLD); + } + + if (pid < 0) + ksft_exit_fail_perror("failed to fork"); + + if (waitpid(pid, &status, 0) != pid) + perr_and_exit("failed to wait for the tracee %d\n", pid); + + /* Stop at the entry point of the syscall */ + resume_and_wait_tracee(pid, PTRACE_SYSCALL); + + /* Check tracee regs before the syscall */ + if (ptrace(PTRACE_GETREGSET, pid, NT_PRSTATUS, &iov)) + perr_and_exit("failed to get tracee registers\n"); + if (ptrace(PTRACE_GETREGSET, pid, NT_RISCV_ORIG_A0, &a0_iov)) + perr_and_exit("failed to get tracee registers\n"); + if (orig_a0 != A0_OLD) + perr_and_exit("unexpected orig_a0: 0x%lx\n", orig_a0); + + /* Modify a0/orig_a0 for the syscall */ + switch (opt) { + case A0_MODIFY: + regs.a0 = A0_NEW; + break; + case ORIG_A0_MODIFY: + orig_a0 = A0_NEW; + break; + } + + if (ptrace(PTRACE_SETREGSET, pid, NT_PRSTATUS, &iov)) + perr_and_exit("failed to set tracee registers\n"); + if (ptrace(PTRACE_SETREGSET, pid, NT_RISCV_ORIG_A0, &a0_iov)) + perr_and_exit("failed to set tracee registers\n"); + + if (ptrace(PTRACE_GET_SYSCALL_INFO, pid, PTRACE_SYSCALL_INFO_ENTRY, &syscall_info_entry)) + perr_and_exit("failed to get syscall info of entry\n"); + result->orig_a0 = syscall_info_entry->entry.args[0]; + if (ptrace(PTRACE_GET_SYSCALL_INFO, pid, PTRACE_SYSCALL_INFO_EXIT, &syscall_info_exit)) + perr_and_exit("failed to get syscall info of exit\n"); + result->a0 = syscall_info_exit->exit.rval; + + /* Resume the tracee */ + ptrace(PTRACE_CONT, pid, 0, 0); + if (waitpid(pid, &status, 0) != pid) + perr_and_exit("failed to wait for the tracee\n"); + +} + +TEST(ptrace_modify_a0) +{ + struct a0_regs result; + + ptrace_test(A0_MODIFY, &result); + + /* The modification of a0 cannot affect the first argument of the syscall */ + EXPECT_EQ(A0_OLD, result.orig_a0); + EXPECT_EQ(A0_NEW, result.a0); +} + +TEST(ptrace_modify_orig_a0) +{ + struct a0_regs result; + + ptrace_test(ORIG_A0_MODIFY, &result); + + /* Only modify orig_a0 to change the first argument of the syscall */ + EXPECT_EQ(A0_NEW, result.orig_a0); + /* a0 will keep default value, orig_a0 or -ENOSYS, depends on internal. */ + EXPECT_NE(A0_NEW, result.a0); +} + +TEST_HARNESS_MAIN
On Thu, Dec 26, 2024 at 06:52:52PM +0800, Celeste Liu wrote:
This test checks that orig_a0 allows a syscall argument to be modified, and that changing a0 does not change the syscall argument.
Co-developed-by: Quan Zhou zhouquan@iscas.ac.cn Signed-off-by: Quan Zhou zhouquan@iscas.ac.cn Co-developed-by: Charlie Jenkins charlie@rivosinc.com Signed-off-by: Charlie Jenkins charlie@rivosinc.com Reviewed-by: Björn Töpel bjorn@rivosinc.com Signed-off-by: Celeste Liu uwu@coelacanthus.name
[...]
diff --git a/tools/testing/selftests/riscv/abi/ptrace.c b/tools/testing/selftests/riscv/abi/ptrace.c new file mode 100644 index 0000000000000000000000000000000000000000..023695352215bb5de3f91c1a6f5ea3b4f9373ff9 --- /dev/null +++ b/tools/testing/selftests/riscv/abi/ptrace.c
[...]
- if (ptrace(PTRACE_GET_SYSCALL_INFO, pid, PTRACE_SYSCALL_INFO_ENTRY, &syscall_info_entry))
perr_and_exit("failed to get syscall info of entry\n");- result->orig_a0 = syscall_info_entry->entry.args[0];
- if (ptrace(PTRACE_GET_SYSCALL_INFO, pid, PTRACE_SYSCALL_INFO_EXIT, &syscall_info_exit))
perr_and_exit("failed to get syscall info of exit\n");- result->a0 = syscall_info_exit->exit.rval;
I'm sorry but this is not how PTRACE_GET_SYSCALL_INFO should be used.
PTRACE_GET_SYSCALL_INFO operation takes a pointer and a size, and in this example instead of size you pass constants 1 and 2, which essentially means that both syscall_info_entry->entry.args[0] and syscall_info_exit->exit.rval are not going to be assigned and would just contain some garbage from the stack.
Also, PTRACE_GET_SYSCALL_INFO operation returns the number of bytes available to be written by the kernel, which is always nonzero on any PTRACE_GET_SYSCALL_INFO-capable kernel. In other words, this example will always end up with perr_and_exit() call.
I wonder how this test was tested before the submission.
On 2024-12-26 21:35, Dmitry V. Levin wrote:
On Thu, Dec 26, 2024 at 06:52:52PM +0800, Celeste Liu wrote:
This test checks that orig_a0 allows a syscall argument to be modified, and that changing a0 does not change the syscall argument.
Co-developed-by: Quan Zhou zhouquan@iscas.ac.cn Signed-off-by: Quan Zhou zhouquan@iscas.ac.cn Co-developed-by: Charlie Jenkins charlie@rivosinc.com Signed-off-by: Charlie Jenkins charlie@rivosinc.com Reviewed-by: Björn Töpel bjorn@rivosinc.com Signed-off-by: Celeste Liu uwu@coelacanthus.name
[...]
diff --git a/tools/testing/selftests/riscv/abi/ptrace.c b/tools/testing/selftests/riscv/abi/ptrace.c new file mode 100644 index 0000000000000000000000000000000000000000..023695352215bb5de3f91c1a6f5ea3b4f9373ff9 --- /dev/null +++ b/tools/testing/selftests/riscv/abi/ptrace.c
[...]
- if (ptrace(PTRACE_GET_SYSCALL_INFO, pid, PTRACE_SYSCALL_INFO_ENTRY, &syscall_info_entry))
perr_and_exit("failed to get syscall info of entry\n");- result->orig_a0 = syscall_info_entry->entry.args[0];
- if (ptrace(PTRACE_GET_SYSCALL_INFO, pid, PTRACE_SYSCALL_INFO_EXIT, &syscall_info_exit))
perr_and_exit("failed to get syscall info of exit\n");- result->a0 = syscall_info_exit->exit.rval;
I'm sorry but this is not how PTRACE_GET_SYSCALL_INFO should be used.
PTRACE_GET_SYSCALL_INFO operation takes a pointer and a size, and in this example instead of size you pass constants 1 and 2, which essentially means that both syscall_info_entry->entry.args[0] and syscall_info_exit->exit.rval are not going to be assigned and would just contain some garbage from the stack.
Also, PTRACE_GET_SYSCALL_INFO operation returns the number of bytes available to be written by the kernel, which is always nonzero on any PTRACE_GET_SYSCALL_INFO-capable kernel. In other words, this example will always end up with perr_and_exit() call.
I wonder how this test was tested before the submission.
Oops... It seems I forget sync the code to test board so it runs with the old code... The code is completely not tested... I'm so sorry for my mistake.
I will correct it and test it carefully later...
linux-kselftest-mirror@lists.linaro.org
-
Celeste Liu -
Dmitry V. Levin