diff options
Diffstat (limited to 'arch/um/os-Linux/skas/process.c')
-rw-r--r-- | arch/um/os-Linux/skas/process.c | 498 |
1 files changed, 367 insertions, 131 deletions
diff --git a/arch/um/os-Linux/skas/process.c b/arch/um/os-Linux/skas/process.c index f683cfc9e51a..e42ffac23e3c 100644 --- a/arch/um/os-Linux/skas/process.c +++ b/arch/um/os-Linux/skas/process.c @@ -1,5 +1,6 @@ // SPDX-License-Identifier: GPL-2.0 /* + * Copyright (C) 2021 Benjamin Berg <benjamin@sipsolutions.net> * Copyright (C) 2015 Thomas Meyer (thomas@m3y3r.de) * Copyright (C) 2002- 2007 Jeff Dike (jdike@{addtoit,linux.intel}.com) */ @@ -15,6 +16,7 @@ #include <sys/mman.h> #include <sys/wait.h> #include <sys/stat.h> +#include <sys/socket.h> #include <asm/unistd.h> #include <as-layout.h> #include <init.h> @@ -25,8 +27,11 @@ #include <registers.h> #include <skas.h> #include <sysdep/stub.h> +#include <sysdep/mcontext.h> +#include <linux/futex.h> #include <linux/threads.h> #include <timetravel.h> +#include <asm-generic/rwonce.h> #include "../internal.h" int is_skas_winch(int pid, int fd, void *data) @@ -142,6 +147,105 @@ bad_wait: fatal_sigsegv(); } +void wait_stub_done_seccomp(struct mm_id *mm_idp, int running, int wait_sigsys) +{ + struct stub_data *data = (void *)mm_idp->stack; + int ret; + + do { + const char byte = 0; + struct iovec iov = { + .iov_base = (void *)&byte, + .iov_len = sizeof(byte), + }; + union { + char data[CMSG_SPACE(sizeof(mm_idp->syscall_fd_map))]; + struct cmsghdr align; + } ctrl; + struct msghdr msgh = { + .msg_iov = &iov, + .msg_iovlen = 1, + }; + + if (!running) { + if (mm_idp->syscall_fd_num) { + unsigned int fds_size = + sizeof(int) * mm_idp->syscall_fd_num; + struct cmsghdr *cmsg; + + msgh.msg_control = ctrl.data; + msgh.msg_controllen = CMSG_SPACE(fds_size); + cmsg = CMSG_FIRSTHDR(&msgh); + cmsg->cmsg_level = SOL_SOCKET; + cmsg->cmsg_type = SCM_RIGHTS; + cmsg->cmsg_len = CMSG_LEN(fds_size); + memcpy(CMSG_DATA(cmsg), mm_idp->syscall_fd_map, + fds_size); + + CATCH_EINTR(syscall(__NR_sendmsg, mm_idp->sock, + &msgh, 0)); + } + + data->signal = 0; + data->futex = FUTEX_IN_CHILD; + CATCH_EINTR(syscall(__NR_futex, &data->futex, + FUTEX_WAKE, 1, NULL, NULL, 0)); + } + + do { + /* + * We need to check whether the child is still alive + * before and after the FUTEX_WAIT call. Before, in + * case it just died but we still updated data->futex + * to FUTEX_IN_CHILD. And after, in case it died while + * we were waiting (and SIGCHLD woke us up, see the + * IRQ handler in mmu.c). + * + * Either way, if PID is negative, then we have no + * choice but to kill the task. + */ + if (__READ_ONCE(mm_idp->pid) < 0) + goto out_kill; + + ret = syscall(__NR_futex, &data->futex, + FUTEX_WAIT, FUTEX_IN_CHILD, + NULL, NULL, 0); + if (ret < 0 && errno != EINTR && errno != EAGAIN) { + printk(UM_KERN_ERR "%s : FUTEX_WAIT failed, errno = %d\n", + __func__, errno); + goto out_kill; + } + } while (data->futex == FUTEX_IN_CHILD); + + if (__READ_ONCE(mm_idp->pid) < 0) + goto out_kill; + + running = 0; + + /* We may receive a SIGALRM before SIGSYS, iterate again. */ + } while (wait_sigsys && data->signal == SIGALRM); + + if (data->mctx_offset > sizeof(data->sigstack) - sizeof(mcontext_t)) { + printk(UM_KERN_ERR "%s : invalid mcontext offset", __func__); + goto out_kill; + } + + if (wait_sigsys && data->signal != SIGSYS) { + printk(UM_KERN_ERR "%s : expected SIGSYS but got %d", + __func__, data->signal); + goto out_kill; + } + + return; + +out_kill: + printk(UM_KERN_ERR "%s : failed to wait for stub, pid = %d, errno = %d\n", + __func__, mm_idp->pid, errno); + /* This is not true inside start_userspace */ + if (current_mm_id() == mm_idp) + fatal_sigsegv(); +} + extern unsigned long current_stub_stack(void); static void get_skas_faultinfo(int pid, struct faultinfo *fi) @@ -163,12 +267,6 @@ static void get_skas_faultinfo(int pid, struct faultinfo *fi) memcpy(fi, (void *)current_stub_stack(), sizeof(*fi)); } -static void handle_segv(int pid, struct uml_pt_regs *regs) -{ - get_skas_faultinfo(pid, ®s->faultinfo); - segv(regs->faultinfo, 0, 1, NULL); -} - static void handle_trap(int pid, struct uml_pt_regs *regs) { if ((UPT_IP(regs) >= STUB_START) && (UPT_IP(regs) < STUB_END)) @@ -181,50 +279,80 @@ extern char __syscall_stub_start[]; static int stub_exe_fd; -static int userspace_tramp(void *stack) +struct tramp_data { + struct stub_data *stub_data; + /* 0 is inherited, 1 is the kernel side */ + int sockpair[2]; +}; + +#ifndef CLOSE_RANGE_CLOEXEC +#define CLOSE_RANGE_CLOEXEC (1U << 2) +#endif + +static int userspace_tramp(void *data) { + struct tramp_data *tramp_data = data; char *const argv[] = { "uml-userspace", NULL }; - int pipe_fds[2]; unsigned long long offset; struct stub_init_data init_data = { + .seccomp = using_seccomp, .stub_start = STUB_START, - .segv_handler = STUB_CODE + - (unsigned long) stub_segv_handler - - (unsigned long) __syscall_stub_start, }; struct iomem_region *iomem; int ret; + if (using_seccomp) { + init_data.signal_handler = STUB_CODE + + (unsigned long) stub_signal_interrupt - + (unsigned long) __syscall_stub_start; + init_data.signal_restorer = STUB_CODE + + (unsigned long) stub_signal_restorer - + (unsigned long) __syscall_stub_start; + } else { + init_data.signal_handler = STUB_CODE + + (unsigned long) stub_segv_handler - + (unsigned long) __syscall_stub_start; + init_data.signal_restorer = 0; + } + init_data.stub_code_fd = phys_mapping(uml_to_phys(__syscall_stub_start), &offset); init_data.stub_code_offset = MMAP_OFFSET(offset); - init_data.stub_data_fd = phys_mapping(uml_to_phys(stack), &offset); + init_data.stub_data_fd = phys_mapping(uml_to_phys(tramp_data->stub_data), + &offset); init_data.stub_data_offset = MMAP_OFFSET(offset); - /* Set CLOEXEC on all FDs and then unset on all memory related FDs */ - close_range(0, ~0U, CLOSE_RANGE_CLOEXEC); + /* + * Avoid leaking unneeded FDs to the stub by setting CLOEXEC on all FDs + * and then unsetting it on all memory related FDs. + * This is not strictly necessary from a safety perspective. + */ + syscall(__NR_close_range, 0, ~0U, CLOSE_RANGE_CLOEXEC); fcntl(init_data.stub_data_fd, F_SETFD, 0); - for (iomem = iomem_regions; iomem; iomem = iomem->next) - fcntl(iomem->fd, F_SETFD, 0); - /* Create a pipe for init_data (no CLOEXEC) and dup2 to STDIN */ - if (pipe(pipe_fds)) - exit(2); + /* In SECCOMP mode, these FDs are passed when needed */ + if (!using_seccomp) { + for (iomem = iomem_regions; iomem; iomem = iomem->next) + fcntl(iomem->fd, F_SETFD, 0); + } - if (dup2(pipe_fds[0], 0) < 0) + /* dup2 signaling FD/socket to STDIN */ + if (dup2(tramp_data->sockpair[0], 0) < 0) exit(3); - close(pipe_fds[0]); + close(tramp_data->sockpair[0]); /* Write init_data and close write side */ - ret = write(pipe_fds[1], &init_data, sizeof(init_data)); - close(pipe_fds[1]); + ret = write(tramp_data->sockpair[1], &init_data, sizeof(init_data)); + close(tramp_data->sockpair[1]); if (ret != sizeof(init_data)) exit(4); - execveat(stub_exe_fd, "", argv, NULL, AT_EMPTY_PATH); + /* Raw execveat for compatibility with older libc versions */ + syscall(__NR_execveat, stub_exe_fd, (unsigned long)"", + (unsigned long)argv, NULL, AT_EMPTY_PATH); exit(5); } @@ -305,11 +433,12 @@ static int __init init_stub_exe_fd(void) } __initcall(init_stub_exe_fd); +int using_seccomp; int userspace_pid[NR_CPUS]; /** * start_userspace() - prepare a new userspace process - * @stub_stack: pointer to the stub stack. + * @mm_id: The corresponding struct mm_id * * Setups a new temporary stack page that is used while userspace_tramp() runs * Clones the kernel process into a new userspace process, with FDs only. @@ -318,11 +447,15 @@ int userspace_pid[NR_CPUS]; * when negative: an error number. * FIXME: can PIDs become negative?! */ -int start_userspace(unsigned long stub_stack) +int start_userspace(struct mm_id *mm_id) { + struct stub_data *proc_data = (void *)mm_id->stack; + struct tramp_data tramp_data = { + .stub_data = proc_data, + }; void *stack; unsigned long sp; - int pid, status, n, err; + int status, n, err; /* setup a temporary stack page */ stack = mmap(NULL, UM_KERN_PAGE_SIZE, @@ -338,40 +471,55 @@ int start_userspace(unsigned long stub_stack) /* set stack pointer to the end of the stack page, so it can grow downwards */ sp = (unsigned long)stack + UM_KERN_PAGE_SIZE; - /* clone into new userspace process */ - pid = clone(userspace_tramp, (void *) sp, + /* socket pair for init data and SECCOMP FD passing (no CLOEXEC here) */ + if (socketpair(AF_UNIX, SOCK_STREAM, 0, tramp_data.sockpair)) { + err = -errno; + printk(UM_KERN_ERR "%s : socketpair failed, errno = %d\n", + __func__, errno); + return err; + } + + if (using_seccomp) + proc_data->futex = FUTEX_IN_CHILD; + + mm_id->pid = clone(userspace_tramp, (void *) sp, CLONE_VFORK | CLONE_VM | SIGCHLD, - (void *)stub_stack); - if (pid < 0) { + (void *)&tramp_data); + if (mm_id->pid < 0) { err = -errno; printk(UM_KERN_ERR "%s : clone failed, errno = %d\n", __func__, errno); - return err; + goto out_close; } - do { - CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED | __WALL)); - if (n < 0) { + if (using_seccomp) { + wait_stub_done_seccomp(mm_id, 1, 1); + } else { + do { + CATCH_EINTR(n = waitpid(mm_id->pid, &status, + WUNTRACED | __WALL)); + if (n < 0) { + err = -errno; + printk(UM_KERN_ERR "%s : wait failed, errno = %d\n", + __func__, errno); + goto out_kill; + } + } while (WIFSTOPPED(status) && (WSTOPSIG(status) == SIGALRM)); + + if (!WIFSTOPPED(status) || (WSTOPSIG(status) != SIGSTOP)) { + err = -EINVAL; + printk(UM_KERN_ERR "%s : expected SIGSTOP, got status = %d\n", + __func__, status); + goto out_kill; + } + + if (ptrace(PTRACE_SETOPTIONS, mm_id->pid, NULL, + (void *) PTRACE_O_TRACESYSGOOD) < 0) { err = -errno; - printk(UM_KERN_ERR "%s : wait failed, errno = %d\n", + printk(UM_KERN_ERR "%s : PTRACE_SETOPTIONS failed, errno = %d\n", __func__, errno); goto out_kill; } - } while (WIFSTOPPED(status) && (WSTOPSIG(status) == SIGALRM)); - - if (!WIFSTOPPED(status) || (WSTOPSIG(status) != SIGSTOP)) { - err = -EINVAL; - printk(UM_KERN_ERR "%s : expected SIGSTOP, got status = %d\n", - __func__, status); - goto out_kill; - } - - if (ptrace(PTRACE_SETOPTIONS, pid, NULL, - (void *) PTRACE_O_TRACESYSGOOD) < 0) { - err = -errno; - printk(UM_KERN_ERR "%s : PTRACE_SETOPTIONS failed, errno = %d\n", - __func__, errno); - goto out_kill; } if (munmap(stack, UM_KERN_PAGE_SIZE) < 0) { @@ -381,10 +529,22 @@ int start_userspace(unsigned long stub_stack) goto out_kill; } - return pid; + close(tramp_data.sockpair[0]); + if (using_seccomp) + mm_id->sock = tramp_data.sockpair[1]; + else + close(tramp_data.sockpair[1]); + + return 0; + +out_kill: + os_kill_ptraced_process(mm_id->pid, 1); +out_close: + close(tramp_data.sockpair[0]); + close(tramp_data.sockpair[1]); + + mm_id->pid = -1; - out_kill: - os_kill_ptraced_process(pid, 1); return err; } @@ -394,7 +554,9 @@ extern unsigned long tt_extra_sched_jiffies; void userspace(struct uml_pt_regs *regs) { int err, status, op, pid = userspace_pid[0]; - siginfo_t si; + siginfo_t si_ptrace; + siginfo_t *si; + int sig; /* Handle any immediate reschedules or signals */ interrupt_end(); @@ -427,103 +589,177 @@ void userspace(struct uml_pt_regs *regs) current_mm_sync(); - /* Flush out any pending syscalls */ - err = syscall_stub_flush(current_mm_id()); - if (err) { - if (err == -ENOMEM) - report_enomem(); + if (using_seccomp) { + struct mm_id *mm_id = current_mm_id(); + struct stub_data *proc_data = (void *) mm_id->stack; + int ret; - printk(UM_KERN_ERR "%s - Error flushing stub syscalls: %d", - __func__, -err); - fatal_sigsegv(); - } + ret = set_stub_state(regs, proc_data, singlestepping()); + if (ret) { + printk(UM_KERN_ERR "%s - failed to set regs: %d", + __func__, ret); + fatal_sigsegv(); + } - /* - * This can legitimately fail if the process loads a - * bogus value into a segment register. It will - * segfault and PTRACE_GETREGS will read that value - * out of the process. However, PTRACE_SETREGS will - * fail. In this case, there is nothing to do but - * just kill the process. - */ - if (ptrace(PTRACE_SETREGS, pid, 0, regs->gp)) { - printk(UM_KERN_ERR "%s - ptrace set regs failed, errno = %d\n", - __func__, errno); - fatal_sigsegv(); - } + /* Must have been reset by the syscall caller */ + if (proc_data->restart_wait != 0) + panic("Programming error: Flag to only run syscalls in child was not cleared!"); - if (put_fp_registers(pid, regs->fp)) { - printk(UM_KERN_ERR "%s - ptrace set fp regs failed, errno = %d\n", - __func__, errno); - fatal_sigsegv(); - } + /* Mark pending syscalls for flushing */ + proc_data->syscall_data_len = mm_id->syscall_data_len; - if (singlestepping()) - op = PTRACE_SYSEMU_SINGLESTEP; - else - op = PTRACE_SYSEMU; + wait_stub_done_seccomp(mm_id, 0, 0); - if (ptrace(op, pid, 0, 0)) { - printk(UM_KERN_ERR "%s - ptrace continue failed, op = %d, errno = %d\n", - __func__, op, errno); - fatal_sigsegv(); - } + sig = proc_data->signal; - CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED | __WALL)); - if (err < 0) { - printk(UM_KERN_ERR "%s - wait failed, errno = %d\n", - __func__, errno); - fatal_sigsegv(); - } + if (sig == SIGTRAP && proc_data->err != 0) { + printk(UM_KERN_ERR "%s - Error flushing stub syscalls", + __func__); + syscall_stub_dump_error(mm_id); + mm_id->syscall_data_len = proc_data->err; + fatal_sigsegv(); + } - regs->is_user = 1; - if (ptrace(PTRACE_GETREGS, pid, 0, regs->gp)) { - printk(UM_KERN_ERR "%s - PTRACE_GETREGS failed, errno = %d\n", - __func__, errno); - fatal_sigsegv(); - } + mm_id->syscall_data_len = 0; + mm_id->syscall_fd_num = 0; - if (get_fp_registers(pid, regs->fp)) { - printk(UM_KERN_ERR "%s - get_fp_registers failed, errno = %d\n", - __func__, errno); - fatal_sigsegv(); - } + ret = get_stub_state(regs, proc_data, NULL); + if (ret) { + printk(UM_KERN_ERR "%s - failed to get regs: %d", + __func__, ret); + fatal_sigsegv(); + } - UPT_SYSCALL_NR(regs) = -1; /* Assume: It's not a syscall */ + if (proc_data->si_offset > sizeof(proc_data->sigstack) - sizeof(*si)) + panic("%s - Invalid siginfo offset from child", + __func__); + si = (void *)&proc_data->sigstack[proc_data->si_offset]; - if (WIFSTOPPED(status)) { - int sig = WSTOPSIG(status); + regs->is_user = 1; - /* These signal handlers need the si argument. - * The SIGIO and SIGALARM handlers which constitute the - * majority of invocations, do not use it. + /* Fill in ORIG_RAX and extract fault information */ + PT_SYSCALL_NR(regs->gp) = si->si_syscall; + if (sig == SIGSEGV) { + mcontext_t *mcontext = (void *)&proc_data->sigstack[proc_data->mctx_offset]; + + GET_FAULTINFO_FROM_MC(regs->faultinfo, mcontext); + } + } else { + /* Flush out any pending syscalls */ + err = syscall_stub_flush(current_mm_id()); + if (err) { + if (err == -ENOMEM) + report_enomem(); + + printk(UM_KERN_ERR "%s - Error flushing stub syscalls: %d", + __func__, -err); + fatal_sigsegv(); + } + + /* + * This can legitimately fail if the process loads a + * bogus value into a segment register. It will + * segfault and PTRACE_GETREGS will read that value + * out of the process. However, PTRACE_SETREGS will + * fail. In this case, there is nothing to do but + * just kill the process. */ - switch (sig) { - case SIGSEGV: - case SIGTRAP: - case SIGILL: - case SIGBUS: - case SIGFPE: - case SIGWINCH: - ptrace(PTRACE_GETSIGINFO, pid, 0, (struct siginfo *)&si); - break; + if (ptrace(PTRACE_SETREGS, pid, 0, regs->gp)) { + printk(UM_KERN_ERR "%s - ptrace set regs failed, errno = %d\n", + __func__, errno); + fatal_sigsegv(); } - switch (sig) { - case SIGSEGV: - if (PTRACE_FULL_FAULTINFO) { + if (put_fp_registers(pid, regs->fp)) { + printk(UM_KERN_ERR "%s - ptrace set fp regs failed, errno = %d\n", + __func__, errno); + fatal_sigsegv(); + } + + if (singlestepping()) + op = PTRACE_SYSEMU_SINGLESTEP; + else + op = PTRACE_SYSEMU; + + if (ptrace(op, pid, 0, 0)) { + printk(UM_KERN_ERR "%s - ptrace continue failed, op = %d, errno = %d\n", + __func__, op, errno); + fatal_sigsegv(); + } + + CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED | __WALL)); + if (err < 0) { + printk(UM_KERN_ERR "%s - wait failed, errno = %d\n", + __func__, errno); + fatal_sigsegv(); + } + + regs->is_user = 1; + if (ptrace(PTRACE_GETREGS, pid, 0, regs->gp)) { + printk(UM_KERN_ERR "%s - PTRACE_GETREGS failed, errno = %d\n", + __func__, errno); + fatal_sigsegv(); + } + + if (get_fp_registers(pid, regs->fp)) { + printk(UM_KERN_ERR "%s - get_fp_registers failed, errno = %d\n", + __func__, errno); + fatal_sigsegv(); + } + + if (WIFSTOPPED(status)) { + sig = WSTOPSIG(status); + + /* + * These signal handlers need the si argument + * and SIGSEGV needs the faultinfo. + * The SIGIO and SIGALARM handlers which constitute + * the majority of invocations, do not use it. + */ + switch (sig) { + case SIGSEGV: get_skas_faultinfo(pid, ®s->faultinfo); - (*sig_info[SIGSEGV])(SIGSEGV, (struct siginfo *)&si, - regs); + fallthrough; + case SIGTRAP: + case SIGILL: + case SIGBUS: + case SIGFPE: + case SIGWINCH: + ptrace(PTRACE_GETSIGINFO, pid, 0, + (struct siginfo *)&si_ptrace); + si = &si_ptrace; + break; + default: + si = NULL; + break; } - else handle_segv(pid, regs); + } else { + sig = 0; + } + } + + UPT_SYSCALL_NR(regs) = -1; /* Assume: It's not a syscall */ + + if (sig) { + switch (sig) { + case SIGSEGV: + if (using_seccomp || PTRACE_FULL_FAULTINFO) + (*sig_info[SIGSEGV])(SIGSEGV, + (struct siginfo *)si, + regs, NULL); + else + segv(regs->faultinfo, 0, 1, NULL, NULL); + + break; + case SIGSYS: + handle_syscall(regs); break; case SIGTRAP + 0x80: handle_trap(pid, regs); break; case SIGTRAP: - relay_signal(SIGTRAP, (struct siginfo *)&si, regs); + relay_signal(SIGTRAP, (struct siginfo *)si, regs, NULL); break; case SIGALRM: break; @@ -533,7 +769,7 @@ void userspace(struct uml_pt_regs *regs) case SIGFPE: case SIGWINCH: block_signals_trace(); - (*sig_info[sig])(sig, (struct siginfo *)&si, regs); + (*sig_info[sig])(sig, (struct siginfo *)si, regs, NULL); unblock_signals_trace(); break; default: |