um: Implement kernel side of SECCOMP based process handling

This adds the kernel side of the seccomp based process handling.

Co-authored-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
Signed-off-by: Benjamin Berg <benjamin.berg@intel.com>
Link: https://patch.msgid.link/20250602130052.545733-6-benjamin@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

3 files changed, 299 insertions, 117 deletions

diff --git a/arch/um/os-Linux/internal.h b/arch/um/os-Linux/internal.h
index 317fca190c2b..5d8d3b0817a9 100644
--- a/arch/um/os-Linux/internal.h
+++ b/arch/um/os-Linux/internal.h
@@ -2,6 +2,9 @@
 #ifndef __UM_OS_LINUX_INTERNAL_H
 #define __UM_OS_LINUX_INTERNAL_H
 
+#include <mm_id.h>
+#include <stub-data.h>
+
 /*
  * elf_aux.c
  */
@@ -16,5 +19,5 @@ void check_tmpexec(void);
  * skas/process.c
  */
 void wait_stub_done(int pid);
-
+void wait_stub_done_seccomp(struct mm_id *mm_idp, int running, int wait_sigsys);
 #endif /* __UM_OS_LINUX_INTERNAL_H */
diff --git a/arch/um/os-Linux/skas/mem.c b/arch/um/os-Linux/skas/mem.c
index d7f1814b0e5a..31bf3a52047a 100644
--- a/arch/um/os-Linux/skas/mem.c
+++ b/arch/um/os-Linux/skas/mem.c
@@ -80,27 +80,32 @@ static inline long do_syscall_stub(struct mm_id *mm_idp)
 	int n, i;
 	int err, pid = mm_idp->pid;
 
-	n = ptrace_setregs(pid, syscall_regs);
-	if (n < 0) {
-		printk(UM_KERN_ERR "Registers - \n");
-		for (i = 0; i < MAX_REG_NR; i++)
-			printk(UM_KERN_ERR "\t%d\t0x%lx\n", i, syscall_regs[i]);
-		panic("%s : PTRACE_SETREGS failed, errno = %d\n",
-		      __func__, -n);
-	}
-
 	/* Inform process how much we have filled in. */
 	proc_data->syscall_data_len = mm_idp->syscall_data_len;
 
-	err = ptrace(PTRACE_CONT, pid, 0, 0);
-	if (err)
-		panic("Failed to continue stub, pid = %d, errno = %d\n", pid,
-		      errno);
-
-	wait_stub_done(pid);
+	if (using_seccomp) {
+		proc_data->restart_wait = 1;
+		wait_stub_done_seccomp(mm_idp, 0, 1);
+	} else {
+		n = ptrace_setregs(pid, syscall_regs);
+		if (n < 0) {
+			printk(UM_KERN_ERR "Registers -\n");
+			for (i = 0; i < MAX_REG_NR; i++)
+				printk(UM_KERN_ERR "\t%d\t0x%lx\n", i, syscall_regs[i]);
+			panic("%s : PTRACE_SETREGS failed, errno = %d\n",
+			      __func__, -n);
+		}
+
+		err = ptrace(PTRACE_CONT, pid, 0, 0);
+		if (err)
+			panic("Failed to continue stub, pid = %d, errno = %d\n",
+			      pid, errno);
+
+		wait_stub_done(pid);
+	}
 
 	/*
-	 * proc_data->err will be non-zero if there was an (unexpected) error.
+	 * proc_data->err will be negative if there was an (unexpected) error.
 	 * In that case, syscall_data_len points to the last executed syscall,
 	 * otherwise it will be zero (but we do not need to rely on that).
 	 */
diff --git a/arch/um/os-Linux/skas/process.c b/arch/um/os-Linux/skas/process.c
index 150e4f6ba633..e1aaa144e273 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)
  */
@@ -25,8 +26,11 @@
 #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 +146,73 @@ 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 {
+		if (!running) {
+			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)
@@ -185,14 +256,26 @@ static int userspace_tramp(void *stack)
 	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);
@@ -323,8 +406,9 @@ 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;
 	void *stack;
 	unsigned long sp;
 	int pid, status, n, err;
@@ -343,10 +427,13 @@ 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;
 
+	if (using_seccomp)
+		proc_data->futex = FUTEX_IN_CHILD;
+
 	/* clone into new userspace process */
 	pid = clone(userspace_tramp, (void *) sp,
 		    CLONE_VFORK | CLONE_VM | SIGCHLD,
-		    (void *)stub_stack);
+		    (void *)mm_id->stack);
 	if (pid < 0) {
 		err = -errno;
 		printk(UM_KERN_ERR "%s : clone failed, errno = %d\n",
@@ -354,29 +441,34 @@ int start_userspace(unsigned long stub_stack)
 		return err;
 	}
 
-	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(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, 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) {
@@ -386,6 +478,8 @@ int start_userspace(unsigned long stub_stack)
 		goto out_kill;
 	}
 
+	mm_id->pid = pid;
+
 	return pid;
 
  out_kill:
@@ -399,7 +493,9 @@ 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();
@@ -432,104 +528,182 @@ 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!");
+
+			/* Mark pending syscalls for flushing */
+			proc_data->syscall_data_len = mm_id->syscall_data_len;
+			mm_id->syscall_data_len = 0;
+
+			proc_data->signal = 0;
+			proc_data->futex = FUTEX_IN_CHILD;
+			CATCH_EINTR(syscall(__NR_futex, &proc_data->futex,
+					    FUTEX_WAKE, 1, NULL, NULL, 0));
+			do {
+				ret = syscall(__NR_futex, &proc_data->futex,
+					      FUTEX_WAIT, FUTEX_IN_CHILD, NULL, NULL, 0);
+			} while ((ret == -1 && errno == EINTR) ||
+				 proc_data->futex == FUTEX_IN_CHILD);
+
+			sig = proc_data->signal;
+
+			if (sig == SIGTRAP && proc_data->err != 0) {
+				printk(UM_KERN_ERR "%s - Error flushing stub syscalls",
+				       __func__);
+				syscall_stub_dump_error(mm_id);
+				fatal_sigsegv();
+			}
 
-		if (put_fp_registers(pid, regs->fp)) {
-			printk(UM_KERN_ERR "%s - ptrace set fp regs 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();
+			}
 
-		if (singlestepping())
-			op = PTRACE_SYSEMU_SINGLESTEP;
-		else
-			op = PTRACE_SYSEMU;
+			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 (ptrace(op, pid, 0, 0)) {
-			printk(UM_KERN_ERR "%s - ptrace continue failed, op = %d, errno = %d\n",
-			       __func__, op, errno);
-			fatal_sigsegv();
-		}
+			regs->is_user = 1;
 
-		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();
-		}
+			/* 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];
 
-		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();
-		}
+				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();
+			}
 
-		if (get_fp_registers(pid, regs->fp)) {
-			printk(UM_KERN_ERR "%s -  get_fp_registers failed, errno = %d\n",
-			       __func__, errno);
-			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();
+			}
 
-		UPT_SYSCALL_NR(regs) = -1; /* Assume: It's not a syscall */
+			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 (WIFSTOPPED(status)) {
-			int sig = WSTOPSIG(status);
+			if (singlestepping())
+				op = PTRACE_SYSEMU_SINGLESTEP;
+			else
+				op = PTRACE_SYSEMU;
 
-			/* These signal handlers need the si argument.
-			 * The SIGIO and SIGALARM handlers which constitute the
-			 * majority of invocations, do not use it.
-			 */
-			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(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,
+							   &regs->faultinfo);
+					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 {
+				sig = 0;
+			}
+		}
+
+		UPT_SYSCALL_NR(regs) = -1; /* Assume: It's not a syscall */
+
+		if (sig) {
 			switch (sig) {
 			case SIGSEGV:
-				get_skas_faultinfo(pid, &regs->faultinfo);
-
-				if (PTRACE_FULL_FAULTINFO)
-					(*sig_info[SIGSEGV])(SIGSEGV, (struct siginfo *)&si,
+				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, NULL);
+				relay_signal(SIGTRAP, (struct siginfo *)si, regs, NULL);
 				break;
 			case SIGALRM:
 				break;
@@ -539,7 +713,7 @@ void userspace(struct uml_pt_regs *regs)
 			case SIGFPE:
 			case SIGWINCH:
 				block_signals_trace();
-				(*sig_info[sig])(sig, (struct siginfo *)&si, regs, NULL);
+				(*sig_info[sig])(sig, (struct siginfo *)si, regs, NULL);
 				unblock_signals_trace();
 				break;
 			default: