um: Track userspace children dying in SECCOMP mode

When in seccomp mode, we would hang forever on the futex if a child has
died unexpectedly. In contrast, ptrace mode will notice it and kill the
corresponding thread when it fails to run it.

Fix this issue using a new IRQ that is fired after a SIGCHLD and keeping
an (internal) list of all MMs. In the IRQ handler, find the affected MM
and set its PID to -1 as well as the futex variable to FUTEX_IN_KERN.

This, together with futex returning -EINTR after the signal is
sufficient to implement a race-free detection of a child dying.

Note that this also enables IRQ handling while starting a userspace
process. This should be safe and SECCOMP requires the IRQ in case the
process does not come up properly.

Signed-off-by: Benjamin Berg <benjamin@sipsolutions.net>
Signed-off-by: Benjamin Berg <benjamin.berg@intel.com>
Link: https://patch.msgid.link/20250602130052.545733-5-benjamin@sipsolutions.net
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

2 files changed, 83 insertions, 5 deletions

diff --git a/arch/um/kernel/irq.c b/arch/um/kernel/irq.c
index abe8f30a521c..f1787be3983c 100644
--- a/arch/um/kernel/irq.c
+++ b/arch/um/kernel/irq.c
@@ -690,3 +690,9 @@ void __init init_IRQ(void)
 	/* Initialize EPOLL Loop */
 	os_setup_epoll();
 }
+
+extern void sigchld_handler(int sig, struct siginfo *unused_si,
+			    struct uml_pt_regs *regs, void *mc)
+{
+	do_IRQ(SIGCHLD_IRQ, regs);
+}
diff --git a/arch/um/kernel/skas/mmu.c b/arch/um/kernel/skas/mmu.c
index 0eb5a1d3ba70..1e146a0f9549 100644
--- a/arch/um/kernel/skas/mmu.c
+++ b/arch/um/kernel/skas/mmu.c
@@ -8,6 +8,7 @@
 #include <linux/sched/signal.h>
 #include <linux/slab.h>
 
+#include <shared/irq_kern.h>
 #include <asm/pgalloc.h>
 #include <asm/sections.h>
 #include <asm/mmu_context.h>
@@ -19,6 +20,9 @@
 /* Ensure the stub_data struct covers the allocated area */
 static_assert(sizeof(struct stub_data) == STUB_DATA_PAGES * UM_KERN_PAGE_SIZE);
 
+spinlock_t mm_list_lock;
+struct list_head mm_list;
+
 int init_new_context(struct task_struct *task, struct mm_struct *mm)
 {
 	struct mm_id *new_id = &mm->context.id;
@@ -31,10 +35,12 @@ int init_new_context(struct task_struct *task, struct mm_struct *mm)
 
 	new_id->stack = stack;
 
-	block_signals_trace();
-	new_id->pid = start_userspace(stack);
-	unblock_signals_trace();
+	scoped_guard(spinlock_irqsave, &mm_list_lock) {
+		/* Insert into list, used for lookups when the child dies */
+		list_add(&mm->context.list, &mm_list);
+	}
 
+	new_id->pid = start_userspace(stack);
 	if (new_id->pid < 0) {
 		ret = new_id->pid;
 		goto out_free;
@@ -60,13 +66,79 @@ void destroy_context(struct mm_struct *mm)
 	 * zero, resulting in a kill(0), which will result in the
 	 * whole UML suddenly dying.  Also, cover negative and
 	 * 1 cases, since they shouldn't happen either.
+	 *
+	 * Negative cases happen if the child died unexpectedly.
 	 */
-	if (mmu->id.pid < 2) {
+	if (mmu->id.pid >= 0 && mmu->id.pid < 2) {
 		printk(KERN_ERR "corrupt mm_context - pid = %d\n",
 		       mmu->id.pid);
 		return;
 	}
-	os_kill_ptraced_process(mmu->id.pid, 1);
+
+	if (mmu->id.pid > 0) {
+		os_kill_ptraced_process(mmu->id.pid, 1);
+		mmu->id.pid = -1;
+	}
 
 	free_pages(mmu->id.stack, ilog2(STUB_DATA_PAGES));
+
+	guard(spinlock_irqsave)(&mm_list_lock);
+
+	list_del(&mm->context.list);
+}
+
+static irqreturn_t mm_sigchld_irq(int irq, void* dev)
+{
+	struct mm_context *mm_context;
+	pid_t pid;
+
+	guard(spinlock)(&mm_list_lock);
+
+	while ((pid = os_reap_child()) > 0) {
+		/*
+		* A child died, check if we have an MM with the PID. This is
+		* only relevant in SECCOMP mode (as ptrace will fail anyway).
+		*
+		* See wait_stub_done_seccomp for more details.
+		*/
+		list_for_each_entry(mm_context, &mm_list, list) {
+			if (mm_context->id.pid == pid) {
+				struct stub_data *stub_data;
+				printk("Unexpectedly lost MM child! Affected tasks will segfault.");
+
+				/* Marks the MM as dead */
+				mm_context->id.pid = -1;
+
+				/*
+				 * NOTE: If SMP is implemented, a futex_wake
+				 * needs to be added here.
+				 */
+				stub_data = (void *)mm_context->id.stack;
+				stub_data->futex = FUTEX_IN_KERN;
+
+				/*
+				 * NOTE: Currently executing syscalls by
+				 * affected tasks may finish normally.
+				 */
+				break;
+			}
+		}
+	}
+
+	return IRQ_HANDLED;
+}
+
+static int __init init_child_tracking(void)
+{
+	int err;
+
+	spin_lock_init(&mm_list_lock);
+	INIT_LIST_HEAD(&mm_list);
+
+	err = request_irq(SIGCHLD_IRQ, mm_sigchld_irq, 0, "SIGCHLD", NULL);
+	if (err < 0)
+		panic("Failed to register SIGCHLD IRQ: %d", err);
+
+	return 0;
 }
+early_initcall(init_child_tracking)