ARM: 9485/1: mm: acquire mmap write lock around show_pte() for user faultsgrafted

When CONFIG_DEBUG_USER=y, and cmdline "user_debug=31" is set,
a user fault may trigger show_pte() without any lock.
If another thread in the same process concurrently calls munmap(),
the page table pages may be freed while show_pte() is still traversing
them, causing a use-after-free in show_pte().

If CONFIG_ARM_LPAE=y, this may cause a kernel panic if the pages table
of PMD are freed when show_pte() is running.

Acquire mmap_write_lock() around show_pte() for user faults to fix the
contention.

For user faults, additionally restrict that show_pte() is called only
when the addr is a user-space address (addr < TASK_SIZE). This is because
the lock of tsk->mm only protects the virtual memory of user address space,
furthermore, dumping the page tables of a kernel-space address for user
faults is unnecessary and may have security implications.

Keep everything unchanged for kernel faults, because the kernel is
already in the "oops" state, acquiring a lock may risk a deadlock.

Co-developed-by: Qi Xi <xiqi2@huawei.com>

Fixes: 6d021b724481 ("ARM: dump pgd, pmd and pte states on unhandled data abort faults")
Link: https://lore.kernel.org/20260716014022.2823-1-xieyuanbin1@huawei.com
Acked-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
Reviewed-by: Linus Walleij <linusw@kernel.org>
Signed-off-by: Qi Xi <xiqi2@huawei.com>
Signed-off-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

4 files changed, 162 insertions, 0 deletions

diff --git a/samples/acrn/Makefile b/samples/acrn/Makefile
new file mode 100644
index 000000000..c8e3ed978
--- /dev/null
+++ b/samples/acrn/Makefile
@@ -0,0 +1,12 @@
+# SPDX-License-Identifier: GPL-2.0
+
+.PHONY: vm-sample
+
+vm-sample: vm-sample.o payload.o
+	$(CC) $^ -o $@
+
+payload.o: payload.ld guest16.o
+	$(LD) -T $< -o $@
+
+clean:
+	rm *.o vm-sample
diff --git a/samples/acrn/guest.ld b/samples/acrn/guest.ld
new file mode 100644
index 000000000..5127c682b
--- /dev/null
+++ b/samples/acrn/guest.ld
@@ -0,0 +1,9 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+OUTPUT_FORMAT(binary)
+SECTIONS
+{
+        .start : { *(.start) }
+        .text : { *(.text*) }
+        .rodata : { *(.rodata) }
+        .data : { *(.data) }
+}
diff --git a/samples/acrn/payload.ld b/samples/acrn/payload.ld
new file mode 100644
index 000000000..e8d9a498a
--- /dev/null
+++ b/samples/acrn/payload.ld
@@ -0,0 +1,9 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+SECTIONS
+{
+        .payload16 0 : {
+                guest16 = .;
+                guest16.o(.text)
+                guest16_end = .;
+        }
+}
diff --git a/samples/acrn/vm-sample.c b/samples/acrn/vm-sample.c
new file mode 100644
index 000000000..c61e0f914
--- /dev/null
+++ b/samples/acrn/vm-sample.c
@@ -0,0 +1,132 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * A sample program to run a User VM on the ACRN hypervisor
+ *
+ * This sample runs in a Service VM, which is a privileged VM of ACRN.
+ * CONFIG_ACRN_HSM needs to be enabled in the Service VM.
+ *
+ * Guest VM code in guest16.s will be executed after the VM launched.
+ *
+ * Copyright (C) 2020 Intel Corporation. All rights reserved.
+ */
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <signal.h>
+#include <sys/ioctl.h>
+#include <linux/acrn.h>
+
+#define GUEST_MEMORY_SIZE	(1024*1024)
+void *guest_memory;
+
+extern const unsigned char guest16[], guest16_end[];
+static char io_request_page[4096] __attribute__((aligned(4096)));
+static struct acrn_io_request *io_req_buf = (struct acrn_io_request *)io_request_page;
+
+__u16 vcpu_num;
+__u16 vmid;
+
+int hsm_fd;
+int is_running = 1;
+
+void vm_exit(int sig)
+{
+	sig = sig;
+
+	is_running = 0;
+	ioctl(hsm_fd, ACRN_IOCTL_PAUSE_VM, vmid);
+	ioctl(hsm_fd, ACRN_IOCTL_DESTROY_IOREQ_CLIENT, 0);
+}
+
+int main(int argc, char **argv)
+{
+	int vcpu_id, ret;
+	struct acrn_vm_creation create_vm = {0};
+	struct acrn_vm_memmap ram_map = {0};
+	struct acrn_vcpu_regs regs;
+	struct acrn_io_request *io_req;
+	struct acrn_ioreq_notify __attribute__((aligned(8))) notify;
+
+	argc = argc;
+	argv = argv;
+
+	ret = posix_memalign(&guest_memory, 4096, GUEST_MEMORY_SIZE);
+	if (ret < 0) {
+		printf("Not enough memory!\n");
+		return -1;
+	}
+	hsm_fd = open("/dev/acrn_hsm", O_RDWR|O_CLOEXEC);
+
+	create_vm.ioreq_buf = (__u64)io_req_buf;
+	ret = ioctl(hsm_fd, ACRN_IOCTL_CREATE_VM, &create_vm);
+	printf("Created VM! [%d]\n", ret);
+	vcpu_num = create_vm.vcpu_num;
+	vmid = create_vm.vmid;
+
+	/* setup guest memory */
+	ram_map.type = ACRN_MEMMAP_RAM;
+	ram_map.vma_base = (__u64)guest_memory;
+	ram_map.len = GUEST_MEMORY_SIZE;
+	ram_map.user_vm_pa = 0;
+	ram_map.attr = ACRN_MEM_ACCESS_RWX;
+	ret = ioctl(hsm_fd, ACRN_IOCTL_SET_MEMSEG, &ram_map);
+	printf("Set up VM memory! [%d]\n", ret);
+
+	memcpy(guest_memory, guest16, guest16_end-guest16);
+
+	/* setup vcpu registers */
+	memset(&regs, 0, sizeof(regs));
+	regs.vcpu_id = 0;
+	regs.vcpu_regs.rip = 0;
+
+	/* CR0_ET | CR0_NE */
+	regs.vcpu_regs.cr0 = 0x30U;
+	regs.vcpu_regs.cs_ar = 0x009FU;
+	regs.vcpu_regs.cs_sel = 0xF000U;
+	regs.vcpu_regs.cs_limit = 0xFFFFU;
+	regs.vcpu_regs.cs_base = 0 & 0xFFFF0000UL;
+	regs.vcpu_regs.rip = 0 & 0xFFFFUL;
+
+	ret = ioctl(hsm_fd, ACRN_IOCTL_SET_VCPU_REGS, &regs);
+	printf("Set up VM BSP registers! [%d]\n", ret);
+
+	/* create an ioreq client for this VM */
+	ret = ioctl(hsm_fd, ACRN_IOCTL_CREATE_IOREQ_CLIENT, 0);
+	printf("Created IO request client! [%d]\n", ret);
+
+	/* run vm */
+	ret = ioctl(hsm_fd, ACRN_IOCTL_START_VM, vmid);
+	printf("Start VM! [%d]\n", ret);
+
+	signal(SIGINT, vm_exit);
+	while (is_running) {
+		ret = ioctl(hsm_fd, ACRN_IOCTL_ATTACH_IOREQ_CLIENT, 0);
+
+		for (vcpu_id = 0; vcpu_id < vcpu_num; vcpu_id++) {
+			io_req = &io_req_buf[vcpu_id];
+			if ((__sync_add_and_fetch(&io_req->processed, 0) == ACRN_IOREQ_STATE_PROCESSING)
+					&& (!io_req->kernel_handled))
+				if (io_req->type == ACRN_IOREQ_TYPE_PORTIO) {
+					int bytes, port, in;
+
+					port = io_req->reqs.pio_request.address;
+					bytes = io_req->reqs.pio_request.size;
+					in = (io_req->reqs.pio_request.direction == ACRN_IOREQ_DIR_READ);
+					printf("Guest VM %s PIO[%x] with size[%x]\n", in ? "read" : "write", port, bytes);
+
+					notify.vmid = vmid;
+					notify.vcpu = vcpu_id;
+					ioctl(hsm_fd, ACRN_IOCTL_NOTIFY_REQUEST_FINISH, &notify);
+				}
+		}
+	}
+
+	ret = ioctl(hsm_fd, ACRN_IOCTL_DESTROY_VM, NULL);
+	printf("Destroy VM! [%d]\n", ret);
+	close(hsm_fd);
+	free(guest_memory);
+	return 0;
+}