LoongArch: Add kdump support

This patch adds support for kdump. In kdump case the normal kernel will
reserve a region for the crash kernel and jump there on panic.

Arch-specific functions are added to allow for implementing a crash dump
file interface, /proc/vmcore, which can be viewed as a ELF file.

A user-space tool, such as kexec-tools, is responsible for allocating a
separate region for the core's ELF header within the crash kdump kernel
memory and filling it in when executing kexec_load().

Then, its location will be advertised to the crash dump kernel via a
command line argument "elfcorehdr=", and the crash dump kernel will
preserve this region for later use with arch_reserve_vmcore() at boot
time.

At the same time, the crash kdump kernel is also limited within the
"crashkernel" area via a command line argument "mem=", so as not to
destroy the original kernel dump data.

In the crash dump kernel environment, /proc/vmcore is used to access the
primary kernel's memory with copy_oldmem_page().

I tested kdump on LoongArch machines (Loongson-3A5000) and it works as
expected (suggested crashkernel parameter is "crashkernel=512M@2560M"),
you may test it by triggering a crash through /proc/sysrq-trigger:

 $ sudo kexec -p /boot/vmlinux-kdump --reuse-cmdline --append="nr_cpus=1"
 # echo c > /proc/sysrq-trigger
Signed-off-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>

arch/loongarch/Kconfig

@@ -437,6 +437,28 @@ config KEXEC
 
 	  The name comes from the similarity to the exec system call.
 
+config CRASH_DUMP
+	bool "Build kdump crash kernel"
+	help
+	  Generate crash dump after being started by kexec. This should
+	  be normally only set in special crash dump kernels which are
+	  loaded in the main kernel with kexec-tools into a specially
+	  reserved region and then later executed after a crash by
+	  kdump/kexec.
+
+	  For more details see Documentation/admin-guide/kdump/kdump.rst
+
+config PHYSICAL_START
+	hex "Physical address where the kernel is loaded"
+	default "0x90000000a0000000"
+	depends on CRASH_DUMP
+	help
+	  This gives the XKPRANGE address where the kernel is loaded.
+	  If you plan to use kernel for capturing the crash dump change
+	  this value to start of the reserved region (the "X" value as
+	  specified in the "crashkernel=YM@XM" command line boot parameter
+	  passed to the panic-ed kernel).
+
 config SECCOMP
 	bool "Enable seccomp to safely compute untrusted bytecode"
 	depends on PROC_FS

arch/loongarch/Makefile

@@ -69,7 +69,11 @@ endif
 cflags-y += -ffreestanding
 cflags-y += $(call cc-option, -mno-check-zero-division)
 
+ifndef CONFIG_PHYSICAL_START
 load-y		= 0x9000000000200000
+else
+load-y		= $(CONFIG_PHYSICAL_START)
+endif
 bootvars-y	= VMLINUX_LOAD_ADDRESS=$(load-y)
 
 drivers-$(CONFIG_PCI)		+= arch/loongarch/pci/

arch/loongarch/kernel/Makefile

@@ -26,6 +26,7 @@ obj-$(CONFIG_NUMA)		+= numa.o
 obj-$(CONFIG_MAGIC_SYSRQ)	+= sysrq.o
 
 obj-$(CONFIG_KEXEC)		+= machine_kexec.o relocate_kernel.o
+obj-$(CONFIG_CRASH_DUMP)	+= crash_dump.o
 
 obj-$(CONFIG_UNWINDER_GUESS)	+= unwind_guess.o
 obj-$(CONFIG_UNWINDER_PROLOGUE) += unwind_prologue.o

arch/loongarch/kernel/crash_dump.c

+// SPDX-License-Identifier: GPL-2.0
+#include <linux/crash_dump.h>
+#include <linux/io.h>
+#include <linux/uio.h>
+
+ssize_t copy_oldmem_page(struct iov_iter *iter, unsigned long pfn,
+			 size_t csize, unsigned long offset)
+{
+	void *vaddr;
+
+	if (!csize)
+		return 0;
+
+	vaddr = memremap(__pfn_to_phys(pfn), PAGE_SIZE, MEMREMAP_WB);
+	if (!vaddr)
+		return -ENOMEM;
+
+	csize = copy_to_iter(vaddr + offset, csize, iter);
+
+	memunmap(vaddr);
+
+	return csize;
+}

arch/loongarch/kernel/machine_kexec.c

@@ -7,10 +7,15 @@
 #include <linux/compiler.h>
 #include <linux/cpu.h>
 #include <linux/kexec.h>
-#include <linux/mm.h>
+#include <linux/crash_dump.h>
 #include <linux/delay.h>
+#include <linux/irq.h>
 #include <linux/libfdt.h>
+#include <linux/mm.h>
 #include <linux/of_fdt.h>
+#include <linux/reboot.h>
+#include <linux/sched.h>
+#include <linux/sched/task_stack.h>
 
 #include <asm/bootinfo.h>
 #include <asm/cacheflush.h>
@@ -21,6 +26,7 @@
 #define KEXEC_CMDLINE_ADDR	TO_CACHE(0x108000UL)
 
 static unsigned long reboot_code_buffer;
+static cpumask_t cpus_in_crash = CPU_MASK_NONE;
 
 #ifdef CONFIG_SMP
 static void (*relocated_kexec_smp_wait)(void *);
@@ -78,7 +84,7 @@ int machine_kexec_prepare(struct kimage *kimage)
 		return -EINVAL;
 	}
 
-	/* kexec need a safe page to save reboot_code_buffer */
+	/* kexec/kdump need a safe page to save reboot_code_buffer */
 	kimage->control_code_page = virt_to_page((void *)KEXEC_CONTROL_CODE);
 
 	reboot_code_buffer = (unsigned long)page_address(kimage->control_code_page);
@@ -102,7 +108,8 @@ void kexec_reboot(void)
 
 	/*
 	 * We know we were online, and there will be no incoming IPIs at
-	 * this point.
+	 * this point. Mark online again before rebooting so that the crash
+	 * analysis tool will see us correctly.
 	 */
 	set_cpu_online(smp_processor_id(), true);
 
@@ -147,7 +154,74 @@ static void kexec_shutdown_secondary(void *regs)
 
 	kexec_reboot();
 }
-#endif
+
+static void crash_shutdown_secondary(void *passed_regs)
+{
+	int cpu = smp_processor_id();
+	struct pt_regs *regs = passed_regs;
+
+	/*
+	 * If we are passed registers, use those. Otherwise get the
+	 * regs from the last interrupt, which should be correct, as
+	 * we are in an interrupt. But if the regs are not there,
+	 * pull them from the top of the stack. They are probably
+	 * wrong, but we need something to keep from crashing again.
+	 */
+	if (!regs)
+		regs = get_irq_regs();
+	if (!regs)
+		regs = task_pt_regs(current);
+
+	if (!cpu_online(cpu))
+		return;
+
+	/* We won't be sent IPIs any more. */
+	set_cpu_online(cpu, false);
+
+	local_irq_disable();
+	if (!cpumask_test_cpu(cpu, &cpus_in_crash))
+		crash_save_cpu(regs, cpu);
+	cpumask_set_cpu(cpu, &cpus_in_crash);
+
+	while (!atomic_read(&kexec_ready_to_reboot))
+		cpu_relax();
+
+	kexec_reboot();
+}
+
+void crash_smp_send_stop(void)
+{
+	unsigned int ncpus;
+	unsigned long timeout;
+	static int cpus_stopped;
+
+	/*
+	 * This function can be called twice in panic path, but obviously
+	 * we should execute this only once.
+	 */
+	if (cpus_stopped)
+		return;
+
+	cpus_stopped = 1;
+
+	 /* Excluding the panic cpu */
+	ncpus = num_online_cpus() - 1;
+
+	smp_call_function(crash_shutdown_secondary, NULL, 0);
+	smp_wmb();
+
+	/*
+	 * The crash CPU sends an IPI and wait for other CPUs to
+	 * respond. Delay of at least 10 seconds.
+	 */
+	timeout = MSEC_PER_SEC * 10;
+	pr_emerg("Sending IPI to other cpus...\n");
+	while ((cpumask_weight(&cpus_in_crash) < ncpus) && timeout--) {
+		mdelay(1);
+		cpu_relax();
+	}
+}
+#endif /* defined(CONFIG_SMP) */
 
 void machine_shutdown(void)
 {
@@ -165,6 +239,19 @@ void machine_shutdown(void)
 
 void machine_crash_shutdown(struct pt_regs *regs)
 {
+	int crashing_cpu;
+
+	local_irq_disable();
+
+	crashing_cpu = smp_processor_id();
+	crash_save_cpu(regs, crashing_cpu);
+
+#ifdef CONFIG_SMP
+	crash_smp_send_stop();
+#endif
+	cpumask_set_cpu(crashing_cpu, &cpus_in_crash);
+
+	pr_info("Starting crashdump kernel...\n");
 }
 
 void machine_kexec(struct kimage *image)
@@ -178,7 +265,8 @@ void machine_kexec(struct kimage *image)
 
 	start_addr = (unsigned long)phys_to_virt(image->start);
 
-	first_ind_entry = (unsigned long)phys_to_virt(image->head & PAGE_MASK);
+	first_ind_entry = (image->type == KEXEC_TYPE_DEFAULT) ?
+		(unsigned long)phys_to_virt(image->head & PAGE_MASK) : 0;
 
 	/*
 	 * The generic kexec code builds a page list with physical

arch/loongarch/kernel/mem.c

@@ -58,7 +58,4 @@ void __init memblock_init(void)
 	/* Reserve the kernel text/data/bss */
 	memblock_reserve(__pa_symbol(&_text),
 			 __pa_symbol(&_end) - __pa_symbol(&_text));
-
-	/* Reserve the initrd */
-	reserve_initrd_mem();
 }

arch/loongarch/kernel/relocate_kernel.S

@@ -24,6 +24,12 @@ SYM_CODE_START(relocate_new_kernel)
 	 */
 	move		s0, a4
 
+	/*
+	 * In case of a kdump/crash kernel, the indirection page is not
+	 * populated as the kernel is directly copied to a reserved location
+	 */
+	beqz		s0, done
+
 process_entry:
 	PTR_L		s1, s0, 0
 	PTR_ADDI	s0, s0, SZREG

arch/loongarch/kernel/setup.c

@@ -19,6 +19,8 @@
 #include <linux/memblock.h>
 #include <linux/initrd.h>
 #include <linux/ioport.h>
+#include <linux/kexec.h>
+#include <linux/crash_dump.h>
 #include <linux/root_dev.h>
 #include <linux/console.h>
 #include <linux/pfn.h>
@@ -185,8 +187,70 @@ static int __init early_parse_mem(char *p)
 }
 early_param("mem", early_parse_mem);
 
+static void __init arch_reserve_vmcore(void)
+{
+#ifdef CONFIG_PROC_VMCORE
+	u64 i;
+	phys_addr_t start, end;
+
+	if (!is_kdump_kernel())
+		return;
+
+	if (!elfcorehdr_size) {
+		for_each_mem_range(i, &start, &end) {
+			if (elfcorehdr_addr >= start && elfcorehdr_addr < end) {
+				/*
+				 * Reserve from the elf core header to the end of
+				 * the memory segment, that should all be kdump
+				 * reserved memory.
+				 */
+				elfcorehdr_size = end - elfcorehdr_addr;
+				break;
+			}
+		}
+	}
+
+	if (memblock_is_region_reserved(elfcorehdr_addr, elfcorehdr_size)) {
+		pr_warn("elfcorehdr is overlapped\n");
+		return;
+	}
+
+	memblock_reserve(elfcorehdr_addr, elfcorehdr_size);
+
+	pr_info("Reserving %llu KiB of memory at 0x%llx for elfcorehdr\n",
+		elfcorehdr_size >> 10, elfcorehdr_addr);
+#endif
+}
+
+static void __init arch_parse_crashkernel(void)
+{
+#ifdef CONFIG_KEXEC
+	int ret;
+	unsigned long long start;
+	unsigned long long total_mem;
+	unsigned long long crash_base, crash_size;
+
+	total_mem = memblock_phys_mem_size();
+	ret = parse_crashkernel(boot_command_line, total_mem, &crash_size, &crash_base);
+	if (ret < 0 || crash_size <= 0)
+		return;
+
+	start = memblock_phys_alloc_range(crash_size, 1, crash_base, crash_base + crash_size);
+	if (start != crash_base) {
+		pr_warn("Invalid memory region reserved for crash kernel\n");
+		return;
+	}
+
+	crashk_res.start = crash_base;
+	crashk_res.end	 = crash_base + crash_size - 1;
+#endif
+}
+
 void __init platform_init(void)
 {
+	arch_reserve_vmcore();
+	arch_parse_crashkernel();
+
 #ifdef CONFIG_ACPI_TABLE_UPGRADE
 	acpi_table_upgrade();
 #endif
@@ -289,6 +353,15 @@ static void __init resource_init(void)
 		request_resource(res, &data_resource);
 		request_resource(res, &bss_resource);
 	}
+
+#ifdef CONFIG_KEXEC
+	if (crashk_res.start < crashk_res.end) {
+		insert_resource(&iomem_resource, &crashk_res);
+		pr_info("Reserving %ldMB of memory at %ldMB for crashkernel\n",
+			(unsigned long)((crashk_res.end - crashk_res.start + 1) >> 20),
+			(unsigned long)(crashk_res.start  >> 20));
+	}
+#endif
 }
 
 static int __init reserve_memblock_reserved_regions(void)
@@ -350,6 +423,7 @@ void __init setup_arch(char **cmdline_p)
 	memblock_init();
 	pagetable_init();
 	parse_early_param();
+	reserve_initrd_mem();
 
 	platform_init();
 	arch_mem_init(cmdline_p);

arch/loongarch/kernel/traps.c

@@ -10,6 +10,7 @@
 #include <linux/entry-common.h>
 #include <linux/init.h>
 #include <linux/kernel.h>
+#include <linux/kexec.h>
 #include <linux/module.h>
 #include <linux/extable.h>
 #include <linux/mm.h>
@@ -246,6 +247,9 @@ void __noreturn die(const char *str, struct pt_regs *regs)
 
 	oops_exit();
 
+	if (regs && kexec_should_crash(current))
+		crash_kexec(regs);
+
 	if (in_interrupt())
 		panic("Fatal exception in interrupt");