Michael Ellerman | cc53291 | 2005-12-04 18:39:43 +1100 | [diff] [blame] | 1 | /* |
| 2 | * Architecture specific (PPC64) functions for kexec based crash dumps. |
| 3 | * |
| 4 | * Copyright (C) 2005, IBM Corp. |
| 5 | * |
| 6 | * Created by: Haren Myneni |
| 7 | * |
| 8 | * This source code is licensed under the GNU General Public License, |
| 9 | * Version 2. See the file COPYING for more details. |
| 10 | * |
| 11 | */ |
| 12 | |
| 13 | #undef DEBUG |
| 14 | |
| 15 | #include <linux/kernel.h> |
| 16 | #include <linux/smp.h> |
| 17 | #include <linux/reboot.h> |
| 18 | #include <linux/kexec.h> |
| 19 | #include <linux/bootmem.h> |
| 20 | #include <linux/crash_dump.h> |
Michael Ellerman | cc53291 | 2005-12-04 18:39:43 +1100 | [diff] [blame] | 21 | #include <linux/delay.h> |
| 22 | #include <linux/elf.h> |
| 23 | #include <linux/elfcore.h> |
| 24 | #include <linux/init.h> |
| 25 | #include <linux/types.h> |
| 26 | |
| 27 | #include <asm/processor.h> |
| 28 | #include <asm/machdep.h> |
| 29 | #include <asm/kdump.h> |
| 30 | #include <asm/lmb.h> |
| 31 | #include <asm/firmware.h> |
Haren Myneni | f6cc82f | 2006-01-10 19:25:25 -0800 | [diff] [blame^] | 32 | #include <asm/smp.h> |
Michael Ellerman | cc53291 | 2005-12-04 18:39:43 +1100 | [diff] [blame] | 33 | |
| 34 | #ifdef DEBUG |
| 35 | #include <asm/udbg.h> |
| 36 | #define DBG(fmt...) udbg_printf(fmt) |
| 37 | #else |
| 38 | #define DBG(fmt...) |
| 39 | #endif |
| 40 | |
| 41 | /* This keeps a track of which one is crashing cpu. */ |
| 42 | int crashing_cpu = -1; |
| 43 | |
| 44 | static u32 *append_elf_note(u32 *buf, char *name, unsigned type, void *data, |
| 45 | size_t data_len) |
| 46 | { |
| 47 | struct elf_note note; |
| 48 | |
| 49 | note.n_namesz = strlen(name) + 1; |
| 50 | note.n_descsz = data_len; |
| 51 | note.n_type = type; |
| 52 | memcpy(buf, ¬e, sizeof(note)); |
| 53 | buf += (sizeof(note) +3)/4; |
| 54 | memcpy(buf, name, note.n_namesz); |
| 55 | buf += (note.n_namesz + 3)/4; |
| 56 | memcpy(buf, data, note.n_descsz); |
| 57 | buf += (note.n_descsz + 3)/4; |
| 58 | |
| 59 | return buf; |
| 60 | } |
| 61 | |
| 62 | static void final_note(u32 *buf) |
| 63 | { |
| 64 | struct elf_note note; |
| 65 | |
| 66 | note.n_namesz = 0; |
| 67 | note.n_descsz = 0; |
| 68 | note.n_type = 0; |
| 69 | memcpy(buf, ¬e, sizeof(note)); |
| 70 | } |
| 71 | |
| 72 | static void crash_save_this_cpu(struct pt_regs *regs, int cpu) |
| 73 | { |
| 74 | struct elf_prstatus prstatus; |
| 75 | u32 *buf; |
| 76 | |
| 77 | if ((cpu < 0) || (cpu >= NR_CPUS)) |
| 78 | return; |
| 79 | |
| 80 | /* Using ELF notes here is opportunistic. |
| 81 | * I need a well defined structure format |
| 82 | * for the data I pass, and I need tags |
| 83 | * on the data to indicate what information I have |
| 84 | * squirrelled away. ELF notes happen to provide |
| 85 | * all of that that no need to invent something new. |
| 86 | */ |
| 87 | buf = &crash_notes[cpu][0]; |
| 88 | memset(&prstatus, 0, sizeof(prstatus)); |
| 89 | prstatus.pr_pid = current->pid; |
| 90 | elf_core_copy_regs(&prstatus.pr_reg, regs); |
| 91 | buf = append_elf_note(buf, "CORE", NT_PRSTATUS, &prstatus, |
| 92 | sizeof(prstatus)); |
| 93 | final_note(buf); |
| 94 | } |
| 95 | |
| 96 | /* FIXME Merge this with xmon_save_regs ?? */ |
| 97 | static inline void crash_get_current_regs(struct pt_regs *regs) |
| 98 | { |
| 99 | unsigned long tmp1, tmp2; |
| 100 | |
| 101 | __asm__ __volatile__ ( |
| 102 | "std 0,0(%2)\n" |
| 103 | "std 1,8(%2)\n" |
| 104 | "std 2,16(%2)\n" |
| 105 | "std 3,24(%2)\n" |
| 106 | "std 4,32(%2)\n" |
| 107 | "std 5,40(%2)\n" |
| 108 | "std 6,48(%2)\n" |
| 109 | "std 7,56(%2)\n" |
| 110 | "std 8,64(%2)\n" |
| 111 | "std 9,72(%2)\n" |
| 112 | "std 10,80(%2)\n" |
| 113 | "std 11,88(%2)\n" |
| 114 | "std 12,96(%2)\n" |
| 115 | "std 13,104(%2)\n" |
| 116 | "std 14,112(%2)\n" |
| 117 | "std 15,120(%2)\n" |
| 118 | "std 16,128(%2)\n" |
| 119 | "std 17,136(%2)\n" |
| 120 | "std 18,144(%2)\n" |
| 121 | "std 19,152(%2)\n" |
| 122 | "std 20,160(%2)\n" |
| 123 | "std 21,168(%2)\n" |
| 124 | "std 22,176(%2)\n" |
| 125 | "std 23,184(%2)\n" |
| 126 | "std 24,192(%2)\n" |
| 127 | "std 25,200(%2)\n" |
| 128 | "std 26,208(%2)\n" |
| 129 | "std 27,216(%2)\n" |
| 130 | "std 28,224(%2)\n" |
| 131 | "std 29,232(%2)\n" |
| 132 | "std 30,240(%2)\n" |
| 133 | "std 31,248(%2)\n" |
| 134 | "mfmsr %0\n" |
| 135 | "std %0, 264(%2)\n" |
| 136 | "mfctr %0\n" |
| 137 | "std %0, 280(%2)\n" |
| 138 | "mflr %0\n" |
| 139 | "std %0, 288(%2)\n" |
| 140 | "bl 1f\n" |
| 141 | "1: mflr %1\n" |
| 142 | "std %1, 256(%2)\n" |
| 143 | "mtlr %0\n" |
| 144 | "mfxer %0\n" |
| 145 | "std %0, 296(%2)\n" |
| 146 | : "=&r" (tmp1), "=&r" (tmp2) |
| 147 | : "b" (regs)); |
| 148 | } |
| 149 | |
| 150 | /* We may have saved_regs from where the error came from |
| 151 | * or it is NULL if via a direct panic(). |
| 152 | */ |
| 153 | static void crash_save_self(struct pt_regs *saved_regs) |
| 154 | { |
| 155 | struct pt_regs regs; |
| 156 | int cpu; |
| 157 | |
| 158 | cpu = smp_processor_id(); |
| 159 | if (saved_regs) |
| 160 | memcpy(®s, saved_regs, sizeof(regs)); |
| 161 | else |
| 162 | crash_get_current_regs(®s); |
| 163 | crash_save_this_cpu(®s, cpu); |
| 164 | } |
| 165 | |
| 166 | #ifdef CONFIG_SMP |
| 167 | static atomic_t waiting_for_crash_ipi; |
| 168 | |
| 169 | void crash_ipi_callback(struct pt_regs *regs) |
| 170 | { |
| 171 | int cpu = smp_processor_id(); |
| 172 | |
| 173 | if (cpu == crashing_cpu) |
| 174 | return; |
| 175 | |
| 176 | if (!cpu_online(cpu)) |
| 177 | return; |
| 178 | |
| 179 | if (ppc_md.kexec_cpu_down) |
| 180 | ppc_md.kexec_cpu_down(1, 1); |
| 181 | |
| 182 | local_irq_disable(); |
| 183 | |
| 184 | crash_save_this_cpu(regs, cpu); |
| 185 | atomic_dec(&waiting_for_crash_ipi); |
| 186 | kexec_smp_wait(); |
| 187 | /* NOTREACHED */ |
| 188 | } |
| 189 | |
| 190 | static void crash_kexec_prepare_cpus(void) |
| 191 | { |
| 192 | unsigned int msecs; |
| 193 | |
| 194 | atomic_set(&waiting_for_crash_ipi, num_online_cpus() - 1); |
| 195 | |
| 196 | crash_send_ipi(crash_ipi_callback); |
| 197 | smp_wmb(); |
| 198 | |
| 199 | /* |
| 200 | * FIXME: Until we will have the way to stop other CPUSs reliabally, |
| 201 | * the crash CPU will send an IPI and wait for other CPUs to |
| 202 | * respond. If not, proceed the kexec boot even though we failed to |
| 203 | * capture other CPU states. |
| 204 | */ |
| 205 | msecs = 1000000; |
| 206 | while ((atomic_read(&waiting_for_crash_ipi) > 0) && (--msecs > 0)) { |
| 207 | barrier(); |
| 208 | mdelay(1); |
| 209 | } |
| 210 | |
| 211 | /* Would it be better to replace the trap vector here? */ |
| 212 | |
| 213 | /* |
| 214 | * FIXME: In case if we do not get all CPUs, one possibility: ask the |
| 215 | * user to do soft reset such that we get all. |
| 216 | * IPI handler is already set by the panic cpu initially. Therefore, |
| 217 | * all cpus could invoke this handler from die() and the panic CPU |
| 218 | * will call machine_kexec() directly from this handler to do |
| 219 | * kexec boot. |
| 220 | */ |
| 221 | if (atomic_read(&waiting_for_crash_ipi)) |
| 222 | printk(KERN_ALERT "done waiting: %d cpus not responding\n", |
| 223 | atomic_read(&waiting_for_crash_ipi)); |
| 224 | /* Leave the IPI callback set */ |
| 225 | } |
| 226 | #else |
| 227 | static void crash_kexec_prepare_cpus(void) |
| 228 | { |
| 229 | /* |
| 230 | * move the secondarys to us so that we can copy |
| 231 | * the new kernel 0-0x100 safely |
| 232 | * |
| 233 | * do this if kexec in setup.c ? |
| 234 | */ |
| 235 | smp_release_cpus(); |
| 236 | } |
| 237 | |
| 238 | #endif |
| 239 | |
| 240 | void default_machine_crash_shutdown(struct pt_regs *regs) |
| 241 | { |
| 242 | /* |
| 243 | * This function is only called after the system |
| 244 | * has paniced or is otherwise in a critical state. |
| 245 | * The minimum amount of code to allow a kexec'd kernel |
| 246 | * to run successfully needs to happen here. |
| 247 | * |
| 248 | * In practice this means stopping other cpus in |
| 249 | * an SMP system. |
| 250 | * The kernel is broken so disable interrupts. |
| 251 | */ |
| 252 | local_irq_disable(); |
| 253 | |
| 254 | if (ppc_md.kexec_cpu_down) |
| 255 | ppc_md.kexec_cpu_down(1, 0); |
| 256 | |
| 257 | /* |
| 258 | * Make a note of crashing cpu. Will be used in machine_kexec |
| 259 | * such that another IPI will not be sent. |
| 260 | */ |
| 261 | crashing_cpu = smp_processor_id(); |
| 262 | crash_kexec_prepare_cpus(); |
| 263 | crash_save_self(regs); |
| 264 | } |