Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / arch / s390 / kernel / sysinfo.c
diff --git a/kernel/arch/s390/kernel/sysinfo.c b/kernel/arch/s390/kernel/sysinfo.c
new file mode 100644 (file)
index 0000000..99babea
--- /dev/null
@@ -0,0 +1,465 @@
+/*
+ *  Copyright IBM Corp. 2001, 2009
+ *  Author(s): Ulrich Weigand <Ulrich.Weigand@de.ibm.com>,
+ *            Martin Schwidefsky <schwidefsky@de.ibm.com>,
+ */
+
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <asm/ebcdic.h>
+#include <asm/sysinfo.h>
+#include <asm/cpcmd.h>
+#include <asm/topology.h>
+
+/* Sigh, math-emu. Don't ask. */
+#include <asm/sfp-util.h>
+#include <math-emu/soft-fp.h>
+#include <math-emu/single.h>
+
+int topology_max_mnest;
+
+/*
+ * stsi - store system information
+ *
+ * Returns the current configuration level if function code 0 was specified.
+ * Otherwise returns 0 on success or a negative value on error.
+ */
+int stsi(void *sysinfo, int fc, int sel1, int sel2)
+{
+       register int r0 asm("0") = (fc << 28) | sel1;
+       register int r1 asm("1") = sel2;
+       int rc = 0;
+
+       asm volatile(
+               "       stsi    0(%3)\n"
+               "0:     jz      2f\n"
+               "1:     lhi     %1,%4\n"
+               "2:\n"
+               EX_TABLE(0b, 1b)
+               : "+d" (r0), "+d" (rc)
+               : "d" (r1), "a" (sysinfo), "K" (-EOPNOTSUPP)
+               : "cc", "memory");
+       if (rc)
+               return rc;
+       return fc ? 0 : ((unsigned int) r0) >> 28;
+}
+EXPORT_SYMBOL(stsi);
+
+static void stsi_1_1_1(struct seq_file *m, struct sysinfo_1_1_1 *info)
+{
+       int i;
+
+       if (stsi(info, 1, 1, 1))
+               return;
+       EBCASC(info->manufacturer, sizeof(info->manufacturer));
+       EBCASC(info->type, sizeof(info->type));
+       EBCASC(info->model, sizeof(info->model));
+       EBCASC(info->sequence, sizeof(info->sequence));
+       EBCASC(info->plant, sizeof(info->plant));
+       EBCASC(info->model_capacity, sizeof(info->model_capacity));
+       EBCASC(info->model_perm_cap, sizeof(info->model_perm_cap));
+       EBCASC(info->model_temp_cap, sizeof(info->model_temp_cap));
+       seq_printf(m, "Manufacturer:         %-16.16s\n", info->manufacturer);
+       seq_printf(m, "Type:                 %-4.4s\n", info->type);
+       /*
+        * Sigh: the model field has been renamed with System z9
+        * to model_capacity and a new model field has been added
+        * after the plant field. To avoid confusing older programs
+        * the "Model:" prints "model_capacity model" or just
+        * "model_capacity" if the model string is empty .
+        */
+       seq_printf(m, "Model:                %-16.16s", info->model_capacity);
+       if (info->model[0] != '\0')
+               seq_printf(m, " %-16.16s", info->model);
+       seq_putc(m, '\n');
+       seq_printf(m, "Sequence Code:        %-16.16s\n", info->sequence);
+       seq_printf(m, "Plant:                %-4.4s\n", info->plant);
+       seq_printf(m, "Model Capacity:       %-16.16s %08u\n",
+                  info->model_capacity, info->model_cap_rating);
+       if (info->model_perm_cap_rating)
+               seq_printf(m, "Model Perm. Capacity: %-16.16s %08u\n",
+                          info->model_perm_cap,
+                          info->model_perm_cap_rating);
+       if (info->model_temp_cap_rating)
+               seq_printf(m, "Model Temp. Capacity: %-16.16s %08u\n",
+                          info->model_temp_cap,
+                          info->model_temp_cap_rating);
+       if (info->ncr)
+               seq_printf(m, "Nominal Cap. Rating:  %08u\n", info->ncr);
+       if (info->npr)
+               seq_printf(m, "Nominal Perm. Rating: %08u\n", info->npr);
+       if (info->ntr)
+               seq_printf(m, "Nominal Temp. Rating: %08u\n", info->ntr);
+       if (info->cai) {
+               seq_printf(m, "Capacity Adj. Ind.:   %d\n", info->cai);
+               seq_printf(m, "Capacity Ch. Reason:  %d\n", info->ccr);
+               seq_printf(m, "Capacity Transient:   %d\n", info->t);
+       }
+       if (info->p) {
+               for (i = 1; i <= ARRAY_SIZE(info->typepct); i++) {
+                       seq_printf(m, "Type %d Percentage:    %d\n",
+                                  i, info->typepct[i - 1]);
+               }
+       }
+}
+
+static void stsi_15_1_x(struct seq_file *m, struct sysinfo_15_1_x *info)
+{
+       static int max_mnest;
+       int i, rc;
+
+       seq_putc(m, '\n');
+       if (!MACHINE_HAS_TOPOLOGY)
+               return;
+       if (stsi(info, 15, 1, topology_max_mnest))
+               return;
+       seq_printf(m, "CPU Topology HW:     ");
+       for (i = 0; i < TOPOLOGY_NR_MAG; i++)
+               seq_printf(m, " %d", info->mag[i]);
+       seq_putc(m, '\n');
+#ifdef CONFIG_SCHED_MC
+       store_topology(info);
+       seq_printf(m, "CPU Topology SW:     ");
+       for (i = 0; i < TOPOLOGY_NR_MAG; i++)
+               seq_printf(m, " %d", info->mag[i]);
+       seq_putc(m, '\n');
+#endif
+}
+
+static void stsi_1_2_2(struct seq_file *m, struct sysinfo_1_2_2 *info)
+{
+       struct sysinfo_1_2_2_extension *ext;
+       int i;
+
+       if (stsi(info, 1, 2, 2))
+               return;
+       ext = (struct sysinfo_1_2_2_extension *)
+               ((unsigned long) info + info->acc_offset);
+       seq_printf(m, "CPUs Total:           %d\n", info->cpus_total);
+       seq_printf(m, "CPUs Configured:      %d\n", info->cpus_configured);
+       seq_printf(m, "CPUs Standby:         %d\n", info->cpus_standby);
+       seq_printf(m, "CPUs Reserved:        %d\n", info->cpus_reserved);
+       /*
+        * Sigh 2. According to the specification the alternate
+        * capability field is a 32 bit floating point number
+        * if the higher order 8 bits are not zero. Printing
+        * a floating point number in the kernel is a no-no,
+        * always print the number as 32 bit unsigned integer.
+        * The user-space needs to know about the strange
+        * encoding of the alternate cpu capability.
+        */
+       seq_printf(m, "Capability:           %u", info->capability);
+       if (info->format == 1)
+               seq_printf(m, " %u", ext->alt_capability);
+       seq_putc(m, '\n');
+       if (info->nominal_cap)
+               seq_printf(m, "Nominal Capability:   %d\n", info->nominal_cap);
+       if (info->secondary_cap)
+               seq_printf(m, "Secondary Capability: %d\n", info->secondary_cap);
+       for (i = 2; i <= info->cpus_total; i++) {
+               seq_printf(m, "Adjustment %02d-way:    %u",
+                          i, info->adjustment[i-2]);
+               if (info->format == 1)
+                       seq_printf(m, " %u", ext->alt_adjustment[i-2]);
+               seq_putc(m, '\n');
+       }
+}
+
+static void stsi_2_2_2(struct seq_file *m, struct sysinfo_2_2_2 *info)
+{
+       if (stsi(info, 2, 2, 2))
+               return;
+       EBCASC(info->name, sizeof(info->name));
+       seq_putc(m, '\n');
+       seq_printf(m, "LPAR Number:          %d\n", info->lpar_number);
+       seq_printf(m, "LPAR Characteristics: ");
+       if (info->characteristics & LPAR_CHAR_DEDICATED)
+               seq_printf(m, "Dedicated ");
+       if (info->characteristics & LPAR_CHAR_SHARED)
+               seq_printf(m, "Shared ");
+       if (info->characteristics & LPAR_CHAR_LIMITED)
+               seq_printf(m, "Limited ");
+       seq_putc(m, '\n');
+       seq_printf(m, "LPAR Name:            %-8.8s\n", info->name);
+       seq_printf(m, "LPAR Adjustment:      %d\n", info->caf);
+       seq_printf(m, "LPAR CPUs Total:      %d\n", info->cpus_total);
+       seq_printf(m, "LPAR CPUs Configured: %d\n", info->cpus_configured);
+       seq_printf(m, "LPAR CPUs Standby:    %d\n", info->cpus_standby);
+       seq_printf(m, "LPAR CPUs Reserved:   %d\n", info->cpus_reserved);
+       seq_printf(m, "LPAR CPUs Dedicated:  %d\n", info->cpus_dedicated);
+       seq_printf(m, "LPAR CPUs Shared:     %d\n", info->cpus_shared);
+       if (info->mt_installed & 0x80) {
+               seq_printf(m, "LPAR CPUs G-MTID:     %d\n",
+                          info->mt_general & 0x1f);
+               seq_printf(m, "LPAR CPUs S-MTID:     %d\n",
+                          info->mt_installed & 0x1f);
+               seq_printf(m, "LPAR CPUs PS-MTID:    %d\n",
+                          info->mt_psmtid & 0x1f);
+       }
+}
+
+static void print_ext_name(struct seq_file *m, int lvl,
+                          struct sysinfo_3_2_2 *info)
+{
+       if (info->vm[lvl].ext_name_encoding == 0)
+               return;
+       if (info->ext_names[lvl][0] == 0)
+               return;
+       switch (info->vm[lvl].ext_name_encoding) {
+       case 1: /* EBCDIC */
+               EBCASC(info->ext_names[lvl], sizeof(info->ext_names[lvl]));
+               break;
+       case 2: /* UTF-8 */
+               break;
+       default:
+               return;
+       }
+       seq_printf(m, "VM%02d Extended Name:   %-.256s\n", lvl,
+                  info->ext_names[lvl]);
+}
+
+static void print_uuid(struct seq_file *m, int i, struct sysinfo_3_2_2 *info)
+{
+       if (!memcmp(&info->vm[i].uuid, &NULL_UUID_BE, sizeof(uuid_be)))
+               return;
+       seq_printf(m, "VM%02d UUID:            %pUb\n", i, &info->vm[i].uuid);
+}
+
+static void stsi_3_2_2(struct seq_file *m, struct sysinfo_3_2_2 *info)
+{
+       int i;
+
+       if (stsi(info, 3, 2, 2))
+               return;
+       for (i = 0; i < info->count; i++) {
+               EBCASC(info->vm[i].name, sizeof(info->vm[i].name));
+               EBCASC(info->vm[i].cpi, sizeof(info->vm[i].cpi));
+               seq_putc(m, '\n');
+               seq_printf(m, "VM%02d Name:            %-8.8s\n", i, info->vm[i].name);
+               seq_printf(m, "VM%02d Control Program: %-16.16s\n", i, info->vm[i].cpi);
+               seq_printf(m, "VM%02d Adjustment:      %d\n", i, info->vm[i].caf);
+               seq_printf(m, "VM%02d CPUs Total:      %d\n", i, info->vm[i].cpus_total);
+               seq_printf(m, "VM%02d CPUs Configured: %d\n", i, info->vm[i].cpus_configured);
+               seq_printf(m, "VM%02d CPUs Standby:    %d\n", i, info->vm[i].cpus_standby);
+               seq_printf(m, "VM%02d CPUs Reserved:   %d\n", i, info->vm[i].cpus_reserved);
+               print_ext_name(m, i, info);
+               print_uuid(m, i, info);
+       }
+}
+
+static int sysinfo_show(struct seq_file *m, void *v)
+{
+       void *info = (void *)get_zeroed_page(GFP_KERNEL);
+       int level;
+
+       if (!info)
+               return 0;
+       level = stsi(NULL, 0, 0, 0);
+       if (level >= 1)
+               stsi_1_1_1(m, info);
+       if (level >= 1)
+               stsi_15_1_x(m, info);
+       if (level >= 1)
+               stsi_1_2_2(m, info);
+       if (level >= 2)
+               stsi_2_2_2(m, info);
+       if (level >= 3)
+               stsi_3_2_2(m, info);
+       free_page((unsigned long)info);
+       return 0;
+}
+
+static int sysinfo_open(struct inode *inode, struct file *file)
+{
+       return single_open(file, sysinfo_show, NULL);
+}
+
+static const struct file_operations sysinfo_fops = {
+       .open           = sysinfo_open,
+       .read           = seq_read,
+       .llseek         = seq_lseek,
+       .release        = single_release,
+};
+
+static int __init sysinfo_create_proc(void)
+{
+       proc_create("sysinfo", 0444, NULL, &sysinfo_fops);
+       return 0;
+}
+device_initcall(sysinfo_create_proc);
+
+/*
+ * Service levels interface.
+ */
+
+static DECLARE_RWSEM(service_level_sem);
+static LIST_HEAD(service_level_list);
+
+int register_service_level(struct service_level *slr)
+{
+       struct service_level *ptr;
+
+       down_write(&service_level_sem);
+       list_for_each_entry(ptr, &service_level_list, list)
+               if (ptr == slr) {
+                       up_write(&service_level_sem);
+                       return -EEXIST;
+               }
+       list_add_tail(&slr->list, &service_level_list);
+       up_write(&service_level_sem);
+       return 0;
+}
+EXPORT_SYMBOL(register_service_level);
+
+int unregister_service_level(struct service_level *slr)
+{
+       struct service_level *ptr, *next;
+       int rc = -ENOENT;
+
+       down_write(&service_level_sem);
+       list_for_each_entry_safe(ptr, next, &service_level_list, list) {
+               if (ptr != slr)
+                       continue;
+               list_del(&ptr->list);
+               rc = 0;
+               break;
+       }
+       up_write(&service_level_sem);
+       return rc;
+}
+EXPORT_SYMBOL(unregister_service_level);
+
+static void *service_level_start(struct seq_file *m, loff_t *pos)
+{
+       down_read(&service_level_sem);
+       return seq_list_start(&service_level_list, *pos);
+}
+
+static void *service_level_next(struct seq_file *m, void *p, loff_t *pos)
+{
+       return seq_list_next(p, &service_level_list, pos);
+}
+
+static void service_level_stop(struct seq_file *m, void *p)
+{
+       up_read(&service_level_sem);
+}
+
+static int service_level_show(struct seq_file *m, void *p)
+{
+       struct service_level *slr;
+
+       slr = list_entry(p, struct service_level, list);
+       slr->seq_print(m, slr);
+       return 0;
+}
+
+static const struct seq_operations service_level_seq_ops = {
+       .start          = service_level_start,
+       .next           = service_level_next,
+       .stop           = service_level_stop,
+       .show           = service_level_show
+};
+
+static int service_level_open(struct inode *inode, struct file *file)
+{
+       return seq_open(file, &service_level_seq_ops);
+}
+
+static const struct file_operations service_level_ops = {
+       .open           = service_level_open,
+       .read           = seq_read,
+       .llseek         = seq_lseek,
+       .release        = seq_release
+};
+
+static void service_level_vm_print(struct seq_file *m,
+                                  struct service_level *slr)
+{
+       char *query_buffer, *str;
+
+       query_buffer = kmalloc(1024, GFP_KERNEL | GFP_DMA);
+       if (!query_buffer)
+               return;
+       cpcmd("QUERY CPLEVEL", query_buffer, 1024, NULL);
+       str = strchr(query_buffer, '\n');
+       if (str)
+               *str = 0;
+       seq_printf(m, "VM: %s\n", query_buffer);
+       kfree(query_buffer);
+}
+
+static struct service_level service_level_vm = {
+       .seq_print = service_level_vm_print
+};
+
+static __init int create_proc_service_level(void)
+{
+       proc_create("service_levels", 0, NULL, &service_level_ops);
+       if (MACHINE_IS_VM)
+               register_service_level(&service_level_vm);
+       return 0;
+}
+subsys_initcall(create_proc_service_level);
+
+/*
+ * CPU capability might have changed. Therefore recalculate loops_per_jiffy.
+ */
+void s390_adjust_jiffies(void)
+{
+       struct sysinfo_1_2_2 *info;
+       const unsigned int fmil = 0x4b189680;   /* 1e7 as 32-bit float. */
+       FP_DECL_S(SA); FP_DECL_S(SB); FP_DECL_S(SR);
+       FP_DECL_EX;
+       unsigned int capability;
+
+       info = (void *) get_zeroed_page(GFP_KERNEL);
+       if (!info)
+               return;
+
+       if (stsi(info, 1, 2, 2) == 0) {
+               /*
+                * Major sigh. The cpu capability encoding is "special".
+                * If the first 9 bits of info->capability are 0 then it
+                * is a 32 bit unsigned integer in the range 0 .. 2^23.
+                * If the first 9 bits are != 0 then it is a 32 bit float.
+                * In addition a lower value indicates a proportionally
+                * higher cpu capacity. Bogomips are the other way round.
+                * To get to a halfway suitable number we divide 1e7
+                * by the cpu capability number. Yes, that means a floating
+                * point division .. math-emu here we come :-)
+                */
+               FP_UNPACK_SP(SA, &fmil);
+               if ((info->capability >> 23) == 0)
+                       FP_FROM_INT_S(SB, (long) info->capability, 64, long);
+               else
+                       FP_UNPACK_SP(SB, &info->capability);
+               FP_DIV_S(SR, SA, SB);
+               FP_TO_INT_S(capability, SR, 32, 0);
+       } else
+               /*
+                * Really old machine without stsi block for basic
+                * cpu information. Report 42.0 bogomips.
+                */
+               capability = 42;
+       loops_per_jiffy = capability * (500000/HZ);
+       free_page((unsigned long) info);
+}
+
+/*
+ * calibrate the delay loop
+ */
+void calibrate_delay(void)
+{
+       s390_adjust_jiffies();
+       /* Print the good old Bogomips line .. */
+       printk(KERN_DEBUG "Calibrating delay loop (skipped)... "
+              "%lu.%02lu BogoMIPS preset\n", loops_per_jiffy/(500000/HZ),
+              (loops_per_jiffy/(5000/HZ)) % 100);
+}