diff options
Diffstat (limited to 'kernel/arch/s390/mm')
-rw-r--r-- | kernel/arch/s390/mm/Makefile | 10 | ||||
-rw-r--r-- | kernel/arch/s390/mm/cmm.c | 495 | ||||
-rw-r--r-- | kernel/arch/s390/mm/dump_pagetables.c | 225 | ||||
-rw-r--r-- | kernel/arch/s390/mm/extable.c | 81 | ||||
-rw-r--r-- | kernel/arch/s390/mm/extmem.c | 763 | ||||
-rw-r--r-- | kernel/arch/s390/mm/fault.c | 758 | ||||
-rw-r--r-- | kernel/arch/s390/mm/gup.c | 240 | ||||
-rw-r--r-- | kernel/arch/s390/mm/hugetlbpage.c | 212 | ||||
-rw-r--r-- | kernel/arch/s390/mm/init.c | 230 | ||||
-rw-r--r-- | kernel/arch/s390/mm/maccess.c | 208 | ||||
-rw-r--r-- | kernel/arch/s390/mm/mem_detect.c | 61 | ||||
-rw-r--r-- | kernel/arch/s390/mm/mmap.c | 288 | ||||
-rw-r--r-- | kernel/arch/s390/mm/page-states.c | 114 | ||||
-rw-r--r-- | kernel/arch/s390/mm/pageattr.c | 170 | ||||
-rw-r--r-- | kernel/arch/s390/mm/pgtable.c | 1412 | ||||
-rw-r--r-- | kernel/arch/s390/mm/vmem.c | 416 |
16 files changed, 5683 insertions, 0 deletions
diff --git a/kernel/arch/s390/mm/Makefile b/kernel/arch/s390/mm/Makefile new file mode 100644 index 000000000..839592ca2 --- /dev/null +++ b/kernel/arch/s390/mm/Makefile @@ -0,0 +1,10 @@ +# +# Makefile for the linux s390-specific parts of the memory manager. +# + +obj-y := init.o fault.o extmem.o mmap.o vmem.o pgtable.o maccess.o +obj-y += page-states.o gup.o extable.o pageattr.o mem_detect.o + +obj-$(CONFIG_CMM) += cmm.o +obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o +obj-$(CONFIG_S390_PTDUMP) += dump_pagetables.o diff --git a/kernel/arch/s390/mm/cmm.c b/kernel/arch/s390/mm/cmm.c new file mode 100644 index 000000000..79ddd580d --- /dev/null +++ b/kernel/arch/s390/mm/cmm.c @@ -0,0 +1,495 @@ +/* + * Collaborative memory management interface. + * + * Copyright IBM Corp 2003, 2010 + * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>, + * + */ + +#include <linux/errno.h> +#include <linux/fs.h> +#include <linux/init.h> +#include <linux/module.h> +#include <linux/gfp.h> +#include <linux/sched.h> +#include <linux/sysctl.h> +#include <linux/ctype.h> +#include <linux/swap.h> +#include <linux/kthread.h> +#include <linux/oom.h> +#include <linux/suspend.h> +#include <linux/uaccess.h> + +#include <asm/pgalloc.h> +#include <asm/diag.h> + +#ifdef CONFIG_CMM_IUCV +static char *cmm_default_sender = "VMRMSVM"; +#endif +static char *sender; +module_param(sender, charp, 0400); +MODULE_PARM_DESC(sender, + "Guest name that may send SMSG messages (default VMRMSVM)"); + +#include "../../../drivers/s390/net/smsgiucv.h" + +#define CMM_NR_PAGES ((PAGE_SIZE / sizeof(unsigned long)) - 2) + +struct cmm_page_array { + struct cmm_page_array *next; + unsigned long index; + unsigned long pages[CMM_NR_PAGES]; +}; + +static long cmm_pages; +static long cmm_timed_pages; +static volatile long cmm_pages_target; +static volatile long cmm_timed_pages_target; +static long cmm_timeout_pages; +static long cmm_timeout_seconds; +static int cmm_suspended; + +static struct cmm_page_array *cmm_page_list; +static struct cmm_page_array *cmm_timed_page_list; +static DEFINE_SPINLOCK(cmm_lock); + +static struct task_struct *cmm_thread_ptr; +static DECLARE_WAIT_QUEUE_HEAD(cmm_thread_wait); +static DEFINE_TIMER(cmm_timer, NULL, 0, 0); + +static void cmm_timer_fn(unsigned long); +static void cmm_set_timer(void); + +static long cmm_alloc_pages(long nr, long *counter, + struct cmm_page_array **list) +{ + struct cmm_page_array *pa, *npa; + unsigned long addr; + + while (nr) { + addr = __get_free_page(GFP_NOIO); + if (!addr) + break; + spin_lock(&cmm_lock); + pa = *list; + if (!pa || pa->index >= CMM_NR_PAGES) { + /* Need a new page for the page list. */ + spin_unlock(&cmm_lock); + npa = (struct cmm_page_array *) + __get_free_page(GFP_NOIO); + if (!npa) { + free_page(addr); + break; + } + spin_lock(&cmm_lock); + pa = *list; + if (!pa || pa->index >= CMM_NR_PAGES) { + npa->next = pa; + npa->index = 0; + pa = npa; + *list = pa; + } else + free_page((unsigned long) npa); + } + diag10_range(addr >> PAGE_SHIFT, 1); + pa->pages[pa->index++] = addr; + (*counter)++; + spin_unlock(&cmm_lock); + nr--; + } + return nr; +} + +static long cmm_free_pages(long nr, long *counter, struct cmm_page_array **list) +{ + struct cmm_page_array *pa; + unsigned long addr; + + spin_lock(&cmm_lock); + pa = *list; + while (nr) { + if (!pa || pa->index <= 0) + break; + addr = pa->pages[--pa->index]; + if (pa->index == 0) { + pa = pa->next; + free_page((unsigned long) *list); + *list = pa; + } + free_page(addr); + (*counter)--; + nr--; + } + spin_unlock(&cmm_lock); + return nr; +} + +static int cmm_oom_notify(struct notifier_block *self, + unsigned long dummy, void *parm) +{ + unsigned long *freed = parm; + long nr = 256; + + nr = cmm_free_pages(nr, &cmm_timed_pages, &cmm_timed_page_list); + if (nr > 0) + nr = cmm_free_pages(nr, &cmm_pages, &cmm_page_list); + cmm_pages_target = cmm_pages; + cmm_timed_pages_target = cmm_timed_pages; + *freed += 256 - nr; + return NOTIFY_OK; +} + +static struct notifier_block cmm_oom_nb = { + .notifier_call = cmm_oom_notify, +}; + +static int cmm_thread(void *dummy) +{ + int rc; + + while (1) { + rc = wait_event_interruptible(cmm_thread_wait, + (!cmm_suspended && (cmm_pages != cmm_pages_target || + cmm_timed_pages != cmm_timed_pages_target)) || + kthread_should_stop()); + if (kthread_should_stop() || rc == -ERESTARTSYS) { + cmm_pages_target = cmm_pages; + cmm_timed_pages_target = cmm_timed_pages; + break; + } + if (cmm_pages_target > cmm_pages) { + if (cmm_alloc_pages(1, &cmm_pages, &cmm_page_list)) + cmm_pages_target = cmm_pages; + } else if (cmm_pages_target < cmm_pages) { + cmm_free_pages(1, &cmm_pages, &cmm_page_list); + } + if (cmm_timed_pages_target > cmm_timed_pages) { + if (cmm_alloc_pages(1, &cmm_timed_pages, + &cmm_timed_page_list)) + cmm_timed_pages_target = cmm_timed_pages; + } else if (cmm_timed_pages_target < cmm_timed_pages) { + cmm_free_pages(1, &cmm_timed_pages, + &cmm_timed_page_list); + } + if (cmm_timed_pages > 0 && !timer_pending(&cmm_timer)) + cmm_set_timer(); + } + return 0; +} + +static void cmm_kick_thread(void) +{ + wake_up(&cmm_thread_wait); +} + +static void cmm_set_timer(void) +{ + if (cmm_timed_pages_target <= 0 || cmm_timeout_seconds <= 0) { + if (timer_pending(&cmm_timer)) + del_timer(&cmm_timer); + return; + } + if (timer_pending(&cmm_timer)) { + if (mod_timer(&cmm_timer, jiffies + cmm_timeout_seconds*HZ)) + return; + } + cmm_timer.function = cmm_timer_fn; + cmm_timer.data = 0; + cmm_timer.expires = jiffies + cmm_timeout_seconds*HZ; + add_timer(&cmm_timer); +} + +static void cmm_timer_fn(unsigned long ignored) +{ + long nr; + + nr = cmm_timed_pages_target - cmm_timeout_pages; + if (nr < 0) + cmm_timed_pages_target = 0; + else + cmm_timed_pages_target = nr; + cmm_kick_thread(); + cmm_set_timer(); +} + +static void cmm_set_pages(long nr) +{ + cmm_pages_target = nr; + cmm_kick_thread(); +} + +static long cmm_get_pages(void) +{ + return cmm_pages; +} + +static void cmm_add_timed_pages(long nr) +{ + cmm_timed_pages_target += nr; + cmm_kick_thread(); +} + +static long cmm_get_timed_pages(void) +{ + return cmm_timed_pages; +} + +static void cmm_set_timeout(long nr, long seconds) +{ + cmm_timeout_pages = nr; + cmm_timeout_seconds = seconds; + cmm_set_timer(); +} + +static int cmm_skip_blanks(char *cp, char **endp) +{ + char *str; + + for (str = cp; *str == ' ' || *str == '\t'; str++) + ; + *endp = str; + return str != cp; +} + +static struct ctl_table cmm_table[]; + +static int cmm_pages_handler(struct ctl_table *ctl, int write, + void __user *buffer, size_t *lenp, loff_t *ppos) +{ + char buf[16], *p; + unsigned int len; + long nr; + + if (!*lenp || (*ppos && !write)) { + *lenp = 0; + return 0; + } + + if (write) { + len = *lenp; + if (copy_from_user(buf, buffer, + len > sizeof(buf) ? sizeof(buf) : len)) + return -EFAULT; + buf[sizeof(buf) - 1] = '\0'; + cmm_skip_blanks(buf, &p); + nr = simple_strtoul(p, &p, 0); + if (ctl == &cmm_table[0]) + cmm_set_pages(nr); + else + cmm_add_timed_pages(nr); + } else { + if (ctl == &cmm_table[0]) + nr = cmm_get_pages(); + else + nr = cmm_get_timed_pages(); + len = sprintf(buf, "%ld\n", nr); + if (len > *lenp) + len = *lenp; + if (copy_to_user(buffer, buf, len)) + return -EFAULT; + } + *lenp = len; + *ppos += len; + return 0; +} + +static int cmm_timeout_handler(struct ctl_table *ctl, int write, + void __user *buffer, size_t *lenp, loff_t *ppos) +{ + char buf[64], *p; + long nr, seconds; + unsigned int len; + + if (!*lenp || (*ppos && !write)) { + *lenp = 0; + return 0; + } + + if (write) { + len = *lenp; + if (copy_from_user(buf, buffer, + len > sizeof(buf) ? sizeof(buf) : len)) + return -EFAULT; + buf[sizeof(buf) - 1] = '\0'; + cmm_skip_blanks(buf, &p); + nr = simple_strtoul(p, &p, 0); + cmm_skip_blanks(p, &p); + seconds = simple_strtoul(p, &p, 0); + cmm_set_timeout(nr, seconds); + } else { + len = sprintf(buf, "%ld %ld\n", + cmm_timeout_pages, cmm_timeout_seconds); + if (len > *lenp) + len = *lenp; + if (copy_to_user(buffer, buf, len)) + return -EFAULT; + } + *lenp = len; + *ppos += len; + return 0; +} + +static struct ctl_table cmm_table[] = { + { + .procname = "cmm_pages", + .mode = 0644, + .proc_handler = cmm_pages_handler, + }, + { + .procname = "cmm_timed_pages", + .mode = 0644, + .proc_handler = cmm_pages_handler, + }, + { + .procname = "cmm_timeout", + .mode = 0644, + .proc_handler = cmm_timeout_handler, + }, + { } +}; + +static struct ctl_table cmm_dir_table[] = { + { + .procname = "vm", + .maxlen = 0, + .mode = 0555, + .child = cmm_table, + }, + { } +}; + +#ifdef CONFIG_CMM_IUCV +#define SMSG_PREFIX "CMM" +static void cmm_smsg_target(const char *from, char *msg) +{ + long nr, seconds; + + if (strlen(sender) > 0 && strcmp(from, sender) != 0) + return; + if (!cmm_skip_blanks(msg + strlen(SMSG_PREFIX), &msg)) + return; + if (strncmp(msg, "SHRINK", 6) == 0) { + if (!cmm_skip_blanks(msg + 6, &msg)) + return; + nr = simple_strtoul(msg, &msg, 0); + cmm_skip_blanks(msg, &msg); + if (*msg == '\0') + cmm_set_pages(nr); + } else if (strncmp(msg, "RELEASE", 7) == 0) { + if (!cmm_skip_blanks(msg + 7, &msg)) + return; + nr = simple_strtoul(msg, &msg, 0); + cmm_skip_blanks(msg, &msg); + if (*msg == '\0') + cmm_add_timed_pages(nr); + } else if (strncmp(msg, "REUSE", 5) == 0) { + if (!cmm_skip_blanks(msg + 5, &msg)) + return; + nr = simple_strtoul(msg, &msg, 0); + if (!cmm_skip_blanks(msg, &msg)) + return; + seconds = simple_strtoul(msg, &msg, 0); + cmm_skip_blanks(msg, &msg); + if (*msg == '\0') + cmm_set_timeout(nr, seconds); + } +} +#endif + +static struct ctl_table_header *cmm_sysctl_header; + +static int cmm_suspend(void) +{ + cmm_suspended = 1; + cmm_free_pages(cmm_pages, &cmm_pages, &cmm_page_list); + cmm_free_pages(cmm_timed_pages, &cmm_timed_pages, &cmm_timed_page_list); + return 0; +} + +static int cmm_resume(void) +{ + cmm_suspended = 0; + cmm_kick_thread(); + return 0; +} + +static int cmm_power_event(struct notifier_block *this, + unsigned long event, void *ptr) +{ + switch (event) { + case PM_POST_HIBERNATION: + return cmm_resume(); + case PM_HIBERNATION_PREPARE: + return cmm_suspend(); + default: + return NOTIFY_DONE; + } +} + +static struct notifier_block cmm_power_notifier = { + .notifier_call = cmm_power_event, +}; + +static int __init cmm_init(void) +{ + int rc = -ENOMEM; + + cmm_sysctl_header = register_sysctl_table(cmm_dir_table); + if (!cmm_sysctl_header) + goto out_sysctl; +#ifdef CONFIG_CMM_IUCV + /* convert sender to uppercase characters */ + if (sender) { + int len = strlen(sender); + while (len--) + sender[len] = toupper(sender[len]); + } else { + sender = cmm_default_sender; + } + + rc = smsg_register_callback(SMSG_PREFIX, cmm_smsg_target); + if (rc < 0) + goto out_smsg; +#endif + rc = register_oom_notifier(&cmm_oom_nb); + if (rc < 0) + goto out_oom_notify; + rc = register_pm_notifier(&cmm_power_notifier); + if (rc) + goto out_pm; + cmm_thread_ptr = kthread_run(cmm_thread, NULL, "cmmthread"); + if (!IS_ERR(cmm_thread_ptr)) + return 0; + + rc = PTR_ERR(cmm_thread_ptr); + unregister_pm_notifier(&cmm_power_notifier); +out_pm: + unregister_oom_notifier(&cmm_oom_nb); +out_oom_notify: +#ifdef CONFIG_CMM_IUCV + smsg_unregister_callback(SMSG_PREFIX, cmm_smsg_target); +out_smsg: +#endif + unregister_sysctl_table(cmm_sysctl_header); +out_sysctl: + del_timer_sync(&cmm_timer); + return rc; +} +module_init(cmm_init); + +static void __exit cmm_exit(void) +{ + unregister_sysctl_table(cmm_sysctl_header); +#ifdef CONFIG_CMM_IUCV + smsg_unregister_callback(SMSG_PREFIX, cmm_smsg_target); +#endif + unregister_pm_notifier(&cmm_power_notifier); + unregister_oom_notifier(&cmm_oom_nb); + kthread_stop(cmm_thread_ptr); + del_timer_sync(&cmm_timer); + cmm_free_pages(cmm_pages, &cmm_pages, &cmm_page_list); + cmm_free_pages(cmm_timed_pages, &cmm_timed_pages, &cmm_timed_page_list); +} +module_exit(cmm_exit); + +MODULE_LICENSE("GPL"); diff --git a/kernel/arch/s390/mm/dump_pagetables.c b/kernel/arch/s390/mm/dump_pagetables.c new file mode 100644 index 000000000..8556d6be9 --- /dev/null +++ b/kernel/arch/s390/mm/dump_pagetables.c @@ -0,0 +1,225 @@ +#include <linux/seq_file.h> +#include <linux/debugfs.h> +#include <linux/module.h> +#include <linux/mm.h> +#include <asm/sections.h> +#include <asm/pgtable.h> + +static unsigned long max_addr; + +struct addr_marker { + unsigned long start_address; + const char *name; +}; + +enum address_markers_idx { + IDENTITY_NR = 0, + KERNEL_START_NR, + KERNEL_END_NR, + VMEMMAP_NR, + VMALLOC_NR, + MODULES_NR, +}; + +static struct addr_marker address_markers[] = { + [IDENTITY_NR] = {0, "Identity Mapping"}, + [KERNEL_START_NR] = {(unsigned long)&_stext, "Kernel Image Start"}, + [KERNEL_END_NR] = {(unsigned long)&_end, "Kernel Image End"}, + [VMEMMAP_NR] = {0, "vmemmap Area"}, + [VMALLOC_NR] = {0, "vmalloc Area"}, + [MODULES_NR] = {0, "Modules Area"}, + { -1, NULL } +}; + +struct pg_state { + int level; + unsigned int current_prot; + unsigned long start_address; + unsigned long current_address; + const struct addr_marker *marker; +}; + +static void print_prot(struct seq_file *m, unsigned int pr, int level) +{ + static const char * const level_name[] = + { "ASCE", "PGD", "PUD", "PMD", "PTE" }; + + seq_printf(m, "%s ", level_name[level]); + if (pr & _PAGE_INVALID) { + seq_printf(m, "I\n"); + return; + } + seq_printf(m, "%s", pr & _PAGE_PROTECT ? "RO " : "RW "); + seq_putc(m, '\n'); +} + +static void note_page(struct seq_file *m, struct pg_state *st, + unsigned int new_prot, int level) +{ + static const char units[] = "KMGTPE"; + int width = sizeof(unsigned long) * 2; + const char *unit = units; + unsigned int prot, cur; + unsigned long delta; + + /* + * If we have a "break" in the series, we need to flush the state + * that we have now. "break" is either changing perms, levels or + * address space marker. + */ + prot = new_prot; + cur = st->current_prot; + + if (!st->level) { + /* First entry */ + st->current_prot = new_prot; + st->level = level; + st->marker = address_markers; + seq_printf(m, "---[ %s ]---\n", st->marker->name); + } else if (prot != cur || level != st->level || + st->current_address >= st->marker[1].start_address) { + /* Print the actual finished series */ + seq_printf(m, "0x%0*lx-0x%0*lx", + width, st->start_address, + width, st->current_address); + delta = (st->current_address - st->start_address) >> 10; + while (!(delta & 0x3ff) && unit[1]) { + delta >>= 10; + unit++; + } + seq_printf(m, "%9lu%c ", delta, *unit); + print_prot(m, st->current_prot, st->level); + if (st->current_address >= st->marker[1].start_address) { + st->marker++; + seq_printf(m, "---[ %s ]---\n", st->marker->name); + } + st->start_address = st->current_address; + st->current_prot = new_prot; + st->level = level; + } +} + +/* + * The actual page table walker functions. In order to keep the + * implementation of print_prot() short, we only check and pass + * _PAGE_INVALID and _PAGE_PROTECT flags to note_page() if a region, + * segment or page table entry is invalid or read-only. + * After all it's just a hint that the current level being walked + * contains an invalid or read-only entry. + */ +static void walk_pte_level(struct seq_file *m, struct pg_state *st, + pmd_t *pmd, unsigned long addr) +{ + unsigned int prot; + pte_t *pte; + int i; + + for (i = 0; i < PTRS_PER_PTE && addr < max_addr; i++) { + st->current_address = addr; + pte = pte_offset_kernel(pmd, addr); + prot = pte_val(*pte) & (_PAGE_PROTECT | _PAGE_INVALID); + note_page(m, st, prot, 4); + addr += PAGE_SIZE; + } +} + +static void walk_pmd_level(struct seq_file *m, struct pg_state *st, + pud_t *pud, unsigned long addr) +{ + unsigned int prot; + pmd_t *pmd; + int i; + + for (i = 0; i < PTRS_PER_PMD && addr < max_addr; i++) { + st->current_address = addr; + pmd = pmd_offset(pud, addr); + if (!pmd_none(*pmd)) { + if (pmd_large(*pmd)) { + prot = pmd_val(*pmd) & _SEGMENT_ENTRY_PROTECT; + note_page(m, st, prot, 3); + } else + walk_pte_level(m, st, pmd, addr); + } else + note_page(m, st, _PAGE_INVALID, 3); + addr += PMD_SIZE; + } +} + +static void walk_pud_level(struct seq_file *m, struct pg_state *st, + pgd_t *pgd, unsigned long addr) +{ + unsigned int prot; + pud_t *pud; + int i; + + for (i = 0; i < PTRS_PER_PUD && addr < max_addr; i++) { + st->current_address = addr; + pud = pud_offset(pgd, addr); + if (!pud_none(*pud)) + if (pud_large(*pud)) { + prot = pud_val(*pud) & _REGION3_ENTRY_RO; + note_page(m, st, prot, 2); + } else + walk_pmd_level(m, st, pud, addr); + else + note_page(m, st, _PAGE_INVALID, 2); + addr += PUD_SIZE; + } +} + +static void walk_pgd_level(struct seq_file *m) +{ + unsigned long addr = 0; + struct pg_state st; + pgd_t *pgd; + int i; + + memset(&st, 0, sizeof(st)); + for (i = 0; i < PTRS_PER_PGD && addr < max_addr; i++) { + st.current_address = addr; + pgd = pgd_offset_k(addr); + if (!pgd_none(*pgd)) + walk_pud_level(m, &st, pgd, addr); + else + note_page(m, &st, _PAGE_INVALID, 1); + addr += PGDIR_SIZE; + } + /* Flush out the last page */ + st.current_address = max_addr; + note_page(m, &st, 0, 0); +} + +static int ptdump_show(struct seq_file *m, void *v) +{ + walk_pgd_level(m); + return 0; +} + +static int ptdump_open(struct inode *inode, struct file *filp) +{ + return single_open(filp, ptdump_show, NULL); +} + +static const struct file_operations ptdump_fops = { + .open = ptdump_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int pt_dump_init(void) +{ + /* + * Figure out the maximum virtual address being accessible with the + * kernel ASCE. We need this to keep the page table walker functions + * from accessing non-existent entries. + */ + max_addr = (S390_lowcore.kernel_asce & _REGION_ENTRY_TYPE_MASK) >> 2; + max_addr = 1UL << (max_addr * 11 + 31); + address_markers[MODULES_NR].start_address = MODULES_VADDR; + address_markers[VMEMMAP_NR].start_address = (unsigned long) vmemmap; + address_markers[VMALLOC_NR].start_address = VMALLOC_START; + debugfs_create_file("kernel_page_tables", 0400, NULL, NULL, &ptdump_fops); + return 0; +} +device_initcall(pt_dump_init); diff --git a/kernel/arch/s390/mm/extable.c b/kernel/arch/s390/mm/extable.c new file mode 100644 index 000000000..4d1ee8886 --- /dev/null +++ b/kernel/arch/s390/mm/extable.c @@ -0,0 +1,81 @@ +#include <linux/module.h> +#include <linux/sort.h> +#include <asm/uaccess.h> + +/* + * Search one exception table for an entry corresponding to the + * given instruction address, and return the address of the entry, + * or NULL if none is found. + * We use a binary search, and thus we assume that the table is + * already sorted. + */ +const struct exception_table_entry * +search_extable(const struct exception_table_entry *first, + const struct exception_table_entry *last, + unsigned long value) +{ + const struct exception_table_entry *mid; + unsigned long addr; + + while (first <= last) { + mid = ((last - first) >> 1) + first; + addr = extable_insn(mid); + if (addr < value) + first = mid + 1; + else if (addr > value) + last = mid - 1; + else + return mid; + } + return NULL; +} + +/* + * The exception table needs to be sorted so that the binary + * search that we use to find entries in it works properly. + * This is used both for the kernel exception table and for + * the exception tables of modules that get loaded. + * + */ +static int cmp_ex(const void *a, const void *b) +{ + const struct exception_table_entry *x = a, *y = b; + + /* This compare is only valid after normalization. */ + return x->insn - y->insn; +} + +void sort_extable(struct exception_table_entry *start, + struct exception_table_entry *finish) +{ + struct exception_table_entry *p; + int i; + + /* Normalize entries to being relative to the start of the section */ + for (p = start, i = 0; p < finish; p++, i += 8) + p->insn += i; + sort(start, finish - start, sizeof(*start), cmp_ex, NULL); + /* Denormalize all entries */ + for (p = start, i = 0; p < finish; p++, i += 8) + p->insn -= i; +} + +#ifdef CONFIG_MODULES +/* + * If the exception table is sorted, any referring to the module init + * will be at the beginning or the end. + */ +void trim_init_extable(struct module *m) +{ + /* Trim the beginning */ + while (m->num_exentries && + within_module_init(extable_insn(&m->extable[0]), m)) { + m->extable++; + m->num_exentries--; + } + /* Trim the end */ + while (m->num_exentries && + within_module_init(extable_insn(&m->extable[m->num_exentries-1]), m)) + m->num_exentries--; +} +#endif /* CONFIG_MODULES */ diff --git a/kernel/arch/s390/mm/extmem.c b/kernel/arch/s390/mm/extmem.c new file mode 100644 index 000000000..23c496957 --- /dev/null +++ b/kernel/arch/s390/mm/extmem.c @@ -0,0 +1,763 @@ +/* + * Author(s)......: Carsten Otte <cotte@de.ibm.com> + * Rob M van der Heij <rvdheij@nl.ibm.com> + * Steven Shultz <shultzss@us.ibm.com> + * Bugreports.to..: <Linux390@de.ibm.com> + * Copyright IBM Corp. 2002, 2004 + */ + +#define KMSG_COMPONENT "extmem" +#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt + +#include <linux/kernel.h> +#include <linux/string.h> +#include <linux/spinlock.h> +#include <linux/list.h> +#include <linux/slab.h> +#include <linux/module.h> +#include <linux/bootmem.h> +#include <linux/ctype.h> +#include <linux/ioport.h> +#include <asm/page.h> +#include <asm/pgtable.h> +#include <asm/ebcdic.h> +#include <asm/errno.h> +#include <asm/extmem.h> +#include <asm/cpcmd.h> +#include <asm/setup.h> + +#define DCSS_LOADSHR 0x00 +#define DCSS_LOADNSR 0x04 +#define DCSS_PURGESEG 0x08 +#define DCSS_FINDSEG 0x0c +#define DCSS_LOADNOLY 0x10 +#define DCSS_SEGEXT 0x18 +#define DCSS_LOADSHRX 0x20 +#define DCSS_LOADNSRX 0x24 +#define DCSS_FINDSEGX 0x2c +#define DCSS_SEGEXTX 0x38 +#define DCSS_FINDSEGA 0x0c + +struct qrange { + unsigned long start; /* last byte type */ + unsigned long end; /* last byte reserved */ +}; + +struct qout64 { + unsigned long segstart; + unsigned long segend; + int segcnt; + int segrcnt; + struct qrange range[6]; +}; + +struct qrange_old { + unsigned int start; /* last byte type */ + unsigned int end; /* last byte reserved */ +}; + +/* output area format for the Diag x'64' old subcode x'18' */ +struct qout64_old { + int segstart; + int segend; + int segcnt; + int segrcnt; + struct qrange_old range[6]; +}; + +struct qin64 { + char qopcode; + char rsrv1[3]; + char qrcode; + char rsrv2[3]; + char qname[8]; + unsigned int qoutptr; + short int qoutlen; +}; + +struct dcss_segment { + struct list_head list; + char dcss_name[8]; + char res_name[15]; + unsigned long start_addr; + unsigned long end; + atomic_t ref_count; + int do_nonshared; + unsigned int vm_segtype; + struct qrange range[6]; + int segcnt; + struct resource *res; +}; + +static DEFINE_MUTEX(dcss_lock); +static LIST_HEAD(dcss_list); +static char *segtype_string[] = { "SW", "EW", "SR", "ER", "SN", "EN", "SC", + "EW/EN-MIXED" }; +static int loadshr_scode, loadnsr_scode, findseg_scode; +static int segext_scode, purgeseg_scode; +static int scode_set; + +/* set correct Diag x'64' subcodes. */ +static int +dcss_set_subcodes(void) +{ + char *name = kmalloc(8 * sizeof(char), GFP_KERNEL | GFP_DMA); + unsigned long rx, ry; + int rc; + + if (name == NULL) + return -ENOMEM; + + rx = (unsigned long) name; + ry = DCSS_FINDSEGX; + + strcpy(name, "dummy"); + asm volatile( + " diag %0,%1,0x64\n" + "0: ipm %2\n" + " srl %2,28\n" + " j 2f\n" + "1: la %2,3\n" + "2:\n" + EX_TABLE(0b, 1b) + : "+d" (rx), "+d" (ry), "=d" (rc) : : "cc"); + + kfree(name); + /* Diag x'64' new subcodes are supported, set to new subcodes */ + if (rc != 3) { + loadshr_scode = DCSS_LOADSHRX; + loadnsr_scode = DCSS_LOADNSRX; + purgeseg_scode = DCSS_PURGESEG; + findseg_scode = DCSS_FINDSEGX; + segext_scode = DCSS_SEGEXTX; + return 0; + } + /* Diag x'64' new subcodes are not supported, set to old subcodes */ + loadshr_scode = DCSS_LOADNOLY; + loadnsr_scode = DCSS_LOADNSR; + purgeseg_scode = DCSS_PURGESEG; + findseg_scode = DCSS_FINDSEG; + segext_scode = DCSS_SEGEXT; + return 0; +} + +/* + * Create the 8 bytes, ebcdic VM segment name from + * an ascii name. + */ +static void +dcss_mkname(char *name, char *dcss_name) +{ + int i; + + for (i = 0; i < 8; i++) { + if (name[i] == '\0') + break; + dcss_name[i] = toupper(name[i]); + }; + for (; i < 8; i++) + dcss_name[i] = ' '; + ASCEBC(dcss_name, 8); +} + + +/* + * search all segments in dcss_list, and return the one + * namend *name. If not found, return NULL. + */ +static struct dcss_segment * +segment_by_name (char *name) +{ + char dcss_name[9]; + struct list_head *l; + struct dcss_segment *tmp, *retval = NULL; + + BUG_ON(!mutex_is_locked(&dcss_lock)); + dcss_mkname (name, dcss_name); + list_for_each (l, &dcss_list) { + tmp = list_entry (l, struct dcss_segment, list); + if (memcmp(tmp->dcss_name, dcss_name, 8) == 0) { + retval = tmp; + break; + } + } + return retval; +} + + +/* + * Perform a function on a dcss segment. + */ +static inline int +dcss_diag(int *func, void *parameter, + unsigned long *ret1, unsigned long *ret2) +{ + unsigned long rx, ry; + int rc; + + if (scode_set == 0) { + rc = dcss_set_subcodes(); + if (rc < 0) + return rc; + scode_set = 1; + } + rx = (unsigned long) parameter; + ry = (unsigned long) *func; + + /* 64-bit Diag x'64' new subcode, keep in 64-bit addressing mode */ + if (*func > DCSS_SEGEXT) + asm volatile( + " diag %0,%1,0x64\n" + " ipm %2\n" + " srl %2,28\n" + : "+d" (rx), "+d" (ry), "=d" (rc) : : "cc"); + /* 31-bit Diag x'64' old subcode, switch to 31-bit addressing mode */ + else + asm volatile( + " sam31\n" + " diag %0,%1,0x64\n" + " sam64\n" + " ipm %2\n" + " srl %2,28\n" + : "+d" (rx), "+d" (ry), "=d" (rc) : : "cc"); + *ret1 = rx; + *ret2 = ry; + return rc; +} + +static inline int +dcss_diag_translate_rc (int vm_rc) { + if (vm_rc == 44) + return -ENOENT; + return -EIO; +} + + +/* do a diag to get info about a segment. + * fills start_address, end and vm_segtype fields + */ +static int +query_segment_type (struct dcss_segment *seg) +{ + unsigned long dummy, vmrc; + int diag_cc, rc, i; + struct qout64 *qout; + struct qin64 *qin; + + qin = kmalloc(sizeof(*qin), GFP_KERNEL | GFP_DMA); + qout = kmalloc(sizeof(*qout), GFP_KERNEL | GFP_DMA); + if ((qin == NULL) || (qout == NULL)) { + rc = -ENOMEM; + goto out_free; + } + + /* initialize diag input parameters */ + qin->qopcode = DCSS_FINDSEGA; + qin->qoutptr = (unsigned long) qout; + qin->qoutlen = sizeof(struct qout64); + memcpy (qin->qname, seg->dcss_name, 8); + + diag_cc = dcss_diag(&segext_scode, qin, &dummy, &vmrc); + + if (diag_cc < 0) { + rc = diag_cc; + goto out_free; + } + if (diag_cc > 1) { + pr_warning("Querying a DCSS type failed with rc=%ld\n", vmrc); + rc = dcss_diag_translate_rc (vmrc); + goto out_free; + } + + /* Only old format of output area of Diagnose x'64' is supported, + copy data for the new format. */ + if (segext_scode == DCSS_SEGEXT) { + struct qout64_old *qout_old; + qout_old = kzalloc(sizeof(*qout_old), GFP_KERNEL | GFP_DMA); + if (qout_old == NULL) { + rc = -ENOMEM; + goto out_free; + } + memcpy(qout_old, qout, sizeof(struct qout64_old)); + qout->segstart = (unsigned long) qout_old->segstart; + qout->segend = (unsigned long) qout_old->segend; + qout->segcnt = qout_old->segcnt; + qout->segrcnt = qout_old->segrcnt; + + if (qout->segcnt > 6) + qout->segrcnt = 6; + for (i = 0; i < qout->segrcnt; i++) { + qout->range[i].start = + (unsigned long) qout_old->range[i].start; + qout->range[i].end = + (unsigned long) qout_old->range[i].end; + } + kfree(qout_old); + } + if (qout->segcnt > 6) { + rc = -EOPNOTSUPP; + goto out_free; + } + + if (qout->segcnt == 1) { + seg->vm_segtype = qout->range[0].start & 0xff; + } else { + /* multi-part segment. only one type supported here: + - all parts are contiguous + - all parts are either EW or EN type + - maximum 6 parts allowed */ + unsigned long start = qout->segstart >> PAGE_SHIFT; + for (i=0; i<qout->segcnt; i++) { + if (((qout->range[i].start & 0xff) != SEG_TYPE_EW) && + ((qout->range[i].start & 0xff) != SEG_TYPE_EN)) { + rc = -EOPNOTSUPP; + goto out_free; + } + if (start != qout->range[i].start >> PAGE_SHIFT) { + rc = -EOPNOTSUPP; + goto out_free; + } + start = (qout->range[i].end >> PAGE_SHIFT) + 1; + } + seg->vm_segtype = SEG_TYPE_EWEN; + } + + /* analyze diag output and update seg */ + seg->start_addr = qout->segstart; + seg->end = qout->segend; + + memcpy (seg->range, qout->range, 6*sizeof(struct qrange)); + seg->segcnt = qout->segcnt; + + rc = 0; + + out_free: + kfree(qin); + kfree(qout); + return rc; +} + +/* + * get info about a segment + * possible return values: + * -ENOSYS : we are not running on VM + * -EIO : could not perform query diagnose + * -ENOENT : no such segment + * -EOPNOTSUPP: multi-part segment cannot be used with linux + * -ENOMEM : out of memory + * 0 .. 6 : type of segment as defined in include/asm-s390/extmem.h + */ +int +segment_type (char* name) +{ + int rc; + struct dcss_segment seg; + + if (!MACHINE_IS_VM) + return -ENOSYS; + + dcss_mkname(name, seg.dcss_name); + rc = query_segment_type (&seg); + if (rc < 0) + return rc; + return seg.vm_segtype; +} + +/* + * check if segment collides with other segments that are currently loaded + * returns 1 if this is the case, 0 if no collision was found + */ +static int +segment_overlaps_others (struct dcss_segment *seg) +{ + struct list_head *l; + struct dcss_segment *tmp; + + BUG_ON(!mutex_is_locked(&dcss_lock)); + list_for_each(l, &dcss_list) { + tmp = list_entry(l, struct dcss_segment, list); + if ((tmp->start_addr >> 20) > (seg->end >> 20)) + continue; + if ((tmp->end >> 20) < (seg->start_addr >> 20)) + continue; + if (seg == tmp) + continue; + return 1; + } + return 0; +} + +/* + * real segment loading function, called from segment_load + */ +static int +__segment_load (char *name, int do_nonshared, unsigned long *addr, unsigned long *end) +{ + unsigned long start_addr, end_addr, dummy; + struct dcss_segment *seg; + int rc, diag_cc; + + start_addr = end_addr = 0; + seg = kmalloc(sizeof(*seg), GFP_KERNEL | GFP_DMA); + if (seg == NULL) { + rc = -ENOMEM; + goto out; + } + dcss_mkname (name, seg->dcss_name); + rc = query_segment_type (seg); + if (rc < 0) + goto out_free; + + if (loadshr_scode == DCSS_LOADSHRX) { + if (segment_overlaps_others(seg)) { + rc = -EBUSY; + goto out_free; + } + } + + rc = vmem_add_mapping(seg->start_addr, seg->end - seg->start_addr + 1); + + if (rc) + goto out_free; + + seg->res = kzalloc(sizeof(struct resource), GFP_KERNEL); + if (seg->res == NULL) { + rc = -ENOMEM; + goto out_shared; + } + seg->res->flags = IORESOURCE_BUSY | IORESOURCE_MEM; + seg->res->start = seg->start_addr; + seg->res->end = seg->end; + memcpy(&seg->res_name, seg->dcss_name, 8); + EBCASC(seg->res_name, 8); + seg->res_name[8] = '\0'; + strncat(seg->res_name, " (DCSS)", 7); + seg->res->name = seg->res_name; + rc = seg->vm_segtype; + if (rc == SEG_TYPE_SC || + ((rc == SEG_TYPE_SR || rc == SEG_TYPE_ER) && !do_nonshared)) + seg->res->flags |= IORESOURCE_READONLY; + if (request_resource(&iomem_resource, seg->res)) { + rc = -EBUSY; + kfree(seg->res); + goto out_shared; + } + + if (do_nonshared) + diag_cc = dcss_diag(&loadnsr_scode, seg->dcss_name, + &start_addr, &end_addr); + else + diag_cc = dcss_diag(&loadshr_scode, seg->dcss_name, + &start_addr, &end_addr); + if (diag_cc < 0) { + dcss_diag(&purgeseg_scode, seg->dcss_name, + &dummy, &dummy); + rc = diag_cc; + goto out_resource; + } + if (diag_cc > 1) { + pr_warning("Loading DCSS %s failed with rc=%ld\n", name, + end_addr); + rc = dcss_diag_translate_rc(end_addr); + dcss_diag(&purgeseg_scode, seg->dcss_name, + &dummy, &dummy); + goto out_resource; + } + seg->start_addr = start_addr; + seg->end = end_addr; + seg->do_nonshared = do_nonshared; + atomic_set(&seg->ref_count, 1); + list_add(&seg->list, &dcss_list); + *addr = seg->start_addr; + *end = seg->end; + if (do_nonshared) + pr_info("DCSS %s of range %p to %p and type %s loaded as " + "exclusive-writable\n", name, (void*) seg->start_addr, + (void*) seg->end, segtype_string[seg->vm_segtype]); + else { + pr_info("DCSS %s of range %p to %p and type %s loaded in " + "shared access mode\n", name, (void*) seg->start_addr, + (void*) seg->end, segtype_string[seg->vm_segtype]); + } + goto out; + out_resource: + release_resource(seg->res); + kfree(seg->res); + out_shared: + vmem_remove_mapping(seg->start_addr, seg->end - seg->start_addr + 1); + out_free: + kfree(seg); + out: + return rc; +} + +/* + * this function loads a DCSS segment + * name : name of the DCSS + * do_nonshared : 0 indicates that the dcss should be shared with other linux images + * 1 indicates that the dcss should be exclusive for this linux image + * addr : will be filled with start address of the segment + * end : will be filled with end address of the segment + * return values: + * -ENOSYS : we are not running on VM + * -EIO : could not perform query or load diagnose + * -ENOENT : no such segment + * -EOPNOTSUPP: multi-part segment cannot be used with linux + * -ENOSPC : segment cannot be used (overlaps with storage) + * -EBUSY : segment can temporarily not be used (overlaps with dcss) + * -ERANGE : segment cannot be used (exceeds kernel mapping range) + * -EPERM : segment is currently loaded with incompatible permissions + * -ENOMEM : out of memory + * 0 .. 6 : type of segment as defined in include/asm-s390/extmem.h + */ +int +segment_load (char *name, int do_nonshared, unsigned long *addr, + unsigned long *end) +{ + struct dcss_segment *seg; + int rc; + + if (!MACHINE_IS_VM) + return -ENOSYS; + + mutex_lock(&dcss_lock); + seg = segment_by_name (name); + if (seg == NULL) + rc = __segment_load (name, do_nonshared, addr, end); + else { + if (do_nonshared == seg->do_nonshared) { + atomic_inc(&seg->ref_count); + *addr = seg->start_addr; + *end = seg->end; + rc = seg->vm_segtype; + } else { + *addr = *end = 0; + rc = -EPERM; + } + } + mutex_unlock(&dcss_lock); + return rc; +} + +/* + * this function modifies the shared state of a DCSS segment. note that + * name : name of the DCSS + * do_nonshared : 0 indicates that the dcss should be shared with other linux images + * 1 indicates that the dcss should be exclusive for this linux image + * return values: + * -EIO : could not perform load diagnose (segment gone!) + * -ENOENT : no such segment (segment gone!) + * -EAGAIN : segment is in use by other exploiters, try later + * -EINVAL : no segment with the given name is currently loaded - name invalid + * -EBUSY : segment can temporarily not be used (overlaps with dcss) + * 0 : operation succeeded + */ +int +segment_modify_shared (char *name, int do_nonshared) +{ + struct dcss_segment *seg; + unsigned long start_addr, end_addr, dummy; + int rc, diag_cc; + + start_addr = end_addr = 0; + mutex_lock(&dcss_lock); + seg = segment_by_name (name); + if (seg == NULL) { + rc = -EINVAL; + goto out_unlock; + } + if (do_nonshared == seg->do_nonshared) { + pr_info("DCSS %s is already in the requested access " + "mode\n", name); + rc = 0; + goto out_unlock; + } + if (atomic_read (&seg->ref_count) != 1) { + pr_warning("DCSS %s is in use and cannot be reloaded\n", + name); + rc = -EAGAIN; + goto out_unlock; + } + release_resource(seg->res); + if (do_nonshared) + seg->res->flags &= ~IORESOURCE_READONLY; + else + if (seg->vm_segtype == SEG_TYPE_SR || + seg->vm_segtype == SEG_TYPE_ER) + seg->res->flags |= IORESOURCE_READONLY; + + if (request_resource(&iomem_resource, seg->res)) { + pr_warning("DCSS %s overlaps with used memory resources " + "and cannot be reloaded\n", name); + rc = -EBUSY; + kfree(seg->res); + goto out_del_mem; + } + + dcss_diag(&purgeseg_scode, seg->dcss_name, &dummy, &dummy); + if (do_nonshared) + diag_cc = dcss_diag(&loadnsr_scode, seg->dcss_name, + &start_addr, &end_addr); + else + diag_cc = dcss_diag(&loadshr_scode, seg->dcss_name, + &start_addr, &end_addr); + if (diag_cc < 0) { + rc = diag_cc; + goto out_del_res; + } + if (diag_cc > 1) { + pr_warning("Reloading DCSS %s failed with rc=%ld\n", name, + end_addr); + rc = dcss_diag_translate_rc(end_addr); + goto out_del_res; + } + seg->start_addr = start_addr; + seg->end = end_addr; + seg->do_nonshared = do_nonshared; + rc = 0; + goto out_unlock; + out_del_res: + release_resource(seg->res); + kfree(seg->res); + out_del_mem: + vmem_remove_mapping(seg->start_addr, seg->end - seg->start_addr + 1); + list_del(&seg->list); + dcss_diag(&purgeseg_scode, seg->dcss_name, &dummy, &dummy); + kfree(seg); + out_unlock: + mutex_unlock(&dcss_lock); + return rc; +} + +/* + * Decrease the use count of a DCSS segment and remove + * it from the address space if nobody is using it + * any longer. + */ +void +segment_unload(char *name) +{ + unsigned long dummy; + struct dcss_segment *seg; + + if (!MACHINE_IS_VM) + return; + + mutex_lock(&dcss_lock); + seg = segment_by_name (name); + if (seg == NULL) { + pr_err("Unloading unknown DCSS %s failed\n", name); + goto out_unlock; + } + if (atomic_dec_return(&seg->ref_count) != 0) + goto out_unlock; + release_resource(seg->res); + kfree(seg->res); + vmem_remove_mapping(seg->start_addr, seg->end - seg->start_addr + 1); + list_del(&seg->list); + dcss_diag(&purgeseg_scode, seg->dcss_name, &dummy, &dummy); + kfree(seg); +out_unlock: + mutex_unlock(&dcss_lock); +} + +/* + * save segment content permanently + */ +void +segment_save(char *name) +{ + struct dcss_segment *seg; + char cmd1[160]; + char cmd2[80]; + int i, response; + + if (!MACHINE_IS_VM) + return; + + mutex_lock(&dcss_lock); + seg = segment_by_name (name); + + if (seg == NULL) { + pr_err("Saving unknown DCSS %s failed\n", name); + goto out; + } + + sprintf(cmd1, "DEFSEG %s", name); + for (i=0; i<seg->segcnt; i++) { + sprintf(cmd1+strlen(cmd1), " %lX-%lX %s", + seg->range[i].start >> PAGE_SHIFT, + seg->range[i].end >> PAGE_SHIFT, + segtype_string[seg->range[i].start & 0xff]); + } + sprintf(cmd2, "SAVESEG %s", name); + response = 0; + cpcmd(cmd1, NULL, 0, &response); + if (response) { + pr_err("Saving a DCSS failed with DEFSEG response code " + "%i\n", response); + goto out; + } + cpcmd(cmd2, NULL, 0, &response); + if (response) { + pr_err("Saving a DCSS failed with SAVESEG response code " + "%i\n", response); + goto out; + } +out: + mutex_unlock(&dcss_lock); +} + +/* + * print appropriate error message for segment_load()/segment_type() + * return code + */ +void segment_warning(int rc, char *seg_name) +{ + switch (rc) { + case -ENOENT: + pr_err("DCSS %s cannot be loaded or queried\n", seg_name); + break; + case -ENOSYS: + pr_err("DCSS %s cannot be loaded or queried without " + "z/VM\n", seg_name); + break; + case -EIO: + pr_err("Loading or querying DCSS %s resulted in a " + "hardware error\n", seg_name); + break; + case -EOPNOTSUPP: + pr_err("DCSS %s has multiple page ranges and cannot be " + "loaded or queried\n", seg_name); + break; + case -ENOSPC: + pr_err("DCSS %s overlaps with used storage and cannot " + "be loaded\n", seg_name); + break; + case -EBUSY: + pr_err("%s needs used memory resources and cannot be " + "loaded or queried\n", seg_name); + break; + case -EPERM: + pr_err("DCSS %s is already loaded in a different access " + "mode\n", seg_name); + break; + case -ENOMEM: + pr_err("There is not enough memory to load or query " + "DCSS %s\n", seg_name); + break; + case -ERANGE: + pr_err("DCSS %s exceeds the kernel mapping range (%lu) " + "and cannot be loaded\n", seg_name, VMEM_MAX_PHYS); + break; + default: + break; + } +} + +EXPORT_SYMBOL(segment_load); +EXPORT_SYMBOL(segment_unload); +EXPORT_SYMBOL(segment_save); +EXPORT_SYMBOL(segment_type); +EXPORT_SYMBOL(segment_modify_shared); +EXPORT_SYMBOL(segment_warning); diff --git a/kernel/arch/s390/mm/fault.c b/kernel/arch/s390/mm/fault.c new file mode 100644 index 000000000..4c8f5d7f9 --- /dev/null +++ b/kernel/arch/s390/mm/fault.c @@ -0,0 +1,758 @@ +/* + * S390 version + * Copyright IBM Corp. 1999 + * Author(s): Hartmut Penner (hp@de.ibm.com) + * Ulrich Weigand (uweigand@de.ibm.com) + * + * Derived from "arch/i386/mm/fault.c" + * Copyright (C) 1995 Linus Torvalds + */ + +#include <linux/kernel_stat.h> +#include <linux/perf_event.h> +#include <linux/signal.h> +#include <linux/sched.h> +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/string.h> +#include <linux/types.h> +#include <linux/ptrace.h> +#include <linux/mman.h> +#include <linux/mm.h> +#include <linux/compat.h> +#include <linux/smp.h> +#include <linux/kdebug.h> +#include <linux/init.h> +#include <linux/console.h> +#include <linux/module.h> +#include <linux/hardirq.h> +#include <linux/kprobes.h> +#include <linux/uaccess.h> +#include <linux/hugetlb.h> +#include <asm/asm-offsets.h> +#include <asm/pgtable.h> +#include <asm/irq.h> +#include <asm/mmu_context.h> +#include <asm/facility.h> +#include "../kernel/entry.h" + +#define __FAIL_ADDR_MASK -4096L +#define __SUBCODE_MASK 0x0600 +#define __PF_RES_FIELD 0x8000000000000000ULL + +#define VM_FAULT_BADCONTEXT 0x010000 +#define VM_FAULT_BADMAP 0x020000 +#define VM_FAULT_BADACCESS 0x040000 +#define VM_FAULT_SIGNAL 0x080000 +#define VM_FAULT_PFAULT 0x100000 + +static unsigned long store_indication __read_mostly; + +static int __init fault_init(void) +{ + if (test_facility(75)) + store_indication = 0xc00; + return 0; +} +early_initcall(fault_init); + +static inline int notify_page_fault(struct pt_regs *regs) +{ + int ret = 0; + + /* kprobe_running() needs smp_processor_id() */ + if (kprobes_built_in() && !user_mode(regs)) { + preempt_disable(); + if (kprobe_running() && kprobe_fault_handler(regs, 14)) + ret = 1; + preempt_enable(); + } + return ret; +} + + +/* + * Unlock any spinlocks which will prevent us from getting the + * message out. + */ +void bust_spinlocks(int yes) +{ + if (yes) { + oops_in_progress = 1; + } else { + int loglevel_save = console_loglevel; + console_unblank(); + oops_in_progress = 0; + /* + * OK, the message is on the console. Now we call printk() + * without oops_in_progress set so that printk will give klogd + * a poke. Hold onto your hats... + */ + console_loglevel = 15; + printk(" "); + console_loglevel = loglevel_save; + } +} + +/* + * Returns the address space associated with the fault. + * Returns 0 for kernel space and 1 for user space. + */ +static inline int user_space_fault(struct pt_regs *regs) +{ + unsigned long trans_exc_code; + + /* + * The lowest two bits of the translation exception + * identification indicate which paging table was used. + */ + trans_exc_code = regs->int_parm_long & 3; + if (trans_exc_code == 3) /* home space -> kernel */ + return 0; + if (user_mode(regs)) + return 1; + if (trans_exc_code == 2) /* secondary space -> set_fs */ + return current->thread.mm_segment.ar4; + if (current->flags & PF_VCPU) + return 1; + return 0; +} + +static int bad_address(void *p) +{ + unsigned long dummy; + + return probe_kernel_address((unsigned long *)p, dummy); +} + +static void dump_pagetable(unsigned long asce, unsigned long address) +{ + unsigned long *table = __va(asce & PAGE_MASK); + + pr_alert("AS:%016lx ", asce); + switch (asce & _ASCE_TYPE_MASK) { + case _ASCE_TYPE_REGION1: + table = table + ((address >> 53) & 0x7ff); + if (bad_address(table)) + goto bad; + pr_cont("R1:%016lx ", *table); + if (*table & _REGION_ENTRY_INVALID) + goto out; + table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN); + /* fallthrough */ + case _ASCE_TYPE_REGION2: + table = table + ((address >> 42) & 0x7ff); + if (bad_address(table)) + goto bad; + pr_cont("R2:%016lx ", *table); + if (*table & _REGION_ENTRY_INVALID) + goto out; + table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN); + /* fallthrough */ + case _ASCE_TYPE_REGION3: + table = table + ((address >> 31) & 0x7ff); + if (bad_address(table)) + goto bad; + pr_cont("R3:%016lx ", *table); + if (*table & (_REGION_ENTRY_INVALID | _REGION3_ENTRY_LARGE)) + goto out; + table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN); + /* fallthrough */ + case _ASCE_TYPE_SEGMENT: + table = table + ((address >> 20) & 0x7ff); + if (bad_address(table)) + goto bad; + pr_cont("S:%016lx ", *table); + if (*table & (_SEGMENT_ENTRY_INVALID | _SEGMENT_ENTRY_LARGE)) + goto out; + table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN); + } + table = table + ((address >> 12) & 0xff); + if (bad_address(table)) + goto bad; + pr_cont("P:%016lx ", *table); +out: + pr_cont("\n"); + return; +bad: + pr_cont("BAD\n"); +} + +static void dump_fault_info(struct pt_regs *regs) +{ + unsigned long asce; + + pr_alert("Fault in "); + switch (regs->int_parm_long & 3) { + case 3: + pr_cont("home space "); + break; + case 2: + pr_cont("secondary space "); + break; + case 1: + pr_cont("access register "); + break; + case 0: + pr_cont("primary space "); + break; + } + pr_cont("mode while using "); + if (!user_space_fault(regs)) { + asce = S390_lowcore.kernel_asce; + pr_cont("kernel "); + } +#ifdef CONFIG_PGSTE + else if ((current->flags & PF_VCPU) && S390_lowcore.gmap) { + struct gmap *gmap = (struct gmap *)S390_lowcore.gmap; + asce = gmap->asce; + pr_cont("gmap "); + } +#endif + else { + asce = S390_lowcore.user_asce; + pr_cont("user "); + } + pr_cont("ASCE.\n"); + dump_pagetable(asce, regs->int_parm_long & __FAIL_ADDR_MASK); +} + +static inline void report_user_fault(struct pt_regs *regs, long signr) +{ + if ((task_pid_nr(current) > 1) && !show_unhandled_signals) + return; + if (!unhandled_signal(current, signr)) + return; + if (!printk_ratelimit()) + return; + printk(KERN_ALERT "User process fault: interruption code %04x ilc:%d ", + regs->int_code & 0xffff, regs->int_code >> 17); + print_vma_addr(KERN_CONT "in ", regs->psw.addr & PSW_ADDR_INSN); + printk(KERN_CONT "\n"); + printk(KERN_ALERT "failing address: %016lx TEID: %016lx\n", + regs->int_parm_long & __FAIL_ADDR_MASK, regs->int_parm_long); + dump_fault_info(regs); + show_regs(regs); +} + +/* + * Send SIGSEGV to task. This is an external routine + * to keep the stack usage of do_page_fault small. + */ +static noinline void do_sigsegv(struct pt_regs *regs, int si_code) +{ + struct siginfo si; + + report_user_fault(regs, SIGSEGV); + si.si_signo = SIGSEGV; + si.si_code = si_code; + si.si_addr = (void __user *)(regs->int_parm_long & __FAIL_ADDR_MASK); + force_sig_info(SIGSEGV, &si, current); +} + +static noinline void do_no_context(struct pt_regs *regs) +{ + const struct exception_table_entry *fixup; + unsigned long address; + + /* Are we prepared to handle this kernel fault? */ + fixup = search_exception_tables(regs->psw.addr & PSW_ADDR_INSN); + if (fixup) { + regs->psw.addr = extable_fixup(fixup) | PSW_ADDR_AMODE; + return; + } + + /* + * Oops. The kernel tried to access some bad page. We'll have to + * terminate things with extreme prejudice. + */ + address = regs->int_parm_long & __FAIL_ADDR_MASK; + if (!user_space_fault(regs)) + printk(KERN_ALERT "Unable to handle kernel pointer dereference" + " in virtual kernel address space\n"); + else + printk(KERN_ALERT "Unable to handle kernel paging request" + " in virtual user address space\n"); + printk(KERN_ALERT "failing address: %016lx TEID: %016lx\n", + regs->int_parm_long & __FAIL_ADDR_MASK, regs->int_parm_long); + dump_fault_info(regs); + die(regs, "Oops"); + do_exit(SIGKILL); +} + +static noinline void do_low_address(struct pt_regs *regs) +{ + /* Low-address protection hit in kernel mode means + NULL pointer write access in kernel mode. */ + if (regs->psw.mask & PSW_MASK_PSTATE) { + /* Low-address protection hit in user mode 'cannot happen'. */ + die (regs, "Low-address protection"); + do_exit(SIGKILL); + } + + do_no_context(regs); +} + +static noinline void do_sigbus(struct pt_regs *regs) +{ + struct task_struct *tsk = current; + struct siginfo si; + + /* + * Send a sigbus, regardless of whether we were in kernel + * or user mode. + */ + si.si_signo = SIGBUS; + si.si_errno = 0; + si.si_code = BUS_ADRERR; + si.si_addr = (void __user *)(regs->int_parm_long & __FAIL_ADDR_MASK); + force_sig_info(SIGBUS, &si, tsk); +} + +static noinline void do_fault_error(struct pt_regs *regs, int fault) +{ + int si_code; + + switch (fault) { + case VM_FAULT_BADACCESS: + case VM_FAULT_BADMAP: + /* Bad memory access. Check if it is kernel or user space. */ + if (user_mode(regs)) { + /* User mode accesses just cause a SIGSEGV */ + si_code = (fault == VM_FAULT_BADMAP) ? + SEGV_MAPERR : SEGV_ACCERR; + do_sigsegv(regs, si_code); + return; + } + case VM_FAULT_BADCONTEXT: + case VM_FAULT_PFAULT: + do_no_context(regs); + break; + case VM_FAULT_SIGNAL: + if (!user_mode(regs)) + do_no_context(regs); + break; + default: /* fault & VM_FAULT_ERROR */ + if (fault & VM_FAULT_OOM) { + if (!user_mode(regs)) + do_no_context(regs); + else + pagefault_out_of_memory(); + } else if (fault & VM_FAULT_SIGSEGV) { + /* Kernel mode? Handle exceptions or die */ + if (!user_mode(regs)) + do_no_context(regs); + else + do_sigsegv(regs, SEGV_MAPERR); + } else if (fault & VM_FAULT_SIGBUS) { + /* Kernel mode? Handle exceptions or die */ + if (!user_mode(regs)) + do_no_context(regs); + else + do_sigbus(regs); + } else + BUG(); + break; + } +} + +/* + * This routine handles page faults. It determines the address, + * and the problem, and then passes it off to one of the appropriate + * routines. + * + * interruption code (int_code): + * 04 Protection -> Write-Protection (suprression) + * 10 Segment translation -> Not present (nullification) + * 11 Page translation -> Not present (nullification) + * 3b Region third trans. -> Not present (nullification) + */ +static inline int do_exception(struct pt_regs *regs, int access) +{ +#ifdef CONFIG_PGSTE + struct gmap *gmap; +#endif + struct task_struct *tsk; + struct mm_struct *mm; + struct vm_area_struct *vma; + unsigned long trans_exc_code; + unsigned long address; + unsigned int flags; + int fault; + + tsk = current; + /* + * The instruction that caused the program check has + * been nullified. Don't signal single step via SIGTRAP. + */ + clear_pt_regs_flag(regs, PIF_PER_TRAP); + + if (notify_page_fault(regs)) + return 0; + + mm = tsk->mm; + trans_exc_code = regs->int_parm_long; + + /* + * Verify that the fault happened in user space, that + * we are not in an interrupt and that there is a + * user context. + */ + fault = VM_FAULT_BADCONTEXT; + if (unlikely(!user_space_fault(regs) || faulthandler_disabled() || !mm)) + goto out; + + address = trans_exc_code & __FAIL_ADDR_MASK; + perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address); + flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE; + if (user_mode(regs)) + flags |= FAULT_FLAG_USER; + if (access == VM_WRITE || (trans_exc_code & store_indication) == 0x400) + flags |= FAULT_FLAG_WRITE; + down_read(&mm->mmap_sem); + +#ifdef CONFIG_PGSTE + gmap = (current->flags & PF_VCPU) ? + (struct gmap *) S390_lowcore.gmap : NULL; + if (gmap) { + current->thread.gmap_addr = address; + address = __gmap_translate(gmap, address); + if (address == -EFAULT) { + fault = VM_FAULT_BADMAP; + goto out_up; + } + if (gmap->pfault_enabled) + flags |= FAULT_FLAG_RETRY_NOWAIT; + } +#endif + +retry: + fault = VM_FAULT_BADMAP; + vma = find_vma(mm, address); + if (!vma) + goto out_up; + + if (unlikely(vma->vm_start > address)) { + if (!(vma->vm_flags & VM_GROWSDOWN)) + goto out_up; + if (expand_stack(vma, address)) + goto out_up; + } + + /* + * Ok, we have a good vm_area for this memory access, so + * we can handle it.. + */ + fault = VM_FAULT_BADACCESS; + if (unlikely(!(vma->vm_flags & access))) + goto out_up; + + if (is_vm_hugetlb_page(vma)) + address &= HPAGE_MASK; + /* + * If for any reason at all we couldn't handle the fault, + * make sure we exit gracefully rather than endlessly redo + * the fault. + */ + fault = handle_mm_fault(mm, vma, address, flags); + /* No reason to continue if interrupted by SIGKILL. */ + if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current)) { + fault = VM_FAULT_SIGNAL; + goto out; + } + if (unlikely(fault & VM_FAULT_ERROR)) + goto out_up; + + /* + * Major/minor page fault accounting is only done on the + * initial attempt. If we go through a retry, it is extremely + * likely that the page will be found in page cache at that point. + */ + if (flags & FAULT_FLAG_ALLOW_RETRY) { + if (fault & VM_FAULT_MAJOR) { + tsk->maj_flt++; + perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ, 1, + regs, address); + } else { + tsk->min_flt++; + perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN, 1, + regs, address); + } + if (fault & VM_FAULT_RETRY) { +#ifdef CONFIG_PGSTE + if (gmap && (flags & FAULT_FLAG_RETRY_NOWAIT)) { + /* FAULT_FLAG_RETRY_NOWAIT has been set, + * mmap_sem has not been released */ + current->thread.gmap_pfault = 1; + fault = VM_FAULT_PFAULT; + goto out_up; + } +#endif + /* Clear FAULT_FLAG_ALLOW_RETRY to avoid any risk + * of starvation. */ + flags &= ~(FAULT_FLAG_ALLOW_RETRY | + FAULT_FLAG_RETRY_NOWAIT); + flags |= FAULT_FLAG_TRIED; + down_read(&mm->mmap_sem); + goto retry; + } + } +#ifdef CONFIG_PGSTE + if (gmap) { + address = __gmap_link(gmap, current->thread.gmap_addr, + address); + if (address == -EFAULT) { + fault = VM_FAULT_BADMAP; + goto out_up; + } + if (address == -ENOMEM) { + fault = VM_FAULT_OOM; + goto out_up; + } + } +#endif + fault = 0; +out_up: + up_read(&mm->mmap_sem); +out: + return fault; +} + +void do_protection_exception(struct pt_regs *regs) +{ + unsigned long trans_exc_code; + int fault; + + trans_exc_code = regs->int_parm_long; + /* + * Protection exceptions are suppressing, decrement psw address. + * The exception to this rule are aborted transactions, for these + * the PSW already points to the correct location. + */ + if (!(regs->int_code & 0x200)) + regs->psw.addr = __rewind_psw(regs->psw, regs->int_code >> 16); + /* + * Check for low-address protection. This needs to be treated + * as a special case because the translation exception code + * field is not guaranteed to contain valid data in this case. + */ + if (unlikely(!(trans_exc_code & 4))) { + do_low_address(regs); + return; + } + fault = do_exception(regs, VM_WRITE); + if (unlikely(fault)) + do_fault_error(regs, fault); +} +NOKPROBE_SYMBOL(do_protection_exception); + +void do_dat_exception(struct pt_regs *regs) +{ + int access, fault; + + access = VM_READ | VM_EXEC | VM_WRITE; + fault = do_exception(regs, access); + if (unlikely(fault)) + do_fault_error(regs, fault); +} +NOKPROBE_SYMBOL(do_dat_exception); + +#ifdef CONFIG_PFAULT +/* + * 'pfault' pseudo page faults routines. + */ +static int pfault_disable; + +static int __init nopfault(char *str) +{ + pfault_disable = 1; + return 1; +} + +__setup("nopfault", nopfault); + +struct pfault_refbk { + u16 refdiagc; + u16 reffcode; + u16 refdwlen; + u16 refversn; + u64 refgaddr; + u64 refselmk; + u64 refcmpmk; + u64 reserved; +} __attribute__ ((packed, aligned(8))); + +int pfault_init(void) +{ + struct pfault_refbk refbk = { + .refdiagc = 0x258, + .reffcode = 0, + .refdwlen = 5, + .refversn = 2, + .refgaddr = __LC_CURRENT_PID, + .refselmk = 1ULL << 48, + .refcmpmk = 1ULL << 48, + .reserved = __PF_RES_FIELD }; + int rc; + + if (pfault_disable) + return -1; + asm volatile( + " diag %1,%0,0x258\n" + "0: j 2f\n" + "1: la %0,8\n" + "2:\n" + EX_TABLE(0b,1b) + : "=d" (rc) : "a" (&refbk), "m" (refbk) : "cc"); + return rc; +} + +void pfault_fini(void) +{ + struct pfault_refbk refbk = { + .refdiagc = 0x258, + .reffcode = 1, + .refdwlen = 5, + .refversn = 2, + }; + + if (pfault_disable) + return; + asm volatile( + " diag %0,0,0x258\n" + "0:\n" + EX_TABLE(0b,0b) + : : "a" (&refbk), "m" (refbk) : "cc"); +} + +static DEFINE_SPINLOCK(pfault_lock); +static LIST_HEAD(pfault_list); + +static void pfault_interrupt(struct ext_code ext_code, + unsigned int param32, unsigned long param64) +{ + struct task_struct *tsk; + __u16 subcode; + pid_t pid; + + /* + * Get the external interruption subcode & pfault + * initial/completion signal bit. VM stores this + * in the 'cpu address' field associated with the + * external interrupt. + */ + subcode = ext_code.subcode; + if ((subcode & 0xff00) != __SUBCODE_MASK) + return; + inc_irq_stat(IRQEXT_PFL); + /* Get the token (= pid of the affected task). */ + pid = sizeof(void *) == 4 ? param32 : param64; + rcu_read_lock(); + tsk = find_task_by_pid_ns(pid, &init_pid_ns); + if (tsk) + get_task_struct(tsk); + rcu_read_unlock(); + if (!tsk) + return; + spin_lock(&pfault_lock); + if (subcode & 0x0080) { + /* signal bit is set -> a page has been swapped in by VM */ + if (tsk->thread.pfault_wait == 1) { + /* Initial interrupt was faster than the completion + * interrupt. pfault_wait is valid. Set pfault_wait + * back to zero and wake up the process. This can + * safely be done because the task is still sleeping + * and can't produce new pfaults. */ + tsk->thread.pfault_wait = 0; + list_del(&tsk->thread.list); + wake_up_process(tsk); + put_task_struct(tsk); + } else { + /* Completion interrupt was faster than initial + * interrupt. Set pfault_wait to -1 so the initial + * interrupt doesn't put the task to sleep. + * If the task is not running, ignore the completion + * interrupt since it must be a leftover of a PFAULT + * CANCEL operation which didn't remove all pending + * completion interrupts. */ + if (tsk->state == TASK_RUNNING) + tsk->thread.pfault_wait = -1; + } + } else { + /* signal bit not set -> a real page is missing. */ + if (WARN_ON_ONCE(tsk != current)) + goto out; + if (tsk->thread.pfault_wait == 1) { + /* Already on the list with a reference: put to sleep */ + __set_task_state(tsk, TASK_UNINTERRUPTIBLE); + set_tsk_need_resched(tsk); + } else if (tsk->thread.pfault_wait == -1) { + /* Completion interrupt was faster than the initial + * interrupt (pfault_wait == -1). Set pfault_wait + * back to zero and exit. */ + tsk->thread.pfault_wait = 0; + } else { + /* Initial interrupt arrived before completion + * interrupt. Let the task sleep. + * An extra task reference is needed since a different + * cpu may set the task state to TASK_RUNNING again + * before the scheduler is reached. */ + get_task_struct(tsk); + tsk->thread.pfault_wait = 1; + list_add(&tsk->thread.list, &pfault_list); + __set_task_state(tsk, TASK_UNINTERRUPTIBLE); + set_tsk_need_resched(tsk); + } + } +out: + spin_unlock(&pfault_lock); + put_task_struct(tsk); +} + +static int pfault_cpu_notify(struct notifier_block *self, unsigned long action, + void *hcpu) +{ + struct thread_struct *thread, *next; + struct task_struct *tsk; + + switch (action & ~CPU_TASKS_FROZEN) { + case CPU_DEAD: + spin_lock_irq(&pfault_lock); + list_for_each_entry_safe(thread, next, &pfault_list, list) { + thread->pfault_wait = 0; + list_del(&thread->list); + tsk = container_of(thread, struct task_struct, thread); + wake_up_process(tsk); + put_task_struct(tsk); + } + spin_unlock_irq(&pfault_lock); + break; + default: + break; + } + return NOTIFY_OK; +} + +static int __init pfault_irq_init(void) +{ + int rc; + + rc = register_external_irq(EXT_IRQ_CP_SERVICE, pfault_interrupt); + if (rc) + goto out_extint; + rc = pfault_init() == 0 ? 0 : -EOPNOTSUPP; + if (rc) + goto out_pfault; + irq_subclass_register(IRQ_SUBCLASS_SERVICE_SIGNAL); + hotcpu_notifier(pfault_cpu_notify, 0); + return 0; + +out_pfault: + unregister_external_irq(EXT_IRQ_CP_SERVICE, pfault_interrupt); +out_extint: + pfault_disable = 1; + return rc; +} +early_initcall(pfault_irq_init); + +#endif /* CONFIG_PFAULT */ diff --git a/kernel/arch/s390/mm/gup.c b/kernel/arch/s390/mm/gup.c new file mode 100644 index 000000000..1eb41bb30 --- /dev/null +++ b/kernel/arch/s390/mm/gup.c @@ -0,0 +1,240 @@ +/* + * Lockless get_user_pages_fast for s390 + * + * Copyright IBM Corp. 2010 + * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com> + */ +#include <linux/sched.h> +#include <linux/mm.h> +#include <linux/hugetlb.h> +#include <linux/vmstat.h> +#include <linux/pagemap.h> +#include <linux/rwsem.h> +#include <asm/pgtable.h> + +/* + * The performance critical leaf functions are made noinline otherwise gcc + * inlines everything into a single function which results in too much + * register pressure. + */ +static inline int gup_pte_range(pmd_t *pmdp, pmd_t pmd, unsigned long addr, + unsigned long end, int write, struct page **pages, int *nr) +{ + unsigned long mask; + pte_t *ptep, pte; + struct page *page; + + mask = (write ? _PAGE_PROTECT : 0) | _PAGE_INVALID | _PAGE_SPECIAL; + + ptep = ((pte_t *) pmd_deref(pmd)) + pte_index(addr); + do { + pte = *ptep; + barrier(); + if ((pte_val(pte) & mask) != 0) + return 0; + VM_BUG_ON(!pfn_valid(pte_pfn(pte))); + page = pte_page(pte); + if (!page_cache_get_speculative(page)) + return 0; + if (unlikely(pte_val(pte) != pte_val(*ptep))) { + put_page(page); + return 0; + } + pages[*nr] = page; + (*nr)++; + + } while (ptep++, addr += PAGE_SIZE, addr != end); + + return 1; +} + +static inline int gup_huge_pmd(pmd_t *pmdp, pmd_t pmd, unsigned long addr, + unsigned long end, int write, struct page **pages, int *nr) +{ + unsigned long mask, result; + struct page *head, *page, *tail; + int refs; + + result = write ? 0 : _SEGMENT_ENTRY_PROTECT; + mask = result | _SEGMENT_ENTRY_INVALID; + if ((pmd_val(pmd) & mask) != result) + return 0; + VM_BUG_ON(!pfn_valid(pmd_val(pmd) >> PAGE_SHIFT)); + + refs = 0; + head = pmd_page(pmd); + page = head + ((addr & ~PMD_MASK) >> PAGE_SHIFT); + tail = page; + do { + VM_BUG_ON(compound_head(page) != head); + pages[*nr] = page; + (*nr)++; + page++; + refs++; + } while (addr += PAGE_SIZE, addr != end); + + if (!page_cache_add_speculative(head, refs)) { + *nr -= refs; + return 0; + } + + if (unlikely(pmd_val(pmd) != pmd_val(*pmdp))) { + *nr -= refs; + while (refs--) + put_page(head); + return 0; + } + + /* + * Any tail page need their mapcount reference taken before we + * return. + */ + while (refs--) { + if (PageTail(tail)) + get_huge_page_tail(tail); + tail++; + } + + return 1; +} + + +static inline int gup_pmd_range(pud_t *pudp, pud_t pud, unsigned long addr, + unsigned long end, int write, struct page **pages, int *nr) +{ + unsigned long next; + pmd_t *pmdp, pmd; + + pmdp = (pmd_t *) pudp; + if ((pud_val(pud) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R3) + pmdp = (pmd_t *) pud_deref(pud); + pmdp += pmd_index(addr); + do { + pmd = *pmdp; + barrier(); + next = pmd_addr_end(addr, end); + /* + * The pmd_trans_splitting() check below explains why + * pmdp_splitting_flush() has to serialize with + * smp_call_function() against our disabled IRQs, to stop + * this gup-fast code from running while we set the + * splitting bit in the pmd. Returning zero will take + * the slow path that will call wait_split_huge_page() + * if the pmd is still in splitting state. + */ + if (pmd_none(pmd) || pmd_trans_splitting(pmd)) + return 0; + if (unlikely(pmd_large(pmd))) { + if (!gup_huge_pmd(pmdp, pmd, addr, next, + write, pages, nr)) + return 0; + } else if (!gup_pte_range(pmdp, pmd, addr, next, + write, pages, nr)) + return 0; + } while (pmdp++, addr = next, addr != end); + + return 1; +} + +static inline int gup_pud_range(pgd_t *pgdp, pgd_t pgd, unsigned long addr, + unsigned long end, int write, struct page **pages, int *nr) +{ + unsigned long next; + pud_t *pudp, pud; + + pudp = (pud_t *) pgdp; + if ((pgd_val(pgd) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R2) + pudp = (pud_t *) pgd_deref(pgd); + pudp += pud_index(addr); + do { + pud = *pudp; + barrier(); + next = pud_addr_end(addr, end); + if (pud_none(pud)) + return 0; + if (!gup_pmd_range(pudp, pud, addr, next, write, pages, nr)) + return 0; + } while (pudp++, addr = next, addr != end); + + return 1; +} + +/* + * Like get_user_pages_fast() except its IRQ-safe in that it won't fall + * back to the regular GUP. + */ +int __get_user_pages_fast(unsigned long start, int nr_pages, int write, + struct page **pages) +{ + struct mm_struct *mm = current->mm; + unsigned long addr, len, end; + unsigned long next, flags; + pgd_t *pgdp, pgd; + int nr = 0; + + start &= PAGE_MASK; + addr = start; + len = (unsigned long) nr_pages << PAGE_SHIFT; + end = start + len; + if ((end <= start) || (end > TASK_SIZE)) + return 0; + /* + * local_irq_save() doesn't prevent pagetable teardown, but does + * prevent the pagetables from being freed on s390. + * + * So long as we atomically load page table pointers versus teardown, + * we can follow the address down to the the page and take a ref on it. + */ + local_irq_save(flags); + pgdp = pgd_offset(mm, addr); + do { + pgd = *pgdp; + barrier(); + next = pgd_addr_end(addr, end); + if (pgd_none(pgd)) + break; + if (!gup_pud_range(pgdp, pgd, addr, next, write, pages, &nr)) + break; + } while (pgdp++, addr = next, addr != end); + local_irq_restore(flags); + + return nr; +} + +/** + * get_user_pages_fast() - pin user pages in memory + * @start: starting user address + * @nr_pages: number of pages from start to pin + * @write: whether pages will be written to + * @pages: array that receives pointers to the pages pinned. + * Should be at least nr_pages long. + * + * Attempt to pin user pages in memory without taking mm->mmap_sem. + * If not successful, it will fall back to taking the lock and + * calling get_user_pages(). + * + * Returns number of pages pinned. This may be fewer than the number + * requested. If nr_pages is 0 or negative, returns 0. If no pages + * were pinned, returns -errno. + */ +int get_user_pages_fast(unsigned long start, int nr_pages, int write, + struct page **pages) +{ + struct mm_struct *mm = current->mm; + int nr, ret; + + start &= PAGE_MASK; + nr = __get_user_pages_fast(start, nr_pages, write, pages); + if (nr == nr_pages) + return nr; + + /* Try to get the remaining pages with get_user_pages */ + start += nr << PAGE_SHIFT; + pages += nr; + ret = get_user_pages_unlocked(current, mm, start, + nr_pages - nr, write, 0, pages); + /* Have to be a bit careful with return values */ + if (nr > 0) + ret = (ret < 0) ? nr : ret + nr; + return ret; +} diff --git a/kernel/arch/s390/mm/hugetlbpage.c b/kernel/arch/s390/mm/hugetlbpage.c new file mode 100644 index 000000000..e617e74b7 --- /dev/null +++ b/kernel/arch/s390/mm/hugetlbpage.c @@ -0,0 +1,212 @@ +/* + * IBM System z Huge TLB Page Support for Kernel. + * + * Copyright IBM Corp. 2007 + * Author(s): Gerald Schaefer <gerald.schaefer@de.ibm.com> + */ + +#include <linux/mm.h> +#include <linux/hugetlb.h> + +static inline pmd_t __pte_to_pmd(pte_t pte) +{ + pmd_t pmd; + + /* + * Convert encoding pte bits pmd bits + * lIR.uswrdy.p dy..R...I...wr + * empty 010.000000.0 -> 00..0...1...00 + * prot-none, clean, old 111.000000.1 -> 00..1...1...00 + * prot-none, clean, young 111.000001.1 -> 01..1...1...00 + * prot-none, dirty, old 111.000010.1 -> 10..1...1...00 + * prot-none, dirty, young 111.000011.1 -> 11..1...1...00 + * read-only, clean, old 111.000100.1 -> 00..1...1...01 + * read-only, clean, young 101.000101.1 -> 01..1...0...01 + * read-only, dirty, old 111.000110.1 -> 10..1...1...01 + * read-only, dirty, young 101.000111.1 -> 11..1...0...01 + * read-write, clean, old 111.001100.1 -> 00..1...1...11 + * read-write, clean, young 101.001101.1 -> 01..1...0...11 + * read-write, dirty, old 110.001110.1 -> 10..0...1...11 + * read-write, dirty, young 100.001111.1 -> 11..0...0...11 + * HW-bits: R read-only, I invalid + * SW-bits: p present, y young, d dirty, r read, w write, s special, + * u unused, l large + */ + if (pte_present(pte)) { + pmd_val(pmd) = pte_val(pte) & PAGE_MASK; + pmd_val(pmd) |= (pte_val(pte) & _PAGE_READ) >> 4; + pmd_val(pmd) |= (pte_val(pte) & _PAGE_WRITE) >> 4; + pmd_val(pmd) |= (pte_val(pte) & _PAGE_INVALID) >> 5; + pmd_val(pmd) |= (pte_val(pte) & _PAGE_PROTECT); + pmd_val(pmd) |= (pte_val(pte) & _PAGE_DIRTY) << 10; + pmd_val(pmd) |= (pte_val(pte) & _PAGE_YOUNG) << 10; + } else + pmd_val(pmd) = _SEGMENT_ENTRY_INVALID; + return pmd; +} + +static inline pte_t __pmd_to_pte(pmd_t pmd) +{ + pte_t pte; + + /* + * Convert encoding pmd bits pte bits + * dy..R...I...wr lIR.uswrdy.p + * empty 00..0...1...00 -> 010.000000.0 + * prot-none, clean, old 00..1...1...00 -> 111.000000.1 + * prot-none, clean, young 01..1...1...00 -> 111.000001.1 + * prot-none, dirty, old 10..1...1...00 -> 111.000010.1 + * prot-none, dirty, young 11..1...1...00 -> 111.000011.1 + * read-only, clean, old 00..1...1...01 -> 111.000100.1 + * read-only, clean, young 01..1...0...01 -> 101.000101.1 + * read-only, dirty, old 10..1...1...01 -> 111.000110.1 + * read-only, dirty, young 11..1...0...01 -> 101.000111.1 + * read-write, clean, old 00..1...1...11 -> 111.001100.1 + * read-write, clean, young 01..1...0...11 -> 101.001101.1 + * read-write, dirty, old 10..0...1...11 -> 110.001110.1 + * read-write, dirty, young 11..0...0...11 -> 100.001111.1 + * HW-bits: R read-only, I invalid + * SW-bits: p present, y young, d dirty, r read, w write, s special, + * u unused, l large + */ + if (pmd_present(pmd)) { + pte_val(pte) = pmd_val(pmd) & _SEGMENT_ENTRY_ORIGIN_LARGE; + pte_val(pte) |= _PAGE_LARGE | _PAGE_PRESENT; + pte_val(pte) |= (pmd_val(pmd) & _SEGMENT_ENTRY_READ) << 4; + pte_val(pte) |= (pmd_val(pmd) & _SEGMENT_ENTRY_WRITE) << 4; + pte_val(pte) |= (pmd_val(pmd) & _SEGMENT_ENTRY_INVALID) << 5; + pte_val(pte) |= (pmd_val(pmd) & _SEGMENT_ENTRY_PROTECT); + pte_val(pte) |= (pmd_val(pmd) & _SEGMENT_ENTRY_DIRTY) >> 10; + pte_val(pte) |= (pmd_val(pmd) & _SEGMENT_ENTRY_YOUNG) >> 10; + } else + pte_val(pte) = _PAGE_INVALID; + return pte; +} + +void set_huge_pte_at(struct mm_struct *mm, unsigned long addr, + pte_t *ptep, pte_t pte) +{ + pmd_t pmd; + + pmd = __pte_to_pmd(pte); + if (!MACHINE_HAS_HPAGE) { + /* Emulated huge ptes loose the dirty and young bit */ + pmd_val(pmd) &= ~_SEGMENT_ENTRY_ORIGIN; + pmd_val(pmd) |= pte_page(pte)[1].index; + } else + pmd_val(pmd) |= _SEGMENT_ENTRY_LARGE; + *(pmd_t *) ptep = pmd; +} + +pte_t huge_ptep_get(pte_t *ptep) +{ + unsigned long origin; + pmd_t pmd; + + pmd = *(pmd_t *) ptep; + if (!MACHINE_HAS_HPAGE && pmd_present(pmd)) { + origin = pmd_val(pmd) & _SEGMENT_ENTRY_ORIGIN; + pmd_val(pmd) &= ~_SEGMENT_ENTRY_ORIGIN; + pmd_val(pmd) |= *(unsigned long *) origin; + /* Emulated huge ptes are young and dirty by definition */ + pmd_val(pmd) |= _SEGMENT_ENTRY_YOUNG | _SEGMENT_ENTRY_DIRTY; + } + return __pmd_to_pte(pmd); +} + +pte_t huge_ptep_get_and_clear(struct mm_struct *mm, + unsigned long addr, pte_t *ptep) +{ + pmd_t *pmdp = (pmd_t *) ptep; + pte_t pte = huge_ptep_get(ptep); + + pmdp_flush_direct(mm, addr, pmdp); + pmd_val(*pmdp) = _SEGMENT_ENTRY_EMPTY; + return pte; +} + +int arch_prepare_hugepage(struct page *page) +{ + unsigned long addr = page_to_phys(page); + pte_t pte; + pte_t *ptep; + int i; + + if (MACHINE_HAS_HPAGE) + return 0; + + ptep = (pte_t *) pte_alloc_one(&init_mm, addr); + if (!ptep) + return -ENOMEM; + + pte_val(pte) = addr; + for (i = 0; i < PTRS_PER_PTE; i++) { + set_pte_at(&init_mm, addr + i * PAGE_SIZE, ptep + i, pte); + pte_val(pte) += PAGE_SIZE; + } + page[1].index = (unsigned long) ptep; + return 0; +} + +void arch_release_hugepage(struct page *page) +{ + pte_t *ptep; + + if (MACHINE_HAS_HPAGE) + return; + + ptep = (pte_t *) page[1].index; + if (!ptep) + return; + clear_table((unsigned long *) ptep, _PAGE_INVALID, + PTRS_PER_PTE * sizeof(pte_t)); + page_table_free(&init_mm, (unsigned long *) ptep); + page[1].index = 0; +} + +pte_t *huge_pte_alloc(struct mm_struct *mm, + unsigned long addr, unsigned long sz) +{ + pgd_t *pgdp; + pud_t *pudp; + pmd_t *pmdp = NULL; + + pgdp = pgd_offset(mm, addr); + pudp = pud_alloc(mm, pgdp, addr); + if (pudp) + pmdp = pmd_alloc(mm, pudp, addr); + return (pte_t *) pmdp; +} + +pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr) +{ + pgd_t *pgdp; + pud_t *pudp; + pmd_t *pmdp = NULL; + + pgdp = pgd_offset(mm, addr); + if (pgd_present(*pgdp)) { + pudp = pud_offset(pgdp, addr); + if (pud_present(*pudp)) + pmdp = pmd_offset(pudp, addr); + } + return (pte_t *) pmdp; +} + +int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep) +{ + return 0; +} + +int pmd_huge(pmd_t pmd) +{ + if (!MACHINE_HAS_HPAGE) + return 0; + + return !!(pmd_val(pmd) & _SEGMENT_ENTRY_LARGE); +} + +int pud_huge(pud_t pud) +{ + return 0; +} diff --git a/kernel/arch/s390/mm/init.c b/kernel/arch/s390/mm/init.c new file mode 100644 index 000000000..80875c43a --- /dev/null +++ b/kernel/arch/s390/mm/init.c @@ -0,0 +1,230 @@ +/* + * S390 version + * Copyright IBM Corp. 1999 + * Author(s): Hartmut Penner (hp@de.ibm.com) + * + * Derived from "arch/i386/mm/init.c" + * Copyright (C) 1995 Linus Torvalds + */ + +#include <linux/signal.h> +#include <linux/sched.h> +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/string.h> +#include <linux/types.h> +#include <linux/ptrace.h> +#include <linux/mman.h> +#include <linux/mm.h> +#include <linux/swap.h> +#include <linux/smp.h> +#include <linux/init.h> +#include <linux/pagemap.h> +#include <linux/bootmem.h> +#include <linux/memory.h> +#include <linux/pfn.h> +#include <linux/poison.h> +#include <linux/initrd.h> +#include <linux/export.h> +#include <linux/gfp.h> +#include <asm/processor.h> +#include <asm/uaccess.h> +#include <asm/pgtable.h> +#include <asm/pgalloc.h> +#include <asm/dma.h> +#include <asm/lowcore.h> +#include <asm/tlb.h> +#include <asm/tlbflush.h> +#include <asm/sections.h> +#include <asm/ctl_reg.h> +#include <asm/sclp.h> + +pgd_t swapper_pg_dir[PTRS_PER_PGD] __attribute__((__aligned__(PAGE_SIZE))); + +unsigned long empty_zero_page, zero_page_mask; +EXPORT_SYMBOL(empty_zero_page); +EXPORT_SYMBOL(zero_page_mask); + +static void __init setup_zero_pages(void) +{ + struct cpuid cpu_id; + unsigned int order; + struct page *page; + int i; + + get_cpu_id(&cpu_id); + switch (cpu_id.machine) { + case 0x9672: /* g5 */ + case 0x2064: /* z900 */ + case 0x2066: /* z900 */ + case 0x2084: /* z990 */ + case 0x2086: /* z990 */ + case 0x2094: /* z9-109 */ + case 0x2096: /* z9-109 */ + order = 0; + break; + case 0x2097: /* z10 */ + case 0x2098: /* z10 */ + case 0x2817: /* z196 */ + case 0x2818: /* z196 */ + order = 2; + break; + case 0x2827: /* zEC12 */ + case 0x2828: /* zEC12 */ + order = 5; + break; + case 0x2964: /* z13 */ + default: + order = 7; + break; + } + /* Limit number of empty zero pages for small memory sizes */ + while (order > 2 && (totalram_pages >> 10) < (1UL << order)) + order--; + + empty_zero_page = __get_free_pages(GFP_KERNEL | __GFP_ZERO, order); + if (!empty_zero_page) + panic("Out of memory in setup_zero_pages"); + + page = virt_to_page((void *) empty_zero_page); + split_page(page, order); + for (i = 1 << order; i > 0; i--) { + mark_page_reserved(page); + page++; + } + + zero_page_mask = ((PAGE_SIZE << order) - 1) & PAGE_MASK; +} + +/* + * paging_init() sets up the page tables + */ +void __init paging_init(void) +{ + unsigned long max_zone_pfns[MAX_NR_ZONES]; + unsigned long pgd_type, asce_bits; + + init_mm.pgd = swapper_pg_dir; + if (VMALLOC_END > (1UL << 42)) { + asce_bits = _ASCE_TYPE_REGION2 | _ASCE_TABLE_LENGTH; + pgd_type = _REGION2_ENTRY_EMPTY; + } else { + asce_bits = _ASCE_TYPE_REGION3 | _ASCE_TABLE_LENGTH; + pgd_type = _REGION3_ENTRY_EMPTY; + } + S390_lowcore.kernel_asce = (__pa(init_mm.pgd) & PAGE_MASK) | asce_bits; + clear_table((unsigned long *) init_mm.pgd, pgd_type, + sizeof(unsigned long)*2048); + vmem_map_init(); + + /* enable virtual mapping in kernel mode */ + __ctl_load(S390_lowcore.kernel_asce, 1, 1); + __ctl_load(S390_lowcore.kernel_asce, 7, 7); + __ctl_load(S390_lowcore.kernel_asce, 13, 13); + arch_local_irq_restore(4UL << (BITS_PER_LONG - 8)); + + sparse_memory_present_with_active_regions(MAX_NUMNODES); + sparse_init(); + memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); + max_zone_pfns[ZONE_DMA] = PFN_DOWN(MAX_DMA_ADDRESS); + max_zone_pfns[ZONE_NORMAL] = max_low_pfn; + free_area_init_nodes(max_zone_pfns); +} + +void __init mem_init(void) +{ + if (MACHINE_HAS_TLB_LC) + cpumask_set_cpu(0, &init_mm.context.cpu_attach_mask); + cpumask_set_cpu(0, mm_cpumask(&init_mm)); + atomic_set(&init_mm.context.attach_count, 1); + + max_mapnr = max_low_pfn; + high_memory = (void *) __va(max_low_pfn * PAGE_SIZE); + + /* Setup guest page hinting */ + cmma_init(); + + /* this will put all low memory onto the freelists */ + free_all_bootmem(); + setup_zero_pages(); /* Setup zeroed pages. */ + + mem_init_print_info(NULL); + printk("Write protected kernel read-only data: %#lx - %#lx\n", + (unsigned long)&_stext, + PFN_ALIGN((unsigned long)&_eshared) - 1); +} + +void free_initmem(void) +{ + free_initmem_default(POISON_FREE_INITMEM); +} + +#ifdef CONFIG_BLK_DEV_INITRD +void __init free_initrd_mem(unsigned long start, unsigned long end) +{ + free_reserved_area((void *)start, (void *)end, POISON_FREE_INITMEM, + "initrd"); +} +#endif + +#ifdef CONFIG_MEMORY_HOTPLUG +int arch_add_memory(int nid, u64 start, u64 size) +{ + unsigned long zone_start_pfn, zone_end_pfn, nr_pages; + unsigned long start_pfn = PFN_DOWN(start); + unsigned long size_pages = PFN_DOWN(size); + struct zone *zone; + int rc; + + rc = vmem_add_mapping(start, size); + if (rc) + return rc; + for_each_zone(zone) { + if (zone_idx(zone) != ZONE_MOVABLE) { + /* Add range within existing zone limits */ + zone_start_pfn = zone->zone_start_pfn; + zone_end_pfn = zone->zone_start_pfn + + zone->spanned_pages; + } else { + /* Add remaining range to ZONE_MOVABLE */ + zone_start_pfn = start_pfn; + zone_end_pfn = start_pfn + size_pages; + } + if (start_pfn < zone_start_pfn || start_pfn >= zone_end_pfn) + continue; + nr_pages = (start_pfn + size_pages > zone_end_pfn) ? + zone_end_pfn - start_pfn : size_pages; + rc = __add_pages(nid, zone, start_pfn, nr_pages); + if (rc) + break; + start_pfn += nr_pages; + size_pages -= nr_pages; + if (!size_pages) + break; + } + if (rc) + vmem_remove_mapping(start, size); + return rc; +} + +unsigned long memory_block_size_bytes(void) +{ + /* + * Make sure the memory block size is always greater + * or equal than the memory increment size. + */ + return max_t(unsigned long, MIN_MEMORY_BLOCK_SIZE, sclp_get_rzm()); +} + +#ifdef CONFIG_MEMORY_HOTREMOVE +int arch_remove_memory(u64 start, u64 size) +{ + /* + * There is no hardware or firmware interface which could trigger a + * hot memory remove on s390. So there is nothing that needs to be + * implemented. + */ + return -EBUSY; +} +#endif +#endif /* CONFIG_MEMORY_HOTPLUG */ diff --git a/kernel/arch/s390/mm/maccess.c b/kernel/arch/s390/mm/maccess.c new file mode 100644 index 000000000..8a993a53f --- /dev/null +++ b/kernel/arch/s390/mm/maccess.c @@ -0,0 +1,208 @@ +/* + * Access kernel memory without faulting -- s390 specific implementation. + * + * Copyright IBM Corp. 2009, 2015 + * + * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>, + * + */ + +#include <linux/uaccess.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/errno.h> +#include <linux/gfp.h> +#include <linux/cpu.h> +#include <asm/ctl_reg.h> +#include <asm/io.h> + +static notrace long s390_kernel_write_odd(void *dst, const void *src, size_t size) +{ + unsigned long aligned, offset, count; + char tmp[8]; + + aligned = (unsigned long) dst & ~7UL; + offset = (unsigned long) dst & 7UL; + size = min(8UL - offset, size); + count = size - 1; + asm volatile( + " bras 1,0f\n" + " mvc 0(1,%4),0(%5)\n" + "0: mvc 0(8,%3),0(%0)\n" + " ex %1,0(1)\n" + " lg %1,0(%3)\n" + " lra %0,0(%0)\n" + " sturg %1,%0\n" + : "+&a" (aligned), "+&a" (count), "=m" (tmp) + : "a" (&tmp), "a" (&tmp[offset]), "a" (src) + : "cc", "memory", "1"); + return size; +} + +/* + * s390_kernel_write - write to kernel memory bypassing DAT + * @dst: destination address + * @src: source address + * @size: number of bytes to copy + * + * This function writes to kernel memory bypassing DAT and possible page table + * write protection. It writes to the destination using the sturg instruction. + * Therefore we have a read-modify-write sequence: the function reads eight + * bytes from destination at an eight byte boundary, modifies the bytes + * requested and writes the result back in a loop. + * + * Note: this means that this function may not be called concurrently on + * several cpus with overlapping words, since this may potentially + * cause data corruption. + */ +void notrace s390_kernel_write(void *dst, const void *src, size_t size) +{ + long copied; + + while (size) { + copied = s390_kernel_write_odd(dst, src, size); + dst += copied; + src += copied; + size -= copied; + } +} + +static int __memcpy_real(void *dest, void *src, size_t count) +{ + register unsigned long _dest asm("2") = (unsigned long) dest; + register unsigned long _len1 asm("3") = (unsigned long) count; + register unsigned long _src asm("4") = (unsigned long) src; + register unsigned long _len2 asm("5") = (unsigned long) count; + int rc = -EFAULT; + + asm volatile ( + "0: mvcle %1,%2,0x0\n" + "1: jo 0b\n" + " lhi %0,0x0\n" + "2:\n" + EX_TABLE(1b,2b) + : "+d" (rc), "+d" (_dest), "+d" (_src), "+d" (_len1), + "+d" (_len2), "=m" (*((long *) dest)) + : "m" (*((long *) src)) + : "cc", "memory"); + return rc; +} + +/* + * Copy memory in real mode (kernel to kernel) + */ +int memcpy_real(void *dest, void *src, size_t count) +{ + unsigned long flags; + int rc; + + if (!count) + return 0; + local_irq_save(flags); + __arch_local_irq_stnsm(0xfbUL); + rc = __memcpy_real(dest, src, count); + local_irq_restore(flags); + return rc; +} + +/* + * Copy memory in absolute mode (kernel to kernel) + */ +void memcpy_absolute(void *dest, void *src, size_t count) +{ + unsigned long cr0, flags, prefix; + + flags = arch_local_irq_save(); + __ctl_store(cr0, 0, 0); + __ctl_clear_bit(0, 28); /* disable lowcore protection */ + prefix = store_prefix(); + if (prefix) { + local_mcck_disable(); + set_prefix(0); + memcpy(dest, src, count); + set_prefix(prefix); + local_mcck_enable(); + } else { + memcpy(dest, src, count); + } + __ctl_load(cr0, 0, 0); + arch_local_irq_restore(flags); +} + +/* + * Copy memory from kernel (real) to user (virtual) + */ +int copy_to_user_real(void __user *dest, void *src, unsigned long count) +{ + int offs = 0, size, rc; + char *buf; + + buf = (char *) __get_free_page(GFP_KERNEL); + if (!buf) + return -ENOMEM; + rc = -EFAULT; + while (offs < count) { + size = min(PAGE_SIZE, count - offs); + if (memcpy_real(buf, src + offs, size)) + goto out; + if (copy_to_user(dest + offs, buf, size)) + goto out; + offs += size; + } + rc = 0; +out: + free_page((unsigned long) buf); + return rc; +} + +/* + * Check if physical address is within prefix or zero page + */ +static int is_swapped(unsigned long addr) +{ + unsigned long lc; + int cpu; + + if (addr < sizeof(struct _lowcore)) + return 1; + for_each_online_cpu(cpu) { + lc = (unsigned long) lowcore_ptr[cpu]; + if (addr > lc + sizeof(struct _lowcore) - 1 || addr < lc) + continue; + return 1; + } + return 0; +} + +/* + * Convert a physical pointer for /dev/mem access + * + * For swapped prefix pages a new buffer is returned that contains a copy of + * the absolute memory. The buffer size is maximum one page large. + */ +void *xlate_dev_mem_ptr(phys_addr_t addr) +{ + void *bounce = (void *) addr; + unsigned long size; + + get_online_cpus(); + preempt_disable(); + if (is_swapped(addr)) { + size = PAGE_SIZE - (addr & ~PAGE_MASK); + bounce = (void *) __get_free_page(GFP_ATOMIC); + if (bounce) + memcpy_absolute(bounce, (void *) addr, size); + } + preempt_enable(); + put_online_cpus(); + return bounce; +} + +/* + * Free converted buffer for /dev/mem access (if necessary) + */ +void unxlate_dev_mem_ptr(phys_addr_t addr, void *buf) +{ + if ((void *) addr != buf) + free_page((unsigned long) buf); +} diff --git a/kernel/arch/s390/mm/mem_detect.c b/kernel/arch/s390/mm/mem_detect.c new file mode 100644 index 000000000..0f3604395 --- /dev/null +++ b/kernel/arch/s390/mm/mem_detect.c @@ -0,0 +1,61 @@ +/* + * Copyright IBM Corp. 2008, 2009 + * + * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com> + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/memblock.h> +#include <linux/init.h> +#include <linux/debugfs.h> +#include <linux/seq_file.h> +#include <asm/ipl.h> +#include <asm/sclp.h> +#include <asm/setup.h> + +#define ADDR2G (1ULL << 31) + +#define CHUNK_READ_WRITE 0 +#define CHUNK_READ_ONLY 1 + +static inline void memblock_physmem_add(phys_addr_t start, phys_addr_t size) +{ + memblock_add_range(&memblock.memory, start, size, 0, 0); + memblock_add_range(&memblock.physmem, start, size, 0, 0); +} + +void __init detect_memory_memblock(void) +{ + unsigned long long memsize, rnmax, rzm; + unsigned long addr, size; + int type; + + rzm = sclp_get_rzm(); + rnmax = sclp_get_rnmax(); + memsize = rzm * rnmax; + if (!rzm) + rzm = 1ULL << 17; + max_physmem_end = memsize; + addr = 0; + /* keep memblock lists close to the kernel */ + memblock_set_bottom_up(true); + do { + size = 0; + type = tprot(addr); + do { + size += rzm; + if (max_physmem_end && addr + size >= max_physmem_end) + break; + } while (type == tprot(addr + size)); + if (type == CHUNK_READ_WRITE || type == CHUNK_READ_ONLY) { + if (max_physmem_end && (addr + size > max_physmem_end)) + size = max_physmem_end - addr; + memblock_physmem_add(addr, size); + } + addr += size; + } while (addr < max_physmem_end); + memblock_set_bottom_up(false); + if (!max_physmem_end) + max_physmem_end = memblock_end_of_DRAM(); +} diff --git a/kernel/arch/s390/mm/mmap.c b/kernel/arch/s390/mm/mmap.c new file mode 100644 index 000000000..6e552af08 --- /dev/null +++ b/kernel/arch/s390/mm/mmap.c @@ -0,0 +1,288 @@ +/* + * flexible mmap layout support + * + * Copyright 2003-2004 Red Hat Inc., Durham, North Carolina. + * All Rights Reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * + * Started by Ingo Molnar <mingo@elte.hu> + */ + +#include <linux/personality.h> +#include <linux/mm.h> +#include <linux/mman.h> +#include <linux/module.h> +#include <linux/random.h> +#include <linux/compat.h> +#include <linux/security.h> +#include <asm/pgalloc.h> + +unsigned long mmap_rnd_mask; +static unsigned long mmap_align_mask; + +static unsigned long stack_maxrandom_size(void) +{ + if (!(current->flags & PF_RANDOMIZE)) + return 0; + if (current->personality & ADDR_NO_RANDOMIZE) + return 0; + return STACK_RND_MASK << PAGE_SHIFT; +} + +/* + * Top of mmap area (just below the process stack). + * + * Leave at least a ~32 MB hole. + */ +#define MIN_GAP (32*1024*1024) +#define MAX_GAP (STACK_TOP/6*5) + +static inline int mmap_is_legacy(void) +{ + if (current->personality & ADDR_COMPAT_LAYOUT) + return 1; + if (rlimit(RLIMIT_STACK) == RLIM_INFINITY) + return 1; + return sysctl_legacy_va_layout; +} + +unsigned long arch_mmap_rnd(void) +{ + if (is_32bit_task()) + return (get_random_int() & 0x7ff) << PAGE_SHIFT; + else + return (get_random_int() & mmap_rnd_mask) << PAGE_SHIFT; +} + +static unsigned long mmap_base_legacy(unsigned long rnd) +{ + return TASK_UNMAPPED_BASE + rnd; +} + +static inline unsigned long mmap_base(unsigned long rnd) +{ + unsigned long gap = rlimit(RLIMIT_STACK); + + if (gap < MIN_GAP) + gap = MIN_GAP; + else if (gap > MAX_GAP) + gap = MAX_GAP; + gap &= PAGE_MASK; + return STACK_TOP - stack_maxrandom_size() - rnd - gap; +} + +unsigned long +arch_get_unmapped_area(struct file *filp, unsigned long addr, + unsigned long len, unsigned long pgoff, unsigned long flags) +{ + struct mm_struct *mm = current->mm; + struct vm_area_struct *vma; + struct vm_unmapped_area_info info; + int do_color_align; + + if (len > TASK_SIZE - mmap_min_addr) + return -ENOMEM; + + if (flags & MAP_FIXED) + return addr; + + if (addr) { + addr = PAGE_ALIGN(addr); + vma = find_vma(mm, addr); + if (TASK_SIZE - len >= addr && addr >= mmap_min_addr && + (!vma || addr + len <= vma->vm_start)) + return addr; + } + + do_color_align = 0; + if (filp || (flags & MAP_SHARED)) + do_color_align = !is_32bit_task(); + + info.flags = 0; + info.length = len; + info.low_limit = mm->mmap_base; + info.high_limit = TASK_SIZE; + info.align_mask = do_color_align ? (mmap_align_mask << PAGE_SHIFT) : 0; + info.align_offset = pgoff << PAGE_SHIFT; + return vm_unmapped_area(&info); +} + +unsigned long +arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0, + const unsigned long len, const unsigned long pgoff, + const unsigned long flags) +{ + struct vm_area_struct *vma; + struct mm_struct *mm = current->mm; + unsigned long addr = addr0; + struct vm_unmapped_area_info info; + int do_color_align; + + /* requested length too big for entire address space */ + if (len > TASK_SIZE - mmap_min_addr) + return -ENOMEM; + + if (flags & MAP_FIXED) + return addr; + + /* requesting a specific address */ + if (addr) { + addr = PAGE_ALIGN(addr); + vma = find_vma(mm, addr); + if (TASK_SIZE - len >= addr && addr >= mmap_min_addr && + (!vma || addr + len <= vma->vm_start)) + return addr; + } + + do_color_align = 0; + if (filp || (flags & MAP_SHARED)) + do_color_align = !is_32bit_task(); + + info.flags = VM_UNMAPPED_AREA_TOPDOWN; + info.length = len; + info.low_limit = max(PAGE_SIZE, mmap_min_addr); + info.high_limit = mm->mmap_base; + info.align_mask = do_color_align ? (mmap_align_mask << PAGE_SHIFT) : 0; + info.align_offset = pgoff << PAGE_SHIFT; + addr = vm_unmapped_area(&info); + + /* + * A failed mmap() very likely causes application failure, + * so fall back to the bottom-up function here. This scenario + * can happen with large stack limits and large mmap() + * allocations. + */ + if (addr & ~PAGE_MASK) { + VM_BUG_ON(addr != -ENOMEM); + info.flags = 0; + info.low_limit = TASK_UNMAPPED_BASE; + info.high_limit = TASK_SIZE; + addr = vm_unmapped_area(&info); + } + + return addr; +} + +int s390_mmap_check(unsigned long addr, unsigned long len, unsigned long flags) +{ + if (is_compat_task() || (TASK_SIZE >= (1UL << 53))) + return 0; + if (!(flags & MAP_FIXED)) + addr = 0; + if ((addr + len) >= TASK_SIZE) + return crst_table_upgrade(current->mm, 1UL << 53); + return 0; +} + +static unsigned long +s390_get_unmapped_area(struct file *filp, unsigned long addr, + unsigned long len, unsigned long pgoff, unsigned long flags) +{ + struct mm_struct *mm = current->mm; + unsigned long area; + int rc; + + area = arch_get_unmapped_area(filp, addr, len, pgoff, flags); + if (!(area & ~PAGE_MASK)) + return area; + if (area == -ENOMEM && !is_compat_task() && TASK_SIZE < (1UL << 53)) { + /* Upgrade the page table to 4 levels and retry. */ + rc = crst_table_upgrade(mm, 1UL << 53); + if (rc) + return (unsigned long) rc; + area = arch_get_unmapped_area(filp, addr, len, pgoff, flags); + } + return area; +} + +static unsigned long +s390_get_unmapped_area_topdown(struct file *filp, const unsigned long addr, + const unsigned long len, const unsigned long pgoff, + const unsigned long flags) +{ + struct mm_struct *mm = current->mm; + unsigned long area; + int rc; + + area = arch_get_unmapped_area_topdown(filp, addr, len, pgoff, flags); + if (!(area & ~PAGE_MASK)) + return area; + if (area == -ENOMEM && !is_compat_task() && TASK_SIZE < (1UL << 53)) { + /* Upgrade the page table to 4 levels and retry. */ + rc = crst_table_upgrade(mm, 1UL << 53); + if (rc) + return (unsigned long) rc; + area = arch_get_unmapped_area_topdown(filp, addr, len, + pgoff, flags); + } + return area; +} +/* + * This function, called very early during the creation of a new + * process VM image, sets up which VM layout function to use: + */ +void arch_pick_mmap_layout(struct mm_struct *mm) +{ + unsigned long random_factor = 0UL; + + if (current->flags & PF_RANDOMIZE) + random_factor = arch_mmap_rnd(); + + /* + * Fall back to the standard layout if the personality + * bit is set, or if the expected stack growth is unlimited: + */ + if (mmap_is_legacy()) { + mm->mmap_base = mmap_base_legacy(random_factor); + mm->get_unmapped_area = s390_get_unmapped_area; + } else { + mm->mmap_base = mmap_base(random_factor); + mm->get_unmapped_area = s390_get_unmapped_area_topdown; + } +} + +static int __init setup_mmap_rnd(void) +{ + struct cpuid cpu_id; + + get_cpu_id(&cpu_id); + switch (cpu_id.machine) { + case 0x9672: + case 0x2064: + case 0x2066: + case 0x2084: + case 0x2086: + case 0x2094: + case 0x2096: + case 0x2097: + case 0x2098: + case 0x2817: + case 0x2818: + case 0x2827: + case 0x2828: + mmap_rnd_mask = 0x7ffUL; + mmap_align_mask = 0UL; + break; + case 0x2964: /* z13 */ + default: + mmap_rnd_mask = 0x3ff80UL; + mmap_align_mask = 0x7fUL; + break; + } + return 0; +} +early_initcall(setup_mmap_rnd); diff --git a/kernel/arch/s390/mm/page-states.c b/kernel/arch/s390/mm/page-states.c new file mode 100644 index 000000000..a90d45e9d --- /dev/null +++ b/kernel/arch/s390/mm/page-states.c @@ -0,0 +1,114 @@ +/* + * Copyright IBM Corp. 2008 + * + * Guest page hinting for unused pages. + * + * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com> + */ + +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/types.h> +#include <linux/mm.h> +#include <linux/gfp.h> +#include <linux/init.h> + +#define ESSA_SET_STABLE 1 +#define ESSA_SET_UNUSED 2 + +static int cmma_flag = 1; + +static int __init cmma(char *str) +{ + char *parm; + + parm = strstrip(str); + if (strcmp(parm, "yes") == 0 || strcmp(parm, "on") == 0) { + cmma_flag = 1; + return 1; + } + cmma_flag = 0; + if (strcmp(parm, "no") == 0 || strcmp(parm, "off") == 0) + return 1; + return 0; +} +__setup("cmma=", cmma); + +void __init cmma_init(void) +{ + register unsigned long tmp asm("0") = 0; + register int rc asm("1") = -EOPNOTSUPP; + + if (!cmma_flag) + return; + asm volatile( + " .insn rrf,0xb9ab0000,%1,%1,0,0\n" + "0: la %0,0\n" + "1:\n" + EX_TABLE(0b,1b) + : "+&d" (rc), "+&d" (tmp)); + if (rc) + cmma_flag = 0; +} + +static inline void set_page_unstable(struct page *page, int order) +{ + int i, rc; + + for (i = 0; i < (1 << order); i++) + asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0" + : "=&d" (rc) + : "a" (page_to_phys(page + i)), + "i" (ESSA_SET_UNUSED)); +} + +void arch_free_page(struct page *page, int order) +{ + if (!cmma_flag) + return; + set_page_unstable(page, order); +} + +static inline void set_page_stable(struct page *page, int order) +{ + int i, rc; + + for (i = 0; i < (1 << order); i++) + asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0" + : "=&d" (rc) + : "a" (page_to_phys(page + i)), + "i" (ESSA_SET_STABLE)); +} + +void arch_alloc_page(struct page *page, int order) +{ + if (!cmma_flag) + return; + set_page_stable(page, order); +} + +void arch_set_page_states(int make_stable) +{ + unsigned long flags, order, t; + struct list_head *l; + struct page *page; + struct zone *zone; + + if (!cmma_flag) + return; + if (make_stable) + drain_local_pages(NULL); + for_each_populated_zone(zone) { + spin_lock_irqsave(&zone->lock, flags); + for_each_migratetype_order(order, t) { + list_for_each(l, &zone->free_area[order].free_list[t]) { + page = list_entry(l, struct page, lru); + if (make_stable) + set_page_stable(page, order); + else + set_page_unstable(page, order); + } + } + spin_unlock_irqrestore(&zone->lock, flags); + } +} diff --git a/kernel/arch/s390/mm/pageattr.c b/kernel/arch/s390/mm/pageattr.c new file mode 100644 index 000000000..749c98407 --- /dev/null +++ b/kernel/arch/s390/mm/pageattr.c @@ -0,0 +1,170 @@ +/* + * Copyright IBM Corp. 2011 + * Author(s): Jan Glauber <jang@linux.vnet.ibm.com> + */ +#include <linux/hugetlb.h> +#include <linux/module.h> +#include <linux/mm.h> +#include <asm/cacheflush.h> +#include <asm/facility.h> +#include <asm/pgtable.h> +#include <asm/page.h> + +#if PAGE_DEFAULT_KEY +static inline unsigned long sske_frame(unsigned long addr, unsigned char skey) +{ + asm volatile(".insn rrf,0xb22b0000,%[skey],%[addr],9,0" + : [addr] "+a" (addr) : [skey] "d" (skey)); + return addr; +} + +void __storage_key_init_range(unsigned long start, unsigned long end) +{ + unsigned long boundary, size; + + while (start < end) { + if (MACHINE_HAS_EDAT1) { + /* set storage keys for a 1MB frame */ + size = 1UL << 20; + boundary = (start + size) & ~(size - 1); + if (boundary <= end) { + do { + start = sske_frame(start, PAGE_DEFAULT_KEY); + } while (start < boundary); + continue; + } + } + page_set_storage_key(start, PAGE_DEFAULT_KEY, 0); + start += PAGE_SIZE; + } +} +#endif + +static pte_t *walk_page_table(unsigned long addr) +{ + pgd_t *pgdp; + pud_t *pudp; + pmd_t *pmdp; + pte_t *ptep; + + pgdp = pgd_offset_k(addr); + if (pgd_none(*pgdp)) + return NULL; + pudp = pud_offset(pgdp, addr); + if (pud_none(*pudp) || pud_large(*pudp)) + return NULL; + pmdp = pmd_offset(pudp, addr); + if (pmd_none(*pmdp) || pmd_large(*pmdp)) + return NULL; + ptep = pte_offset_kernel(pmdp, addr); + if (pte_none(*ptep)) + return NULL; + return ptep; +} + +static void change_page_attr(unsigned long addr, int numpages, + pte_t (*set) (pte_t)) +{ + pte_t *ptep, pte; + int i; + + for (i = 0; i < numpages; i++) { + ptep = walk_page_table(addr); + if (WARN_ON_ONCE(!ptep)) + break; + pte = *ptep; + pte = set(pte); + __ptep_ipte(addr, ptep); + *ptep = pte; + addr += PAGE_SIZE; + } +} + +int set_memory_ro(unsigned long addr, int numpages) +{ + change_page_attr(addr, numpages, pte_wrprotect); + return 0; +} + +int set_memory_rw(unsigned long addr, int numpages) +{ + change_page_attr(addr, numpages, pte_mkwrite); + return 0; +} + +/* not possible */ +int set_memory_nx(unsigned long addr, int numpages) +{ + return 0; +} + +int set_memory_x(unsigned long addr, int numpages) +{ + return 0; +} + +#ifdef CONFIG_DEBUG_PAGEALLOC + +static void ipte_range(pte_t *pte, unsigned long address, int nr) +{ + int i; + + if (test_facility(13)) { + __ptep_ipte_range(address, nr - 1, pte); + return; + } + for (i = 0; i < nr; i++) { + __ptep_ipte(address, pte); + address += PAGE_SIZE; + pte++; + } +} + +void __kernel_map_pages(struct page *page, int numpages, int enable) +{ + unsigned long address; + int nr, i, j; + pgd_t *pgd; + pud_t *pud; + pmd_t *pmd; + pte_t *pte; + + for (i = 0; i < numpages;) { + address = page_to_phys(page + i); + pgd = pgd_offset_k(address); + pud = pud_offset(pgd, address); + pmd = pmd_offset(pud, address); + pte = pte_offset_kernel(pmd, address); + nr = (unsigned long)pte >> ilog2(sizeof(long)); + nr = PTRS_PER_PTE - (nr & (PTRS_PER_PTE - 1)); + nr = min(numpages - i, nr); + if (enable) { + for (j = 0; j < nr; j++) { + pte_val(*pte) = __pa(address); + address += PAGE_SIZE; + pte++; + } + } else { + ipte_range(pte, address, nr); + } + i += nr; + } +} + +#ifdef CONFIG_HIBERNATION +bool kernel_page_present(struct page *page) +{ + unsigned long addr; + int cc; + + addr = page_to_phys(page); + asm volatile( + " lra %1,0(%1)\n" + " ipm %0\n" + " srl %0,28" + : "=d" (cc), "+a" (addr) : : "cc"); + return cc == 0; +} +#endif /* CONFIG_HIBERNATION */ + +#endif /* CONFIG_DEBUG_PAGEALLOC */ diff --git a/kernel/arch/s390/mm/pgtable.c b/kernel/arch/s390/mm/pgtable.c new file mode 100644 index 000000000..b33f66110 --- /dev/null +++ b/kernel/arch/s390/mm/pgtable.c @@ -0,0 +1,1412 @@ +/* + * Copyright IBM Corp. 2007, 2011 + * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com> + */ + +#include <linux/sched.h> +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/gfp.h> +#include <linux/mm.h> +#include <linux/swap.h> +#include <linux/smp.h> +#include <linux/highmem.h> +#include <linux/pagemap.h> +#include <linux/spinlock.h> +#include <linux/module.h> +#include <linux/quicklist.h> +#include <linux/rcupdate.h> +#include <linux/slab.h> +#include <linux/swapops.h> +#include <linux/sysctl.h> +#include <linux/ksm.h> +#include <linux/mman.h> + +#include <asm/pgtable.h> +#include <asm/pgalloc.h> +#include <asm/tlb.h> +#include <asm/tlbflush.h> +#include <asm/mmu_context.h> + +#define ALLOC_ORDER 2 +#define FRAG_MASK 0x03 + +unsigned long *crst_table_alloc(struct mm_struct *mm) +{ + struct page *page = alloc_pages(GFP_KERNEL, ALLOC_ORDER); + + if (!page) + return NULL; + return (unsigned long *) page_to_phys(page); +} + +void crst_table_free(struct mm_struct *mm, unsigned long *table) +{ + free_pages((unsigned long) table, ALLOC_ORDER); +} + +static void __crst_table_upgrade(void *arg) +{ + struct mm_struct *mm = arg; + + if (current->active_mm == mm) { + clear_user_asce(); + set_user_asce(mm); + } + __tlb_flush_local(); +} + +int crst_table_upgrade(struct mm_struct *mm, unsigned long limit) +{ + unsigned long *table, *pgd; + unsigned long entry; + int flush; + + BUG_ON(limit > (1UL << 53)); + flush = 0; +repeat: + table = crst_table_alloc(mm); + if (!table) + return -ENOMEM; + spin_lock_bh(&mm->page_table_lock); + if (mm->context.asce_limit < limit) { + pgd = (unsigned long *) mm->pgd; + if (mm->context.asce_limit <= (1UL << 31)) { + entry = _REGION3_ENTRY_EMPTY; + mm->context.asce_limit = 1UL << 42; + mm->context.asce_bits = _ASCE_TABLE_LENGTH | + _ASCE_USER_BITS | + _ASCE_TYPE_REGION3; + } else { + entry = _REGION2_ENTRY_EMPTY; + mm->context.asce_limit = 1UL << 53; + mm->context.asce_bits = _ASCE_TABLE_LENGTH | + _ASCE_USER_BITS | + _ASCE_TYPE_REGION2; + } + crst_table_init(table, entry); + pgd_populate(mm, (pgd_t *) table, (pud_t *) pgd); + mm->pgd = (pgd_t *) table; + mm->task_size = mm->context.asce_limit; + table = NULL; + flush = 1; + } + spin_unlock_bh(&mm->page_table_lock); + if (table) + crst_table_free(mm, table); + if (mm->context.asce_limit < limit) + goto repeat; + if (flush) + on_each_cpu(__crst_table_upgrade, mm, 0); + return 0; +} + +void crst_table_downgrade(struct mm_struct *mm, unsigned long limit) +{ + pgd_t *pgd; + + if (current->active_mm == mm) { + clear_user_asce(); + __tlb_flush_mm(mm); + } + while (mm->context.asce_limit > limit) { + pgd = mm->pgd; + switch (pgd_val(*pgd) & _REGION_ENTRY_TYPE_MASK) { + case _REGION_ENTRY_TYPE_R2: + mm->context.asce_limit = 1UL << 42; + mm->context.asce_bits = _ASCE_TABLE_LENGTH | + _ASCE_USER_BITS | + _ASCE_TYPE_REGION3; + break; + case _REGION_ENTRY_TYPE_R3: + mm->context.asce_limit = 1UL << 31; + mm->context.asce_bits = _ASCE_TABLE_LENGTH | + _ASCE_USER_BITS | + _ASCE_TYPE_SEGMENT; + break; + default: + BUG(); + } + mm->pgd = (pgd_t *) (pgd_val(*pgd) & _REGION_ENTRY_ORIGIN); + mm->task_size = mm->context.asce_limit; + crst_table_free(mm, (unsigned long *) pgd); + } + if (current->active_mm == mm) + set_user_asce(mm); +} + +#ifdef CONFIG_PGSTE + +/** + * gmap_alloc - allocate a guest address space + * @mm: pointer to the parent mm_struct + * @limit: maximum size of the gmap address space + * + * Returns a guest address space structure. + */ +struct gmap *gmap_alloc(struct mm_struct *mm, unsigned long limit) +{ + struct gmap *gmap; + struct page *page; + unsigned long *table; + unsigned long etype, atype; + + if (limit < (1UL << 31)) { + limit = (1UL << 31) - 1; + atype = _ASCE_TYPE_SEGMENT; + etype = _SEGMENT_ENTRY_EMPTY; + } else if (limit < (1UL << 42)) { + limit = (1UL << 42) - 1; + atype = _ASCE_TYPE_REGION3; + etype = _REGION3_ENTRY_EMPTY; + } else if (limit < (1UL << 53)) { + limit = (1UL << 53) - 1; + atype = _ASCE_TYPE_REGION2; + etype = _REGION2_ENTRY_EMPTY; + } else { + limit = -1UL; + atype = _ASCE_TYPE_REGION1; + etype = _REGION1_ENTRY_EMPTY; + } + gmap = kzalloc(sizeof(struct gmap), GFP_KERNEL); + if (!gmap) + goto out; + INIT_LIST_HEAD(&gmap->crst_list); + INIT_RADIX_TREE(&gmap->guest_to_host, GFP_KERNEL); + INIT_RADIX_TREE(&gmap->host_to_guest, GFP_ATOMIC); + spin_lock_init(&gmap->guest_table_lock); + gmap->mm = mm; + page = alloc_pages(GFP_KERNEL, ALLOC_ORDER); + if (!page) + goto out_free; + page->index = 0; + list_add(&page->lru, &gmap->crst_list); + table = (unsigned long *) page_to_phys(page); + crst_table_init(table, etype); + gmap->table = table; + gmap->asce = atype | _ASCE_TABLE_LENGTH | + _ASCE_USER_BITS | __pa(table); + gmap->asce_end = limit; + down_write(&mm->mmap_sem); + list_add(&gmap->list, &mm->context.gmap_list); + up_write(&mm->mmap_sem); + return gmap; + +out_free: + kfree(gmap); +out: + return NULL; +} +EXPORT_SYMBOL_GPL(gmap_alloc); + +static void gmap_flush_tlb(struct gmap *gmap) +{ + if (MACHINE_HAS_IDTE) + __tlb_flush_asce(gmap->mm, gmap->asce); + else + __tlb_flush_global(); +} + +static void gmap_radix_tree_free(struct radix_tree_root *root) +{ + struct radix_tree_iter iter; + unsigned long indices[16]; + unsigned long index; + void **slot; + int i, nr; + + /* A radix tree is freed by deleting all of its entries */ + index = 0; + do { + nr = 0; + radix_tree_for_each_slot(slot, root, &iter, index) { + indices[nr] = iter.index; + if (++nr == 16) + break; + } + for (i = 0; i < nr; i++) { + index = indices[i]; + radix_tree_delete(root, index); + } + } while (nr > 0); +} + +/** + * gmap_free - free a guest address space + * @gmap: pointer to the guest address space structure + */ +void gmap_free(struct gmap *gmap) +{ + struct page *page, *next; + + /* Flush tlb. */ + if (MACHINE_HAS_IDTE) + __tlb_flush_asce(gmap->mm, gmap->asce); + else + __tlb_flush_global(); + + /* Free all segment & region tables. */ + list_for_each_entry_safe(page, next, &gmap->crst_list, lru) + __free_pages(page, ALLOC_ORDER); + gmap_radix_tree_free(&gmap->guest_to_host); + gmap_radix_tree_free(&gmap->host_to_guest); + down_write(&gmap->mm->mmap_sem); + list_del(&gmap->list); + up_write(&gmap->mm->mmap_sem); + kfree(gmap); +} +EXPORT_SYMBOL_GPL(gmap_free); + +/** + * gmap_enable - switch primary space to the guest address space + * @gmap: pointer to the guest address space structure + */ +void gmap_enable(struct gmap *gmap) +{ + S390_lowcore.gmap = (unsigned long) gmap; +} +EXPORT_SYMBOL_GPL(gmap_enable); + +/** + * gmap_disable - switch back to the standard primary address space + * @gmap: pointer to the guest address space structure + */ +void gmap_disable(struct gmap *gmap) +{ + S390_lowcore.gmap = 0UL; +} +EXPORT_SYMBOL_GPL(gmap_disable); + +/* + * gmap_alloc_table is assumed to be called with mmap_sem held + */ +static int gmap_alloc_table(struct gmap *gmap, unsigned long *table, + unsigned long init, unsigned long gaddr) +{ + struct page *page; + unsigned long *new; + + /* since we dont free the gmap table until gmap_free we can unlock */ + page = alloc_pages(GFP_KERNEL, ALLOC_ORDER); + if (!page) + return -ENOMEM; + new = (unsigned long *) page_to_phys(page); + crst_table_init(new, init); + spin_lock(&gmap->mm->page_table_lock); + if (*table & _REGION_ENTRY_INVALID) { + list_add(&page->lru, &gmap->crst_list); + *table = (unsigned long) new | _REGION_ENTRY_LENGTH | + (*table & _REGION_ENTRY_TYPE_MASK); + page->index = gaddr; + page = NULL; + } + spin_unlock(&gmap->mm->page_table_lock); + if (page) + __free_pages(page, ALLOC_ORDER); + return 0; +} + +/** + * __gmap_segment_gaddr - find virtual address from segment pointer + * @entry: pointer to a segment table entry in the guest address space + * + * Returns the virtual address in the guest address space for the segment + */ +static unsigned long __gmap_segment_gaddr(unsigned long *entry) +{ + struct page *page; + unsigned long offset, mask; + + offset = (unsigned long) entry / sizeof(unsigned long); + offset = (offset & (PTRS_PER_PMD - 1)) * PMD_SIZE; + mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); + page = virt_to_page((void *)((unsigned long) entry & mask)); + return page->index + offset; +} + +/** + * __gmap_unlink_by_vmaddr - unlink a single segment via a host address + * @gmap: pointer to the guest address space structure + * @vmaddr: address in the host process address space + * + * Returns 1 if a TLB flush is required + */ +static int __gmap_unlink_by_vmaddr(struct gmap *gmap, unsigned long vmaddr) +{ + unsigned long *entry; + int flush = 0; + + spin_lock(&gmap->guest_table_lock); + entry = radix_tree_delete(&gmap->host_to_guest, vmaddr >> PMD_SHIFT); + if (entry) { + flush = (*entry != _SEGMENT_ENTRY_INVALID); + *entry = _SEGMENT_ENTRY_INVALID; + } + spin_unlock(&gmap->guest_table_lock); + return flush; +} + +/** + * __gmap_unmap_by_gaddr - unmap a single segment via a guest address + * @gmap: pointer to the guest address space structure + * @gaddr: address in the guest address space + * + * Returns 1 if a TLB flush is required + */ +static int __gmap_unmap_by_gaddr(struct gmap *gmap, unsigned long gaddr) +{ + unsigned long vmaddr; + + vmaddr = (unsigned long) radix_tree_delete(&gmap->guest_to_host, + gaddr >> PMD_SHIFT); + return vmaddr ? __gmap_unlink_by_vmaddr(gmap, vmaddr) : 0; +} + +/** + * gmap_unmap_segment - unmap segment from the guest address space + * @gmap: pointer to the guest address space structure + * @to: address in the guest address space + * @len: length of the memory area to unmap + * + * Returns 0 if the unmap succeeded, -EINVAL if not. + */ +int gmap_unmap_segment(struct gmap *gmap, unsigned long to, unsigned long len) +{ + unsigned long off; + int flush; + + if ((to | len) & (PMD_SIZE - 1)) + return -EINVAL; + if (len == 0 || to + len < to) + return -EINVAL; + + flush = 0; + down_write(&gmap->mm->mmap_sem); + for (off = 0; off < len; off += PMD_SIZE) + flush |= __gmap_unmap_by_gaddr(gmap, to + off); + up_write(&gmap->mm->mmap_sem); + if (flush) + gmap_flush_tlb(gmap); + return 0; +} +EXPORT_SYMBOL_GPL(gmap_unmap_segment); + +/** + * gmap_mmap_segment - map a segment to the guest address space + * @gmap: pointer to the guest address space structure + * @from: source address in the parent address space + * @to: target address in the guest address space + * @len: length of the memory area to map + * + * Returns 0 if the mmap succeeded, -EINVAL or -ENOMEM if not. + */ +int gmap_map_segment(struct gmap *gmap, unsigned long from, + unsigned long to, unsigned long len) +{ + unsigned long off; + int flush; + + if ((from | to | len) & (PMD_SIZE - 1)) + return -EINVAL; + if (len == 0 || from + len < from || to + len < to || + from + len > TASK_MAX_SIZE || to + len > gmap->asce_end) + return -EINVAL; + + flush = 0; + down_write(&gmap->mm->mmap_sem); + for (off = 0; off < len; off += PMD_SIZE) { + /* Remove old translation */ + flush |= __gmap_unmap_by_gaddr(gmap, to + off); + /* Store new translation */ + if (radix_tree_insert(&gmap->guest_to_host, + (to + off) >> PMD_SHIFT, + (void *) from + off)) + break; + } + up_write(&gmap->mm->mmap_sem); + if (flush) + gmap_flush_tlb(gmap); + if (off >= len) + return 0; + gmap_unmap_segment(gmap, to, len); + return -ENOMEM; +} +EXPORT_SYMBOL_GPL(gmap_map_segment); + +/** + * __gmap_translate - translate a guest address to a user space address + * @gmap: pointer to guest mapping meta data structure + * @gaddr: guest address + * + * Returns user space address which corresponds to the guest address or + * -EFAULT if no such mapping exists. + * This function does not establish potentially missing page table entries. + * The mmap_sem of the mm that belongs to the address space must be held + * when this function gets called. + */ +unsigned long __gmap_translate(struct gmap *gmap, unsigned long gaddr) +{ + unsigned long vmaddr; + + vmaddr = (unsigned long) + radix_tree_lookup(&gmap->guest_to_host, gaddr >> PMD_SHIFT); + return vmaddr ? (vmaddr | (gaddr & ~PMD_MASK)) : -EFAULT; +} +EXPORT_SYMBOL_GPL(__gmap_translate); + +/** + * gmap_translate - translate a guest address to a user space address + * @gmap: pointer to guest mapping meta data structure + * @gaddr: guest address + * + * Returns user space address which corresponds to the guest address or + * -EFAULT if no such mapping exists. + * This function does not establish potentially missing page table entries. + */ +unsigned long gmap_translate(struct gmap *gmap, unsigned long gaddr) +{ + unsigned long rc; + + down_read(&gmap->mm->mmap_sem); + rc = __gmap_translate(gmap, gaddr); + up_read(&gmap->mm->mmap_sem); + return rc; +} +EXPORT_SYMBOL_GPL(gmap_translate); + +/** + * gmap_unlink - disconnect a page table from the gmap shadow tables + * @gmap: pointer to guest mapping meta data structure + * @table: pointer to the host page table + * @vmaddr: vm address associated with the host page table + */ +static void gmap_unlink(struct mm_struct *mm, unsigned long *table, + unsigned long vmaddr) +{ + struct gmap *gmap; + int flush; + + list_for_each_entry(gmap, &mm->context.gmap_list, list) { + flush = __gmap_unlink_by_vmaddr(gmap, vmaddr); + if (flush) + gmap_flush_tlb(gmap); + } +} + +/** + * gmap_link - set up shadow page tables to connect a host to a guest address + * @gmap: pointer to guest mapping meta data structure + * @gaddr: guest address + * @vmaddr: vm address + * + * Returns 0 on success, -ENOMEM for out of memory conditions, and -EFAULT + * if the vm address is already mapped to a different guest segment. + * The mmap_sem of the mm that belongs to the address space must be held + * when this function gets called. + */ +int __gmap_link(struct gmap *gmap, unsigned long gaddr, unsigned long vmaddr) +{ + struct mm_struct *mm; + unsigned long *table; + spinlock_t *ptl; + pgd_t *pgd; + pud_t *pud; + pmd_t *pmd; + int rc; + + /* Create higher level tables in the gmap page table */ + table = gmap->table; + if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION1) { + table += (gaddr >> 53) & 0x7ff; + if ((*table & _REGION_ENTRY_INVALID) && + gmap_alloc_table(gmap, table, _REGION2_ENTRY_EMPTY, + gaddr & 0xffe0000000000000UL)) + return -ENOMEM; + table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN); + } + if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION2) { + table += (gaddr >> 42) & 0x7ff; + if ((*table & _REGION_ENTRY_INVALID) && + gmap_alloc_table(gmap, table, _REGION3_ENTRY_EMPTY, + gaddr & 0xfffffc0000000000UL)) + return -ENOMEM; + table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN); + } + if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION3) { + table += (gaddr >> 31) & 0x7ff; + if ((*table & _REGION_ENTRY_INVALID) && + gmap_alloc_table(gmap, table, _SEGMENT_ENTRY_EMPTY, + gaddr & 0xffffffff80000000UL)) + return -ENOMEM; + table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN); + } + table += (gaddr >> 20) & 0x7ff; + /* Walk the parent mm page table */ + mm = gmap->mm; + pgd = pgd_offset(mm, vmaddr); + VM_BUG_ON(pgd_none(*pgd)); + pud = pud_offset(pgd, vmaddr); + VM_BUG_ON(pud_none(*pud)); + pmd = pmd_offset(pud, vmaddr); + VM_BUG_ON(pmd_none(*pmd)); + /* large pmds cannot yet be handled */ + if (pmd_large(*pmd)) + return -EFAULT; + /* Link gmap segment table entry location to page table. */ + rc = radix_tree_preload(GFP_KERNEL); + if (rc) + return rc; + ptl = pmd_lock(mm, pmd); + spin_lock(&gmap->guest_table_lock); + if (*table == _SEGMENT_ENTRY_INVALID) { + rc = radix_tree_insert(&gmap->host_to_guest, + vmaddr >> PMD_SHIFT, table); + if (!rc) + *table = pmd_val(*pmd); + } else + rc = 0; + spin_unlock(&gmap->guest_table_lock); + spin_unlock(ptl); + radix_tree_preload_end(); + return rc; +} + +/** + * gmap_fault - resolve a fault on a guest address + * @gmap: pointer to guest mapping meta data structure + * @gaddr: guest address + * @fault_flags: flags to pass down to handle_mm_fault() + * + * Returns 0 on success, -ENOMEM for out of memory conditions, and -EFAULT + * if the vm address is already mapped to a different guest segment. + */ +int gmap_fault(struct gmap *gmap, unsigned long gaddr, + unsigned int fault_flags) +{ + unsigned long vmaddr; + int rc; + + down_read(&gmap->mm->mmap_sem); + vmaddr = __gmap_translate(gmap, gaddr); + if (IS_ERR_VALUE(vmaddr)) { + rc = vmaddr; + goto out_up; + } + if (fixup_user_fault(current, gmap->mm, vmaddr, fault_flags)) { + rc = -EFAULT; + goto out_up; + } + rc = __gmap_link(gmap, gaddr, vmaddr); +out_up: + up_read(&gmap->mm->mmap_sem); + return rc; +} +EXPORT_SYMBOL_GPL(gmap_fault); + +static void gmap_zap_swap_entry(swp_entry_t entry, struct mm_struct *mm) +{ + if (!non_swap_entry(entry)) + dec_mm_counter(mm, MM_SWAPENTS); + else if (is_migration_entry(entry)) { + struct page *page = migration_entry_to_page(entry); + + if (PageAnon(page)) + dec_mm_counter(mm, MM_ANONPAGES); + else + dec_mm_counter(mm, MM_FILEPAGES); + } + free_swap_and_cache(entry); +} + +/* + * this function is assumed to be called with mmap_sem held + */ +void __gmap_zap(struct gmap *gmap, unsigned long gaddr) +{ + unsigned long vmaddr, ptev, pgstev; + pte_t *ptep, pte; + spinlock_t *ptl; + pgste_t pgste; + + /* Find the vm address for the guest address */ + vmaddr = (unsigned long) radix_tree_lookup(&gmap->guest_to_host, + gaddr >> PMD_SHIFT); + if (!vmaddr) + return; + vmaddr |= gaddr & ~PMD_MASK; + /* Get pointer to the page table entry */ + ptep = get_locked_pte(gmap->mm, vmaddr, &ptl); + if (unlikely(!ptep)) + return; + pte = *ptep; + if (!pte_swap(pte)) + goto out_pte; + /* Zap unused and logically-zero pages */ + pgste = pgste_get_lock(ptep); + pgstev = pgste_val(pgste); + ptev = pte_val(pte); + if (((pgstev & _PGSTE_GPS_USAGE_MASK) == _PGSTE_GPS_USAGE_UNUSED) || + ((pgstev & _PGSTE_GPS_ZERO) && (ptev & _PAGE_INVALID))) { + gmap_zap_swap_entry(pte_to_swp_entry(pte), gmap->mm); + pte_clear(gmap->mm, vmaddr, ptep); + } + pgste_set_unlock(ptep, pgste); +out_pte: + pte_unmap_unlock(ptep, ptl); +} +EXPORT_SYMBOL_GPL(__gmap_zap); + +void gmap_discard(struct gmap *gmap, unsigned long from, unsigned long to) +{ + unsigned long gaddr, vmaddr, size; + struct vm_area_struct *vma; + + down_read(&gmap->mm->mmap_sem); + for (gaddr = from; gaddr < to; + gaddr = (gaddr + PMD_SIZE) & PMD_MASK) { + /* Find the vm address for the guest address */ + vmaddr = (unsigned long) + radix_tree_lookup(&gmap->guest_to_host, + gaddr >> PMD_SHIFT); + if (!vmaddr) + continue; + vmaddr |= gaddr & ~PMD_MASK; + /* Find vma in the parent mm */ + vma = find_vma(gmap->mm, vmaddr); + size = min(to - gaddr, PMD_SIZE - (gaddr & ~PMD_MASK)); + zap_page_range(vma, vmaddr, size, NULL); + } + up_read(&gmap->mm->mmap_sem); +} +EXPORT_SYMBOL_GPL(gmap_discard); + +static LIST_HEAD(gmap_notifier_list); +static DEFINE_SPINLOCK(gmap_notifier_lock); + +/** + * gmap_register_ipte_notifier - register a pte invalidation callback + * @nb: pointer to the gmap notifier block + */ +void gmap_register_ipte_notifier(struct gmap_notifier *nb) +{ + spin_lock(&gmap_notifier_lock); + list_add(&nb->list, &gmap_notifier_list); + spin_unlock(&gmap_notifier_lock); +} +EXPORT_SYMBOL_GPL(gmap_register_ipte_notifier); + +/** + * gmap_unregister_ipte_notifier - remove a pte invalidation callback + * @nb: pointer to the gmap notifier block + */ +void gmap_unregister_ipte_notifier(struct gmap_notifier *nb) +{ + spin_lock(&gmap_notifier_lock); + list_del_init(&nb->list); + spin_unlock(&gmap_notifier_lock); +} +EXPORT_SYMBOL_GPL(gmap_unregister_ipte_notifier); + +/** + * gmap_ipte_notify - mark a range of ptes for invalidation notification + * @gmap: pointer to guest mapping meta data structure + * @gaddr: virtual address in the guest address space + * @len: size of area + * + * Returns 0 if for each page in the given range a gmap mapping exists and + * the invalidation notification could be set. If the gmap mapping is missing + * for one or more pages -EFAULT is returned. If no memory could be allocated + * -ENOMEM is returned. This function establishes missing page table entries. + */ +int gmap_ipte_notify(struct gmap *gmap, unsigned long gaddr, unsigned long len) +{ + unsigned long addr; + spinlock_t *ptl; + pte_t *ptep, entry; + pgste_t pgste; + int rc = 0; + + if ((gaddr & ~PAGE_MASK) || (len & ~PAGE_MASK)) + return -EINVAL; + down_read(&gmap->mm->mmap_sem); + while (len) { + /* Convert gmap address and connect the page tables */ + addr = __gmap_translate(gmap, gaddr); + if (IS_ERR_VALUE(addr)) { + rc = addr; + break; + } + /* Get the page mapped */ + if (fixup_user_fault(current, gmap->mm, addr, FAULT_FLAG_WRITE)) { + rc = -EFAULT; + break; + } + rc = __gmap_link(gmap, gaddr, addr); + if (rc) + break; + /* Walk the process page table, lock and get pte pointer */ + ptep = get_locked_pte(gmap->mm, addr, &ptl); + VM_BUG_ON(!ptep); + /* Set notification bit in the pgste of the pte */ + entry = *ptep; + if ((pte_val(entry) & (_PAGE_INVALID | _PAGE_PROTECT)) == 0) { + pgste = pgste_get_lock(ptep); + pgste_val(pgste) |= PGSTE_IN_BIT; + pgste_set_unlock(ptep, pgste); + gaddr += PAGE_SIZE; + len -= PAGE_SIZE; + } + pte_unmap_unlock(ptep, ptl); + } + up_read(&gmap->mm->mmap_sem); + return rc; +} +EXPORT_SYMBOL_GPL(gmap_ipte_notify); + +/** + * gmap_do_ipte_notify - call all invalidation callbacks for a specific pte. + * @mm: pointer to the process mm_struct + * @addr: virtual address in the process address space + * @pte: pointer to the page table entry + * + * This function is assumed to be called with the page table lock held + * for the pte to notify. + */ +void gmap_do_ipte_notify(struct mm_struct *mm, unsigned long vmaddr, pte_t *pte) +{ + unsigned long offset, gaddr; + unsigned long *table; + struct gmap_notifier *nb; + struct gmap *gmap; + + offset = ((unsigned long) pte) & (255 * sizeof(pte_t)); + offset = offset * (4096 / sizeof(pte_t)); + spin_lock(&gmap_notifier_lock); + list_for_each_entry(gmap, &mm->context.gmap_list, list) { + table = radix_tree_lookup(&gmap->host_to_guest, + vmaddr >> PMD_SHIFT); + if (!table) + continue; + gaddr = __gmap_segment_gaddr(table) + offset; + list_for_each_entry(nb, &gmap_notifier_list, list) + nb->notifier_call(gmap, gaddr); + } + spin_unlock(&gmap_notifier_lock); +} +EXPORT_SYMBOL_GPL(gmap_do_ipte_notify); + +static inline int page_table_with_pgste(struct page *page) +{ + return atomic_read(&page->_mapcount) == 0; +} + +static inline unsigned long *page_table_alloc_pgste(struct mm_struct *mm) +{ + struct page *page; + unsigned long *table; + + page = alloc_page(GFP_KERNEL|__GFP_REPEAT); + if (!page) + return NULL; + if (!pgtable_page_ctor(page)) { + __free_page(page); + return NULL; + } + atomic_set(&page->_mapcount, 0); + table = (unsigned long *) page_to_phys(page); + clear_table(table, _PAGE_INVALID, PAGE_SIZE/2); + clear_table(table + PTRS_PER_PTE, 0, PAGE_SIZE/2); + return table; +} + +static inline void page_table_free_pgste(unsigned long *table) +{ + struct page *page; + + page = pfn_to_page(__pa(table) >> PAGE_SHIFT); + pgtable_page_dtor(page); + atomic_set(&page->_mapcount, -1); + __free_page(page); +} + +int set_guest_storage_key(struct mm_struct *mm, unsigned long addr, + unsigned long key, bool nq) +{ + spinlock_t *ptl; + pgste_t old, new; + pte_t *ptep; + + down_read(&mm->mmap_sem); +retry: + ptep = get_locked_pte(mm, addr, &ptl); + if (unlikely(!ptep)) { + up_read(&mm->mmap_sem); + return -EFAULT; + } + if (!(pte_val(*ptep) & _PAGE_INVALID) && + (pte_val(*ptep) & _PAGE_PROTECT)) { + pte_unmap_unlock(ptep, ptl); + if (fixup_user_fault(current, mm, addr, FAULT_FLAG_WRITE)) { + up_read(&mm->mmap_sem); + return -EFAULT; + } + goto retry; + } + + new = old = pgste_get_lock(ptep); + pgste_val(new) &= ~(PGSTE_GR_BIT | PGSTE_GC_BIT | + PGSTE_ACC_BITS | PGSTE_FP_BIT); + pgste_val(new) |= (key & (_PAGE_CHANGED | _PAGE_REFERENCED)) << 48; + pgste_val(new) |= (key & (_PAGE_ACC_BITS | _PAGE_FP_BIT)) << 56; + if (!(pte_val(*ptep) & _PAGE_INVALID)) { + unsigned long address, bits, skey; + + address = pte_val(*ptep) & PAGE_MASK; + skey = (unsigned long) page_get_storage_key(address); + bits = skey & (_PAGE_CHANGED | _PAGE_REFERENCED); + skey = key & (_PAGE_ACC_BITS | _PAGE_FP_BIT); + /* Set storage key ACC and FP */ + page_set_storage_key(address, skey, !nq); + /* Merge host changed & referenced into pgste */ + pgste_val(new) |= bits << 52; + } + /* changing the guest storage key is considered a change of the page */ + if ((pgste_val(new) ^ pgste_val(old)) & + (PGSTE_ACC_BITS | PGSTE_FP_BIT | PGSTE_GR_BIT | PGSTE_GC_BIT)) + pgste_val(new) |= PGSTE_UC_BIT; + + pgste_set_unlock(ptep, new); + pte_unmap_unlock(ptep, ptl); + up_read(&mm->mmap_sem); + return 0; +} +EXPORT_SYMBOL(set_guest_storage_key); + +unsigned long get_guest_storage_key(struct mm_struct *mm, unsigned long addr) +{ + spinlock_t *ptl; + pgste_t pgste; + pte_t *ptep; + uint64_t physaddr; + unsigned long key = 0; + + down_read(&mm->mmap_sem); + ptep = get_locked_pte(mm, addr, &ptl); + if (unlikely(!ptep)) { + up_read(&mm->mmap_sem); + return -EFAULT; + } + pgste = pgste_get_lock(ptep); + + if (pte_val(*ptep) & _PAGE_INVALID) { + key |= (pgste_val(pgste) & PGSTE_ACC_BITS) >> 56; + key |= (pgste_val(pgste) & PGSTE_FP_BIT) >> 56; + key |= (pgste_val(pgste) & PGSTE_GR_BIT) >> 48; + key |= (pgste_val(pgste) & PGSTE_GC_BIT) >> 48; + } else { + physaddr = pte_val(*ptep) & PAGE_MASK; + key = page_get_storage_key(physaddr); + + /* Reflect guest's logical view, not physical */ + if (pgste_val(pgste) & PGSTE_GR_BIT) + key |= _PAGE_REFERENCED; + if (pgste_val(pgste) & PGSTE_GC_BIT) + key |= _PAGE_CHANGED; + } + + pgste_set_unlock(ptep, pgste); + pte_unmap_unlock(ptep, ptl); + up_read(&mm->mmap_sem); + return key; +} +EXPORT_SYMBOL(get_guest_storage_key); + +static int page_table_allocate_pgste_min = 0; +static int page_table_allocate_pgste_max = 1; +int page_table_allocate_pgste = 0; +EXPORT_SYMBOL(page_table_allocate_pgste); + +static struct ctl_table page_table_sysctl[] = { + { + .procname = "allocate_pgste", + .data = &page_table_allocate_pgste, + .maxlen = sizeof(int), + .mode = S_IRUGO | S_IWUSR, + .proc_handler = proc_dointvec, + .extra1 = &page_table_allocate_pgste_min, + .extra2 = &page_table_allocate_pgste_max, + }, + { } +}; + +static struct ctl_table page_table_sysctl_dir[] = { + { + .procname = "vm", + .maxlen = 0, + .mode = 0555, + .child = page_table_sysctl, + }, + { } +}; + +static int __init page_table_register_sysctl(void) +{ + return register_sysctl_table(page_table_sysctl_dir) ? 0 : -ENOMEM; +} +__initcall(page_table_register_sysctl); + +#else /* CONFIG_PGSTE */ + +static inline int page_table_with_pgste(struct page *page) +{ + return 0; +} + +static inline unsigned long *page_table_alloc_pgste(struct mm_struct *mm) +{ + return NULL; +} + +static inline void page_table_free_pgste(unsigned long *table) +{ +} + +static inline void gmap_unlink(struct mm_struct *mm, unsigned long *table, + unsigned long vmaddr) +{ +} + +#endif /* CONFIG_PGSTE */ + +static inline unsigned int atomic_xor_bits(atomic_t *v, unsigned int bits) +{ + unsigned int old, new; + + do { + old = atomic_read(v); + new = old ^ bits; + } while (atomic_cmpxchg(v, old, new) != old); + return new; +} + +/* + * page table entry allocation/free routines. + */ +unsigned long *page_table_alloc(struct mm_struct *mm) +{ + unsigned long *uninitialized_var(table); + struct page *uninitialized_var(page); + unsigned int mask, bit; + + if (mm_alloc_pgste(mm)) + return page_table_alloc_pgste(mm); + /* Allocate fragments of a 4K page as 1K/2K page table */ + spin_lock_bh(&mm->context.list_lock); + mask = FRAG_MASK; + if (!list_empty(&mm->context.pgtable_list)) { + page = list_first_entry(&mm->context.pgtable_list, + struct page, lru); + table = (unsigned long *) page_to_phys(page); + mask = atomic_read(&page->_mapcount); + mask = mask | (mask >> 4); + } + if ((mask & FRAG_MASK) == FRAG_MASK) { + spin_unlock_bh(&mm->context.list_lock); + page = alloc_page(GFP_KERNEL|__GFP_REPEAT); + if (!page) + return NULL; + if (!pgtable_page_ctor(page)) { + __free_page(page); + return NULL; + } + atomic_set(&page->_mapcount, 1); + table = (unsigned long *) page_to_phys(page); + clear_table(table, _PAGE_INVALID, PAGE_SIZE); + spin_lock_bh(&mm->context.list_lock); + list_add(&page->lru, &mm->context.pgtable_list); + } else { + for (bit = 1; mask & bit; bit <<= 1) + table += PTRS_PER_PTE; + mask = atomic_xor_bits(&page->_mapcount, bit); + if ((mask & FRAG_MASK) == FRAG_MASK) + list_del(&page->lru); + } + spin_unlock_bh(&mm->context.list_lock); + return table; +} + +void page_table_free(struct mm_struct *mm, unsigned long *table) +{ + struct page *page; + unsigned int bit, mask; + + page = pfn_to_page(__pa(table) >> PAGE_SHIFT); + if (page_table_with_pgste(page)) + return page_table_free_pgste(table); + /* Free 1K/2K page table fragment of a 4K page */ + bit = 1 << ((__pa(table) & ~PAGE_MASK)/(PTRS_PER_PTE*sizeof(pte_t))); + spin_lock_bh(&mm->context.list_lock); + if ((atomic_read(&page->_mapcount) & FRAG_MASK) != FRAG_MASK) + list_del(&page->lru); + mask = atomic_xor_bits(&page->_mapcount, bit); + if (mask & FRAG_MASK) + list_add(&page->lru, &mm->context.pgtable_list); + spin_unlock_bh(&mm->context.list_lock); + if (mask == 0) { + pgtable_page_dtor(page); + atomic_set(&page->_mapcount, -1); + __free_page(page); + } +} + +static void __page_table_free_rcu(void *table, unsigned bit) +{ + struct page *page; + + if (bit == FRAG_MASK) + return page_table_free_pgste(table); + /* Free 1K/2K page table fragment of a 4K page */ + page = pfn_to_page(__pa(table) >> PAGE_SHIFT); + if (atomic_xor_bits(&page->_mapcount, bit) == 0) { + pgtable_page_dtor(page); + atomic_set(&page->_mapcount, -1); + __free_page(page); + } +} + +void page_table_free_rcu(struct mmu_gather *tlb, unsigned long *table, + unsigned long vmaddr) +{ + struct mm_struct *mm; + struct page *page; + unsigned int bit, mask; + + mm = tlb->mm; + page = pfn_to_page(__pa(table) >> PAGE_SHIFT); + if (page_table_with_pgste(page)) { + gmap_unlink(mm, table, vmaddr); + table = (unsigned long *) (__pa(table) | FRAG_MASK); + tlb_remove_table(tlb, table); + return; + } + bit = 1 << ((__pa(table) & ~PAGE_MASK) / (PTRS_PER_PTE*sizeof(pte_t))); + spin_lock_bh(&mm->context.list_lock); + if ((atomic_read(&page->_mapcount) & FRAG_MASK) != FRAG_MASK) + list_del(&page->lru); + mask = atomic_xor_bits(&page->_mapcount, bit | (bit << 4)); + if (mask & FRAG_MASK) + list_add_tail(&page->lru, &mm->context.pgtable_list); + spin_unlock_bh(&mm->context.list_lock); + table = (unsigned long *) (__pa(table) | (bit << 4)); + tlb_remove_table(tlb, table); +} + +static void __tlb_remove_table(void *_table) +{ + const unsigned long mask = (FRAG_MASK << 4) | FRAG_MASK; + void *table = (void *)((unsigned long) _table & ~mask); + unsigned type = (unsigned long) _table & mask; + + if (type) + __page_table_free_rcu(table, type); + else + free_pages((unsigned long) table, ALLOC_ORDER); +} + +static void tlb_remove_table_smp_sync(void *arg) +{ + /* Simply deliver the interrupt */ +} + +static void tlb_remove_table_one(void *table) +{ + /* + * This isn't an RCU grace period and hence the page-tables cannot be + * assumed to be actually RCU-freed. + * + * It is however sufficient for software page-table walkers that rely + * on IRQ disabling. See the comment near struct mmu_table_batch. + */ + smp_call_function(tlb_remove_table_smp_sync, NULL, 1); + __tlb_remove_table(table); +} + +static void tlb_remove_table_rcu(struct rcu_head *head) +{ + struct mmu_table_batch *batch; + int i; + + batch = container_of(head, struct mmu_table_batch, rcu); + + for (i = 0; i < batch->nr; i++) + __tlb_remove_table(batch->tables[i]); + + free_page((unsigned long)batch); +} + +void tlb_table_flush(struct mmu_gather *tlb) +{ + struct mmu_table_batch **batch = &tlb->batch; + + if (*batch) { + call_rcu_sched(&(*batch)->rcu, tlb_remove_table_rcu); + *batch = NULL; + } +} + +void tlb_remove_table(struct mmu_gather *tlb, void *table) +{ + struct mmu_table_batch **batch = &tlb->batch; + + tlb->mm->context.flush_mm = 1; + if (*batch == NULL) { + *batch = (struct mmu_table_batch *) + __get_free_page(GFP_NOWAIT | __GFP_NOWARN); + if (*batch == NULL) { + __tlb_flush_mm_lazy(tlb->mm); + tlb_remove_table_one(table); + return; + } + (*batch)->nr = 0; + } + (*batch)->tables[(*batch)->nr++] = table; + if ((*batch)->nr == MAX_TABLE_BATCH) + tlb_flush_mmu(tlb); +} + +#ifdef CONFIG_TRANSPARENT_HUGEPAGE +static inline void thp_split_vma(struct vm_area_struct *vma) +{ + unsigned long addr; + + for (addr = vma->vm_start; addr < vma->vm_end; addr += PAGE_SIZE) + follow_page(vma, addr, FOLL_SPLIT); +} + +static inline void thp_split_mm(struct mm_struct *mm) +{ + struct vm_area_struct *vma; + + for (vma = mm->mmap; vma != NULL; vma = vma->vm_next) { + thp_split_vma(vma); + vma->vm_flags &= ~VM_HUGEPAGE; + vma->vm_flags |= VM_NOHUGEPAGE; + } + mm->def_flags |= VM_NOHUGEPAGE; +} +#else +static inline void thp_split_mm(struct mm_struct *mm) +{ +} +#endif /* CONFIG_TRANSPARENT_HUGEPAGE */ + +/* + * switch on pgstes for its userspace process (for kvm) + */ +int s390_enable_sie(void) +{ + struct mm_struct *mm = current->mm; + + /* Do we have pgstes? if yes, we are done */ + if (mm_has_pgste(mm)) + return 0; + /* Fail if the page tables are 2K */ + if (!mm_alloc_pgste(mm)) + return -EINVAL; + down_write(&mm->mmap_sem); + mm->context.has_pgste = 1; + /* split thp mappings and disable thp for future mappings */ + thp_split_mm(mm); + up_write(&mm->mmap_sem); + return 0; +} +EXPORT_SYMBOL_GPL(s390_enable_sie); + +/* + * Enable storage key handling from now on and initialize the storage + * keys with the default key. + */ +static int __s390_enable_skey(pte_t *pte, unsigned long addr, + unsigned long next, struct mm_walk *walk) +{ + unsigned long ptev; + pgste_t pgste; + + pgste = pgste_get_lock(pte); + /* + * Remove all zero page mappings, + * after establishing a policy to forbid zero page mappings + * following faults for that page will get fresh anonymous pages + */ + if (is_zero_pfn(pte_pfn(*pte))) { + ptep_flush_direct(walk->mm, addr, pte); + pte_val(*pte) = _PAGE_INVALID; + } + /* Clear storage key */ + pgste_val(pgste) &= ~(PGSTE_ACC_BITS | PGSTE_FP_BIT | + PGSTE_GR_BIT | PGSTE_GC_BIT); + ptev = pte_val(*pte); + if (!(ptev & _PAGE_INVALID) && (ptev & _PAGE_WRITE)) + page_set_storage_key(ptev & PAGE_MASK, PAGE_DEFAULT_KEY, 1); + pgste_set_unlock(pte, pgste); + return 0; +} + +int s390_enable_skey(void) +{ + struct mm_walk walk = { .pte_entry = __s390_enable_skey }; + struct mm_struct *mm = current->mm; + struct vm_area_struct *vma; + int rc = 0; + + down_write(&mm->mmap_sem); + if (mm_use_skey(mm)) + goto out_up; + + mm->context.use_skey = 1; + for (vma = mm->mmap; vma; vma = vma->vm_next) { + if (ksm_madvise(vma, vma->vm_start, vma->vm_end, + MADV_UNMERGEABLE, &vma->vm_flags)) { + mm->context.use_skey = 0; + rc = -ENOMEM; + goto out_up; + } + } + mm->def_flags &= ~VM_MERGEABLE; + + walk.mm = mm; + walk_page_range(0, TASK_SIZE, &walk); + +out_up: + up_write(&mm->mmap_sem); + return rc; +} +EXPORT_SYMBOL_GPL(s390_enable_skey); + +/* + * Reset CMMA state, make all pages stable again. + */ +static int __s390_reset_cmma(pte_t *pte, unsigned long addr, + unsigned long next, struct mm_walk *walk) +{ + pgste_t pgste; + + pgste = pgste_get_lock(pte); + pgste_val(pgste) &= ~_PGSTE_GPS_USAGE_MASK; + pgste_set_unlock(pte, pgste); + return 0; +} + +void s390_reset_cmma(struct mm_struct *mm) +{ + struct mm_walk walk = { .pte_entry = __s390_reset_cmma }; + + down_write(&mm->mmap_sem); + walk.mm = mm; + walk_page_range(0, TASK_SIZE, &walk); + up_write(&mm->mmap_sem); +} +EXPORT_SYMBOL_GPL(s390_reset_cmma); + +/* + * Test and reset if a guest page is dirty + */ +bool gmap_test_and_clear_dirty(unsigned long address, struct gmap *gmap) +{ + pte_t *pte; + spinlock_t *ptl; + bool dirty = false; + + pte = get_locked_pte(gmap->mm, address, &ptl); + if (unlikely(!pte)) + return false; + + if (ptep_test_and_clear_user_dirty(gmap->mm, address, pte)) + dirty = true; + + spin_unlock(ptl); + return dirty; +} +EXPORT_SYMBOL_GPL(gmap_test_and_clear_dirty); + +#ifdef CONFIG_TRANSPARENT_HUGEPAGE +int pmdp_clear_flush_young(struct vm_area_struct *vma, unsigned long address, + pmd_t *pmdp) +{ + VM_BUG_ON(address & ~HPAGE_PMD_MASK); + /* No need to flush TLB + * On s390 reference bits are in storage key and never in TLB */ + return pmdp_test_and_clear_young(vma, address, pmdp); +} + +int pmdp_set_access_flags(struct vm_area_struct *vma, + unsigned long address, pmd_t *pmdp, + pmd_t entry, int dirty) +{ + VM_BUG_ON(address & ~HPAGE_PMD_MASK); + + entry = pmd_mkyoung(entry); + if (dirty) + entry = pmd_mkdirty(entry); + if (pmd_same(*pmdp, entry)) + return 0; + pmdp_invalidate(vma, address, pmdp); + set_pmd_at(vma->vm_mm, address, pmdp, entry); + return 1; +} + +static void pmdp_splitting_flush_sync(void *arg) +{ + /* Simply deliver the interrupt */ +} + +void pmdp_splitting_flush(struct vm_area_struct *vma, unsigned long address, + pmd_t *pmdp) +{ + VM_BUG_ON(address & ~HPAGE_PMD_MASK); + if (!test_and_set_bit(_SEGMENT_ENTRY_SPLIT_BIT, + (unsigned long *) pmdp)) { + /* need to serialize against gup-fast (IRQ disabled) */ + smp_call_function(pmdp_splitting_flush_sync, NULL, 1); + } +} + +void pgtable_trans_huge_deposit(struct mm_struct *mm, pmd_t *pmdp, + pgtable_t pgtable) +{ + struct list_head *lh = (struct list_head *) pgtable; + + assert_spin_locked(pmd_lockptr(mm, pmdp)); + + /* FIFO */ + if (!pmd_huge_pte(mm, pmdp)) + INIT_LIST_HEAD(lh); + else + list_add(lh, (struct list_head *) pmd_huge_pte(mm, pmdp)); + pmd_huge_pte(mm, pmdp) = pgtable; +} + +pgtable_t pgtable_trans_huge_withdraw(struct mm_struct *mm, pmd_t *pmdp) +{ + struct list_head *lh; + pgtable_t pgtable; + pte_t *ptep; + + assert_spin_locked(pmd_lockptr(mm, pmdp)); + + /* FIFO */ + pgtable = pmd_huge_pte(mm, pmdp); + lh = (struct list_head *) pgtable; + if (list_empty(lh)) + pmd_huge_pte(mm, pmdp) = NULL; + else { + pmd_huge_pte(mm, pmdp) = (pgtable_t) lh->next; + list_del(lh); + } + ptep = (pte_t *) pgtable; + pte_val(*ptep) = _PAGE_INVALID; + ptep++; + pte_val(*ptep) = _PAGE_INVALID; + return pgtable; +} +#endif /* CONFIG_TRANSPARENT_HUGEPAGE */ diff --git a/kernel/arch/s390/mm/vmem.c b/kernel/arch/s390/mm/vmem.c new file mode 100644 index 000000000..ef7d6c8fe --- /dev/null +++ b/kernel/arch/s390/mm/vmem.c @@ -0,0 +1,416 @@ +/* + * Copyright IBM Corp. 2006 + * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com> + */ + +#include <linux/bootmem.h> +#include <linux/pfn.h> +#include <linux/mm.h> +#include <linux/module.h> +#include <linux/list.h> +#include <linux/hugetlb.h> +#include <linux/slab.h> +#include <linux/memblock.h> +#include <asm/pgalloc.h> +#include <asm/pgtable.h> +#include <asm/setup.h> +#include <asm/tlbflush.h> +#include <asm/sections.h> + +static DEFINE_MUTEX(vmem_mutex); + +struct memory_segment { + struct list_head list; + unsigned long start; + unsigned long size; +}; + +static LIST_HEAD(mem_segs); + +static void __ref *vmem_alloc_pages(unsigned int order) +{ + if (slab_is_available()) + return (void *)__get_free_pages(GFP_KERNEL, order); + return alloc_bootmem_pages((1 << order) * PAGE_SIZE); +} + +static inline pud_t *vmem_pud_alloc(void) +{ + pud_t *pud = NULL; + + pud = vmem_alloc_pages(2); + if (!pud) + return NULL; + clear_table((unsigned long *) pud, _REGION3_ENTRY_EMPTY, PAGE_SIZE * 4); + return pud; +} + +static inline pmd_t *vmem_pmd_alloc(void) +{ + pmd_t *pmd = NULL; + + pmd = vmem_alloc_pages(2); + if (!pmd) + return NULL; + clear_table((unsigned long *) pmd, _SEGMENT_ENTRY_EMPTY, PAGE_SIZE * 4); + return pmd; +} + +static pte_t __ref *vmem_pte_alloc(unsigned long address) +{ + pte_t *pte; + + if (slab_is_available()) + pte = (pte_t *) page_table_alloc(&init_mm); + else + pte = alloc_bootmem_align(PTRS_PER_PTE * sizeof(pte_t), + PTRS_PER_PTE * sizeof(pte_t)); + if (!pte) + return NULL; + clear_table((unsigned long *) pte, _PAGE_INVALID, + PTRS_PER_PTE * sizeof(pte_t)); + return pte; +} + +/* + * Add a physical memory range to the 1:1 mapping. + */ +static int vmem_add_mem(unsigned long start, unsigned long size, int ro) +{ + unsigned long end = start + size; + unsigned long address = start; + pgd_t *pg_dir; + pud_t *pu_dir; + pmd_t *pm_dir; + pte_t *pt_dir; + int ret = -ENOMEM; + + while (address < end) { + pg_dir = pgd_offset_k(address); + if (pgd_none(*pg_dir)) { + pu_dir = vmem_pud_alloc(); + if (!pu_dir) + goto out; + pgd_populate(&init_mm, pg_dir, pu_dir); + } + pu_dir = pud_offset(pg_dir, address); +#ifndef CONFIG_DEBUG_PAGEALLOC + if (MACHINE_HAS_EDAT2 && pud_none(*pu_dir) && address && + !(address & ~PUD_MASK) && (address + PUD_SIZE <= end)) { + pud_val(*pu_dir) = __pa(address) | + _REGION_ENTRY_TYPE_R3 | _REGION3_ENTRY_LARGE | + (ro ? _REGION_ENTRY_PROTECT : 0); + address += PUD_SIZE; + continue; + } +#endif + if (pud_none(*pu_dir)) { + pm_dir = vmem_pmd_alloc(); + if (!pm_dir) + goto out; + pud_populate(&init_mm, pu_dir, pm_dir); + } + pm_dir = pmd_offset(pu_dir, address); +#ifndef CONFIG_DEBUG_PAGEALLOC + if (MACHINE_HAS_EDAT1 && pmd_none(*pm_dir) && address && + !(address & ~PMD_MASK) && (address + PMD_SIZE <= end)) { + pmd_val(*pm_dir) = __pa(address) | + _SEGMENT_ENTRY | _SEGMENT_ENTRY_LARGE | + _SEGMENT_ENTRY_YOUNG | + (ro ? _SEGMENT_ENTRY_PROTECT : 0); + address += PMD_SIZE; + continue; + } +#endif + if (pmd_none(*pm_dir)) { + pt_dir = vmem_pte_alloc(address); + if (!pt_dir) + goto out; + pmd_populate(&init_mm, pm_dir, pt_dir); + } + + pt_dir = pte_offset_kernel(pm_dir, address); + pte_val(*pt_dir) = __pa(address) | + pgprot_val(ro ? PAGE_KERNEL_RO : PAGE_KERNEL); + address += PAGE_SIZE; + } + ret = 0; +out: + return ret; +} + +/* + * Remove a physical memory range from the 1:1 mapping. + * Currently only invalidates page table entries. + */ +static void vmem_remove_range(unsigned long start, unsigned long size) +{ + unsigned long end = start + size; + unsigned long address = start; + pgd_t *pg_dir; + pud_t *pu_dir; + pmd_t *pm_dir; + pte_t *pt_dir; + pte_t pte; + + pte_val(pte) = _PAGE_INVALID; + while (address < end) { + pg_dir = pgd_offset_k(address); + if (pgd_none(*pg_dir)) { + address += PGDIR_SIZE; + continue; + } + pu_dir = pud_offset(pg_dir, address); + if (pud_none(*pu_dir)) { + address += PUD_SIZE; + continue; + } + if (pud_large(*pu_dir)) { + pud_clear(pu_dir); + address += PUD_SIZE; + continue; + } + pm_dir = pmd_offset(pu_dir, address); + if (pmd_none(*pm_dir)) { + address += PMD_SIZE; + continue; + } + if (pmd_large(*pm_dir)) { + pmd_clear(pm_dir); + address += PMD_SIZE; + continue; + } + pt_dir = pte_offset_kernel(pm_dir, address); + *pt_dir = pte; + address += PAGE_SIZE; + } + flush_tlb_kernel_range(start, end); +} + +/* + * Add a backed mem_map array to the virtual mem_map array. + */ +int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) +{ + unsigned long address = start; + pgd_t *pg_dir; + pud_t *pu_dir; + pmd_t *pm_dir; + pte_t *pt_dir; + int ret = -ENOMEM; + + for (address = start; address < end;) { + pg_dir = pgd_offset_k(address); + if (pgd_none(*pg_dir)) { + pu_dir = vmem_pud_alloc(); + if (!pu_dir) + goto out; + pgd_populate(&init_mm, pg_dir, pu_dir); + } + + pu_dir = pud_offset(pg_dir, address); + if (pud_none(*pu_dir)) { + pm_dir = vmem_pmd_alloc(); + if (!pm_dir) + goto out; + pud_populate(&init_mm, pu_dir, pm_dir); + } + + pm_dir = pmd_offset(pu_dir, address); + if (pmd_none(*pm_dir)) { + /* Use 1MB frames for vmemmap if available. We always + * use large frames even if they are only partially + * used. + * Otherwise we would have also page tables since + * vmemmap_populate gets called for each section + * separately. */ + if (MACHINE_HAS_EDAT1) { + void *new_page; + + new_page = vmemmap_alloc_block(PMD_SIZE, node); + if (!new_page) + goto out; + pmd_val(*pm_dir) = __pa(new_page) | + _SEGMENT_ENTRY | _SEGMENT_ENTRY_LARGE; + address = (address + PMD_SIZE) & PMD_MASK; + continue; + } + pt_dir = vmem_pte_alloc(address); + if (!pt_dir) + goto out; + pmd_populate(&init_mm, pm_dir, pt_dir); + } else if (pmd_large(*pm_dir)) { + address = (address + PMD_SIZE) & PMD_MASK; + continue; + } + + pt_dir = pte_offset_kernel(pm_dir, address); + if (pte_none(*pt_dir)) { + void *new_page; + + new_page = vmemmap_alloc_block(PAGE_SIZE, node); + if (!new_page) + goto out; + pte_val(*pt_dir) = + __pa(new_page) | pgprot_val(PAGE_KERNEL); + } + address += PAGE_SIZE; + } + ret = 0; +out: + return ret; +} + +void vmemmap_free(unsigned long start, unsigned long end) +{ +} + +/* + * Add memory segment to the segment list if it doesn't overlap with + * an already present segment. + */ +static int insert_memory_segment(struct memory_segment *seg) +{ + struct memory_segment *tmp; + + if (seg->start + seg->size > VMEM_MAX_PHYS || + seg->start + seg->size < seg->start) + return -ERANGE; + + list_for_each_entry(tmp, &mem_segs, list) { + if (seg->start >= tmp->start + tmp->size) + continue; + if (seg->start + seg->size <= tmp->start) + continue; + return -ENOSPC; + } + list_add(&seg->list, &mem_segs); + return 0; +} + +/* + * Remove memory segment from the segment list. + */ +static void remove_memory_segment(struct memory_segment *seg) +{ + list_del(&seg->list); +} + +static void __remove_shared_memory(struct memory_segment *seg) +{ + remove_memory_segment(seg); + vmem_remove_range(seg->start, seg->size); +} + +int vmem_remove_mapping(unsigned long start, unsigned long size) +{ + struct memory_segment *seg; + int ret; + + mutex_lock(&vmem_mutex); + + ret = -ENOENT; + list_for_each_entry(seg, &mem_segs, list) { + if (seg->start == start && seg->size == size) + break; + } + + if (seg->start != start || seg->size != size) + goto out; + + ret = 0; + __remove_shared_memory(seg); + kfree(seg); +out: + mutex_unlock(&vmem_mutex); + return ret; +} + +int vmem_add_mapping(unsigned long start, unsigned long size) +{ + struct memory_segment *seg; + int ret; + + mutex_lock(&vmem_mutex); + ret = -ENOMEM; + seg = kzalloc(sizeof(*seg), GFP_KERNEL); + if (!seg) + goto out; + seg->start = start; + seg->size = size; + + ret = insert_memory_segment(seg); + if (ret) + goto out_free; + + ret = vmem_add_mem(start, size, 0); + if (ret) + goto out_remove; + goto out; + +out_remove: + __remove_shared_memory(seg); +out_free: + kfree(seg); +out: + mutex_unlock(&vmem_mutex); + return ret; +} + +/* + * map whole physical memory to virtual memory (identity mapping) + * we reserve enough space in the vmalloc area for vmemmap to hotplug + * additional memory segments. + */ +void __init vmem_map_init(void) +{ + unsigned long ro_start, ro_end; + struct memblock_region *reg; + phys_addr_t start, end; + + ro_start = PFN_ALIGN((unsigned long)&_stext); + ro_end = (unsigned long)&_eshared & PAGE_MASK; + for_each_memblock(memory, reg) { + start = reg->base; + end = reg->base + reg->size - 1; + if (start >= ro_end || end <= ro_start) + vmem_add_mem(start, end - start, 0); + else if (start >= ro_start && end <= ro_end) + vmem_add_mem(start, end - start, 1); + else if (start >= ro_start) { + vmem_add_mem(start, ro_end - start, 1); + vmem_add_mem(ro_end, end - ro_end, 0); + } else if (end < ro_end) { + vmem_add_mem(start, ro_start - start, 0); + vmem_add_mem(ro_start, end - ro_start, 1); + } else { + vmem_add_mem(start, ro_start - start, 0); + vmem_add_mem(ro_start, ro_end - ro_start, 1); + vmem_add_mem(ro_end, end - ro_end, 0); + } + } +} + +/* + * Convert memblock.memory to a memory segment list so there is a single + * list that contains all memory segments. + */ +static int __init vmem_convert_memory_chunk(void) +{ + struct memblock_region *reg; + struct memory_segment *seg; + + mutex_lock(&vmem_mutex); + for_each_memblock(memory, reg) { + seg = kzalloc(sizeof(*seg), GFP_KERNEL); + if (!seg) + panic("Out of memory...\n"); + seg->start = reg->base; + seg->size = reg->size; + insert_memory_segment(seg); + } + mutex_unlock(&vmem_mutex); + return 0; +} + +core_initcall(vmem_convert_memory_chunk); |