diff options
Diffstat (limited to 'arch/x86/mm/kmmio.c')
| -rw-r--r-- | arch/x86/mm/kmmio.c | 173 |
1 files changed, 104 insertions, 69 deletions
diff --git a/arch/x86/mm/kmmio.c b/arch/x86/mm/kmmio.c index 6a518dd08a3..637ab34ed63 100644 --- a/arch/x86/mm/kmmio.c +++ b/arch/x86/mm/kmmio.c @@ -5,11 +5,12 @@ * 2008 Pekka Paalanen <pq@iki.fi> */ +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + #include <linux/list.h> #include <linux/rculist.h> #include <linux/spinlock.h> #include <linux/hash.h> -#include <linux/init.h> #include <linux/module.h> #include <linux/kernel.h> #include <linux/uaccess.h> @@ -19,6 +20,7 @@ #include <linux/kdebug.h> #include <linux/mutex.h> #include <linux/io.h> +#include <linux/slab.h> #include <asm/cacheflush.h> #include <asm/tlbflush.h> #include <linux/errno.h> @@ -32,7 +34,7 @@ struct kmmio_fault_page { struct list_head list; struct kmmio_fault_page *release_next; unsigned long page; /* location of the fault page */ - bool old_presence; /* page presence prior to arming */ + pteval_t old_presence; /* page presence prior to arming */ bool armed; /* @@ -42,6 +44,8 @@ struct kmmio_fault_page { * Protected by kmmio_lock, when linked into kmmio_page_table. */ int count; + + bool scheduled_for_release; }; struct kmmio_delayed_release { @@ -87,7 +91,7 @@ static struct kmmio_probe *get_kmmio_probe(unsigned long addr) { struct kmmio_probe *p; list_for_each_entry_rcu(p, &kmmio_probes, list) { - if (addr >= p->addr && addr <= (p->addr + p->len)) + if (addr >= p->addr && addr < (p->addr + p->len)) return p; } return NULL; @@ -97,60 +101,62 @@ static struct kmmio_probe *get_kmmio_probe(unsigned long addr) static struct kmmio_fault_page *get_kmmio_fault_page(unsigned long page) { struct list_head *head; - struct kmmio_fault_page *p; + struct kmmio_fault_page *f; page &= PAGE_MASK; head = kmmio_page_list(page); - list_for_each_entry_rcu(p, head, list) { - if (p->page == page) - return p; + list_for_each_entry_rcu(f, head, list) { + if (f->page == page) + return f; } return NULL; } -static void set_pmd_presence(pmd_t *pmd, bool present, bool *old) +static void clear_pmd_presence(pmd_t *pmd, bool clear, pmdval_t *old) { pmdval_t v = pmd_val(*pmd); - *old = !!(v & _PAGE_PRESENT); - v &= ~_PAGE_PRESENT; - if (present) - v |= _PAGE_PRESENT; + if (clear) { + *old = v & _PAGE_PRESENT; + v &= ~_PAGE_PRESENT; + } else /* presume this has been called with clear==true previously */ + v |= *old; set_pmd(pmd, __pmd(v)); } -static void set_pte_presence(pte_t *pte, bool present, bool *old) +static void clear_pte_presence(pte_t *pte, bool clear, pteval_t *old) { pteval_t v = pte_val(*pte); - *old = !!(v & _PAGE_PRESENT); - v &= ~_PAGE_PRESENT; - if (present) - v |= _PAGE_PRESENT; + if (clear) { + *old = v & _PAGE_PRESENT; + v &= ~_PAGE_PRESENT; + } else /* presume this has been called with clear==true previously */ + v |= *old; set_pte_atomic(pte, __pte(v)); } -static int set_page_presence(unsigned long addr, bool present, bool *old) +static int clear_page_presence(struct kmmio_fault_page *f, bool clear) { unsigned int level; - pte_t *pte = lookup_address(addr, &level); + pte_t *pte = lookup_address(f->page, &level); if (!pte) { - pr_err("kmmio: no pte for page 0x%08lx\n", addr); + pr_err("no pte for page 0x%08lx\n", f->page); return -1; } switch (level) { case PG_LEVEL_2M: - set_pmd_presence((pmd_t *)pte, present, old); + clear_pmd_presence((pmd_t *)pte, clear, &f->old_presence); break; case PG_LEVEL_4K: - set_pte_presence(pte, present, old); + clear_pte_presence(pte, clear, &f->old_presence); break; default: - pr_err("kmmio: unexpected page level 0x%x.\n", level); + pr_err("unexpected page level 0x%x.\n", level); return -1; } - __flush_tlb_one(addr); + __flush_tlb_one(f->page); return 0; } @@ -168,13 +174,14 @@ static int set_page_presence(unsigned long addr, bool present, bool *old) static int arm_kmmio_fault_page(struct kmmio_fault_page *f) { int ret; - WARN_ONCE(f->armed, KERN_ERR "kmmio page already armed.\n"); + WARN_ONCE(f->armed, KERN_ERR pr_fmt("kmmio page already armed.\n")); if (f->armed) { - pr_warning("kmmio double-arm: page 0x%08lx, ref %d, old %d\n", - f->page, f->count, f->old_presence); + pr_warning("double-arm: page 0x%08lx, ref %d, old %d\n", + f->page, f->count, !!f->old_presence); } - ret = set_page_presence(f->page, false, &f->old_presence); - WARN_ONCE(ret < 0, KERN_ERR "kmmio arming 0x%08lx failed.\n", f->page); + ret = clear_page_presence(f, true); + WARN_ONCE(ret < 0, KERN_ERR pr_fmt("arming 0x%08lx failed.\n"), + f->page); f->armed = true; return ret; } @@ -182,8 +189,7 @@ static int arm_kmmio_fault_page(struct kmmio_fault_page *f) /** Restore the given page to saved presence state. */ static void disarm_kmmio_fault_page(struct kmmio_fault_page *f) { - bool tmp; - int ret = set_page_presence(f->page, f->old_presence, &tmp); + int ret = clear_page_presence(f, false); WARN_ONCE(ret < 0, KERN_ERR "kmmio disarming 0x%08lx failed.\n", f->page); f->armed = false; @@ -202,7 +208,7 @@ static void disarm_kmmio_fault_page(struct kmmio_fault_page *f) */ /* * Interrupts are disabled on entry as trap3 is an interrupt gate - * and they remain disabled thorough out this function. + * and they remain disabled throughout this function. */ int kmmio_handler(struct pt_regs *regs, unsigned long addr) { @@ -239,24 +245,21 @@ int kmmio_handler(struct pt_regs *regs, unsigned long addr) * condition needs handling by do_page_fault(), the * page really not being present is the most common. */ - pr_debug("kmmio: secondary hit for 0x%08lx CPU %d.\n", - addr, smp_processor_id()); + pr_debug("secondary hit for 0x%08lx CPU %d.\n", + addr, smp_processor_id()); if (!faultpage->old_presence) - pr_info("kmmio: unexpected secondary hit for " - "address 0x%08lx on CPU %d.\n", addr, - smp_processor_id()); + pr_info("unexpected secondary hit for address 0x%08lx on CPU %d.\n", + addr, smp_processor_id()); } else { /* * Prevent overwriting already in-flight context. * This should not happen, let's hope disarming at * least prevents a panic. */ - pr_emerg("kmmio: recursive probe hit on CPU %d, " - "for address 0x%08lx. Ignoring.\n", - smp_processor_id(), addr); - pr_emerg("kmmio: previous hit was at 0x%08lx.\n", - ctx->addr); + pr_emerg("recursive probe hit on CPU %d, for address 0x%08lx. Ignoring.\n", + smp_processor_id(), addr); + pr_emerg("previous hit was at 0x%08lx.\n", ctx->addr); disarm_kmmio_fault_page(faultpage); } goto no_kmmio_ctx; @@ -301,7 +304,7 @@ no_kmmio: /* * Interrupts are disabled on entry as trap1 is an interrupt gate - * and they remain disabled thorough out this function. + * and they remain disabled throughout this function. * This must always get called as the pair to kmmio_handler(). */ static int post_kmmio_handler(unsigned long condition, struct pt_regs *regs) @@ -310,8 +313,13 @@ static int post_kmmio_handler(unsigned long condition, struct pt_regs *regs) struct kmmio_context *ctx = &get_cpu_var(kmmio_ctx); if (!ctx->active) { - pr_warning("kmmio: spurious debug trap on CPU %d.\n", - smp_processor_id()); + /* + * debug traps without an active context are due to either + * something external causing them (f.e. using a debugger while + * mmio tracing enabled), or erroneous behaviour + */ + pr_warning("unexpected debug trap on CPU %d.\n", + smp_processor_id()); goto out; } @@ -391,8 +399,11 @@ static void release_kmmio_fault_page(unsigned long page, BUG_ON(f->count < 0); if (!f->count) { disarm_kmmio_fault_page(f); - f->release_next = *release_list; - *release_list = f; + if (!f->scheduled_for_release) { + f->release_next = *release_list; + *release_list = f; + f->scheduled_for_release = true; + } } } @@ -419,7 +430,7 @@ int register_kmmio_probe(struct kmmio_probe *p) list_add_rcu(&p->list, &kmmio_probes); while (size < size_lim) { if (add_kmmio_fault_page(p->addr + size)) - pr_err("kmmio: Unable to set page fault.\n"); + pr_err("Unable to set page fault.\n"); size += PAGE_SIZE; } out: @@ -439,12 +450,12 @@ static void rcu_free_kmmio_fault_pages(struct rcu_head *head) head, struct kmmio_delayed_release, rcu); - struct kmmio_fault_page *p = dr->release_list; - while (p) { - struct kmmio_fault_page *next = p->release_next; - BUG_ON(p->count); - kfree(p); - p = next; + struct kmmio_fault_page *f = dr->release_list; + while (f) { + struct kmmio_fault_page *next = f->release_next; + BUG_ON(f->count); + kfree(f); + f = next; } kfree(dr); } @@ -453,19 +464,21 @@ static void remove_kmmio_fault_pages(struct rcu_head *head) { struct kmmio_delayed_release *dr = container_of(head, struct kmmio_delayed_release, rcu); - struct kmmio_fault_page *p = dr->release_list; + struct kmmio_fault_page *f = dr->release_list; struct kmmio_fault_page **prevp = &dr->release_list; unsigned long flags; spin_lock_irqsave(&kmmio_lock, flags); - while (p) { - if (!p->count) { - list_del_rcu(&p->list); - prevp = &p->release_next; + while (f) { + if (!f->count) { + list_del_rcu(&f->list); + prevp = &f->release_next; } else { - *prevp = p->release_next; + *prevp = f->release_next; + f->release_next = NULL; + f->scheduled_for_release = false; } - p = p->release_next; + f = *prevp; } spin_unlock_irqrestore(&kmmio_lock, flags); @@ -484,7 +497,7 @@ static void remove_kmmio_fault_pages(struct rcu_head *head) * 2. remove_kmmio_fault_pages() * Remove the pages from kmmio_page_table. * 3. rcu_free_kmmio_fault_pages() - * Actally free the kmmio_fault_page structs as with RCU. + * Actually free the kmmio_fault_page structs as with RCU. */ void unregister_kmmio_probe(struct kmmio_probe *p) { @@ -503,9 +516,12 @@ void unregister_kmmio_probe(struct kmmio_probe *p) kmmio_count--; spin_unlock_irqrestore(&kmmio_lock, flags); + if (!release_list) + return; + drelease = kmalloc(sizeof(*drelease), GFP_ATOMIC); if (!drelease) { - pr_crit("kmmio: leaking kmmio_fault_page objects.\n"); + pr_crit("leaking kmmio_fault_page objects.\n"); return; } drelease->release_list = release_list; @@ -528,14 +544,21 @@ void unregister_kmmio_probe(struct kmmio_probe *p) } EXPORT_SYMBOL(unregister_kmmio_probe); -static int kmmio_die_notifier(struct notifier_block *nb, unsigned long val, - void *args) +static int +kmmio_die_notifier(struct notifier_block *nb, unsigned long val, void *args) { struct die_args *arg = args; + unsigned long* dr6_p = (unsigned long *)ERR_PTR(arg->err); - if (val == DIE_DEBUG && (arg->err & DR_STEP)) - if (post_kmmio_handler(arg->err, arg->regs) == 1) + if (val == DIE_DEBUG && (*dr6_p & DR_STEP)) + if (post_kmmio_handler(*dr6_p, arg->regs) == 1) { + /* + * Reset the BS bit in dr6 (pointed by args->err) to + * denote completion of processing + */ + *dr6_p &= ~DR_STEP; return NOTIFY_STOP; + } return NOTIFY_DONE; } @@ -544,11 +567,23 @@ static struct notifier_block nb_die = { .notifier_call = kmmio_die_notifier }; -static int __init init_kmmio(void) +int kmmio_init(void) { int i; + for (i = 0; i < KMMIO_PAGE_TABLE_SIZE; i++) INIT_LIST_HEAD(&kmmio_page_table[i]); + return register_die_notifier(&nb_die); } -fs_initcall(init_kmmio); /* should be before device_initcall() */ + +void kmmio_cleanup(void) +{ + int i; + + unregister_die_notifier(&nb_die); + for (i = 0; i < KMMIO_PAGE_TABLE_SIZE; i++) { + WARN_ONCE(!list_empty(&kmmio_page_table[i]), + KERN_ERR "kmmio_page_table not empty at cleanup, any further tracing will leak memory.\n"); + } +} |