diff options
Diffstat (limited to 'arch/x86/mm')
-rw-r--r-- | arch/x86/mm/init_64.c | 2 | ||||
-rw-r--r-- | arch/x86/mm/iomap_32.c | 11 | ||||
-rw-r--r-- | arch/x86/mm/ioremap.c | 19 | ||||
-rw-r--r-- | arch/x86/mm/kmmio.c | 164 | ||||
-rw-r--r-- | arch/x86/mm/numa_64.c | 2 | ||||
-rw-r--r-- | arch/x86/mm/pageattr.c | 30 | ||||
-rw-r--r-- | arch/x86/mm/pat.c | 85 | ||||
-rw-r--r-- | arch/x86/mm/testmmiotrace.c | 70 |
8 files changed, 241 insertions, 142 deletions
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c index e6d36b490250..b1352250096e 100644 --- a/arch/x86/mm/init_64.c +++ b/arch/x86/mm/init_64.c @@ -714,6 +714,8 @@ unsigned long __init_refok init_memory_mapping(unsigned long start, pos = start_pfn << PAGE_SHIFT; end_pfn = ((pos + (PMD_SIZE - 1)) >> PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT); + if (end_pfn > (end >> PAGE_SHIFT)) + end_pfn = end >> PAGE_SHIFT; if (start_pfn < end_pfn) { nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0); pos = end_pfn << PAGE_SHIFT; diff --git a/arch/x86/mm/iomap_32.c b/arch/x86/mm/iomap_32.c index ca53224fc56c..04102d42ff42 100644 --- a/arch/x86/mm/iomap_32.c +++ b/arch/x86/mm/iomap_32.c @@ -20,6 +20,17 @@ #include <asm/pat.h> #include <linux/module.h> +int is_io_mapping_possible(resource_size_t base, unsigned long size) +{ +#ifndef CONFIG_X86_PAE + /* There is no way to map greater than 1 << 32 address without PAE */ + if (base + size > 0x100000000ULL) + return 0; +#endif + return 1; +} +EXPORT_SYMBOL_GPL(is_io_mapping_possible); + /* Map 'pfn' using fixed map 'type' and protections 'prot' */ void * diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c index af750ab973b6..f45d5e29a72e 100644 --- a/arch/x86/mm/ioremap.c +++ b/arch/x86/mm/ioremap.c @@ -134,25 +134,6 @@ int page_is_ram(unsigned long pagenr) return 0; } -int pagerange_is_ram(unsigned long start, unsigned long end) -{ - int ram_page = 0, not_rampage = 0; - unsigned long page_nr; - - for (page_nr = (start >> PAGE_SHIFT); page_nr < (end >> PAGE_SHIFT); - ++page_nr) { - if (page_is_ram(page_nr)) - ram_page = 1; - else - not_rampage = 1; - - if (ram_page == not_rampage) - return -1; - } - - return ram_page; -} - /* * Fix up the linear direct mapping of the kernel to avoid cache attribute * conflicts. diff --git a/arch/x86/mm/kmmio.c b/arch/x86/mm/kmmio.c index 93d82038af4b..6a518dd08a36 100644 --- a/arch/x86/mm/kmmio.c +++ b/arch/x86/mm/kmmio.c @@ -32,11 +32,14 @@ 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 */ + bool armed; /* * Number of times this page has been registered as a part * of a probe. If zero, page is disarmed and this may be freed. - * Used only by writers (RCU). + * Used only by writers (RCU) and post_kmmio_handler(). + * Protected by kmmio_lock, when linked into kmmio_page_table. */ int count; }; @@ -105,57 +108,85 @@ static struct kmmio_fault_page *get_kmmio_fault_page(unsigned long page) return NULL; } -static void set_page_present(unsigned long addr, bool present, - unsigned int *pglevel) +static void set_pmd_presence(pmd_t *pmd, bool present, bool *old) +{ + pmdval_t v = pmd_val(*pmd); + *old = !!(v & _PAGE_PRESENT); + v &= ~_PAGE_PRESENT; + if (present) + v |= _PAGE_PRESENT; + set_pmd(pmd, __pmd(v)); +} + +static void set_pte_presence(pte_t *pte, bool present, bool *old) +{ + pteval_t v = pte_val(*pte); + *old = !!(v & _PAGE_PRESENT); + v &= ~_PAGE_PRESENT; + if (present) + v |= _PAGE_PRESENT; + set_pte_atomic(pte, __pte(v)); +} + +static int set_page_presence(unsigned long addr, bool present, bool *old) { - pteval_t pteval; - pmdval_t pmdval; unsigned int level; - pmd_t *pmd; pte_t *pte = lookup_address(addr, &level); if (!pte) { pr_err("kmmio: no pte for page 0x%08lx\n", addr); - return; + return -1; } - if (pglevel) - *pglevel = level; - switch (level) { case PG_LEVEL_2M: - pmd = (pmd_t *)pte; - pmdval = pmd_val(*pmd) & ~_PAGE_PRESENT; - if (present) - pmdval |= _PAGE_PRESENT; - set_pmd(pmd, __pmd(pmdval)); + set_pmd_presence((pmd_t *)pte, present, old); break; - case PG_LEVEL_4K: - pteval = pte_val(*pte) & ~_PAGE_PRESENT; - if (present) - pteval |= _PAGE_PRESENT; - set_pte_atomic(pte, __pte(pteval)); + set_pte_presence(pte, present, old); break; - default: pr_err("kmmio: unexpected page level 0x%x.\n", level); - return; + return -1; } __flush_tlb_one(addr); + return 0; } -/** Mark the given page as not present. Access to it will trigger a fault. */ -static void arm_kmmio_fault_page(unsigned long page, unsigned int *pglevel) +/* + * Mark the given page as not present. Access to it will trigger a fault. + * + * Struct kmmio_fault_page is protected by RCU and kmmio_lock, but the + * protection is ignored here. RCU read lock is assumed held, so the struct + * will not disappear unexpectedly. Furthermore, the caller must guarantee, + * that double arming the same virtual address (page) cannot occur. + * + * Double disarming on the other hand is allowed, and may occur when a fault + * and mmiotrace shutdown happen simultaneously. + */ +static int arm_kmmio_fault_page(struct kmmio_fault_page *f) { - set_page_present(page & PAGE_MASK, false, pglevel); + int ret; + WARN_ONCE(f->armed, KERN_ERR "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); + } + ret = set_page_presence(f->page, false, &f->old_presence); + WARN_ONCE(ret < 0, KERN_ERR "kmmio arming 0x%08lx failed.\n", f->page); + f->armed = true; + return ret; } -/** Mark the given page as present. */ -static void disarm_kmmio_fault_page(unsigned long page, unsigned int *pglevel) +/** Restore the given page to saved presence state. */ +static void disarm_kmmio_fault_page(struct kmmio_fault_page *f) { - set_page_present(page & PAGE_MASK, true, pglevel); + bool tmp; + int ret = set_page_presence(f->page, f->old_presence, &tmp); + WARN_ONCE(ret < 0, + KERN_ERR "kmmio disarming 0x%08lx failed.\n", f->page); + f->armed = false; } /* @@ -202,28 +233,32 @@ int kmmio_handler(struct pt_regs *regs, unsigned long addr) ctx = &get_cpu_var(kmmio_ctx); if (ctx->active) { - disarm_kmmio_fault_page(faultpage->page, NULL); if (addr == ctx->addr) { /* - * On SMP we sometimes get recursive probe hits on the - * same address. Context is already saved, fall out. + * A second fault on the same page means some other + * condition needs handling by do_page_fault(), the + * page really not being present is the most common. */ - pr_debug("kmmio: duplicate probe hit on CPU %d, for " - "address 0x%08lx.\n", - smp_processor_id(), addr); - ret = 1; - goto no_kmmio_ctx; - } - /* - * 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, " + pr_debug("kmmio: 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()); + } 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("kmmio: previous hit was at 0x%08lx.\n", + ctx->addr); + disarm_kmmio_fault_page(faultpage); + } goto no_kmmio_ctx; } ctx->active++; @@ -244,7 +279,7 @@ int kmmio_handler(struct pt_regs *regs, unsigned long addr) regs->flags &= ~X86_EFLAGS_IF; /* Now we set present bit in PTE and single step. */ - disarm_kmmio_fault_page(ctx->fpage->page, NULL); + disarm_kmmio_fault_page(ctx->fpage); /* * If another cpu accesses the same page while we are stepping, @@ -275,7 +310,7 @@ 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_debug("kmmio: spurious debug trap on CPU %d.\n", + pr_warning("kmmio: spurious debug trap on CPU %d.\n", smp_processor_id()); goto out; } @@ -283,7 +318,11 @@ static int post_kmmio_handler(unsigned long condition, struct pt_regs *regs) if (ctx->probe && ctx->probe->post_handler) ctx->probe->post_handler(ctx->probe, condition, regs); - arm_kmmio_fault_page(ctx->fpage->page, NULL); + /* Prevent racing against release_kmmio_fault_page(). */ + spin_lock(&kmmio_lock); + if (ctx->fpage->count) + arm_kmmio_fault_page(ctx->fpage); + spin_unlock(&kmmio_lock); regs->flags &= ~X86_EFLAGS_TF; regs->flags |= ctx->saved_flags; @@ -315,20 +354,24 @@ static int add_kmmio_fault_page(unsigned long page) f = get_kmmio_fault_page(page); if (f) { if (!f->count) - arm_kmmio_fault_page(f->page, NULL); + arm_kmmio_fault_page(f); f->count++; return 0; } - f = kmalloc(sizeof(*f), GFP_ATOMIC); + f = kzalloc(sizeof(*f), GFP_ATOMIC); if (!f) return -1; f->count = 1; f->page = page; - list_add_rcu(&f->list, kmmio_page_list(f->page)); - arm_kmmio_fault_page(f->page, NULL); + if (arm_kmmio_fault_page(f)) { + kfree(f); + return -1; + } + + list_add_rcu(&f->list, kmmio_page_list(f->page)); return 0; } @@ -347,7 +390,7 @@ static void release_kmmio_fault_page(unsigned long page, f->count--; BUG_ON(f->count < 0); if (!f->count) { - disarm_kmmio_fault_page(f->page, NULL); + disarm_kmmio_fault_page(f); f->release_next = *release_list; *release_list = f; } @@ -408,23 +451,24 @@ static void rcu_free_kmmio_fault_pages(struct rcu_head *head) 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_delayed_release *dr = + container_of(head, struct kmmio_delayed_release, rcu); struct kmmio_fault_page *p = 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) + if (!p->count) { list_del_rcu(&p->list); - else + prevp = &p->release_next; + } else { *prevp = p->release_next; - prevp = &p->release_next; + } p = p->release_next; } spin_unlock_irqrestore(&kmmio_lock, flags); + /* This is the real RCU destroy call. */ call_rcu(&dr->rcu, rcu_free_kmmio_fault_pages); } diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c index 71a14f89f89e..f3516da035d1 100644 --- a/arch/x86/mm/numa_64.c +++ b/arch/x86/mm/numa_64.c @@ -145,7 +145,7 @@ int __init compute_hash_shift(struct bootnode *nodes, int numnodes, return shift; } -int early_pfn_to_nid(unsigned long pfn) +int __meminit __early_pfn_to_nid(unsigned long pfn) { return phys_to_nid(pfn << PAGE_SHIFT); } diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c index 84ba74820ad6..7be47d1a97e4 100644 --- a/arch/x86/mm/pageattr.c +++ b/arch/x86/mm/pageattr.c @@ -508,18 +508,13 @@ static int split_large_page(pte_t *kpte, unsigned long address) #endif /* - * Install the new, split up pagetable. Important details here: + * Install the new, split up pagetable. * - * On Intel the NX bit of all levels must be cleared to make a - * page executable. See section 4.13.2 of Intel 64 and IA-32 - * Architectures Software Developer's Manual). - * - * Mark the entry present. The current mapping might be - * set to not present, which we preserved above. + * We use the standard kernel pagetable protections for the new + * pagetable protections, the actual ptes set above control the + * primary protection behavior: */ - ref_prot = pte_pgprot(pte_mkexec(pte_clrhuge(*kpte))); - pgprot_val(ref_prot) |= _PAGE_PRESENT; - __set_pmd_pte(kpte, address, mk_pte(base, ref_prot)); + __set_pmd_pte(kpte, address, mk_pte(base, __pgprot(_KERNPG_TABLE))); base = NULL; out_unlock: @@ -575,7 +570,6 @@ static int __change_page_attr(struct cpa_data *cpa, int primary) address = cpa->vaddr[cpa->curpage]; else address = *cpa->vaddr; - repeat: kpte = lookup_address(address, &level); if (!kpte) @@ -812,6 +806,13 @@ static int change_page_attr_set_clr(unsigned long *addr, int numpages, vm_unmap_aliases(); + /* + * If we're called with lazy mmu updates enabled, the + * in-memory pte state may be stale. Flush pending updates to + * bring them up to date. + */ + arch_flush_lazy_mmu_mode(); + cpa.vaddr = addr; cpa.numpages = numpages; cpa.mask_set = mask_set; @@ -854,6 +855,13 @@ static int change_page_attr_set_clr(unsigned long *addr, int numpages, } else cpa_flush_all(cache); + /* + * If we've been called with lazy mmu updates enabled, then + * make sure that everything gets flushed out before we + * return. + */ + arch_flush_lazy_mmu_mode(); + out: return ret; } diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c index 7b61036427df..e0ab173b6974 100644 --- a/arch/x86/mm/pat.c +++ b/arch/x86/mm/pat.c @@ -11,6 +11,7 @@ #include <linux/bootmem.h> #include <linux/debugfs.h> #include <linux/kernel.h> +#include <linux/module.h> #include <linux/gfp.h> #include <linux/mm.h> #include <linux/fs.h> @@ -211,6 +212,33 @@ chk_conflict(struct memtype *new, struct memtype *entry, unsigned long *type) static struct memtype *cached_entry; static u64 cached_start; +static int pat_pagerange_is_ram(unsigned long start, unsigned long end) +{ + int ram_page = 0, not_rampage = 0; + unsigned long page_nr; + + for (page_nr = (start >> PAGE_SHIFT); page_nr < (end >> PAGE_SHIFT); + ++page_nr) { + /* + * For legacy reasons, physical address range in the legacy ISA + * region is tracked as non-RAM. This will allow users of + * /dev/mem to map portions of legacy ISA region, even when + * some of those portions are listed(or not even listed) with + * different e820 types(RAM/reserved/..) + */ + if (page_nr >= (ISA_END_ADDRESS >> PAGE_SHIFT) && + page_is_ram(page_nr)) + ram_page = 1; + else + not_rampage = 1; + + if (ram_page == not_rampage) + return -1; + } + + return ram_page; +} + /* * For RAM pages, mark the pages as non WB memory type using * PageNonWB (PG_arch_1). We allow only one set_memory_uc() or @@ -336,20 +364,12 @@ int reserve_memtype(u64 start, u64 end, unsigned long req_type, if (new_type) *new_type = actual_type; - /* - * For legacy reasons, some parts of the physical address range in the - * legacy 1MB region is treated as non-RAM (even when listed as RAM in - * the e820 tables). So we will track the memory attributes of this - * legacy 1MB region using the linear memtype_list always. - */ - if (end >= ISA_END_ADDRESS) { - is_range_ram = pagerange_is_ram(start, end); - if (is_range_ram == 1) - return reserve_ram_pages_type(start, end, req_type, - new_type); - else if (is_range_ram < 0) - return -EINVAL; - } + is_range_ram = pat_pagerange_is_ram(start, end); + if (is_range_ram == 1) + return reserve_ram_pages_type(start, end, req_type, + new_type); + else if (is_range_ram < 0) + return -EINVAL; new = kmalloc(sizeof(struct memtype), GFP_KERNEL); if (!new) @@ -446,19 +466,11 @@ int free_memtype(u64 start, u64 end) if (is_ISA_range(start, end - 1)) return 0; - /* - * For legacy reasons, some parts of the physical address range in the - * legacy 1MB region is treated as non-RAM (even when listed as RAM in - * the e820 tables). So we will track the memory attributes of this - * legacy 1MB region using the linear memtype_list always. - */ - if (end >= ISA_END_ADDRESS) { - is_range_ram = pagerange_is_ram(start, end); - if (is_range_ram == 1) - return free_ram_pages_type(start, end); - else if (is_range_ram < 0) - return -EINVAL; - } + is_range_ram = pat_pagerange_is_ram(start, end); + if (is_range_ram == 1) + return free_ram_pages_type(start, end); + else if (is_range_ram < 0) + return -EINVAL; spin_lock(&memtype_lock); list_for_each_entry(entry, &memtype_list, nd) { @@ -626,17 +638,13 @@ static int reserve_pfn_range(u64 paddr, unsigned long size, pgprot_t *vma_prot, unsigned long flags; unsigned long want_flags = (pgprot_val(*vma_prot) & _PAGE_CACHE_MASK); - is_ram = pagerange_is_ram(paddr, paddr + size); + is_ram = pat_pagerange_is_ram(paddr, paddr + size); - if (is_ram != 0) { - /* - * For mapping RAM pages, drivers need to call - * set_memory_[uc|wc|wb] directly, for reserve and free, before - * setting up the PTE. - */ - WARN_ON_ONCE(1); - return 0; - } + /* + * reserve_pfn_range() doesn't support RAM pages. + */ + if (is_ram != 0) + return -EINVAL; ret = reserve_memtype(paddr, paddr + size, want_flags, &flags); if (ret) @@ -693,7 +701,7 @@ static void free_pfn_range(u64 paddr, unsigned long size) { int is_ram; - is_ram = pagerange_is_ram(paddr, paddr + size); + is_ram = pat_pagerange_is_ram(paddr, paddr + size); if (is_ram == 0) free_memtype(paddr, paddr + size); } @@ -861,6 +869,7 @@ pgprot_t pgprot_writecombine(pgprot_t prot) else return pgprot_noncached(prot); } +EXPORT_SYMBOL_GPL(pgprot_writecombine); #if defined(CONFIG_DEBUG_FS) && defined(CONFIG_X86_PAT) diff --git a/arch/x86/mm/testmmiotrace.c b/arch/x86/mm/testmmiotrace.c index ab50a8d7402c..427fd1b56df5 100644 --- a/arch/x86/mm/testmmiotrace.c +++ b/arch/x86/mm/testmmiotrace.c @@ -1,5 +1,5 @@ /* - * Written by Pekka Paalanen, 2008 <pq@iki.fi> + * Written by Pekka Paalanen, 2008-2009 <pq@iki.fi> */ #include <linux/module.h> #include <linux/io.h> @@ -9,35 +9,74 @@ static unsigned long mmio_address; module_param(mmio_address, ulong, 0); -MODULE_PARM_DESC(mmio_address, "Start address of the mapping of 16 kB."); +MODULE_PARM_DESC(mmio_address, " Start address of the mapping of 16 kB " + "(or 8 MB if read_far is non-zero)."); + +static unsigned long read_far = 0x400100; +module_param(read_far, ulong, 0); +MODULE_PARM_DESC(read_far, " Offset of a 32-bit read within 8 MB " + "(default: 0x400100)."); + +static unsigned v16(unsigned i) +{ + return i * 12 + 7; +} + +static unsigned v32(unsigned i) +{ + return i * 212371 + 13; +} static void do_write_test(void __iomem *p) { unsigned int i; + pr_info(MODULE_NAME ": write test.\n"); mmiotrace_printk("Write test.\n"); + for (i = 0; i < 256; i++) iowrite8(i, p + i); + for (i = 1024; i < (5 * 1024); i += 2) - iowrite16(i * 12 + 7, p + i); + iowrite16(v16(i), p + i); + for (i = (5 * 1024); i < (16 * 1024); i += 4) - iowrite32(i * 212371 + 13, p + i); + iowrite32(v32(i), p + i); } static void do_read_test(void __iomem *p) { unsigned int i; + unsigned errs[3] = { 0 }; + pr_info(MODULE_NAME ": read test.\n"); mmiotrace_printk("Read test.\n"); + for (i = 0; i < 256; i++) - ioread8(p + i); + if (ioread8(p + i) != i) + ++errs[0]; + for (i = 1024; i < (5 * 1024); i += 2) - ioread16(p + i); + if (ioread16(p + i) != v16(i)) + ++errs[1]; + for (i = (5 * 1024); i < (16 * 1024); i += 4) - ioread32(p + i); + if (ioread32(p + i) != v32(i)) + ++errs[2]; + + mmiotrace_printk("Read errors: 8-bit %d, 16-bit %d, 32-bit %d.\n", + errs[0], errs[1], errs[2]); } -static void do_test(void) +static void do_read_far_test(void __iomem *p) { - void __iomem *p = ioremap_nocache(mmio_address, 0x4000); + pr_info(MODULE_NAME ": read far test.\n"); + mmiotrace_printk("Read far test.\n"); + + ioread32(p + read_far); +} + +static void do_test(unsigned long size) +{ + void __iomem *p = ioremap_nocache(mmio_address, size); if (!p) { pr_err(MODULE_NAME ": could not ioremap, aborting.\n"); return; @@ -45,11 +84,15 @@ static void do_test(void) mmiotrace_printk("ioremap returned %p.\n", p); do_write_test(p); do_read_test(p); + if (read_far && read_far < size - 4) + do_read_far_test(p); iounmap(p); } static int __init init(void) { + unsigned long size = (read_far) ? (8 << 20) : (16 << 10); + if (mmio_address == 0) { pr_err(MODULE_NAME ": you have to use the module argument " "mmio_address.\n"); @@ -58,10 +101,11 @@ static int __init init(void) return -ENXIO; } - pr_warning(MODULE_NAME ": WARNING: mapping 16 kB @ 0x%08lx " - "in PCI address space, and writing " - "rubbish in there.\n", mmio_address); - do_test(); + pr_warning(MODULE_NAME ": WARNING: mapping %lu kB @ 0x%08lx in PCI " + "address space, and writing 16 kB of rubbish in there.\n", + size >> 10, mmio_address); + do_test(size); + pr_info(MODULE_NAME ": All done.\n"); return 0; } |