diff options
Diffstat (limited to 'arch/x86/mm/fault.c')
| -rw-r--r-- | arch/x86/mm/fault.c | 93 |
1 files changed, 87 insertions, 6 deletions
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c index 35f1498e9832..82bf37a5c9ec 100644 --- a/arch/x86/mm/fault.c +++ b/arch/x86/mm/fault.c @@ -190,6 +190,53 @@ static inline pmd_t *vmalloc_sync_one(pgd_t *pgd, unsigned long address) return pmd_k; } +/* + * Handle a fault on the vmalloc or module mapping area + * + * This is needed because there is a race condition between the time + * when the vmalloc mapping code updates the PMD to the point in time + * where it synchronizes this update with the other page-tables in the + * system. + * + * In this race window another thread/CPU can map an area on the same + * PMD, finds it already present and does not synchronize it with the + * rest of the system yet. As a result v[mz]alloc might return areas + * which are not mapped in every page-table in the system, causing an + * unhandled page-fault when they are accessed. + */ +static noinline int vmalloc_fault(unsigned long address) +{ + unsigned long pgd_paddr; + pmd_t *pmd_k; + pte_t *pte_k; + + /* Make sure we are in vmalloc area: */ + if (!(address >= VMALLOC_START && address < VMALLOC_END)) + return -1; + + /* + * Synchronize this task's top level page-table + * with the 'reference' page table. + * + * Do _not_ use "current" here. We might be inside + * an interrupt in the middle of a task switch.. + */ + pgd_paddr = read_cr3_pa(); + pmd_k = vmalloc_sync_one(__va(pgd_paddr), address); + if (!pmd_k) + return -1; + + if (pmd_large(*pmd_k)) + return 0; + + pte_k = pte_offset_kernel(pmd_k, address); + if (!pte_present(*pte_k)) + return -1; + + return 0; +} +NOKPROBE_SYMBOL(vmalloc_fault); + void arch_sync_kernel_mappings(unsigned long start, unsigned long end) { unsigned long addr; @@ -1081,7 +1128,7 @@ access_error(unsigned long error_code, struct vm_area_struct *vma) return 0; } -static int fault_in_kernel_space(unsigned long address) +bool fault_in_kernel_space(unsigned long address) { /* * On 64-bit systems, the vsyscall page is at an address above @@ -1110,6 +1157,37 @@ do_kern_addr_fault(struct pt_regs *regs, unsigned long hw_error_code, */ WARN_ON_ONCE(hw_error_code & X86_PF_PK); +#ifdef CONFIG_X86_32 + /* + * We can fault-in kernel-space virtual memory on-demand. The + * 'reference' page table is init_mm.pgd. + * + * NOTE! We MUST NOT take any locks for this case. We may + * be in an interrupt or a critical region, and should + * only copy the information from the master page table, + * nothing more. + * + * Before doing this on-demand faulting, ensure that the + * fault is not any of the following: + * 1. A fault on a PTE with a reserved bit set. + * 2. A fault caused by a user-mode access. (Do not demand- + * fault kernel memory due to user-mode accesses). + * 3. A fault caused by a page-level protection violation. + * (A demand fault would be on a non-present page which + * would have X86_PF_PROT==0). + * + * This is only needed to close a race condition on x86-32 in + * the vmalloc mapping/unmapping code. See the comment above + * vmalloc_fault() for details. On x86-64 the race does not + * exist as the vmalloc mappings don't need to be synchronized + * there. + */ + if (!(hw_error_code & (X86_PF_RSVD | X86_PF_USER | X86_PF_PROT))) { + if (vmalloc_fault(address) >= 0) + return; + } +#endif + /* Was the fault spurious, caused by lazy TLB invalidation? */ if (spurious_kernel_fault(hw_error_code, address)) return; @@ -1368,11 +1446,14 @@ DEFINE_IDTENTRY_RAW_ERRORCODE(exc_page_fault) prefetchw(¤t->mm->mmap_lock); /* - * KVM has two types of events that are, logically, interrupts, but - * are unfortunately delivered using the #PF vector. These events are - * "you just accessed valid memory, but the host doesn't have it right - * now, so I'll put you to sleep if you continue" and "that memory - * you tried to access earlier is available now." + * KVM uses #PF vector to deliver 'page not present' events to guests + * (asynchronous page fault mechanism). The event happens when a + * userspace task is trying to access some valid (from guest's point of + * view) memory which is not currently mapped by the host (e.g. the + * memory is swapped out). Note, the corresponding "page ready" event + * which is injected when the memory becomes available, is delived via + * an interrupt mechanism and not a #PF exception + * (see arch/x86/kernel/kvm.c: sysvec_kvm_asyncpf_interrupt()). * * We are relying on the interrupted context being sane (valid RSP, * relevant locks not held, etc.), which is fine as long as the |