X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=arch%2Fi386%2Fkernel%2Fvm86.c;h=d5b775c00647fed42f62464e9542bcdbbb3c68e1;hb=refs%2Fheads%2Fvserver;hp=4484e47a77c57506960d4ec3cca23abeed48b669;hpb=9bf4aaab3e101692164d49b7ca357651eb691cb6;p=linux-2.6.git diff --git a/arch/i386/kernel/vm86.c b/arch/i386/kernel/vm86.c index 4484e47a7..d5b775c00 100644 --- a/arch/i386/kernel/vm86.c +++ b/arch/i386/kernel/vm86.c @@ -4,7 +4,7 @@ * Copyright (C) 1994 Linus Torvalds * * 29 dec 2001 - Fixed oopses caused by unchecked access to the vm86 - * stack - Manfred Spraul + * stack - Manfred Spraul * * 22 mar 2002 - Manfred detected the stackfaults, but didn't handle * them correctly. Now the emulation will be in a @@ -30,7 +30,7 @@ * */ -#include +#include #include #include #include @@ -42,6 +42,8 @@ #include #include #include +#include +#include #include #include @@ -71,10 +73,10 @@ /* * 8- and 16-bit register defines.. */ -#define AL(regs) (((unsigned char *)&((regs)->eax))[0]) -#define AH(regs) (((unsigned char *)&((regs)->eax))[1]) -#define IP(regs) (*(unsigned short *)&((regs)->eip)) -#define SP(regs) (*(unsigned short *)&((regs)->esp)) +#define AL(regs) (((unsigned char *)&((regs)->pt.eax))[0]) +#define AH(regs) (((unsigned char *)&((regs)->pt.eax))[1]) +#define IP(regs) (*(unsigned short *)&((regs)->pt.eip)) +#define SP(regs) (*(unsigned short *)&((regs)->pt.esp)) /* * virtual flags (16 and 32-bit versions) @@ -88,15 +90,44 @@ #define SAFE_MASK (0xDD5) #define RETURN_MASK (0xDFF) -#define VM86_REGS_PART2 orig_eax -#define VM86_REGS_SIZE1 \ - ( (unsigned)( & (((struct kernel_vm86_regs *)0)->VM86_REGS_PART2) ) ) -#define VM86_REGS_SIZE2 (sizeof(struct kernel_vm86_regs) - VM86_REGS_SIZE1) +/* convert kernel_vm86_regs to vm86_regs */ +static int copy_vm86_regs_to_user(struct vm86_regs __user *user, + const struct kernel_vm86_regs *regs) +{ + int ret = 0; + + /* kernel_vm86_regs is missing xfs, so copy everything up to + (but not including) xgs, and then rest after xgs. */ + ret += copy_to_user(user, regs, offsetof(struct kernel_vm86_regs, pt.xgs)); + ret += copy_to_user(&user->__null_gs, ®s->pt.xgs, + sizeof(struct kernel_vm86_regs) - + offsetof(struct kernel_vm86_regs, pt.xgs)); + + return ret; +} + +/* convert vm86_regs to kernel_vm86_regs */ +static int copy_vm86_regs_from_user(struct kernel_vm86_regs *regs, + const struct vm86_regs __user *user, + unsigned extra) +{ + int ret = 0; + + ret += copy_from_user(regs, user, offsetof(struct kernel_vm86_regs, pt.xgs)); + ret += copy_from_user(®s->pt.xgs, &user->__null_gs, + sizeof(struct kernel_vm86_regs) - + offsetof(struct kernel_vm86_regs, pt.xgs) + + extra); + + return ret; +} struct pt_regs * FASTCALL(save_v86_state(struct kernel_vm86_regs * regs)); struct pt_regs * fastcall save_v86_state(struct kernel_vm86_regs * regs) { +#ifndef CONFIG_X86_NO_TSS struct tss_struct *tss; +#endif struct pt_regs *ret; unsigned long tmp; @@ -111,64 +142,59 @@ struct pt_regs * fastcall save_v86_state(struct kernel_vm86_regs * regs) printk("no vm86_info: BAD\n"); do_exit(SIGSEGV); } - set_flags(regs->eflags, VEFLAGS, VIF_MASK | current->thread.v86mask); - tmp = copy_to_user(¤t->thread.vm86_info->regs,regs, VM86_REGS_SIZE1); - tmp += copy_to_user(¤t->thread.vm86_info->regs.VM86_REGS_PART2, - ®s->VM86_REGS_PART2, VM86_REGS_SIZE2); + set_flags(regs->pt.eflags, VEFLAGS, VIF_MASK | current->thread.v86mask); + tmp = copy_vm86_regs_to_user(¤t->thread.vm86_info->regs,regs); tmp += put_user(current->thread.screen_bitmap,¤t->thread.vm86_info->screen_bitmap); if (tmp) { printk("vm86: could not access userspace vm86_info\n"); do_exit(SIGSEGV); } - tss = init_tss + get_cpu(); +#ifndef CONFIG_X86_NO_TSS + tss = &per_cpu(init_tss, get_cpu()); +#endif current->thread.esp0 = current->thread.saved_esp0; current->thread.sysenter_cs = __KERNEL_CS; load_esp0(tss, ¤t->thread); current->thread.saved_esp0 = 0; +#ifndef CONFIG_X86_NO_TSS put_cpu(); +#endif - loadsegment(fs, current->thread.saved_fs); - loadsegment(gs, current->thread.saved_gs); ret = KVM86->regs32; + + loadsegment(fs, current->thread.saved_fs); + ret->xgs = current->thread.saved_gs; + return ret; } -static void mark_screen_rdonly(struct task_struct * tsk) +static void mark_screen_rdonly(struct mm_struct *mm) { pgd_t *pgd; + pud_t *pud; pmd_t *pmd; - pte_t *pte, *mapped; + pte_t *pte; + spinlock_t *ptl; int i; - preempt_disable(); - spin_lock(&tsk->mm->page_table_lock); - pgd = pgd_offset(tsk->mm, 0xA0000); - if (pgd_none(*pgd)) + pgd = pgd_offset(mm, 0xA0000); + if (pgd_none_or_clear_bad(pgd)) goto out; - if (pgd_bad(*pgd)) { - pgd_ERROR(*pgd); - pgd_clear(pgd); + pud = pud_offset(pgd, 0xA0000); + if (pud_none_or_clear_bad(pud)) goto out; - } - pmd = pmd_offset(pgd, 0xA0000); - if (pmd_none(*pmd)) - goto out; - if (pmd_bad(*pmd)) { - pmd_ERROR(*pmd); - pmd_clear(pmd); + pmd = pmd_offset(pud, 0xA0000); + if (pmd_none_or_clear_bad(pmd)) goto out; - } - pte = mapped = pte_offset_map(pmd, 0xA0000); + pte = pte_offset_map_lock(mm, pmd, 0xA0000, &ptl); for (i = 0; i < 32; i++) { if (pte_present(*pte)) set_pte(pte, pte_wrprotect(*pte)); pte++; } - pte_unmap(mapped); + pte_unmap_unlock(pte, ptl); out: - spin_unlock(&tsk->mm->page_table_lock); - preempt_enable(); flush_tlb(); } @@ -191,9 +217,9 @@ asmlinkage int sys_vm86old(struct pt_regs regs) tsk = current; if (tsk->thread.saved_esp0) goto out; - tmp = copy_from_user(&info, v86, VM86_REGS_SIZE1); - tmp += copy_from_user(&info.regs.VM86_REGS_PART2, &v86->regs.VM86_REGS_PART2, - (long)&info.vm86plus - (long)&info.regs.VM86_REGS_PART2); + tmp = copy_vm86_regs_from_user(&info.regs, &v86->regs, + offsetof(struct kernel_vm86_struct, vm86plus) - + sizeof(info.regs)); ret = -EFAULT; if (tmp) goto out; @@ -228,7 +254,7 @@ asmlinkage int sys_vm86(struct pt_regs regs) goto out; case VM86_PLUS_INSTALL_CHECK: /* NOTE: on old vm86 stuff this will return the error - from verify_area(), because the subfunction is + from access_ok(), because the subfunction is interpreted as (invalid) address to vm86_struct. So the installation check works. */ @@ -241,9 +267,9 @@ asmlinkage int sys_vm86(struct pt_regs regs) if (tsk->thread.saved_esp0) goto out; v86 = (struct vm86plus_struct __user *)regs.ecx; - tmp = copy_from_user(&info, v86, VM86_REGS_SIZE1); - tmp += copy_from_user(&info.regs.VM86_REGS_PART2, &v86->regs.VM86_REGS_PART2, - (long)&info.regs32 - (long)&info.regs.VM86_REGS_PART2); + tmp = copy_vm86_regs_from_user(&info.regs, &v86->regs, + offsetof(struct kernel_vm86_struct, regs32) - + sizeof(info.regs)); ret = -EFAULT; if (tmp) goto out; @@ -259,15 +285,18 @@ out: static void do_sys_vm86(struct kernel_vm86_struct *info, struct task_struct *tsk) { +#ifndef CONFIG_X86_NO_TSS struct tss_struct *tss; +#endif /* * make sure the vm86() system call doesn't try to do anything silly */ - info->regs.__null_ds = 0; - info->regs.__null_es = 0; + info->regs.pt.xds = 0; + info->regs.pt.xes = 0; + info->regs.pt.xgs = 0; -/* we are clearing fs,gs later just before "jmp resume_userspace", - * because starting with Linux 2.1.x they aren't no longer saved/restored +/* we are clearing fs later just before "jmp resume_userspace", + * because it is not saved/restored. */ /* @@ -275,10 +304,10 @@ static void do_sys_vm86(struct kernel_vm86_struct *info, struct task_struct *tsk * has set it up safely, so this makes sure interrupt etc flags are * inherited from protected mode. */ - VEFLAGS = info->regs.eflags; - info->regs.eflags &= SAFE_MASK; - info->regs.eflags |= info->regs32->eflags & ~SAFE_MASK; - info->regs.eflags |= VM_MASK; + VEFLAGS = info->regs.pt.eflags; + info->regs.pt.eflags &= SAFE_MASK; + info->regs.pt.eflags |= info->regs32->eflags & ~SAFE_MASK; + info->regs.pt.eflags |= VM_MASK; switch (info->cpu_type) { case CPU_286: @@ -300,26 +329,35 @@ static void do_sys_vm86(struct kernel_vm86_struct *info, struct task_struct *tsk */ info->regs32->eax = 0; tsk->thread.saved_esp0 = tsk->thread.esp0; - asm volatile("movl %%fs,%0":"=m" (tsk->thread.saved_fs)); - asm volatile("movl %%gs,%0":"=m" (tsk->thread.saved_gs)); + savesegment(fs, tsk->thread.saved_fs); + tsk->thread.saved_gs = info->regs32->xgs; - tss = init_tss + get_cpu(); +#ifndef CONFIG_X86_NO_TSS + tss = &per_cpu(init_tss, get_cpu()); +#endif tsk->thread.esp0 = (unsigned long) &info->VM86_TSS_ESP0; if (cpu_has_sep) tsk->thread.sysenter_cs = 0; load_esp0(tss, &tsk->thread); +#ifndef CONFIG_X86_NO_TSS put_cpu(); +#endif tsk->thread.screen_bitmap = info->screen_bitmap; if (info->flags & VM86_SCREEN_BITMAP) - mark_screen_rdonly(tsk); + mark_screen_rdonly(tsk->mm); + + /*call audit_syscall_exit since we do not exit via the normal paths */ + if (unlikely(current->audit_context)) + audit_syscall_exit(AUDITSC_RESULT(0), 0); + __asm__ __volatile__( - "xorl %%eax,%%eax; movl %%eax,%%fs; movl %%eax,%%gs\n\t" "movl %0,%%esp\n\t" "movl %1,%%ebp\n\t" + "mov %2, %%fs\n\t" "jmp resume_userspace" : /* no outputs */ - :"r" (&info->regs), "r" (tsk->thread_info) : "ax"); + :"r" (&info->regs), "r" (task_thread_info(tsk)), "r" (0)); /* we never return here */ } @@ -349,12 +387,12 @@ static inline void clear_IF(struct kernel_vm86_regs * regs) static inline void clear_TF(struct kernel_vm86_regs * regs) { - regs->eflags &= ~TF_MASK; + regs->pt.eflags &= ~TF_MASK; } static inline void clear_AC(struct kernel_vm86_regs * regs) { - regs->eflags &= ~AC_MASK; + regs->pt.eflags &= ~AC_MASK; } /* It is correct to call set_IF(regs) from the set_vflags_* @@ -371,7 +409,7 @@ static inline void clear_AC(struct kernel_vm86_regs * regs) static inline void set_vflags_long(unsigned long eflags, struct kernel_vm86_regs * regs) { set_flags(VEFLAGS, eflags, current->thread.v86mask); - set_flags(regs->eflags, eflags, SAFE_MASK); + set_flags(regs->pt.eflags, eflags, SAFE_MASK); if (eflags & IF_MASK) set_IF(regs); else @@ -381,7 +419,7 @@ static inline void set_vflags_long(unsigned long eflags, struct kernel_vm86_regs static inline void set_vflags_short(unsigned short flags, struct kernel_vm86_regs * regs) { set_flags(VFLAGS, flags, current->thread.v86mask); - set_flags(regs->eflags, flags, SAFE_MASK); + set_flags(regs->pt.eflags, flags, SAFE_MASK); if (flags & IF_MASK) set_IF(regs); else @@ -390,7 +428,7 @@ static inline void set_vflags_short(unsigned short flags, struct kernel_vm86_reg static inline unsigned long get_vflags(struct kernel_vm86_regs * regs) { - unsigned long flags = regs->eflags & RETURN_MASK; + unsigned long flags = regs->pt.eflags & RETURN_MASK; if (VEFLAGS & VIF_MASK) flags |= IF_MASK; @@ -494,7 +532,7 @@ static void do_int(struct kernel_vm86_regs *regs, int i, unsigned long __user *intr_ptr; unsigned long segoffs; - if (regs->cs == BIOSSEG) + if (regs->pt.xcs == BIOSSEG) goto cannot_handle; if (is_revectored(i, &KVM86->int_revectored)) goto cannot_handle; @@ -506,9 +544,9 @@ static void do_int(struct kernel_vm86_regs *regs, int i, if ((segoffs >> 16) == BIOSSEG) goto cannot_handle; pushw(ssp, sp, get_vflags(regs), cannot_handle); - pushw(ssp, sp, regs->cs, cannot_handle); + pushw(ssp, sp, regs->pt.xcs, cannot_handle); pushw(ssp, sp, IP(regs), cannot_handle); - regs->cs = segoffs >> 16; + regs->pt.xcs = segoffs >> 16; SP(regs) -= 6; IP(regs) = segoffs & 0xffff; clear_TF(regs); @@ -525,18 +563,11 @@ int handle_vm86_trap(struct kernel_vm86_regs * regs, long error_code, int trapno if (VMPI.is_vm86pus) { if ( (trapno==3) || (trapno==1) ) return_to_32bit(regs, VM86_TRAP + (trapno << 8)); - do_int(regs, trapno, (unsigned char __user *) (regs->ss << 4), SP(regs)); + do_int(regs, trapno, (unsigned char __user *) (regs->pt.xss << 4), SP(regs)); return 0; } if (trapno !=1) return 1; /* we let this handle by the calling routine */ - if (current->ptrace & PT_PTRACED) { - unsigned long flags; - spin_lock_irqsave(¤t->sighand->siglock, flags); - sigdelset(¤t->blocked, SIGTRAP); - recalc_sigpending(); - spin_unlock_irqrestore(¤t->sighand->siglock, flags); - } send_sig(SIGTRAP, current, 1); current->thread.trap_no = trapno; current->thread.error_code = error_code; @@ -548,7 +579,7 @@ void handle_vm86_fault(struct kernel_vm86_regs * regs, long error_code) unsigned char opcode; unsigned char __user *csp; unsigned char __user *ssp; - unsigned short ip, sp; + unsigned short ip, sp, orig_flags; int data32, pref_done; #define CHECK_IF_IN_TRAP \ @@ -557,10 +588,14 @@ void handle_vm86_fault(struct kernel_vm86_regs * regs, long error_code) #define VM86_FAULT_RETURN do { \ if (VMPI.force_return_for_pic && (VEFLAGS & (IF_MASK | VIF_MASK))) \ return_to_32bit(regs, VM86_PICRETURN); \ + if (orig_flags & TF_MASK) \ + handle_vm86_trap(regs, 0, 1); \ return; } while (0) - csp = (unsigned char __user *) (regs->cs << 4); - ssp = (unsigned char __user *) (regs->ss << 4); + orig_flags = *(unsigned short *)®s->pt.eflags; + + csp = (unsigned char __user *) (regs->pt.xcs << 4); + ssp = (unsigned char __user *) (regs->pt.xss << 4); sp = SP(regs); ip = IP(regs); @@ -647,7 +682,7 @@ void handle_vm86_fault(struct kernel_vm86_regs * regs, long error_code) SP(regs) += 6; } IP(regs) = newip; - regs->cs = newcs; + regs->pt.xcs = newcs; CHECK_IF_IN_TRAP; if (data32) { set_vflags_long(newflags, regs); @@ -704,14 +739,14 @@ static struct vm86_irqs { int sig; } vm86_irqs[16]; -static spinlock_t irqbits_lock = SPIN_LOCK_UNLOCKED; +static DEFINE_SPINLOCK(irqbits_lock); static int irqbits; #define ALLOWED_SIGS ( 1 /* 0 = don't send a signal */ \ | (1 << SIGUSR1) | (1 << SIGUSR2) | (1 << SIGIO) | (1 << SIGURG) \ | (1 << SIGUNUSED) ) -static irqreturn_t irq_handler(int intno, void *dev_id, struct pt_regs * regs) +static irqreturn_t irq_handler(int intno, void *dev_id) { int irq_bit; unsigned long flags; @@ -723,7 +758,14 @@ static irqreturn_t irq_handler(int intno, void *dev_id, struct pt_regs * regs) irqbits |= irq_bit; if (vm86_irqs[intno].sig) send_sig(vm86_irqs[intno].sig, vm86_irqs[intno].tsk, 1); - /* else user will poll for IRQs */ + /* + * IRQ will be re-enabled when user asks for the irq (whether + * polling or as a result of the signal) + */ + disable_irq_nosync(intno); + spin_unlock_irqrestore(&irqbits_lock, flags); + return IRQ_HANDLED; + out: spin_unlock_irqrestore(&irqbits_lock, flags); return IRQ_NONE; @@ -741,7 +783,7 @@ static inline void free_vm86_irq(int irqnumber) spin_unlock_irqrestore(&irqbits_lock, flags); } -void release_x86_irqs(struct task_struct *task) +void release_vm86_irqs(struct task_struct *task) { int i; for (i = FIRST_VM86_IRQ ; i <= LAST_VM86_IRQ; i++) @@ -753,17 +795,20 @@ static inline int get_and_reset_irq(int irqnumber) { int bit; unsigned long flags; + int ret = 0; if (invalid_vm86_irq(irqnumber)) return 0; if (vm86_irqs[irqnumber].tsk != current) return 0; spin_lock_irqsave(&irqbits_lock, flags); bit = irqbits & (1 << irqnumber); irqbits &= ~bit; + if (bit) { + enable_irq(irqnumber); + ret = 1; + } + spin_unlock_irqrestore(&irqbits_lock, flags); - if (!bit) - return 0; - enable_irq(irqnumber); - return 1; + return ret; }