Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261407AbVCMSVZ (ORCPT ); Sun, 13 Mar 2005 13:21:25 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261414AbVCMSVZ (ORCPT ); Sun, 13 Mar 2005 13:21:25 -0500 Received: from mail.aknet.ru ([217.67.122.194]:19983 "EHLO mail.aknet.ru") by vger.kernel.org with ESMTP id S261407AbVCMSUY (ORCPT ); Sun, 13 Mar 2005 13:20:24 -0500 Message-ID: <42348474.7040808@aknet.ru> Date: Sun, 13 Mar 2005 21:20:36 +0300 From: Stas Sergeev User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.3) Gecko/20041020 X-Accept-Language: ru, en-us, en MIME-Version: 1.0 To: Alan Cox Cc: Linux kernel , Linus Torvalds , Petr Vandrovec , Denis Vlasenko , Pavel Machek Subject: [patch] x86: fix ESP corruption CPU bug Content-Type: multipart/mixed; boundary="------------040102020901030503060706" Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 13645 Lines: 445 This is a multi-part message in MIME format. --------------040102020901030503060706 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Hi Alan. Attached patch works around the corruption of the high word of the ESP register, which is the official bug of x86 CPUs. The bug triggers only when the one is using the 16bit stack segment, and is described here: http://www.intel.com/design/intarch/specupdt/27287402.PDF Patch helps running many apps under dosemu, and, according to the comments found in Wine sources, also helps Wine. Also the patch closes the "information leak", which is that the half of the kernel's %esp value gets leaked to user-space. The initial discussion about the problem can be found here: http://lkml.org/lkml/2004/9/16/254 On a later stages the development of that patch was driven mainly by Linus, but it was in private. This patch adds only 6 instructions for the fast kernel-->user return path, 6 instructions on an exception path and 5 instructions on an NMI path. All the rest of the patch gets executed only for the two apps in that world, namely, dosemu and wine (and VMWare?). That's why it should be relatevely easy to make sure the patch doesn't do any harm for the normal apps, and so it is safe, and probably fits into an -ac tree. How it works: - Allocates the per-cpu data for 16bit stacks and sets the per-cpu GDT entries for them. - On a return to userspace, checks whether the SS is from LDT and is 16 bit. - If it is, the iret frame gets copied to the 16bit per-cpu stack and stack gets switched to the 16bit one, while the higher word of %esp gets preloaded with the proper value. - On an exceptions the check is performed to see if we are on a 16bit stack, and if we are - switch to the 32bit one. Alan, would it be possible to apply that patch to an -ac tree? Acked-by: Linus Torvalds Acked-by: Petr Vandrovec Signed-off-by: Stas Sergeev --------------040102020901030503060706 Content-Type: text/x-patch; name="linux-2.6.11-ac2-stk7.diff" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="linux-2.6.11-ac2-stk7.diff" diff -ur linux-2.6.11-ac2/arch/i386/kernel/cpu/common.c linux-2.6.11-ac2-stk/arch/i386/kernel/cpu/common.c --- linux-2.6.11-ac2/arch/i386/kernel/cpu/common.c 2005-03-13 17:59:45.000000000 +0300 +++ linux-2.6.11-ac2-stk/arch/i386/kernel/cpu/common.c 2005-03-13 18:17:47.000000000 +0300 @@ -21,6 +21,10 @@ DEFINE_PER_CPU(struct desc_struct, cpu_gdt_table[GDT_ENTRIES]); EXPORT_PER_CPU_SYMBOL(cpu_gdt_table); +unsigned char cpu_16bit_stack[CPU_16BIT_STACK_SIZE]; +DEFINE_PER_CPU(unsigned char, cpu_16bit_stack[CPU_16BIT_STACK_SIZE]); +EXPORT_PER_CPU_SYMBOL(cpu_16bit_stack); + static int cachesize_override __initdata = -1; static int disable_x86_fxsr __initdata = 0; static int disable_x86_serial_nr __initdata = 1; @@ -557,6 +561,7 @@ int cpu = smp_processor_id(); struct tss_struct * t = &per_cpu(init_tss, cpu); struct thread_struct *thread = ¤t->thread; + __u32 stk16_off = (__u32)&per_cpu(cpu_16bit_stack, cpu); if (cpu_test_and_set(cpu, cpu_initialized)) { printk(KERN_WARNING "CPU#%d already initialized!\n", cpu); @@ -579,6 +584,13 @@ */ memcpy(&per_cpu(cpu_gdt_table, cpu), cpu_gdt_table, GDT_SIZE); + + /* Set up GDT entry for 16bit stack */ + *(__u64 *)&(per_cpu(cpu_gdt_table, cpu)[GDT_ENTRY_ESPFIX_SS]) |= + ((((__u64)stk16_off) << 16) & 0x000000ffffff0000ULL) | + ((((__u64)stk16_off) << 32) & 0xff00000000000000ULL) | + (CPU_16BIT_STACK_SIZE - 1); + cpu_gdt_descr[cpu].size = GDT_SIZE - 1; cpu_gdt_descr[cpu].address = (unsigned long)&per_cpu(cpu_gdt_table, cpu); diff -ur linux-2.6.11-ac2/arch/i386/kernel/entry.S linux-2.6.11-ac2-stk/arch/i386/kernel/entry.S --- linux-2.6.11-ac2/arch/i386/kernel/entry.S 2005-03-13 17:59:45.000000000 +0300 +++ linux-2.6.11-ac2-stk/arch/i386/kernel/entry.S 2005-03-13 18:17:47.000000000 +0300 @@ -47,6 +47,7 @@ #include #include #include +#include #include "irq_vectors.h" #define nr_syscalls ((syscall_table_size)/4) @@ -78,7 +79,7 @@ #define preempt_stop cli #else #define preempt_stop -#define resume_kernel restore_all +#define resume_kernel restore_nocheck #endif #define SAVE_ALL \ @@ -122,24 +123,6 @@ .previous -#define RESTORE_ALL \ - RESTORE_REGS \ - addl $4, %esp; \ -1: iret; \ -.section .fixup,"ax"; \ -2: sti; \ - movl $(__USER_DS), %edx; \ - movl %edx, %ds; \ - movl %edx, %es; \ - movl $11,%eax; \ - call do_exit; \ -.previous; \ -.section __ex_table,"a";\ - .align 4; \ - .long 1b,2b; \ -.previous - - ENTRY(ret_from_fork) pushl %eax call schedule_tail @@ -163,7 +146,7 @@ movl EFLAGS(%esp), %eax # mix EFLAGS and CS movb CS(%esp), %al testl $(VM_MASK | 3), %eax - jz resume_kernel # returning to kernel or vm86-space + jz resume_kernel ENTRY(resume_userspace) cli # make sure we don't miss an interrupt # setting need_resched or sigpending @@ -178,7 +161,7 @@ ENTRY(resume_kernel) cli cmpl $0,TI_preempt_count(%ebp) # non-zero preempt_count ? - jnz restore_all + jnz restore_nocheck need_resched: movl TI_flags(%ebp), %ecx # need_resched set ? testb $_TIF_NEED_RESCHED, %cl @@ -189,6 +172,31 @@ jmp need_resched #endif +ldt_ss: + larl OLDSS(%esp), %eax + jnz restore_nocheck + testl $0x00400000, %eax # returning to 32bit stack? + jnz restore_nocheck # allright, normal return + /* If returning to userspace with 16bit stack, + * try to fix the higher word of ESP, as the CPU + * won't restore it. + * This is an "official" bug of all the x86-compatible + * CPUs, which we can try to work around to make + * dosemu and wine happy. */ + subl $8, %esp # reserve space for switch16 pointer + cli + movl %esp, %eax + /* Set up the 16bit stack frame with switch32 pointer on top, + * and a switch16 pointer on top of the current frame. */ + call setup_x86_bogus_stack + RESTORE_REGS + lss 20+4(%esp), %esp # switch to 16bit stack +1: iret +.section __ex_table,"a" + .align 4 + .long 1b,iret_exc +.previous + /* SYSENTER_RETURN points to after the "sysenter" instruction in the vsyscall page. See vsyscall-sysentry.S, which defines the symbol. */ @@ -257,8 +265,31 @@ movl TI_flags(%ebp), %ecx testw $_TIF_ALLWORK_MASK, %cx # current->work jne syscall_exit_work + restore_all: - RESTORE_ALL + movl EFLAGS(%esp), %eax # mix EFLAGS, SS and CS + movb OLDSS(%esp), %ah + movb CS(%esp), %al + andl $(VM_MASK | (4 << 8) | 3), %eax + cmpl $((4 << 8) | 3), %eax + je ldt_ss # returning to user-space with LDT SS +restore_nocheck: + RESTORE_REGS + addl $4, %esp +1: iret +.section .fixup,"ax" +iret_exc: + sti + movl $__USER_DS, %edx + movl %edx, %ds + movl %edx, %es + movl $11,%eax + call do_exit +.previous +.section __ex_table,"a" + .align 4 + .long 1b,iret_exc +.previous # perform work that needs to be done immediately before resumption ALIGN @@ -334,6 +365,27 @@ movl $-ENOSYS,EAX(%esp) jmp resume_userspace +#define FIXUP_ESPFIX_STACK \ + movl %esp, %eax; \ + /* switch to 32bit stack using the pointer on top of 16bit stack */ \ + lss %ss:CPU_16BIT_STACK_SIZE-8, %esp; \ + /* copy data from 16bit stack to 32bit stack */ \ + call fixup_x86_bogus_stack; \ + /* put ESP to the proper location */ \ + movl %eax, %esp; +#define UNWIND_ESPFIX_STACK \ + pushl %eax; \ + movl %ss, %eax; \ + /* see if on 16bit stack */ \ + cmpw $__ESPFIX_SS, %ax; \ + jne 28f; \ + movl $__KERNEL_DS, %edx; \ + movl %edx, %ds; \ + movl %edx, %es; \ + /* switch to 32bit stack */ \ + FIXUP_ESPFIX_STACK \ +28: popl %eax; + /* * Build the entry stubs and pointer table with * some assembler magic. @@ -388,7 +440,9 @@ pushl %ecx pushl %ebx cld - movl %es, %ecx + pushl %es + UNWIND_ESPFIX_STACK + popl %ecx movl ES(%esp), %edi # get the function address movl ORIG_EAX(%esp), %edx # get the error code movl %eax, ORIG_EAX(%esp) @@ -470,6 +524,11 @@ * fault happened on the sysenter path. */ ENTRY(nmi) + pushl %eax + movl %ss, %eax + cmpw $__ESPFIX_SS, %ax + popl %eax + je nmi_16bit_stack cmpl $sysenter_entry,(%esp) je nmi_stack_fixup pushl %eax @@ -489,7 +548,7 @@ xorl %edx,%edx # zero error code movl %esp,%eax # pt_regs pointer call do_nmi - RESTORE_ALL + jmp restore_all nmi_stack_fixup: FIX_STACK(12,nmi_stack_correct, 1) @@ -505,6 +564,29 @@ FIX_STACK(24,nmi_stack_correct, 1) jmp nmi_stack_correct +nmi_16bit_stack: + /* create the pointer to lss back */ + pushl %ss + pushl %esp + movzwl %sp, %esp + addw $4, (%esp) + /* copy the iret frame of 12 bytes */ + .rept 3 + pushl 16(%esp) + .endr + pushl %eax + SAVE_ALL + FIXUP_ESPFIX_STACK # %eax == %esp + xorl %edx,%edx # zero error code + call do_nmi + RESTORE_REGS + lss 12+4(%esp), %esp # back to 16bit stack +1: iret +.section __ex_table,"a" + .align 4 + .long 1b,iret_exc +.previous + ENTRY(int3) pushl $-1 # mark this as an int SAVE_ALL diff -ur linux-2.6.11-ac2/arch/i386/kernel/head.S linux-2.6.11-ac2-stk/arch/i386/kernel/head.S --- linux-2.6.11-ac2/arch/i386/kernel/head.S 2005-03-13 17:56:19.000000000 +0300 +++ linux-2.6.11-ac2-stk/arch/i386/kernel/head.S 2005-03-13 18:17:47.000000000 +0300 @@ -512,7 +512,7 @@ .quad 0x00009a0000000000 /* 0xc0 APM CS 16 code (16 bit) */ .quad 0x0040920000000000 /* 0xc8 APM DS data */ - .quad 0x0000000000000000 /* 0xd0 - unused */ + .quad 0x0000920000000000 /* 0xd0 - ESPFIX 16-bit SS */ .quad 0x0000000000000000 /* 0xd8 - unused */ .quad 0x0000000000000000 /* 0xe0 - unused */ .quad 0x0000000000000000 /* 0xe8 - unused */ diff -ur linux-2.6.11-ac2/arch/i386/kernel/traps.c linux-2.6.11-ac2-stk/arch/i386/kernel/traps.c --- linux-2.6.11-ac2/arch/i386/kernel/traps.c 2005-03-13 17:59:45.000000000 +0300 +++ linux-2.6.11-ac2-stk/arch/i386/kernel/traps.c 2005-03-13 18:17:47.000000000 +0300 @@ -895,6 +895,51 @@ #endif } +fastcall void setup_x86_bogus_stack(unsigned char * stk) +{ + unsigned long *switch16_ptr, *switch32_ptr; + struct pt_regs *regs; + unsigned long stack_top, stack_bot; + unsigned short iret_frame16_off; + int cpu = smp_processor_id(); + /* reserve the space on 32bit stack for the magic switch16 pointer */ + memmove(stk, stk + 8, sizeof(struct pt_regs)); + switch16_ptr = (unsigned long *)(stk + sizeof(struct pt_regs)); + regs = (struct pt_regs *)stk; + /* now the switch32 on 16bit stack */ + stack_bot = (unsigned long)&per_cpu(cpu_16bit_stack, cpu); + stack_top = stack_bot + CPU_16BIT_STACK_SIZE; + switch32_ptr = (unsigned long *)(stack_top - 8); + iret_frame16_off = CPU_16BIT_STACK_SIZE - 8 - 20; + /* copy iret frame on 16bit stack */ + memcpy((void *)(stack_bot + iret_frame16_off), ®s->eip, 20); + /* fill in the switch pointers */ + switch16_ptr[0] = (regs->esp & 0xffff0000) | iret_frame16_off; + switch16_ptr[1] = __ESPFIX_SS; + switch32_ptr[0] = (unsigned long)stk + sizeof(struct pt_regs) + + 8 - CPU_16BIT_STACK_SIZE; + switch32_ptr[1] = __KERNEL_DS; +} + +fastcall unsigned char * fixup_x86_bogus_stack(unsigned short sp) +{ + unsigned long *switch32_ptr; + unsigned char *stack16, *stack32; + unsigned long stack_top, stack_bot; + int len; + int cpu = smp_processor_id(); + stack_bot = (unsigned long)&per_cpu(cpu_16bit_stack, cpu); + stack_top = stack_bot + CPU_16BIT_STACK_SIZE; + switch32_ptr = (unsigned long *)(stack_top - 8); + /* copy the data from 16bit stack to 32bit stack */ + len = CPU_16BIT_STACK_SIZE - 8 - sp; + stack16 = (unsigned char *)(stack_bot + sp); + stack32 = (unsigned char *) + (switch32_ptr[0] + CPU_16BIT_STACK_SIZE - 8 - len); + memcpy(stack32, stack16, len); + return stack32; +} + /* * 'math_state_restore()' saves the current math information in the * old math state array, and gets the new ones from the current task diff -ur linux-2.6.11-ac2/include/asm-i386/desc.h linux-2.6.11-ac2-stk/include/asm-i386/desc.h --- linux-2.6.11-ac2/include/asm-i386/desc.h 2005-03-13 17:57:12.000000000 +0300 +++ linux-2.6.11-ac2-stk/include/asm-i386/desc.h 2005-03-13 18:17:47.000000000 +0300 @@ -4,6 +4,8 @@ #include #include +#define CPU_16BIT_STACK_SIZE 1024 + #ifndef __ASSEMBLY__ #include @@ -15,6 +17,9 @@ extern struct desc_struct cpu_gdt_table[GDT_ENTRIES]; DECLARE_PER_CPU(struct desc_struct, cpu_gdt_table[GDT_ENTRIES]); +extern unsigned char cpu_16bit_stack[CPU_16BIT_STACK_SIZE]; +DECLARE_PER_CPU(unsigned char, cpu_16bit_stack[CPU_16BIT_STACK_SIZE]); + struct Xgt_desc_struct { unsigned short size; unsigned long address __attribute__((packed)); diff -ur linux-2.6.11-ac2/include/asm-i386/segment.h linux-2.6.11-ac2-stk/include/asm-i386/segment.h --- linux-2.6.11-ac2/include/asm-i386/segment.h 2005-01-04 03:59:37.000000000 +0300 +++ linux-2.6.11-ac2-stk/include/asm-i386/segment.h 2005-03-13 18:17:47.000000000 +0300 @@ -38,7 +38,7 @@ * 24 - APM BIOS support * 25 - APM BIOS support * - * 26 - unused + * 26 - ESPFIX small SS * 27 - unused * 28 - unused * 29 - unused @@ -71,6 +71,9 @@ #define GDT_ENTRY_PNPBIOS_BASE (GDT_ENTRY_KERNEL_BASE + 6) #define GDT_ENTRY_APMBIOS_BASE (GDT_ENTRY_KERNEL_BASE + 11) +#define GDT_ENTRY_ESPFIX_SS (GDT_ENTRY_KERNEL_BASE + 14) +#define __ESPFIX_SS (GDT_ENTRY_ESPFIX_SS * 8) + #define GDT_ENTRY_DOUBLEFAULT_TSS 31 /* --------------040102020901030503060706-- - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/