From: Ingo Molnar DESC 4G/4G: remove debug code EDESC The debug code in do_page_fault gets in the way of BUG handling, and triggers in the boot-time buggy wp test probe. DESC 4g4g: pmd fix EDESC From: Ingo Molnar DESC 4g/4g: fixes from Bill EDESC From: William Lee Irwin III In case a more explicit description is required (only difference from before is ripping out touch_all_pages() for CONFIG_DISCONTIGMEM): (a) fixes oops on page->lru list poison in pgd_dtor() on PAE (b) fixes pmd corruption due to userspace teardown not zeroing out the kernel pmd entries (c) fixes pgd_ctor() performance problems on larger/PAE systems (d) fixes touch_all_pages() boot-time triplefault on i386 discontig (e) preserves pgd and pmd preconstruction (f) preserves pgtable.c's #ifdef-lessness Successfully tested on 28x/56GB (sorry, I lost a node to a temporary hardware failure) with XKVA enabled. Without this patch (and the cpumask_arith.h fixes) no runtime whatsoever is possible (in fact, the touch_all_pages() issue takes it out well prior to console_init()). vs. 2.6.0-test2-mm4 DESC 4g4g: fpu emulation fix EDESC From: Ingo Molnar , Raphael Kubo da Costa the attached patch makes the i386 FPU-emu code compile. It boots on a newer CPU (fpu-emu code inactive but present). But it might even work on a real i386, if no other bugs get in the way. DESC 4g/4g usercopy atomicity fix EDESC We're not allowed to do down_read() on the filemap_copy_from_user() path: the caller has an atomic kmap. Just fail the copy and let the filemap_copy_from_user() slow path recover. DESC 4G/4G: remove debug code EDESC The debug code in do_page_fault gets in the way of BUG handling, and triggers in the boot-time buggy wp test probe. DESC 4g4g: pmd fix EDESC From: Ingo Molnar DESC 4g/4g: fixes from Bill EDESC From: William Lee Irwin III In case a more explicit description is required (only difference from before is ripping out touch_all_pages() for CONFIG_DISCONTIGMEM): (a) fixes oops on page->lru list poison in pgd_dtor() on PAE (b) fixes pmd corruption due to userspace teardown not zeroing out the kernel pmd entries (c) fixes pgd_ctor() performance problems on larger/PAE systems (d) fixes touch_all_pages() boot-time triplefault on i386 discontig (e) preserves pgd and pmd preconstruction (f) preserves pgtable.c's #ifdef-lessness Successfully tested on 28x/56GB (sorry, I lost a node to a temporary hardware failure) with XKVA enabled. Without this patch (and the cpumask_arith.h fixes) no runtime whatsoever is possible (in fact, the touch_all_pages() issue takes it out well prior to console_init()). vs. 2.6.0-test2-mm4 DESC 4g4g: fpu emulation fix EDESC From: Ingo Molnar , Raphael Kubo da Costa the attached patch makes the i386 FPU-emu code compile. It boots on a newer CPU (fpu-emu code inactive but present). But it might even work on a real i386, if no other bugs get in the way. DESC 4g/4g usercopy atomicity fix EDESC We're not allowed to do down_read() on the filemap_copy_from_user() path: the caller has an atomic kmap. Just fail the copy and let the filemap_copy_from_user() slow path recover. DESC 4G/4G preempt on vstack EDESC From: Hugh Dickins Repeated -j3 kernel builds, run in tandem on dual PIII, have been collapsing recently on -mm with 4G/4G split, SMP and preemption. Typically 'make' fails with Error 139 because 'as' or another got SIGSEGV; maybe within ten minutes, maybe after ten hours. This patch seems to fix that (ran successfully overnight on test4-mm1, will run over the weekend on test4-mm3-1). Please cast a critical eye over it, I expect Ingo or someone else will find it can be improved. The problem is that a task may be preempted just after it has entered kernelspace, while using the transitional "virtual stack" i.e. %esp pointing to high per-cpu kmap of the kernel stack. If the task resumes on another cpu, that %esp needs to be repointed into the new cpu's kmap. The corresponding returns to userspace look okay to me: interrupts are disabled over the critical points. And in general no copy is taken of %esp while on the virtual stack e.g. setting pointer to pt_regs is and must be done after switching to real stack. But there's one place in __SWITCH_KERNELSPACE itself where we need to check and repeat if moved. DESC 4G/4G: even number of kmap types EDESC From: Hugh Dickins Extend the warning comment in kmap_types.h: if you add an odd number of new kmaps, KM_VSTACK0 becomes misaligned on odd numbered cpus. I've not added a corresponding BUG_ON to entry_trampoline.c, having an aversion to errors which trigger too early to be seen. We should do better, perhaps #error, or robustifying the layout; but for now just comment. DESC 4g4g: fix __get_user in slab EDESC From: Kirill Korotaev The __get_user() in slab is against a kernel address and will crash in XKVA mode. Fix that, and accidentally repair some whitespace. DESC 4g4g: Remove extra .data.idt section definition EDESC From: Zwane Mwaikambo Remove extra .data.idt section definition DESC 4g/4g linker error (overlapping sections) EDESC From: Zwane Mwaikambo I got this on a NUMAQ 32way/32G kernel build with an external patch, i had to bump the PAGE_SIZE multiple (which is currently 2) to fix it, but how does this patch look instead? ld: section __ex_table [022cd000 -> 022cd207] overlaps section .entry.text [022cb000 -> 022cdfdf] ld: section .rodata [022cd220 -> 02314f61] overlaps section .entry.text [022cb000 -> 022cdfdf] The resultant image is the same (checked with objdump -t before and after) and compiles with my patch too. 02562000 g *ABS* 00000000 __entry_tramp_start 0256373c g *ABS* 00000000 __entry_tramp_end 02564000 g *ABS* 00000000 _etext 02564000 l d __ex_table 00000000 02564178 g *ABS* 00000000 __stop___ex_table 02564000 g *ABS* 00000000 __start___ex_table DESC 4G/4G: remove debug code EDESC The debug code in do_page_fault gets in the way of BUG handling, and triggers in the boot-time buggy wp test probe. DESC 4g4g: pmd fix EDESC From: Ingo Molnar DESC 4g/4g: fixes from Bill EDESC From: William Lee Irwin III In case a more explicit description is required (only difference from before is ripping out touch_all_pages() for CONFIG_DISCONTIGMEM): (a) fixes oops on page->lru list poison in pgd_dtor() on PAE (b) fixes pmd corruption due to userspace teardown not zeroing out the kernel pmd entries (c) fixes pgd_ctor() performance problems on larger/PAE systems (d) fixes touch_all_pages() boot-time triplefault on i386 discontig (e) preserves pgd and pmd preconstruction (f) preserves pgtable.c's #ifdef-lessness Successfully tested on 28x/56GB (sorry, I lost a node to a temporary hardware failure) with XKVA enabled. Without this patch (and the cpumask_arith.h fixes) no runtime whatsoever is possible (in fact, the touch_all_pages() issue takes it out well prior to console_init()). vs. 2.6.0-test2-mm4 DESC 4g4g: fpu emulation fix EDESC From: Ingo Molnar , Raphael Kubo da Costa the attached patch makes the i386 FPU-emu code compile. It boots on a newer CPU (fpu-emu code inactive but present). But it might even work on a real i386, if no other bugs get in the way. DESC 4g4g: show_registers() fix EDESC From: Ingo Molnar ok, the patch below does the trick. It works fine on 4G and non-4G kernels as well and dumps the expected code. DESC 4g/4g usercopy atomicity fix EDESC We're not allowed to do down_read() on the filemap_copy_from_user() path: the caller has an atomic kmap. Just fail the copy and let the filemap_copy_from_user() slow path recover. DESC 4g4g: debug flags fix EDESC From: Manfred Spraul The GD bit in dr7 is on, thus every write access to a debug register causes a debug interrupt. There seem to be two bugs: a) the code is entry.S does spurious reloads: The call to do_notify_resume() overwrites %cl, thus the test of %cl against $_TIF_DB7 is spuriously true. b) someone loads 0xfff into db7. This includes the GD bit, which the kernel doesn't handle. Could you add the attached patch to your tree? It fixes the first part of the bug. I don't understand yet where the 0xfff comes from. With it applied, I could successfully boot 2.6.0-test3-mm1 on my 2xpIII with your .config DESC 4g4g: Fix wrong asm-offsets entry EDESC From: Manfred Spraul I found the 2nd error that caused the db7 corruption with 2.4.0-test3-mm1: There is a copy and paste error in asm-offsets.c that generates a wrong TI_task value. The result of the error is that wrong values are loaded into db7 at syscall return with 4g/4g kernels. DESC cyclone time fixmap fix EDESC From: Dave Hansen The presence of FIX_CYCLONE_TIMER depends on CONFIG_X86_CYCLONE. DESC 4G/4G preempt on vstack EDESC From: Hugh Dickins Repeated -j3 kernel builds, run in tandem on dual PIII, have been collapsing recently on -mm with 4G/4G split, SMP and preemption. Typically 'make' fails with Error 139 because 'as' or another got SIGSEGV; maybe within ten minutes, maybe after ten hours. This patch seems to fix that (ran successfully overnight on test4-mm1, will run over the weekend on test4-mm3-1). Please cast a critical eye over it, I expect Ingo or someone else will find it can be improved. The problem is that a task may be preempted just after it has entered kernelspace, while using the transitional "virtual stack" i.e. %esp pointing to high per-cpu kmap of the kernel stack. If the task resumes on another cpu, that %esp needs to be repointed into the new cpu's kmap. The corresponding returns to userspace look okay to me: interrupts are disabled over the critical points. And in general no copy is taken of %esp while on the virtual stack e.g. setting pointer to pt_regs is and must be done after switching to real stack. But there's one place in __SWITCH_KERNELSPACE itself where we need to check and repeat if moved. DESC 4G/4G: even number of kmap types EDESC From: Hugh Dickins Extend the warning comment in kmap_types.h: if you add an odd number of new kmaps, KM_VSTACK0 becomes misaligned on odd numbered cpus. I've not added a corresponding BUG_ON to entry_trampoline.c, having an aversion to errors which trigger too early to be seen. We should do better, perhaps #error, or robustifying the layout; but for now just comment. DESC 4g4g: fix __get_user in slab EDESC From: Kirill Korotaev The __get_user() in slab is against a kernel address and will crash in XKVA mode. Fix that, and accidentally repair some whitespace. DESC 4g4g: Remove extra .data.idt section definition EDESC From: Zwane Mwaikambo Remove extra .data.idt section definition DESC 4g/4g linker error (overlapping sections) EDESC From: Zwane Mwaikambo I got this on a NUMAQ 32way/32G kernel build with an external patch, i had to bump the PAGE_SIZE multiple (which is currently 2) to fix it, but how does this patch look instead? ld: section __ex_table [022cd000 -> 022cd207] overlaps section .entry.text [022cb000 -> 022cdfdf] ld: section .rodata [022cd220 -> 02314f61] overlaps section .entry.text [022cb000 -> 022cdfdf] The resultant image is the same (checked with objdump -t before and after) and compiles with my patch too. 02562000 g *ABS* 00000000 __entry_tramp_start 0256373c g *ABS* 00000000 __entry_tramp_end 02564000 g *ABS* 00000000 _etext 02564000 l d __ex_table 00000000 02564178 g *ABS* 00000000 __stop___ex_table 02564000 g *ABS* 00000000 __start___ex_table DESC 4G/4G: remove debug code EDESC The debug code in do_page_fault gets in the way of BUG handling, and triggers in the boot-time buggy wp test probe. DESC 4g4g: pmd fix EDESC From: Ingo Molnar DESC 4g/4g: fixes from Bill EDESC From: William Lee Irwin III In case a more explicit description is required (only difference from before is ripping out touch_all_pages() for CONFIG_DISCONTIGMEM): (a) fixes oops on page->lru list poison in pgd_dtor() on PAE (b) fixes pmd corruption due to userspace teardown not zeroing out the kernel pmd entries (c) fixes pgd_ctor() performance problems on larger/PAE systems (d) fixes touch_all_pages() boot-time triplefault on i386 discontig (e) preserves pgd and pmd preconstruction (f) preserves pgtable.c's #ifdef-lessness Successfully tested on 28x/56GB (sorry, I lost a node to a temporary hardware failure) with XKVA enabled. Without this patch (and the cpumask_arith.h fixes) no runtime whatsoever is possible (in fact, the touch_all_pages() issue takes it out well prior to console_init()). vs. 2.6.0-test2-mm4 DESC 4g4g: fpu emulation fix EDESC From: Ingo Molnar , Raphael Kubo da Costa the attached patch makes the i386 FPU-emu code compile. It boots on a newer CPU (fpu-emu code inactive but present). But it might even work on a real i386, if no other bugs get in the way. DESC 4g4g: show_registers() fix EDESC From: Ingo Molnar ok, the patch below does the trick. It works fine on 4G and non-4G kernels as well and dumps the expected code. DESC 4g/4g usercopy atomicity fix EDESC We're not allowed to do down_read() on the filemap_copy_from_user() path: the caller has an atomic kmap. Just fail the copy and let the filemap_copy_from_user() slow path recover. DESC 4g4g: debug flags fix EDESC From: Manfred Spraul The GD bit in dr7 is on, thus every write access to a debug register causes a debug interrupt. There seem to be two bugs: a) the code is entry.S does spurious reloads: The call to do_notify_resume() overwrites %cl, thus the test of %cl against $_TIF_DB7 is spuriously true. b) someone loads 0xfff into db7. This includes the GD bit, which the kernel doesn't handle. Could you add the attached patch to your tree? It fixes the first part of the bug. I don't understand yet where the 0xfff comes from. With it applied, I could successfully boot 2.6.0-test3-mm1 on my 2xpIII with your .config DESC 4g4g: Fix wrong asm-offsets entry EDESC From: Manfred Spraul I found the 2nd error that caused the db7 corruption with 2.4.0-test3-mm1: There is a copy and paste error in asm-offsets.c that generates a wrong TI_task value. The result of the error is that wrong values are loaded into db7 at syscall return with 4g/4g kernels. DESC cyclone time fixmap fix EDESC From: Dave Hansen The presence of FIX_CYCLONE_TIMER depends on CONFIG_X86_CYCLONE. DESC use direct_copy_{to,from}_user for kernel access in mm/usercopy.c EDESC From: Manfred Spraul 4G/4G+CONFIG_DEBUG_PAGEALLOC crashes with my bochs setup. It's caused by copy_mount_options: the function must perform a copy_from_user without proper boundary checks [stupid ABI] and the user access functions in mm/usercopy.c do not create exception handler entries if fs==KERNEL_DS. The attached patch fixes that by using direct_copy_{from,to}_user instead of memcpy. I've triggered the crash with the debug option, but I'd bet that - given enough users and setups - someone will trigger it with AGP GART memory. DESC 4G/4G might_sleep warning fix EDESC From: Hugh Dickins Manfred's fix for copy_mount_options with 4G/4G + CONFIG_DEBUG_PAGEALLOC makes trouble for do_execve with 4G/4G + CONFIG_DEBUG_SPINLOCK_SLEEP: prepare_binrpm.. vfs_read.. file_read_actor.. (kmap_atomic) put_user_size.. __might_sleep. These underscores work for my case: for yours, Manfred? DESC 4g/4g pagetable accounting fix EDESC From: Hugh Dickins The problem is that PKMAP_BASE is lower than TASK_SIZE, so pgd_ctor (non-PAE) or kpmd_ctor (PAE) copies init_mm's entry for PKMAP_BASE into the task's pgd/pmd, and so it gets "freed" below clean_page_tables. DESC Fix 4G/4G and WP test lockup EDESC From: Zwane Mwaikambo It looks like when we do the WP test and trigger a (write) protection fault, the 4G/4G page fault handling path doesn't expect this kind of fault and instead results in recursive fault handling (or so it appears). DESC 4G/4G KERNEL_DS usercopy again EDESC From: Hugh Dickins 4G/4G get_user_size and put_user_size are still not quite right in the KERNEL_DS case: Manfred found the original memcpys wrong, we fixed them to direct_copys then __direct_copys; now use by kdb's md command proves we shouldn't assume success 0, but return the actual residue; and to perfectly imitate non-4G/4G behaviour, I suppose get_user_size should even memset to 0 in the KERNEL_DS case of residue. DESC Fix 4G/4G X11/vm86 oops EDESC From: Zwane Mwaikambo The 4G/4G page fault handling path doesn't appear to handle faults happening whilst in vm86. The regs->xcs != __USER_CS so it confused the in kernel test. However i'm still debugging the X11 triple fault in test9-mm3 Unable to handle kernel paging request at virtual address 00002000 printing eip: 00007341 *pde = 00000000 Oops: 0004 [#1] SMP DEBUG_PAGEALLOC CPU: 0 EIP: c000:[<00007341>] Not tainted VLI EFLAGS: 00033246 EIP is at 0x7341 eax: 32454256 ebx: 00000000 ecx: 00000000 edx: 00000000 esi: 00000000 edi: 00002000 ebp: 00000fd6 esp: 087bbf24 ds: 0000 es: 0000 ss: 0068 Process X (pid: 939, threadinfo=087ba000 task=0891c690) Stack: 00000fcb 00000100 00000000 0000c000 00000000 00000000 00000000 00000000 00000005 ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff ffffffff Call Trace: DESC Fix 4G/4G athlon triplefault EDESC From: Matt Mackall Fix triple faulting on some boxes with 4G/4G DESC 4g4g SEP fix EDESC From: Ingo Molnar the attached patch solves a 'init hang' problem i saw with the 4G/4G patch. SEP is not fully implemented under 4G/4G and even though we disable AT_SYSINFO, it still triggers problems ... so get rid of it explicitly. DESC Fix 4G/4G split fix for pre-pentiumII machines EDESC From: Arjan van de Ven As per bug 1705; machines earlier than pII don't have sysenter support; the 4g/4g split code tests for sysenter support correctly before writing the relevant MSR everywhere except for one place; the patch below adds the test. The person with the machine confirmed it working as per bugzilla. DESC 4g/4g PAE ACPI low mappings fix EDESC From: Ingo Molnar - Teach acpi to not access physical memory instead of virtual. It dies with PAE and the 4g split. - the reason why PAE triggers the bug and !PAE doesnt is the following: there's a small window between pagetable_init() and zap_low_mappings() in which window eg. the ACPI code runs - in this window the low mappings still exist. In the PAE case the low mappings dont exist at all (they are not set up). The patch below makes the code more consistent and zaps the low mappings earlier. --- arch/i386/Kconfig | 48 ++++ arch/i386/boot/setup.S | 2 arch/i386/kernel/Makefile | 2 arch/i386/kernel/acpi/boot.c | 2 arch/i386/kernel/asm-offsets.c | 15 + arch/i386/kernel/cpu/common.c | 6 arch/i386/kernel/cpu/intel.c | 11 arch/i386/kernel/doublefault.c | 7 arch/i386/kernel/entry.S | 238 ++++++++++++++++---- arch/i386/kernel/entry_trampoline.c | 75 ++++++ arch/i386/kernel/head.S | 49 +++- arch/i386/kernel/i386_ksyms.c | 7 arch/i386/kernel/i387.c | 24 +- arch/i386/kernel/init_task.c | 4 arch/i386/kernel/ldt.c | 132 +++++++---- arch/i386/kernel/mpparse.c | 2 arch/i386/kernel/process.c | 50 ++++ arch/i386/kernel/reboot.c | 11 arch/i386/kernel/signal.c | 96 ++++---- arch/i386/kernel/smp.c | 22 - arch/i386/kernel/sysenter.c | 5 arch/i386/kernel/traps.c | 80 +++++-- arch/i386/kernel/vm86.c | 4 arch/i386/kernel/vmlinux.lds.S | 47 +++- arch/i386/kernel/vsyscall-sysenter.S | 5 arch/i386/kernel/vsyscall.lds | 2 arch/i386/lib/checksum.S | 6 arch/i386/lib/getuser.S | 7 arch/i386/lib/usercopy.c | 11 arch/i386/math-emu/fpu_system.h | 3 arch/i386/mm/fault.c | 27 ++ arch/i386/mm/init.c | 400 +++++++++++++++++------------------ arch/i386/mm/pgtable.c | 123 +++++++++- drivers/block/scsi_ioctl.c | 2 include/asm-i386/atomic_kmap.h | 95 ++++++++ include/asm-i386/checksum.h | 30 ++ include/asm-i386/desc.h | 37 --- include/asm-i386/fixmap.h | 37 ++- include/asm-i386/highmem.h | 11 include/asm-i386/kmap_types.h | 54 ++-- include/asm-i386/mmu.h | 5 include/asm-i386/mmu_context.h | 12 - include/asm-i386/page.h | 28 +- include/asm-i386/pgtable.h | 18 + include/asm-i386/processor.h | 26 +- include/asm-i386/string.h | 23 ++ include/asm-i386/thread_info.h | 20 - include/asm-i386/tlbflush.h | 9 include/asm-i386/uaccess.h | 127 +++++++++-- mm/Makefile | 3 mm/memory.c | 19 + mm/slab.c | 34 ++ mm/usercopy.c | 277 ++++++++++++++++++++++++ 53 files changed, 1779 insertions(+), 611 deletions(-) diff -puN arch/i386/boot/setup.S~4g-2.6.0-test2-mm2-A5 arch/i386/boot/setup.S --- 25/arch/i386/boot/setup.S~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/boot/setup.S 2004-01-20 21:32:48.000000000 -0800 @@ -164,7 +164,7 @@ cmd_line_ptr: .long 0 # (Header versio # can be located anywhere in # low memory 0x10000 or higher. -ramdisk_max: .long MAXMEM-1 # (Header version 0x0203 or later) +ramdisk_max: .long __MAXMEM-1 # (Header version 0x0203 or later) # The highest safe address for # the contents of an initrd diff -puN arch/i386/Kconfig~4g-2.6.0-test2-mm2-A5 arch/i386/Kconfig --- 25/arch/i386/Kconfig~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/Kconfig 2004-01-20 21:32:48.000000000 -0800 @@ -403,6 +403,54 @@ config X86_OOSTORE depends on CPU_ONLY_WINCHIP default y +config X86_4G + bool "4 GB kernel-space and 4 GB user-space virtual memory support" + help + This option is only useful for systems that have more than 1 GB + of RAM. + + The default kernel VM layout leaves 1 GB of virtual memory for + kernel-space mappings, and 3 GB of VM for user-space applications. + This option ups both the kernel-space VM and the user-space VM to + 4 GB. + + The cost of this option is additional TLB flushes done at + system-entry points that transition from user-mode into kernel-mode. + I.e. system calls and page faults, and IRQs that interrupt user-mode + code. There's also additional overhead to kernel operations that copy + memory to/from user-space. The overhead from this is hard to tell and + depends on the workload - it can be anything from no visible overhead + to 20-30% overhead. A good rule of thumb is to count with a runtime + overhead of 20%. + + The upside is the much increased kernel-space VM, which more than + quadruples the maximum amount of RAM supported. Kernels compiled with + this option boot on 64GB of RAM and still have more than 3.1 GB of + 'lowmem' left. Another bonus is that highmem IO bouncing decreases, + if used with drivers that still use bounce-buffers. + + There's also a 33% increase in user-space VM size - database + applications might see a boost from this. + + But the cost of the TLB flushes and the runtime overhead has to be + weighed against the bonuses offered by the larger VM spaces. The + dividing line depends on the actual workload - there might be 4 GB + systems that benefit from this option. Systems with less than 4 GB + of RAM will rarely see a benefit from this option - but it's not + out of question, the exact circumstances have to be considered. + +config X86_SWITCH_PAGETABLES + def_bool X86_4G + +config X86_4G_VM_LAYOUT + def_bool X86_4G + +config X86_UACCESS_INDIRECT + def_bool X86_4G + +config X86_HIGH_ENTRY + def_bool X86_4G + config HPET_TIMER bool "HPET Timer Support" help diff -puN arch/i386/kernel/asm-offsets.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/asm-offsets.c --- 25/arch/i386/kernel/asm-offsets.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/asm-offsets.c 2004-01-20 21:32:48.000000000 -0800 @@ -4,9 +4,11 @@ * to extract and format the required data. */ +#include #include #include #include "sigframe.h" +#include #define DEFINE(sym, val) \ asm volatile("\n->" #sym " %0 " #val : : "i" (val)) @@ -28,4 +30,17 @@ void foo(void) DEFINE(RT_SIGFRAME_sigcontext, offsetof (struct rt_sigframe, uc.uc_mcontext)); + DEFINE(TI_task, offsetof (struct thread_info, task)); + DEFINE(TI_exec_domain, offsetof (struct thread_info, exec_domain)); + DEFINE(TI_flags, offsetof (struct thread_info, flags)); + DEFINE(TI_preempt_count, offsetof (struct thread_info, preempt_count)); + DEFINE(TI_addr_limit, offsetof (struct thread_info, addr_limit)); + DEFINE(TI_real_stack, offsetof (struct thread_info, real_stack)); + DEFINE(TI_virtual_stack, offsetof (struct thread_info, virtual_stack)); + DEFINE(TI_user_pgd, offsetof (struct thread_info, user_pgd)); + + DEFINE(FIX_ENTRY_TRAMPOLINE_0_addr, __fix_to_virt(FIX_ENTRY_TRAMPOLINE_0)); + DEFINE(FIX_VSYSCALL_addr, __fix_to_virt(FIX_VSYSCALL)); + DEFINE(PAGE_SIZE_asm, PAGE_SIZE); + DEFINE(task_thread_db7, offsetof (struct task_struct, thread.debugreg[7])); } diff -puN arch/i386/kernel/cpu/common.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/cpu/common.c --- 25/arch/i386/kernel/cpu/common.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/cpu/common.c 2004-01-20 21:32:48.000000000 -0800 @@ -514,12 +514,16 @@ void __init cpu_init (void) set_tss_desc(cpu,t); cpu_gdt_table[cpu][GDT_ENTRY_TSS].b &= 0xfffffdff; load_TR_desc(); - load_LDT(&init_mm.context); + if (cpu) + load_LDT(&init_mm.context); /* Set up doublefault TSS pointer in the GDT */ __set_tss_desc(cpu, GDT_ENTRY_DOUBLEFAULT_TSS, &doublefault_tss); cpu_gdt_table[cpu][GDT_ENTRY_DOUBLEFAULT_TSS].b &= 0xfffffdff; + if (cpu) + trap_init_virtual_GDT(); + /* Clear %fs and %gs. */ asm volatile ("xorl %eax, %eax; movl %eax, %fs; movl %eax, %gs"); diff -puN arch/i386/kernel/cpu/intel.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/cpu/intel.c --- 25/arch/i386/kernel/cpu/intel.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/cpu/intel.c 2004-01-20 21:32:48.000000000 -0800 @@ -10,6 +10,7 @@ #include #include #include +#include #include "cpu.h" @@ -19,8 +20,6 @@ #include #endif -extern int trap_init_f00f_bug(void); - #ifdef CONFIG_X86_INTEL_USERCOPY /* * Alignment at which movsl is preferred for bulk memory copies. @@ -165,7 +164,7 @@ static void __init init_intel(struct cpu c->f00f_bug = 1; if ( !f00f_workaround_enabled ) { - trap_init_f00f_bug(); + trap_init_virtual_IDT(); printk(KERN_NOTICE "Intel Pentium with F0 0F bug - workaround enabled.\n"); f00f_workaround_enabled = 1; } @@ -248,6 +247,12 @@ static void __init init_intel(struct cpu /* SEP CPUID bug: Pentium Pro reports SEP but doesn't have it until model 3 mask 3 */ if ((c->x86<<8 | c->x86_model<<4 | c->x86_mask) < 0x633) clear_bit(X86_FEATURE_SEP, c->x86_capability); + /* + * FIXME: SEP is disabled for 4G/4G for now: + */ +#ifdef CONFIG_X86_HIGH_ENTRY + clear_bit(X86_FEATURE_SEP, c->x86_capability); +#endif /* Names for the Pentium II/Celeron processors detectable only by also checking the cache size. diff -puN arch/i386/kernel/doublefault.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/doublefault.c --- 25/arch/i386/kernel/doublefault.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/doublefault.c 2004-01-20 21:32:48.000000000 -0800 @@ -7,12 +7,13 @@ #include #include #include +#include #define DOUBLEFAULT_STACKSIZE (1024) static unsigned long doublefault_stack[DOUBLEFAULT_STACKSIZE]; #define STACK_START (unsigned long)(doublefault_stack+DOUBLEFAULT_STACKSIZE) -#define ptr_ok(x) ((x) > 0xc0000000 && (x) < 0xc1000000) +#define ptr_ok(x) (((x) > __PAGE_OFFSET && (x) < (__PAGE_OFFSET + 0x01000000)) || ((x) >= FIXADDR_START)) static void doublefault_fn(void) { @@ -38,8 +39,8 @@ static void doublefault_fn(void) printk("eax = %08lx, ebx = %08lx, ecx = %08lx, edx = %08lx\n", t->eax, t->ebx, t->ecx, t->edx); - printk("esi = %08lx, edi = %08lx\n", - t->esi, t->edi); + printk("esi = %08lx, edi = %08lx, ebp = %08lx\n", + t->esi, t->edi, t->ebp); } } diff -puN arch/i386/kernel/entry.S~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/entry.S --- 25/arch/i386/kernel/entry.S~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/entry.S 2004-01-20 21:32:48.000000000 -0800 @@ -43,8 +43,10 @@ #include #include #include +#include #include #include +#include #include #include #include "irq_vectors.h" @@ -99,7 +101,102 @@ TSS_ESP0_OFFSET = (4 - 0x200) #define resume_kernel restore_all #endif -#define SAVE_ALL \ +#ifdef CONFIG_X86_HIGH_ENTRY + +#ifdef CONFIG_X86_SWITCH_PAGETABLES + +#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP) +/* + * If task is preempted in __SWITCH_KERNELSPACE, and moved to another cpu, + * __switch_to repoints %esp to the appropriate virtual stack; but %ebp is + * left stale, so we must check whether to repeat the real stack calculation. + */ +#define repeat_if_esp_changed \ + xorl %esp, %ebp; \ + testl $0xffffe000, %ebp; \ + jnz 0b +#else +#define repeat_if_esp_changed +#endif + +/* clobbers ebx, edx and ebp */ + +#define __SWITCH_KERNELSPACE \ + cmpl $0xff000000, %esp; \ + jb 1f; \ + \ + /* \ + * switch pagetables and load the real stack, \ + * keep the stack offset: \ + */ \ + \ + movl $swapper_pg_dir-__PAGE_OFFSET, %edx; \ + \ + /* GET_THREAD_INFO(%ebp) intermixed */ \ +0: \ + movl %esp, %ebp; \ + movl %esp, %ebx; \ + andl $0xffffe000, %ebp; \ + andl $0x00001fff, %ebx; \ + orl TI_real_stack(%ebp), %ebx; \ + repeat_if_esp_changed; \ + \ + movl %edx, %cr3; \ + movl %ebx, %esp; \ +1: + +#endif + + +#define __SWITCH_USERSPACE \ + /* interrupted any of the user return paths? */ \ + \ + movl EIP(%esp), %eax; \ + \ + cmpl $int80_ret_start_marker, %eax; \ + jb 33f; /* nope - continue with sysexit check */\ + cmpl $int80_ret_end_marker, %eax; \ + jb 22f; /* yes - switch to virtual stack */ \ +33: \ + cmpl $sysexit_ret_start_marker, %eax; \ + jb 44f; /* nope - continue with user check */ \ + cmpl $sysexit_ret_end_marker, %eax; \ + jb 22f; /* yes - switch to virtual stack */ \ + /* return to userspace? */ \ +44: \ + movl EFLAGS(%esp),%ecx; \ + movb CS(%esp),%cl; \ + testl $(VM_MASK | 3),%ecx; \ + jz 2f; \ +22: \ + /* \ + * switch to the virtual stack, then switch to \ + * the userspace pagetables. \ + */ \ + \ + GET_THREAD_INFO(%ebp); \ + movl TI_virtual_stack(%ebp), %edx; \ + movl TI_user_pgd(%ebp), %ecx; \ + \ + movl %esp, %ebx; \ + andl $0x1fff, %ebx; \ + orl %ebx, %edx; \ +int80_ret_start_marker: \ + movl %edx, %esp; \ + movl %ecx, %cr3; \ + \ + __RESTORE_ALL; \ +int80_ret_end_marker: \ +2: + +#else /* !CONFIG_X86_HIGH_ENTRY */ + +#define __SWITCH_KERNELSPACE +#define __SWITCH_USERSPACE + +#endif + +#define __SAVE_ALL \ cld; \ pushl %es; \ pushl %ds; \ @@ -112,10 +209,9 @@ TSS_ESP0_OFFSET = (4 - 0x200) pushl %ebx; \ movl $(__USER_DS), %edx; \ movl %edx, %ds; \ - movl %edx, %es; \ - STACK_OVERFLOW_TEST + movl %edx, %es; -#define RESTORE_INT_REGS \ +#define __RESTORE_INT_REGS \ popl %ebx; \ popl %ecx; \ popl %edx; \ @@ -124,29 +220,28 @@ TSS_ESP0_OFFSET = (4 - 0x200) popl %ebp; \ popl %eax -#define RESTORE_REGS \ - RESTORE_INT_REGS; \ -1: popl %ds; \ -2: popl %es; \ +#define __RESTORE_REGS \ + __RESTORE_INT_REGS; \ +111: popl %ds; \ +222: popl %es; \ .section .fixup,"ax"; \ -3: movl $0,(%esp); \ - jmp 1b; \ -4: movl $0,(%esp); \ - jmp 2b; \ +444: movl $0,(%esp); \ + jmp 111b; \ +555: movl $0,(%esp); \ + jmp 222b; \ .previous; \ .section __ex_table,"a";\ .align 4; \ - .long 1b,3b; \ - .long 2b,4b; \ + .long 111b,444b;\ + .long 222b,555b;\ .previous - -#define RESTORE_ALL \ - RESTORE_REGS \ +#define __RESTORE_ALL \ + __RESTORE_REGS \ addl $4, %esp; \ -1: iret; \ +333: iret; \ .section .fixup,"ax"; \ -2: sti; \ +666: sti; \ movl $(__USER_DS), %edx; \ movl %edx, %ds; \ movl %edx, %es; \ @@ -155,10 +250,19 @@ TSS_ESP0_OFFSET = (4 - 0x200) .previous; \ .section __ex_table,"a";\ .align 4; \ - .long 1b,2b; \ + .long 333b,666b;\ .previous +#define SAVE_ALL \ + __SAVE_ALL; \ + __SWITCH_KERNELSPACE; \ + STACK_OVERFLOW_TEST; + +#define RESTORE_ALL \ + __SWITCH_USERSPACE; \ + __RESTORE_ALL; +.section .entry.text,"ax" ENTRY(lcall7) pushfl # We get a different stack layout with call @@ -176,7 +280,7 @@ do_lcall: movl %edx,EIP(%ebp) # Now we move them to their "normal" places movl %ecx,CS(%ebp) # andl $-8192, %ebp # GET_THREAD_INFO - movl TI_EXEC_DOMAIN(%ebp), %edx # Get the execution domain + movl TI_exec_domain(%ebp), %edx # Get the execution domain call *4(%edx) # Call the lcall7 handler for the domain addl $4, %esp popl %eax @@ -221,7 +325,7 @@ ENTRY(resume_userspace) cli # make sure we don't miss an interrupt # setting need_resched or sigpending # between sampling and the iret - movl TI_FLAGS(%ebp), %ecx + movl TI_flags(%ebp), %ecx andl $_TIF_WORK_MASK, %ecx # is there any work to be done on # int/exception return? jne work_pending @@ -229,18 +333,18 @@ ENTRY(resume_userspace) #ifdef CONFIG_PREEMPT ENTRY(resume_kernel) - cmpl $0,TI_PRE_COUNT(%ebp) # non-zero preempt_count ? + cmpl $0,TI_preempt_count(%ebp) # non-zero preempt_count ? jnz restore_all need_resched: - movl TI_FLAGS(%ebp), %ecx # need_resched set ? + movl TI_flags(%ebp), %ecx # need_resched set ? testb $_TIF_NEED_RESCHED, %cl jz restore_all testl $IF_MASK,EFLAGS(%esp) # interrupts off (exception path) ? jz restore_all - movl $PREEMPT_ACTIVE,TI_PRE_COUNT(%ebp) + movl $PREEMPT_ACTIVE,TI_preempt_count(%ebp) sti call schedule - movl $0,TI_PRE_COUNT(%ebp) + movl $0,TI_preempt_count(%ebp) cli jmp need_resched #endif @@ -259,37 +363,50 @@ sysenter_past_esp: pushl $(__USER_CS) pushl $SYSENTER_RETURN -/* - * Load the potential sixth argument from user stack. - * Careful about security. - */ - cmpl $__PAGE_OFFSET-3,%ebp - jae syscall_fault -1: movl (%ebp),%ebp -.section __ex_table,"a" - .align 4 - .long 1b,syscall_fault -.previous - pushl %eax SAVE_ALL GET_THREAD_INFO(%ebp) cmpl $(nr_syscalls), %eax jae syscall_badsys - testb $_TIF_SYSCALL_TRACE,TI_FLAGS(%ebp) + testb $_TIF_SYSCALL_TRACE,TI_flags(%ebp) jnz syscall_trace_entry call *sys_call_table(,%eax,4) movl %eax,EAX(%esp) cli - movl TI_FLAGS(%ebp), %ecx + movl TI_flags(%ebp), %ecx testw $_TIF_ALLWORK_MASK, %cx jne syscall_exit_work + +#ifdef CONFIG_X86_SWITCH_PAGETABLES + + GET_THREAD_INFO(%ebp) + movl TI_virtual_stack(%ebp), %edx + movl TI_user_pgd(%ebp), %ecx + movl %esp, %ebx + andl $0x1fff, %ebx + orl %ebx, %edx +sysexit_ret_start_marker: + movl %edx, %esp + movl %ecx, %cr3 +#endif + /* + * only ebx is not restored by the userspace sysenter vsyscall + * code, it assumes it to be callee-saved. + */ + movl EBX(%esp), %ebx + /* if something modifies registers it must also disable sysexit */ + movl EIP(%esp), %edx movl OLDESP(%esp), %ecx + sti sysexit +#ifdef CONFIG_X86_SWITCH_PAGETABLES +sysexit_ret_end_marker: + nop +#endif # system call handler stub @@ -300,7 +417,7 @@ ENTRY(system_call) cmpl $(nr_syscalls), %eax jae syscall_badsys # system call tracing in operation - testb $_TIF_SYSCALL_TRACE,TI_FLAGS(%ebp) + testb $_TIF_SYSCALL_TRACE,TI_flags(%ebp) jnz syscall_trace_entry syscall_call: call *sys_call_table(,%eax,4) @@ -309,7 +426,7 @@ syscall_exit: cli # make sure we don't miss an interrupt # setting need_resched or sigpending # between sampling and the iret - movl TI_FLAGS(%ebp), %ecx + movl TI_flags(%ebp), %ecx testw $_TIF_ALLWORK_MASK, %cx # current->work jne syscall_exit_work restore_all: @@ -319,7 +436,7 @@ restore_all: testl $(VM_MASK | 3), %eax jz resume_kernelX # returning to kernel or vm86-space - cmpl $0,TI_PRE_COUNT(%ebx) # non-zero preempt_count ? + cmpl $0,TI_preempt_count(%ebp) # non-zero preempt_count ? jz resume_kernelX int $3 @@ -338,7 +455,7 @@ work_resched: cli # make sure we don't miss an interrupt # setting need_resched or sigpending # between sampling and the iret - movl TI_FLAGS(%ebp), %ecx + movl TI_flags(%ebp), %ecx andl $_TIF_WORK_MASK, %ecx # is there any work to be done other # than syscall tracing? jz restore_all @@ -353,6 +470,22 @@ work_notifysig: # deal with pending s # vm86-space xorl %edx, %edx call do_notify_resume + +#if CONFIG_X86_HIGH_ENTRY + /* + * Reload db7 if necessary: + */ + movl TI_flags(%ebp), %ecx + testb $_TIF_DB7, %cl + jnz work_db7 + + jmp restore_all + +work_db7: + movl TI_task(%ebp), %edx; + movl task_thread_db7(%edx), %edx; + movl %edx, %db7; +#endif jmp restore_all ALIGN @@ -408,7 +541,7 @@ syscall_badsys: */ .data ENTRY(interrupt) -.text +.previous vector=0 ENTRY(irq_entries_start) @@ -418,7 +551,7 @@ ENTRY(irq_entries_start) jmp common_interrupt .data .long 1b -.text +.previous vector=vector+1 .endr @@ -459,12 +592,17 @@ error_code: movl ES(%esp), %edi # get the function address movl %eax, ORIG_EAX(%esp) movl %ecx, ES(%esp) - movl %esp, %edx pushl %esi # push the error code - pushl %edx # push the pt_regs pointer movl $(__USER_DS), %edx movl %edx, %ds movl %edx, %es + +/* clobbers edx, ebx and ebp */ + __SWITCH_KERNELSPACE + + leal 4(%esp), %edx # prepare pt_regs + pushl %edx # push pt_regs + call *%edi addl $8, %esp jmp ret_from_exception @@ -555,7 +693,7 @@ nmi_stack_correct: pushl %edx call do_nmi addl $8, %esp - RESTORE_ALL + jmp restore_all nmi_stack_fixup: FIX_STACK(12,nmi_stack_correct, 1) @@ -632,6 +770,8 @@ ENTRY(spurious_interrupt_bug) pushl $do_spurious_interrupt_bug jmp error_code +.previous + .data ENTRY(sys_call_table) .long sys_restart_syscall /* 0 - old "setup()" system call, used for restarting */ diff -puN /dev/null arch/i386/kernel/entry_trampoline.c --- /dev/null 2002-08-30 16:31:37.000000000 -0700 +++ 25-akpm/arch/i386/kernel/entry_trampoline.c 2004-01-20 21:32:48.000000000 -0800 @@ -0,0 +1,75 @@ +/* + * linux/arch/i386/kernel/entry_trampoline.c + * + * (C) Copyright 2003 Ingo Molnar + * + * This file contains the needed support code for 4GB userspace + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +extern char __entry_tramp_start, __entry_tramp_end, __start___entry_text; + +void __init init_entry_mappings(void) +{ +#ifdef CONFIG_X86_HIGH_ENTRY + void *tramp; + + /* + * We need a high IDT and GDT for the 4G/4G split: + */ + trap_init_virtual_IDT(); + + __set_fixmap(FIX_ENTRY_TRAMPOLINE_0, __pa((unsigned long)&__entry_tramp_start), PAGE_KERNEL); + __set_fixmap(FIX_ENTRY_TRAMPOLINE_1, __pa((unsigned long)&__entry_tramp_start) + PAGE_SIZE, PAGE_KERNEL); + tramp = (void *)fix_to_virt(FIX_ENTRY_TRAMPOLINE_0); + + printk("mapped 4G/4G trampoline to %p.\n", tramp); + BUG_ON((void *)&__start___entry_text != tramp); + /* + * Virtual kernel stack: + */ + BUG_ON(__kmap_atomic_vaddr(KM_VSTACK0) & 8191); + BUG_ON(sizeof(struct desc_struct)*NR_CPUS*GDT_ENTRIES > 2*PAGE_SIZE); + BUG_ON((unsigned int)&__entry_tramp_end - (unsigned int)&__entry_tramp_start > 2*PAGE_SIZE); + + /* + * set up the initial thread's virtual stack related + * fields: + */ + current->thread.stack_page0 = virt_to_page((char *)current->thread_info); + current->thread.stack_page1 = virt_to_page((char *)current->thread_info + PAGE_SIZE); + current->thread_info->virtual_stack = (void *)__kmap_atomic_vaddr(KM_VSTACK0); + + __kunmap_atomic_type(KM_VSTACK0); + __kunmap_atomic_type(KM_VSTACK1); + __kmap_atomic(current->thread.stack_page0, KM_VSTACK0); + __kmap_atomic(current->thread.stack_page1, KM_VSTACK1); + +#endif + printk("current: %p\n", current); + printk("current->thread_info: %p\n", current->thread_info); + current->thread_info->real_stack = (void *)current->thread_info; + current->thread_info->user_pgd = NULL; + current->thread.esp0 = (unsigned long)current->thread_info->real_stack + THREAD_SIZE; +} + + + +void __init entry_trampoline_setup(void) +{ + /* + * old IRQ entries set up by the boot code will still hang + * around - they are a sign of hw trouble anyway, now they'll + * produce a double fault message. + */ + trap_init_virtual_GDT(); +} diff -puN arch/i386/kernel/head.S~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/head.S --- 25/arch/i386/kernel/head.S~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/head.S 2004-01-20 21:32:48.000000000 -0800 @@ -16,6 +16,7 @@ #include #include #include +#include #define OLD_CL_MAGIC_ADDR 0x90020 #define OLD_CL_MAGIC 0xA33F @@ -330,7 +331,7 @@ ENTRY(stack_start) /* This is the default interrupt "handler" :-) */ int_msg: - .asciz "Unknown interrupt\n" + .asciz "Unknown interrupt or fault at EIP %p %p %p\n" ALIGN ignore_int: cld @@ -342,9 +343,17 @@ ignore_int: movl $(__KERNEL_DS),%eax movl %eax,%ds movl %eax,%es + pushl 16(%esp) + pushl 24(%esp) + pushl 32(%esp) + pushl 40(%esp) pushl $int_msg call printk popl %eax + popl %eax + popl %eax + popl %eax + popl %eax popl %ds popl %es popl %edx @@ -377,23 +386,27 @@ cpu_gdt_descr: .fill NR_CPUS-1,8,0 # space for the other GDT descriptors /* - * This is initialized to create an identity-mapping at 0-8M (for bootup - * purposes) and another mapping of the 0-8M area at virtual address + * This is initialized to create an identity-mapping at 0-16M (for bootup + * purposes) and another mapping of the 0-16M area at virtual address * PAGE_OFFSET. */ .org 0x1000 ENTRY(swapper_pg_dir) .long 0x00102007 .long 0x00103007 - .fill BOOT_USER_PGD_PTRS-2,4,0 - /* default: 766 entries */ + .long 0x00104007 + .long 0x00105007 + .fill BOOT_USER_PGD_PTRS-4,4,0 + /* default: 764 entries */ .long 0x00102007 .long 0x00103007 - /* default: 254 entries */ - .fill BOOT_KERNEL_PGD_PTRS-2,4,0 + .long 0x00104007 + .long 0x00105007 + /* default: 252 entries */ + .fill BOOT_KERNEL_PGD_PTRS-4,4,0 /* - * The page tables are initialized to only 8MB here - the final page + * The page tables are initialized to only 16MB here - the final page * tables are set up later depending on memory size. */ .org 0x2000 @@ -402,15 +415,21 @@ ENTRY(pg0) .org 0x3000 ENTRY(pg1) +.org 0x4000 +ENTRY(pg2) + +.org 0x5000 +ENTRY(pg3) + /* * empty_zero_page must immediately follow the page tables ! (The * initialization loop counts until empty_zero_page) */ -.org 0x4000 +.org 0x6000 ENTRY(empty_zero_page) -.org 0x5000 +.org 0x7000 /* * Real beginning of normal "text" segment @@ -419,12 +438,12 @@ ENTRY(stext) ENTRY(_stext) /* - * This starts the data section. Note that the above is all - * in the text section because it has alignment requirements - * that we cannot fulfill any other way. + * This starts the data section. */ .data +.align PAGE_SIZE_asm + /* * The Global Descriptor Table contains 28 quadwords, per-CPU. */ @@ -439,7 +458,9 @@ ENTRY(boot_gdt_table) .quad 0x00cf9a000000ffff /* kernel 4GB code at 0x00000000 */ .quad 0x00cf92000000ffff /* kernel 4GB data at 0x00000000 */ #endif - .align L1_CACHE_BYTES + +.align PAGE_SIZE_asm + ENTRY(cpu_gdt_table) .quad 0x0000000000000000 /* NULL descriptor */ .quad 0x0000000000000000 /* 0x0b reserved */ diff -puN arch/i386/kernel/i386_ksyms.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/i386_ksyms.c --- 25/arch/i386/kernel/i386_ksyms.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/i386_ksyms.c 2004-01-20 21:32:48.000000000 -0800 @@ -97,7 +97,6 @@ EXPORT_SYMBOL_NOVERS(__down_failed_inter EXPORT_SYMBOL_NOVERS(__down_failed_trylock); EXPORT_SYMBOL_NOVERS(__up_wakeup); /* Networking helper routines. */ -EXPORT_SYMBOL(csum_partial_copy_generic); /* Delay loops */ EXPORT_SYMBOL(__ndelay); EXPORT_SYMBOL(__udelay); @@ -111,13 +110,17 @@ EXPORT_SYMBOL_NOVERS(__get_user_4); EXPORT_SYMBOL(strpbrk); EXPORT_SYMBOL(strstr); +#if !defined(CONFIG_X86_UACCESS_INDIRECT) EXPORT_SYMBOL(strncpy_from_user); -EXPORT_SYMBOL(__strncpy_from_user); +EXPORT_SYMBOL(__direct_strncpy_from_user); EXPORT_SYMBOL(clear_user); EXPORT_SYMBOL(__clear_user); EXPORT_SYMBOL(__copy_from_user_ll); EXPORT_SYMBOL(__copy_to_user_ll); EXPORT_SYMBOL(strnlen_user); +#else /* CONFIG_X86_UACCESS_INDIRECT */ +EXPORT_SYMBOL(direct_csum_partial_copy_generic); +#endif EXPORT_SYMBOL(dma_alloc_coherent); EXPORT_SYMBOL(dma_free_coherent); diff -puN arch/i386/kernel/i387.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/i387.c --- 25/arch/i386/kernel/i387.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/i387.c 2004-01-20 21:32:48.000000000 -0800 @@ -218,6 +218,7 @@ void set_fpu_mxcsr( struct task_struct * static int convert_fxsr_to_user( struct _fpstate __user *buf, struct i387_fxsave_struct *fxsave ) { + struct _fpreg tmp[8]; /* 80 bytes scratch area */ unsigned long env[7]; struct _fpreg __user *to; struct _fpxreg *from; @@ -234,23 +235,25 @@ static int convert_fxsr_to_user( struct if ( __copy_to_user( buf, env, 7 * sizeof(unsigned long) ) ) return 1; - to = &buf->_st[0]; + to = tmp; from = (struct _fpxreg *) &fxsave->st_space[0]; for ( i = 0 ; i < 8 ; i++, to++, from++ ) { unsigned long *t = (unsigned long *)to; unsigned long *f = (unsigned long *)from; - if (__put_user(*f, t) || - __put_user(*(f + 1), t + 1) || - __put_user(from->exponent, &to->exponent)) - return 1; + *t = *f; + *(t + 1) = *(f+1); + to->exponent = from->exponent; } + if (copy_to_user(buf->_st, tmp, sizeof(struct _fpreg [8]))) + return 1; return 0; } static int convert_fxsr_from_user( struct i387_fxsave_struct *fxsave, struct _fpstate __user *buf ) { + struct _fpreg tmp[8]; /* 80 bytes scratch area */ unsigned long env[7]; struct _fpxreg *to; struct _fpreg __user *from; @@ -258,6 +261,8 @@ static int convert_fxsr_from_user( struc if ( __copy_from_user( env, buf, 7 * sizeof(long) ) ) return 1; + if (copy_from_user(tmp, buf->_st, sizeof(struct _fpreg [8]))) + return 1; fxsave->cwd = (unsigned short)(env[0] & 0xffff); fxsave->swd = (unsigned short)(env[1] & 0xffff); @@ -269,15 +274,14 @@ static int convert_fxsr_from_user( struc fxsave->fos = env[6]; to = (struct _fpxreg *) &fxsave->st_space[0]; - from = &buf->_st[0]; + from = tmp; for ( i = 0 ; i < 8 ; i++, to++, from++ ) { unsigned long *t = (unsigned long *)to; unsigned long *f = (unsigned long *)from; - if (__get_user(*t, f) || - __get_user(*(t + 1), f + 1) || - __get_user(to->exponent, &from->exponent)) - return 1; + *t = *f; + *(t + 1) = *(f + 1); + to->exponent = from->exponent; } return 0; } diff -puN arch/i386/kernel/init_task.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/init_task.c --- 25/arch/i386/kernel/init_task.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/init_task.c 2004-01-20 21:32:48.000000000 -0800 @@ -26,7 +26,7 @@ EXPORT_SYMBOL(init_mm); */ union thread_union init_thread_union __attribute__((__section__(".data.init_task"))) = - { INIT_THREAD_INFO(init_task) }; + { INIT_THREAD_INFO(init_task, init_thread_union) }; /* * Initial task structure. @@ -44,5 +44,5 @@ EXPORT_SYMBOL(init_task); * section. Since TSS's are completely CPU-local, we want them * on exact cacheline boundaries, to eliminate cacheline ping-pong. */ -struct tss_struct init_tss[NR_CPUS] __cacheline_aligned = { [0 ... NR_CPUS-1] = INIT_TSS }; +struct tss_struct init_tss[NR_CPUS] __attribute__((__section__(".data.tss"))) = { [0 ... NR_CPUS-1] = INIT_TSS }; diff -puN arch/i386/kernel/ldt.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/ldt.c --- 25/arch/i386/kernel/ldt.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/ldt.c 2004-01-20 21:32:48.000000000 -0800 @@ -2,7 +2,7 @@ * linux/kernel/ldt.c * * Copyright (C) 1992 Krishna Balasubramanian and Linus Torvalds - * Copyright (C) 1999 Ingo Molnar + * Copyright (C) 1999, 2003 Ingo Molnar */ #include @@ -18,6 +18,8 @@ #include #include #include +#include +#include #ifdef CONFIG_SMP /* avoids "defined but not used" warnig */ static void flush_ldt(void *null) @@ -29,34 +31,31 @@ static void flush_ldt(void *null) static int alloc_ldt(mm_context_t *pc, int mincount, int reload) { - void *oldldt; - void *newldt; - int oldsize; + int oldsize, newsize, i; if (mincount <= pc->size) return 0; + /* + * LDT got larger - reallocate if necessary. + */ oldsize = pc->size; mincount = (mincount+511)&(~511); - if (mincount*LDT_ENTRY_SIZE > PAGE_SIZE) - newldt = vmalloc(mincount*LDT_ENTRY_SIZE); - else - newldt = kmalloc(mincount*LDT_ENTRY_SIZE, GFP_KERNEL); - - if (!newldt) - return -ENOMEM; - - if (oldsize) - memcpy(newldt, pc->ldt, oldsize*LDT_ENTRY_SIZE); - oldldt = pc->ldt; - memset(newldt+oldsize*LDT_ENTRY_SIZE, 0, (mincount-oldsize)*LDT_ENTRY_SIZE); - pc->ldt = newldt; - wmb(); + newsize = mincount*LDT_ENTRY_SIZE; + for (i = 0; i < newsize; i += PAGE_SIZE) { + int nr = i/PAGE_SIZE; + BUG_ON(i >= 64*1024); + if (!pc->ldt_pages[nr]) { + pc->ldt_pages[nr] = alloc_page(GFP_HIGHUSER); + if (!pc->ldt_pages[nr]) + return -ENOMEM; + clear_highpage(pc->ldt_pages[nr]); + } + } pc->size = mincount; - wmb(); - if (reload) { #ifdef CONFIG_SMP cpumask_t mask; + preempt_disable(); load_LDT(pc); mask = cpumask_of_cpu(smp_processor_id()); @@ -67,21 +66,20 @@ static int alloc_ldt(mm_context_t *pc, i load_LDT(pc); #endif } - if (oldsize) { - if (oldsize*LDT_ENTRY_SIZE > PAGE_SIZE) - vfree(oldldt); - else - kfree(oldldt); - } return 0; } static inline int copy_ldt(mm_context_t *new, mm_context_t *old) { - int err = alloc_ldt(new, old->size, 0); - if (err < 0) + int i, err, size = old->size, nr_pages = (size*LDT_ENTRY_SIZE + PAGE_SIZE-1)/PAGE_SIZE; + + err = alloc_ldt(new, size, 0); + if (err < 0) { + new->size = 0; return err; - memcpy(new->ldt, old->ldt, old->size*LDT_ENTRY_SIZE); + } + for (i = 0; i < nr_pages; i++) + copy_user_highpage(new->ldt_pages[i], old->ldt_pages[i], 0); return 0; } @@ -96,6 +94,7 @@ int init_new_context(struct task_struct init_MUTEX(&mm->context.sem); mm->context.size = 0; + memset(mm->context.ldt_pages, 0, sizeof(struct page *) * MAX_LDT_PAGES); old_mm = current->mm; if (old_mm && old_mm->context.size > 0) { down(&old_mm->context.sem); @@ -107,23 +106,21 @@ int init_new_context(struct task_struct /* * No need to lock the MM as we are the last user + * Do not touch the ldt register, we are already + * in the next thread. */ void destroy_context(struct mm_struct *mm) { - if (mm->context.size) { - if (mm == current->active_mm) - clear_LDT(); - if (mm->context.size*LDT_ENTRY_SIZE > PAGE_SIZE) - vfree(mm->context.ldt); - else - kfree(mm->context.ldt); - mm->context.size = 0; - } + int i, nr_pages = (mm->context.size*LDT_ENTRY_SIZE + PAGE_SIZE-1) / PAGE_SIZE; + + for (i = 0; i < nr_pages; i++) + __free_page(mm->context.ldt_pages[i]); + mm->context.size = 0; } static int read_ldt(void __user * ptr, unsigned long bytecount) { - int err; + int err, i; unsigned long size; struct mm_struct * mm = current->mm; @@ -138,8 +135,25 @@ static int read_ldt(void __user * ptr, u size = bytecount; err = 0; - if (copy_to_user(ptr, mm->context.ldt, size)) - err = -EFAULT; + /* + * This is necessary just in case we got here straight from a + * context-switch where the ptes were set but no tlb flush + * was done yet. We rather avoid doing a TLB flush in the + * context-switch path and do it here instead. + */ + __flush_tlb_global(); + + for (i = 0; i < size; i += PAGE_SIZE) { + int nr = i / PAGE_SIZE, bytes; + char *kaddr = kmap(mm->context.ldt_pages[nr]); + + bytes = size - i; + if (bytes > PAGE_SIZE) + bytes = PAGE_SIZE; + if (copy_to_user(ptr + i, kaddr, size - i)) + err = -EFAULT; + kunmap(mm->context.ldt_pages[nr]); + } up(&mm->context.sem); if (err < 0) return err; @@ -158,7 +172,7 @@ static int read_default_ldt(void __user err = 0; address = &default_ldt[0]; - size = 5*sizeof(struct desc_struct); + size = 5*LDT_ENTRY_SIZE; if (size > bytecount) size = bytecount; @@ -200,7 +214,15 @@ static int write_ldt(void __user * ptr, goto out_unlock; } - lp = (__u32 *) ((ldt_info.entry_number << 3) + (char *) mm->context.ldt); + /* + * No rescheduling allowed from this point to the install. + * + * We do a TLB flush for the same reason as in the read_ldt() path. + */ + preempt_disable(); + __flush_tlb_global(); + lp = (__u32 *) ((ldt_info.entry_number << 3) + + (char *) __kmap_atomic_vaddr(KM_LDT_PAGE0)); /* Allow LDTs to be cleared by the user. */ if (ldt_info.base_addr == 0 && ldt_info.limit == 0) { @@ -221,6 +243,7 @@ install: *lp = entry_1; *(lp+1) = entry_2; error = 0; + preempt_enable(); out_unlock: up(&mm->context.sem); @@ -248,3 +271,26 @@ asmlinkage int sys_modify_ldt(int func, } return ret; } + +/* + * load one particular LDT into the current CPU + */ +void load_LDT_nolock(mm_context_t *pc, int cpu) +{ + struct page **pages = pc->ldt_pages; + int count = pc->size; + int nr_pages, i; + + if (likely(!count)) { + pages = &default_ldt_page; + count = 5; + } + nr_pages = (count*LDT_ENTRY_SIZE + PAGE_SIZE-1) / PAGE_SIZE; + + for (i = 0; i < nr_pages; i++) { + __kunmap_atomic_type(KM_LDT_PAGE0 - i); + __kmap_atomic(pages[i], KM_LDT_PAGE0 - i); + } + set_ldt_desc(cpu, (void *)__kmap_atomic_vaddr(KM_LDT_PAGE0), count); + load_LDT_desc(); +} diff -puN arch/i386/kernel/Makefile~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/Makefile --- 25/arch/i386/kernel/Makefile~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/Makefile 2004-01-20 21:32:48.000000000 -0800 @@ -7,7 +7,7 @@ extra-y := head.o init_task.o vmlinux.ld obj-y := process.o semaphore.o signal.o entry.o traps.o irq.o vm86.o \ ptrace.o i8259.o ioport.o ldt.o setup.o time.o sys_i386.o \ pci-dma.o i386_ksyms.o i387.o dmi_scan.o bootflag.o \ - doublefault.o + doublefault.o entry_trampoline.o obj-y += cpu/ obj-y += timers/ diff -puN arch/i386/kernel/mpparse.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/mpparse.c --- 25/arch/i386/kernel/mpparse.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/mpparse.c 2004-01-20 21:32:48.000000000 -0800 @@ -668,7 +668,7 @@ void __init get_smp_config (void) * Read the physical hardware table. Anything here will * override the defaults. */ - if (!smp_read_mpc((void *)mpf->mpf_physptr)) { + if (!smp_read_mpc((void *)phys_to_virt(mpf->mpf_physptr))) { smp_found_config = 0; printk(KERN_ERR "BIOS bug, MP table errors detected!...\n"); printk(KERN_ERR "... disabling SMP support. (tell your hw vendor)\n"); diff -puN arch/i386/kernel/process.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/process.c --- 25/arch/i386/kernel/process.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/process.c 2004-01-20 21:32:48.000000000 -0800 @@ -47,6 +47,7 @@ #include #include #include +#include #ifdef CONFIG_MATH_EMULATION #include #endif @@ -302,6 +303,9 @@ void flush_thread(void) struct task_struct *tsk = current; memset(tsk->thread.debugreg, 0, sizeof(unsigned long)*8); +#ifdef CONFIG_X86_HIGH_ENTRY + clear_thread_flag(TIF_DB7); +#endif memset(tsk->thread.tls_array, 0, sizeof(tsk->thread.tls_array)); /* * Forget coprocessor state.. @@ -315,9 +319,8 @@ void release_thread(struct task_struct * if (dead_task->mm) { // temporary debugging check if (dead_task->mm->context.size) { - printk("WARNING: dead process %8s still has LDT? <%p/%d>\n", + printk("WARNING: dead process %8s still has LDT? <%d>\n", dead_task->comm, - dead_task->mm->context.ldt, dead_task->mm->context.size); BUG(); } @@ -352,7 +355,17 @@ int copy_thread(int nr, unsigned long cl p->thread.esp = (unsigned long) childregs; p->thread.esp0 = (unsigned long) (childregs+1); + /* + * get the two stack pages, for the virtual stack. + * + * IMPORTANT: this code relies on the fact that the task + * structure is an 8K aligned piece of physical memory. + */ + p->thread.stack_page0 = virt_to_page((unsigned long)p->thread_info); + p->thread.stack_page1 = virt_to_page((unsigned long)p->thread_info + PAGE_SIZE); + p->thread.eip = (unsigned long) ret_from_fork; + p->thread_info->real_stack = p->thread_info; savesegment(fs,p->thread.fs); savesegment(gs,p->thread.gs); @@ -504,10 +517,41 @@ struct task_struct * __switch_to(struct __unlazy_fpu(prev_p); +#ifdef CONFIG_X86_HIGH_ENTRY + /* + * Set the ptes of the virtual stack. (NOTE: a one-page TLB flush is + * needed because otherwise NMIs could interrupt the + * user-return code with a virtual stack and stale TLBs.) + */ + __kunmap_atomic_type(KM_VSTACK0); + __kunmap_atomic_type(KM_VSTACK1); + __kmap_atomic(next->stack_page0, KM_VSTACK0); + __kmap_atomic(next->stack_page1, KM_VSTACK1); + + /* + * NOTE: here we rely on the task being the stack as well + */ + next_p->thread_info->virtual_stack = + (void *)__kmap_atomic_vaddr(KM_VSTACK0); + +#if defined(CONFIG_PREEMPT) && defined(CONFIG_SMP) + /* + * If next was preempted on entry from userspace to kernel, + * and now it's on a different cpu, we need to adjust %esp. + * This assumes that entry.S does not copy %esp while on the + * virtual stack (with interrupts enabled): which is so, + * except within __SWITCH_KERNELSPACE itself. + */ + if (unlikely(next->esp >= TASK_SIZE)) { + next->esp &= THREAD_SIZE - 1; + next->esp |= (unsigned long) next_p->thread_info->virtual_stack; + } +#endif +#endif /* * Reload esp0, LDT and the page table pointer: */ - load_esp0(tss, next); + load_virtual_esp0(tss, next_p); /* * Load the per-thread Thread-Local Storage descriptor. diff -puN arch/i386/kernel/reboot.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/reboot.c --- 25/arch/i386/kernel/reboot.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/reboot.c 2004-01-20 21:32:48.000000000 -0800 @@ -155,12 +155,11 @@ void machine_real_restart(unsigned char CMOS_WRITE(0x00, 0x8f); spin_unlock_irqrestore(&rtc_lock, flags); - /* Remap the kernel at virtual address zero, as well as offset zero - from the kernel segment. This assumes the kernel segment starts at - virtual address PAGE_OFFSET. */ - - memcpy (swapper_pg_dir, swapper_pg_dir + USER_PGD_PTRS, - sizeof (swapper_pg_dir [0]) * KERNEL_PGD_PTRS); + /* + * Remap the first 16 MB of RAM (which includes the kernel image) + * at virtual address zero: + */ + setup_identity_mappings(swapper_pg_dir, 0, 16*1024*1024); /* * Use `swapper_pg_dir' as our page directory. diff -puN arch/i386/kernel/signal.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/signal.c --- 25/arch/i386/kernel/signal.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/signal.c 2004-01-20 21:32:48.000000000 -0800 @@ -128,28 +128,29 @@ sys_sigaltstack(const stack_t __user *us */ static int -restore_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc, int *peax) +restore_sigcontext(struct pt_regs *regs, + struct sigcontext __user *__sc, int *peax) { - unsigned int err = 0; + struct sigcontext scratch; /* 88 bytes of scratch area */ /* Always make any pending restarted system calls return -EINTR */ current_thread_info()->restart_block.fn = do_no_restart_syscall; -#define COPY(x) err |= __get_user(regs->x, &sc->x) + if (copy_from_user(&scratch, __sc, sizeof(scratch))) + return -EFAULT; + +#define COPY(x) regs->x = scratch.x #define COPY_SEG(seg) \ - { unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ + { unsigned short tmp = scratch.seg; \ regs->x##seg = tmp; } #define COPY_SEG_STRICT(seg) \ - { unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ + { unsigned short tmp = scratch.seg; \ regs->x##seg = tmp|3; } #define GET_SEG(seg) \ - { unsigned short tmp; \ - err |= __get_user(tmp, &sc->seg); \ + { unsigned short tmp = scratch.seg; \ loadsegment(seg,tmp); } GET_SEG(gs); @@ -168,27 +169,23 @@ restore_sigcontext(struct pt_regs *regs, COPY_SEG_STRICT(ss); { - unsigned int tmpflags; - err |= __get_user(tmpflags, &sc->eflags); + unsigned int tmpflags = scratch.eflags; regs->eflags = (regs->eflags & ~0x40DD5) | (tmpflags & 0x40DD5); regs->orig_eax = -1; /* disable syscall checks */ } { - struct _fpstate __user * buf; - err |= __get_user(buf, &sc->fpstate); + struct _fpstate * buf = scratch.fpstate; if (buf) { if (verify_area(VERIFY_READ, buf, sizeof(*buf))) - goto badframe; - err |= restore_i387(buf); + return -EFAULT; + if (restore_i387(buf)) + return -EFAULT; } } - err |= __get_user(*peax, &sc->eax); - return err; - -badframe: - return 1; + *peax = scratch.eax; + return 0; } asmlinkage int sys_sigreturn(unsigned long __unused) @@ -266,46 +263,47 @@ badframe: */ static int -setup_sigcontext(struct sigcontext __user *sc, struct _fpstate __user *fpstate, +setup_sigcontext(struct sigcontext __user *__sc, struct _fpstate __user *fpstate, struct pt_regs *regs, unsigned long mask) { - int tmp, err = 0; + struct sigcontext sc; /* 88 bytes of scratch area */ + int tmp; tmp = 0; __asm__("movl %%gs,%0" : "=r"(tmp): "0"(tmp)); - err |= __put_user(tmp, (unsigned int *)&sc->gs); + *(unsigned int *)&sc.gs = tmp; __asm__("movl %%fs,%0" : "=r"(tmp): "0"(tmp)); - err |= __put_user(tmp, (unsigned int *)&sc->fs); - - err |= __put_user(regs->xes, (unsigned int *)&sc->es); - err |= __put_user(regs->xds, (unsigned int *)&sc->ds); - err |= __put_user(regs->edi, &sc->edi); - err |= __put_user(regs->esi, &sc->esi); - err |= __put_user(regs->ebp, &sc->ebp); - err |= __put_user(regs->esp, &sc->esp); - err |= __put_user(regs->ebx, &sc->ebx); - err |= __put_user(regs->edx, &sc->edx); - err |= __put_user(regs->ecx, &sc->ecx); - err |= __put_user(regs->eax, &sc->eax); - err |= __put_user(current->thread.trap_no, &sc->trapno); - err |= __put_user(current->thread.error_code, &sc->err); - err |= __put_user(regs->eip, &sc->eip); - err |= __put_user(regs->xcs, (unsigned int *)&sc->cs); - err |= __put_user(regs->eflags, &sc->eflags); - err |= __put_user(regs->esp, &sc->esp_at_signal); - err |= __put_user(regs->xss, (unsigned int *)&sc->ss); + *(unsigned int *)&sc.fs = tmp; + *(unsigned int *)&sc.es = regs->xes; + *(unsigned int *)&sc.ds = regs->xds; + sc.edi = regs->edi; + sc.esi = regs->esi; + sc.ebp = regs->ebp; + sc.esp = regs->esp; + sc.ebx = regs->ebx; + sc.edx = regs->edx; + sc.ecx = regs->ecx; + sc.eax = regs->eax; + sc.trapno = current->thread.trap_no; + sc.err = current->thread.error_code; + sc.eip = regs->eip; + *(unsigned int *)&sc.cs = regs->xcs; + sc.eflags = regs->eflags; + sc.esp_at_signal = regs->esp; + *(unsigned int *)&sc.ss = regs->xss; tmp = save_i387(fpstate); if (tmp < 0) - err = 1; - else - err |= __put_user(tmp ? fpstate : NULL, &sc->fpstate); + return 1; + sc.fpstate = tmp ? fpstate : NULL; /* non-iBCS2 extensions.. */ - err |= __put_user(mask, &sc->oldmask); - err |= __put_user(current->thread.cr2, &sc->cr2); + sc.oldmask = mask; + sc.cr2 = current->thread.cr2; - return err; + if (copy_to_user(__sc, &sc, sizeof(sc))) + return 1; + return 0; } /* @@ -443,7 +441,7 @@ static void setup_rt_frame(int sig, stru /* Create the ucontext. */ err |= __put_user(0, &frame->uc.uc_flags); err |= __put_user(0, &frame->uc.uc_link); - err |= __put_user(current->sas_ss_sp, &frame->uc.uc_stack.ss_sp); + err |= __put_user(current->sas_ss_sp, (unsigned long *)&frame->uc.uc_stack.ss_sp); err |= __put_user(sas_ss_flags(regs->esp), &frame->uc.uc_stack.ss_flags); err |= __put_user(current->sas_ss_size, &frame->uc.uc_stack.ss_size); diff -puN arch/i386/kernel/smp.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/smp.c --- 25/arch/i386/kernel/smp.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/smp.c 2004-01-20 21:32:48.000000000 -0800 @@ -327,10 +327,12 @@ asmlinkage void smp_invalidate_interrupt if (flush_mm == cpu_tlbstate[cpu].active_mm) { if (cpu_tlbstate[cpu].state == TLBSTATE_OK) { +#ifndef CONFIG_X86_SWITCH_PAGETABLES if (flush_va == FLUSH_ALL) local_flush_tlb(); else __flush_tlb_one(flush_va); +#endif } else leave_mm(cpu); } @@ -396,21 +398,6 @@ static void flush_tlb_others(cpumask_t c spin_unlock(&tlbstate_lock); } -void flush_tlb_current_task(void) -{ - struct mm_struct *mm = current->mm; - cpumask_t cpu_mask; - - preempt_disable(); - cpu_mask = mm->cpu_vm_mask; - cpu_clear(smp_processor_id(), cpu_mask); - - local_flush_tlb(); - if (!cpus_empty(cpu_mask)) - flush_tlb_others(cpu_mask, mm, FLUSH_ALL); - preempt_enable(); -} - void flush_tlb_mm (struct mm_struct * mm) { cpumask_t cpu_mask; @@ -442,7 +429,10 @@ void flush_tlb_page(struct vm_area_struc if (current->active_mm == mm) { if(current->mm) - __flush_tlb_one(va); +#ifndef CONFIG_X86_SWITCH_PAGETABLES + __flush_tlb_one(va) +#endif + ; else leave_mm(smp_processor_id()); } diff -puN arch/i386/kernel/sysenter.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/sysenter.c --- 25/arch/i386/kernel/sysenter.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/sysenter.c 2004-01-20 21:32:48.000000000 -0800 @@ -18,13 +18,18 @@ #include #include #include +#include extern asmlinkage void sysenter_entry(void); void enable_sep_cpu(void *info) { int cpu = get_cpu(); +#ifdef CONFIG_X86_HIGH_ENTRY + struct tss_struct *tss = (struct tss_struct *) __fix_to_virt(FIX_TSS_0) + cpu; +#else struct tss_struct *tss = init_tss + cpu; +#endif tss->ss1 = __KERNEL_CS; tss->esp1 = sizeof(struct tss_struct) + (unsigned long) tss; diff -puN arch/i386/kernel/traps.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/traps.c --- 25/arch/i386/kernel/traps.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/traps.c 2004-01-20 21:32:48.000000000 -0800 @@ -54,12 +54,8 @@ #include "mach_traps.h" -asmlinkage int system_call(void); -asmlinkage void lcall7(void); -asmlinkage void lcall27(void); - -struct desc_struct default_ldt[] = { { 0, 0 }, { 0, 0 }, { 0, 0 }, - { 0, 0 }, { 0, 0 } }; +struct desc_struct default_ldt[] __attribute__((__section__(".data.default_ldt"))) = { { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 } }; +struct page *default_ldt_page; /* Do we ignore FPU interrupts ? */ char ignore_fpu_irq = 0; @@ -108,6 +104,7 @@ static void set_intr_usr_gate(unsigned i */ void breakpoint(void) { + init_entry_mappings(); set_intr_usr_gate(3,&int3); /* disable ints on trap */ set_intr_gate(1,&debug); set_intr_gate(14,&page_fault); @@ -238,9 +235,11 @@ void show_registers(struct pt_regs *regs eip = (u8 *)regs->eip - 43; for (i = 0; i < 64; i++, eip++) { - unsigned char c; + unsigned char c = 0xff; + + if ((user_mode(regs) && get_user(c, eip)) || + (!user_mode(regs) && __direct_get_user(c, eip))) { - if (eip < (u8 *)PAGE_OFFSET || __get_user(c, eip)) { printk(" Bad EIP value."); break; } @@ -602,10 +601,18 @@ asmlinkage void do_debug(struct pt_regs if (regs->eflags & X86_EFLAGS_IF) local_irq_enable(); - /* Mask out spurious debug traps due to lazy DR7 setting */ + /* + * Mask out spurious debug traps due to lazy DR7 setting or + * due to 4G/4G kernel mode: + */ if (condition & (DR_TRAP0|DR_TRAP1|DR_TRAP2|DR_TRAP3)) { if (!tsk->thread.debugreg[7]) goto clear_dr7; + if (!user_mode(regs)) { + // restore upon return-to-userspace: + set_thread_flag(TIF_DB7); + goto clear_dr7; + } } if (regs->eflags & VM_MASK) @@ -869,19 +876,53 @@ asmlinkage void math_emulate(long arg) #endif /* CONFIG_MATH_EMULATION */ -#ifdef CONFIG_X86_F00F_BUG -void __init trap_init_f00f_bug(void) +void __init trap_init_virtual_IDT(void) { - __set_fixmap(FIX_F00F_IDT, __pa(&idt_table), PAGE_KERNEL_RO); - /* - * Update the IDT descriptor and reload the IDT so that - * it uses the read-only mapped virtual address. + * "idt" is magic - it overlaps the idt_descr + * variable so that updating idt will automatically + * update the idt descriptor.. */ - idt_descr.address = fix_to_virt(FIX_F00F_IDT); + __set_fixmap(FIX_IDT, __pa(&idt_table), PAGE_KERNEL_RO); + idt_descr.address = __fix_to_virt(FIX_IDT); + __asm__ __volatile__("lidt %0" : : "m" (idt_descr)); } + +void __init trap_init_virtual_GDT(void) +{ + int cpu = smp_processor_id(); + struct Xgt_desc_struct *gdt_desc = cpu_gdt_descr + cpu; + struct Xgt_desc_struct tmp_desc = {0, 0}; + struct tss_struct * t; + + __asm__ __volatile__("sgdt %0": "=m" (tmp_desc): :"memory"); + +#ifdef CONFIG_X86_HIGH_ENTRY + if (!cpu) { + __set_fixmap(FIX_GDT_0, __pa(cpu_gdt_table), PAGE_KERNEL); + __set_fixmap(FIX_GDT_1, __pa(cpu_gdt_table) + PAGE_SIZE, PAGE_KERNEL); + __set_fixmap(FIX_TSS_0, __pa(init_tss), PAGE_KERNEL); + __set_fixmap(FIX_TSS_1, __pa(init_tss) + 1*PAGE_SIZE, PAGE_KERNEL); + __set_fixmap(FIX_TSS_2, __pa(init_tss) + 2*PAGE_SIZE, PAGE_KERNEL); + __set_fixmap(FIX_TSS_3, __pa(init_tss) + 3*PAGE_SIZE, PAGE_KERNEL); + } + + gdt_desc->address = __fix_to_virt(FIX_GDT_0) + sizeof(cpu_gdt_table[0]) * cpu; +#else + gdt_desc->address = (unsigned long)cpu_gdt_table[cpu]; #endif + __asm__ __volatile__("lgdt %0": "=m" (*gdt_desc)); + +#ifdef CONFIG_X86_HIGH_ENTRY + t = (struct tss_struct *) __fix_to_virt(FIX_TSS_0) + cpu; +#else + t = init_tss + cpu; +#endif + set_tss_desc(cpu, t); + cpu_gdt_table[cpu][GDT_ENTRY_TSS].b &= 0xfffffdff; + load_TR_desc(); +} #define _set_gate(gate_addr,type,dpl,addr,seg) \ do { \ @@ -908,17 +949,17 @@ void set_intr_gate(unsigned int n, void _set_gate(idt_table+n,14,0,addr,__KERNEL_CS); } -static void __init set_trap_gate(unsigned int n, void *addr) +void __init set_trap_gate(unsigned int n, void *addr) { _set_gate(idt_table+n,15,0,addr,__KERNEL_CS); } -static void __init set_system_gate(unsigned int n, void *addr) +void __init set_system_gate(unsigned int n, void *addr) { _set_gate(idt_table+n,15,3,addr,__KERNEL_CS); } -static void __init set_call_gate(void *a, void *addr) +void __init set_call_gate(void *a, void *addr) { _set_gate(a,12,3,addr,__KERNEL_CS); } @@ -946,6 +987,7 @@ void __init trap_init(void) #ifdef CONFIG_X86_LOCAL_APIC init_apic_mappings(); #endif + init_entry_mappings(); set_trap_gate(0,÷_error); set_intr_gate(1,&debug); diff -puN arch/i386/kernel/vm86.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/vm86.c --- 25/arch/i386/kernel/vm86.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/vm86.c 2004-01-20 21:32:48.000000000 -0800 @@ -125,7 +125,7 @@ struct pt_regs * save_v86_state(struct k tss = init_tss + get_cpu(); current->thread.esp0 = current->thread.saved_esp0; current->thread.sysenter_cs = __KERNEL_CS; - load_esp0(tss, ¤t->thread); + load_virtual_esp0(tss, current); current->thread.saved_esp0 = 0; put_cpu(); @@ -305,7 +305,7 @@ static void do_sys_vm86(struct kernel_vm tsk->thread.esp0 = (unsigned long) &info->VM86_TSS_ESP0; if (cpu_has_sep) tsk->thread.sysenter_cs = 0; - load_esp0(tss, &tsk->thread); + load_virtual_esp0(tss, tsk); put_cpu(); tsk->thread.screen_bitmap = info->screen_bitmap; diff -puN arch/i386/kernel/vmlinux.lds.S~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/vmlinux.lds.S --- 25/arch/i386/kernel/vmlinux.lds.S~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/vmlinux.lds.S 2004-01-20 21:32:48.000000000 -0800 @@ -3,6 +3,9 @@ */ #include +#include +#include +#include OUTPUT_FORMAT("elf32-i386", "elf32-i386", "elf32-i386") OUTPUT_ARCH(i386) @@ -10,7 +13,7 @@ ENTRY(startup_32) jiffies = jiffies_64; SECTIONS { - . = 0xC0000000 + 0x100000; + . = __PAGE_OFFSET + 0x100000; /* read-only */ _text = .; /* Text and read-only data */ .text : { @@ -19,6 +22,19 @@ SECTIONS *(.gnu.warning) } = 0x9090 +#ifdef CONFIG_X86_4G + . = ALIGN(PAGE_SIZE_asm); + __entry_tramp_start = .; + . = FIX_ENTRY_TRAMPOLINE_0_addr; + __start___entry_text = .; + .entry.text : AT (__entry_tramp_start) { *(.entry.text) } + __entry_tramp_end = __entry_tramp_start + SIZEOF(.entry.text); + . = __entry_tramp_end; + . = ALIGN(PAGE_SIZE_asm); +#else + .entry.text : { *(.entry.text) } +#endif + _etext = .; /* End of text section */ . = ALIGN(16); /* Exception table */ @@ -34,15 +50,12 @@ SECTIONS CONSTRUCTORS } - . = ALIGN(4096); + . = ALIGN(PAGE_SIZE_asm); __nosave_begin = .; .data_nosave : { *(.data.nosave) } - . = ALIGN(4096); + . = ALIGN(PAGE_SIZE_asm); __nosave_end = .; - . = ALIGN(4096); - .data.page_aligned : { *(.data.idt) } - . = ALIGN(32); .data.cacheline_aligned : { *(.data.cacheline_aligned) } @@ -52,7 +65,7 @@ SECTIONS .data.init_task : { *(.data.init_task) } /* will be freed after init */ - . = ALIGN(4096); /* Init code and data */ + . = ALIGN(PAGE_SIZE_asm); /* Init code and data */ __init_begin = .; .init.text : { _sinittext = .; @@ -91,7 +104,7 @@ SECTIONS from .altinstructions and .eh_frame */ .exit.text : { *(.exit.text) } .exit.data : { *(.exit.data) } - . = ALIGN(4096); + . = ALIGN(PAGE_SIZE_asm); __initramfs_start = .; .init.ramfs : { *(.init.ramfs) } __initramfs_end = .; @@ -99,10 +112,22 @@ SECTIONS __per_cpu_start = .; .data.percpu : { *(.data.percpu) } __per_cpu_end = .; - . = ALIGN(4096); + . = ALIGN(PAGE_SIZE_asm); __init_end = .; /* freed after init ends here */ - + + . = ALIGN(PAGE_SIZE_asm); + .data.page_aligned_tss : { *(.data.tss) } + + . = ALIGN(PAGE_SIZE_asm); + .data.page_aligned_default_ldt : { *(.data.default_ldt) } + + . = ALIGN(PAGE_SIZE_asm); + .data.page_aligned_idt : { *(.data.idt) } + + . = ALIGN(PAGE_SIZE_asm); + .data.page_aligned_gdt : { *(.data.gdt) } + __bss_start = .; /* BSS */ .bss : { *(.bss) } __bss_stop = .; @@ -122,4 +147,6 @@ SECTIONS .stab.index 0 : { *(.stab.index) } .stab.indexstr 0 : { *(.stab.indexstr) } .comment 0 : { *(.comment) } + + } diff -puN arch/i386/kernel/vsyscall.lds~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/vsyscall.lds --- 25/arch/i386/kernel/vsyscall.lds~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/vsyscall.lds 2004-01-20 21:32:48.000000000 -0800 @@ -5,7 +5,7 @@ */ /* This must match . */ -VSYSCALL_BASE = 0xffffe000; +VSYSCALL_BASE = 0xffffd000; SECTIONS { diff -puN arch/i386/kernel/vsyscall-sysenter.S~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/vsyscall-sysenter.S --- 25/arch/i386/kernel/vsyscall-sysenter.S~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/vsyscall-sysenter.S 2004-01-20 21:32:48.000000000 -0800 @@ -7,6 +7,11 @@ .type __kernel_vsyscall,@function __kernel_vsyscall: .LSTART_vsyscall: + cmpl $192, %eax + jne 1f + int $0x80 + ret +1: push %ecx .Lpush_ecx: push %edx diff -puN arch/i386/lib/checksum.S~4g-2.6.0-test2-mm2-A5 arch/i386/lib/checksum.S --- 25/arch/i386/lib/checksum.S~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/lib/checksum.S 2004-01-20 21:32:48.000000000 -0800 @@ -280,14 +280,14 @@ unsigned int csum_partial_copy_generic ( .previous .align 4 -.globl csum_partial_copy_generic +.globl direct_csum_partial_copy_generic #ifndef CONFIG_X86_USE_PPRO_CHECKSUM #define ARGBASE 16 #define FP 12 -csum_partial_copy_generic: +direct_csum_partial_copy_generic: subl $4,%esp pushl %edi pushl %esi @@ -422,7 +422,7 @@ DST( movb %cl, (%edi) ) #define ARGBASE 12 -csum_partial_copy_generic: +direct_csum_partial_copy_generic: pushl %ebx pushl %edi pushl %esi diff -puN arch/i386/lib/getuser.S~4g-2.6.0-test2-mm2-A5 arch/i386/lib/getuser.S --- 25/arch/i386/lib/getuser.S~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/lib/getuser.S 2004-01-20 21:32:48.000000000 -0800 @@ -9,6 +9,7 @@ * return value. */ #include +#include /* @@ -28,7 +29,7 @@ .globl __get_user_1 __get_user_1: GET_THREAD_INFO(%edx) - cmpl TI_ADDR_LIMIT(%edx),%eax + cmpl TI_addr_limit(%edx),%eax jae bad_get_user 1: movzbl (%eax),%edx xorl %eax,%eax @@ -40,7 +41,7 @@ __get_user_2: addl $1,%eax jc bad_get_user GET_THREAD_INFO(%edx) - cmpl TI_ADDR_LIMIT(%edx),%eax + cmpl TI_addr_limit(%edx),%eax jae bad_get_user 2: movzwl -1(%eax),%edx xorl %eax,%eax @@ -52,7 +53,7 @@ __get_user_4: addl $3,%eax jc bad_get_user GET_THREAD_INFO(%edx) - cmpl TI_ADDR_LIMIT(%edx),%eax + cmpl TI_addr_limit(%edx),%eax jae bad_get_user 3: movl -3(%eax),%edx xorl %eax,%eax diff -puN arch/i386/lib/usercopy.c~4g-2.6.0-test2-mm2-A5 arch/i386/lib/usercopy.c --- 25/arch/i386/lib/usercopy.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/lib/usercopy.c 2004-01-20 21:32:48.000000000 -0800 @@ -76,7 +76,7 @@ do { \ * and returns @count. */ long -__strncpy_from_user(char *dst, const char __user *src, long count) +__direct_strncpy_from_user(char *dst, const char __user *src, long count) { long res; __do_strncpy_from_user(dst, src, count, res); @@ -102,7 +102,7 @@ __strncpy_from_user(char *dst, const cha * and returns @count. */ long -strncpy_from_user(char *dst, const char __user *src, long count) +direct_strncpy_from_user(char *dst, const char __user *src, long count) { long res = -EFAULT; if (access_ok(VERIFY_READ, src, 1)) @@ -147,7 +147,7 @@ do { \ * On success, this will be zero. */ unsigned long -clear_user(void __user *to, unsigned long n) +direct_clear_user(void __user *to, unsigned long n) { might_sleep(); if (access_ok(VERIFY_WRITE, to, n)) @@ -167,7 +167,7 @@ clear_user(void __user *to, unsigned lon * On success, this will be zero. */ unsigned long -__clear_user(void __user *to, unsigned long n) +__direct_clear_user(void __user *to, unsigned long n) { __do_clear_user(to, n); return n; @@ -184,7 +184,7 @@ __clear_user(void __user *to, unsigned l * On exception, returns 0. * If the string is too long, returns a value greater than @n. */ -long strnlen_user(const char __user *s, long n) +long direct_strnlen_user(const char __user *s, long n) { unsigned long mask = -__addr_ok(s); unsigned long res, tmp; @@ -575,3 +575,4 @@ unsigned long __copy_from_user_ll(void * n = __copy_user_zeroing_intel(to, (const void *) from, n); return n; } + diff -puN arch/i386/math-emu/fpu_system.h~4g-2.6.0-test2-mm2-A5 arch/i386/math-emu/fpu_system.h --- 25/arch/i386/math-emu/fpu_system.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/math-emu/fpu_system.h 2004-01-20 21:32:48.000000000 -0800 @@ -15,6 +15,7 @@ #include #include #include +#include /* This sets the pointer FPU_info to point to the argument part of the stack frame of math_emulate() */ @@ -22,7 +23,7 @@ /* s is always from a cpu register, and the cpu does bounds checking * during register load --> no further bounds checks needed */ -#define LDT_DESCRIPTOR(s) (((struct desc_struct *)current->mm->context.ldt)[(s) >> 3]) +#define LDT_DESCRIPTOR(s) (((struct desc_struct *)__kmap_atomic_vaddr(KM_LDT_PAGE0))[(s) >> 3]) #define SEG_D_SIZE(x) ((x).b & (3 << 21)) #define SEG_G_BIT(x) ((x).b & (1 << 23)) #define SEG_GRANULARITY(x) (((x).b & (1 << 23)) ? 4096 : 1) diff -puN arch/i386/mm/fault.c~4g-2.6.0-test2-mm2-A5 arch/i386/mm/fault.c --- 25/arch/i386/mm/fault.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/mm/fault.c 2004-01-20 21:32:48.000000000 -0800 @@ -27,6 +27,7 @@ #include #include #include +#include extern void die(const char *,struct pt_regs *,long); @@ -104,8 +105,17 @@ static inline unsigned long get_segment_ if (seg & (1<<2)) { /* Must lock the LDT while reading it. */ down(¤t->mm->context.sem); +#if 1 + /* horrible hack for 4/4 disabled kernels. + I'm not quite sure what the TLB flush is good for, + it's mindlessly copied from the read_ldt code */ + __flush_tlb_global(); + desc = kmap(current->mm->context.ldt_pages[(seg&~7)/PAGE_SIZE]); + desc = (void *)desc + ((seg & ~7) % PAGE_SIZE); +#else desc = current->mm->context.ldt; desc = (void *)desc + (seg & ~7); +#endif } else { /* Must disable preemption while reading the GDT. */ desc = (u32 *)&cpu_gdt_table[get_cpu()]; @@ -118,6 +128,9 @@ static inline unsigned long get_segment_ (desc[1] & 0xff000000); if (seg & (1<<2)) { +#if 1 + kunmap((void *)((unsigned long)desc & PAGE_MASK)); +#endif up(¤t->mm->context.sem); } else put_cpu(); @@ -243,6 +256,19 @@ asmlinkage void do_page_fault(struct pt_ * (error_code & 4) == 0, and that the fault was not a * protection error (error_code & 1) == 0. */ +#ifdef CONFIG_X86_4G + /* + * On 4/4 all kernels faults are either bugs, vmalloc or prefetch + */ + if (unlikely((regs->xcs & 3) == 0)) { + if (error_code & 3) + goto bad_area_nosemaphore; + + /* If it's vm86 fall through */ + if (!(regs->eflags & VM_MASK)) + goto vmalloc_fault; + } +#else if (unlikely(address >= TASK_SIZE)) { if (!(error_code & 5)) goto vmalloc_fault; @@ -252,6 +278,7 @@ asmlinkage void do_page_fault(struct pt_ */ goto bad_area_nosemaphore; } +#endif mm = tsk->mm; diff -puN arch/i386/mm/init.c~4g-2.6.0-test2-mm2-A5 arch/i386/mm/init.c --- 25/arch/i386/mm/init.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/mm/init.c 2004-01-20 21:32:48.000000000 -0800 @@ -40,125 +40,13 @@ #include #include #include +#include DEFINE_PER_CPU(struct mmu_gather, mmu_gathers); unsigned long highstart_pfn, highend_pfn; static int do_test_wp_bit(void); -/* - * Creates a middle page table and puts a pointer to it in the - * given global directory entry. This only returns the gd entry - * in non-PAE compilation mode, since the middle layer is folded. - */ -static pmd_t * __init one_md_table_init(pgd_t *pgd) -{ - pmd_t *pmd_table; - -#ifdef CONFIG_X86_PAE - pmd_table = (pmd_t *) alloc_bootmem_low_pages(PAGE_SIZE); - set_pgd(pgd, __pgd(__pa(pmd_table) | _PAGE_PRESENT)); - if (pmd_table != pmd_offset(pgd, 0)) - BUG(); -#else - pmd_table = pmd_offset(pgd, 0); -#endif - - return pmd_table; -} - -/* - * Create a page table and place a pointer to it in a middle page - * directory entry. - */ -static pte_t * __init one_page_table_init(pmd_t *pmd) -{ - if (pmd_none(*pmd)) { - pte_t *page_table = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); - set_pmd(pmd, __pmd(__pa(page_table) | _PAGE_TABLE)); - if (page_table != pte_offset_kernel(pmd, 0)) - BUG(); - - return page_table; - } - - return pte_offset_kernel(pmd, 0); -} - -/* - * This function initializes a certain range of kernel virtual memory - * with new bootmem page tables, everywhere page tables are missing in - * the given range. - */ - -/* - * NOTE: The pagetables are allocated contiguous on the physical space - * so we can cache the place of the first one and move around without - * checking the pgd every time. - */ -static void __init page_table_range_init (unsigned long start, unsigned long end, pgd_t *pgd_base) -{ - pgd_t *pgd; - pmd_t *pmd; - int pgd_idx, pmd_idx; - unsigned long vaddr; - - vaddr = start; - pgd_idx = pgd_index(vaddr); - pmd_idx = pmd_index(vaddr); - pgd = pgd_base + pgd_idx; - - for ( ; (pgd_idx < PTRS_PER_PGD) && (vaddr != end); pgd++, pgd_idx++) { - if (pgd_none(*pgd)) - one_md_table_init(pgd); - - pmd = pmd_offset(pgd, vaddr); - for (; (pmd_idx < PTRS_PER_PMD) && (vaddr != end); pmd++, pmd_idx++) { - if (pmd_none(*pmd)) - one_page_table_init(pmd); - - vaddr += PMD_SIZE; - } - pmd_idx = 0; - } -} - -/* - * This maps the physical memory to kernel virtual address space, a total - * of max_low_pfn pages, by creating page tables starting from address - * PAGE_OFFSET. - */ -static void __init kernel_physical_mapping_init(pgd_t *pgd_base) -{ - unsigned long pfn; - pgd_t *pgd; - pmd_t *pmd; - pte_t *pte; - int pgd_idx, pmd_idx, pte_ofs; - - pgd_idx = pgd_index(PAGE_OFFSET); - pgd = pgd_base + pgd_idx; - pfn = 0; - - for (; pgd_idx < PTRS_PER_PGD; pgd++, pgd_idx++) { - pmd = one_md_table_init(pgd); - if (pfn >= max_low_pfn) - continue; - for (pmd_idx = 0; pmd_idx < PTRS_PER_PMD && pfn < max_low_pfn; pmd++, pmd_idx++) { - /* Map with big pages if possible, otherwise create normal page tables. */ - if (cpu_has_pse) { - set_pmd(pmd, pfn_pmd(pfn, PAGE_KERNEL_LARGE)); - pfn += PTRS_PER_PTE; - } else { - pte = one_page_table_init(pmd); - - for (pte_ofs = 0; pte_ofs < PTRS_PER_PTE && pfn < max_low_pfn; pte++, pfn++, pte_ofs++) - set_pte(pte, pfn_pte(pfn, PAGE_KERNEL)); - } - } - } -} - static inline int page_kills_ppro(unsigned long pagenr) { if (pagenr >= 0x70000 && pagenr <= 0x7003F) @@ -206,11 +94,8 @@ static inline int page_is_ram(unsigned l return 0; } -#ifdef CONFIG_HIGHMEM pte_t *kmap_pte; -pgprot_t kmap_prot; -EXPORT_SYMBOL(kmap_prot); EXPORT_SYMBOL(kmap_pte); #define kmap_get_fixmap_pte(vaddr) \ @@ -218,29 +103,7 @@ EXPORT_SYMBOL(kmap_pte); void __init kmap_init(void) { - unsigned long kmap_vstart; - - /* cache the first kmap pte */ - kmap_vstart = __fix_to_virt(FIX_KMAP_BEGIN); - kmap_pte = kmap_get_fixmap_pte(kmap_vstart); - - kmap_prot = PAGE_KERNEL; -} - -void __init permanent_kmaps_init(pgd_t *pgd_base) -{ - pgd_t *pgd; - pmd_t *pmd; - pte_t *pte; - unsigned long vaddr; - - vaddr = PKMAP_BASE; - page_table_range_init(vaddr, vaddr + PAGE_SIZE*LAST_PKMAP, pgd_base); - - pgd = swapper_pg_dir + pgd_index(vaddr); - pmd = pmd_offset(pgd, vaddr); - pte = pte_offset_kernel(pmd, vaddr); - pkmap_page_table = pte; + kmap_pte = kmap_get_fixmap_pte(__fix_to_virt(FIX_KMAP_BEGIN)); } void __init one_highpage_init(struct page *page, int pfn, int bad_ppro) @@ -255,6 +118,8 @@ void __init one_highpage_init(struct pag SetPageReserved(page); } +#ifdef CONFIG_HIGHMEM + #ifndef CONFIG_DISCONTIGMEM void __init set_highmem_pages_init(int bad_ppro) { @@ -266,12 +131,9 @@ void __init set_highmem_pages_init(int b #else extern void set_highmem_pages_init(int); #endif /* !CONFIG_DISCONTIGMEM */ - #else -#define kmap_init() do { } while (0) -#define permanent_kmaps_init(pgd_base) do { } while (0) -#define set_highmem_pages_init(bad_ppro) do { } while (0) -#endif /* CONFIG_HIGHMEM */ +# define set_highmem_pages_init(bad_ppro) do { } while (0) +#endif unsigned long __PAGE_KERNEL = _PAGE_KERNEL; @@ -281,30 +143,125 @@ unsigned long __PAGE_KERNEL = _PAGE_KERN extern void __init remap_numa_kva(void); #endif -static void __init pagetable_init (void) +static __init void prepare_pagetables(pgd_t *pgd_base, unsigned long address) +{ + pgd_t *pgd; + pmd_t *pmd; + pte_t *pte; + + pgd = pgd_base + pgd_index(address); + pmd = pmd_offset(pgd, address); + if (!pmd_present(*pmd)) { + pte = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); + set_pmd(pmd, __pmd(_PAGE_TABLE + __pa(pte))); + } +} + +static void __init fixrange_init (unsigned long start, unsigned long end, pgd_t *pgd_base) +{ + unsigned long vaddr; + + for (vaddr = start; vaddr != end; vaddr += PAGE_SIZE) + prepare_pagetables(pgd_base, vaddr); +} + +void setup_identity_mappings(pgd_t *pgd_base, unsigned long start, unsigned long end) { unsigned long vaddr; - pgd_t *pgd_base = swapper_pg_dir; + pgd_t *pgd; + int i, j, k; + pmd_t *pmd; + pte_t *pte, *pte_base; + + pgd = pgd_base; + + for (i = 0; i < PTRS_PER_PGD; pgd++, i++) { + vaddr = i*PGDIR_SIZE; + if (end && (vaddr >= end)) + break; + pmd = pmd_offset(pgd, 0); + for (j = 0; j < PTRS_PER_PMD; pmd++, j++) { + vaddr = i*PGDIR_SIZE + j*PMD_SIZE; + if (end && (vaddr >= end)) + break; + if (vaddr < start) + continue; + if (cpu_has_pse) { + unsigned long __pe; + set_in_cr4(X86_CR4_PSE); + boot_cpu_data.wp_works_ok = 1; + __pe = _KERNPG_TABLE + _PAGE_PSE + vaddr - start; + /* Make it "global" too if supported */ + if (cpu_has_pge) { + set_in_cr4(X86_CR4_PGE); +#if !defined(CONFIG_X86_SWITCH_PAGETABLES) + __pe += _PAGE_GLOBAL; + __PAGE_KERNEL |= _PAGE_GLOBAL; +#endif + } + set_pmd(pmd, __pmd(__pe)); + continue; + } + if (!pmd_present(*pmd)) + pte_base = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); + else + pte_base = (pte_t *) page_address(pmd_page(*pmd)); + pte = pte_base; + for (k = 0; k < PTRS_PER_PTE; pte++, k++) { + vaddr = i*PGDIR_SIZE + j*PMD_SIZE + k*PAGE_SIZE; + if (end && (vaddr >= end)) + break; + if (vaddr < start) + continue; + *pte = mk_pte_phys(vaddr-start, PAGE_KERNEL); + } + set_pmd(pmd, __pmd(_KERNPG_TABLE + __pa(pte_base))); + } + } +} + +static void __init pagetable_init (void) +{ + unsigned long vaddr, end; + pgd_t *pgd_base; #ifdef CONFIG_X86_PAE int i; - /* Init entries of the first-level page table to the zero page */ - for (i = 0; i < PTRS_PER_PGD; i++) - set_pgd(pgd_base + i, __pgd(__pa(empty_zero_page) | _PAGE_PRESENT)); #endif - /* Enable PSE if available */ - if (cpu_has_pse) { - set_in_cr4(X86_CR4_PSE); - } + /* + * This can be zero as well - no problem, in that case we exit + * the loops anyway due to the PTRS_PER_* conditions. + */ + end = (unsigned long)__va(max_low_pfn*PAGE_SIZE); - /* Enable PGE if available */ - if (cpu_has_pge) { - set_in_cr4(X86_CR4_PGE); - __PAGE_KERNEL |= _PAGE_GLOBAL; + pgd_base = swapper_pg_dir; +#ifdef CONFIG_X86_PAE + /* + * It causes too many problems if there's no proper pmd set up + * for all 4 entries of the PGD - so we allocate all of them. + * PAE systems will not miss this extra 4-8K anyway ... + */ + for (i = 0; i < PTRS_PER_PGD; i++) { + pmd_t *pmd = (pmd_t *) alloc_bootmem_low_pages(PAGE_SIZE); + set_pgd(pgd_base + i, __pgd(__pa(pmd) + 0x1)); } +#endif + /* + * Set up lowmem-sized identity mappings at PAGE_OFFSET: + */ + setup_identity_mappings(pgd_base, PAGE_OFFSET, end); - kernel_physical_mapping_init(pgd_base); + /* + * Add flat-mode identity-mappings - SMP needs it when + * starting up on an AP from real-mode. (In the non-PAE + * case we already have these mappings through head.S.) + * All user-space mappings are explicitly cleared after + * SMP startup. + */ +#if CONFIG_SMP && CONFIG_X86_PAE + setup_identity_mappings(pgd_base, 0, 16*1024*1024); +#endif remap_numa_kva(); /* @@ -312,38 +269,64 @@ static void __init pagetable_init (void) * created - mappings will be set by set_fixmap(): */ vaddr = __fix_to_virt(__end_of_fixed_addresses - 1) & PMD_MASK; - page_table_range_init(vaddr, 0, pgd_base); + fixrange_init(vaddr, 0, pgd_base); - permanent_kmaps_init(pgd_base); +#if CONFIG_HIGHMEM + { + pgd_t *pgd; + pmd_t *pmd; + pte_t *pte; -#ifdef CONFIG_X86_PAE - /* - * Add low memory identity-mappings - SMP needs it when - * starting up on an AP from real-mode. In the non-PAE - * case we already have these mappings through head.S. - * All user-space mappings are explicitly cleared after - * SMP startup. - */ - pgd_base[0] = pgd_base[USER_PTRS_PER_PGD]; + /* + * Permanent kmaps: + */ + vaddr = PKMAP_BASE; + fixrange_init(vaddr, vaddr + PAGE_SIZE*LAST_PKMAP, pgd_base); + + pgd = swapper_pg_dir + pgd_index(vaddr); + pmd = pmd_offset(pgd, vaddr); + pte = pte_offset_kernel(pmd, vaddr); + pkmap_page_table = pte; + } #endif } -void zap_low_mappings (void) +/* + * Clear kernel pagetables in a PMD_SIZE-aligned range. + */ +static void clear_mappings(pgd_t *pgd_base, unsigned long start, unsigned long end) { - int i; + unsigned long vaddr; + pgd_t *pgd; + pmd_t *pmd; + int i, j; + + pgd = pgd_base; + + for (i = 0; i < PTRS_PER_PGD; pgd++, i++) { + vaddr = i*PGDIR_SIZE; + if (end && (vaddr >= end)) + break; + pmd = pmd_offset(pgd, 0); + for (j = 0; j < PTRS_PER_PMD; pmd++, j++) { + vaddr = i*PGDIR_SIZE + j*PMD_SIZE; + if (end && (vaddr >= end)) + break; + if (vaddr < start) + continue; + pmd_clear(pmd); + } + } + flush_tlb_all(); +} + +void __init zap_low_mappings(void) +{ + printk("zapping low mappings.\n"); /* * Zap initial low-memory mappings. - * - * Note that "pgd_clear()" doesn't do it for - * us, because pgd_clear() is a no-op on i386. */ - for (i = 0; i < USER_PTRS_PER_PGD; i++) -#ifdef CONFIG_X86_PAE - set_pgd(swapper_pg_dir+i, __pgd(1 + __pa(empty_zero_page))); -#else - set_pgd(swapper_pg_dir+i, __pgd(0)); -#endif - flush_tlb_all(); + clear_mappings(swapper_pg_dir, 0, 16*1024*1024); } #ifndef CONFIG_DISCONTIGMEM @@ -393,7 +376,15 @@ void __init paging_init(void) set_in_cr4(X86_CR4_PAE); #endif __flush_tlb_all(); - + /* + * Subtle. SMP is doing it's boot stuff late (because it has to + * fork idle threads) - but it also needs low mappings for the + * protected-mode entry to work. We zap these entries only after + * the WP-bit has been tested. + */ +#ifndef CONFIG_SMP + zap_low_mappings(); +#endif kmap_init(); zone_sizes_init(); } @@ -515,22 +506,18 @@ void __init mem_init(void) if (boot_cpu_data.wp_works_ok < 0) test_wp_bit(); - /* - * Subtle. SMP is doing it's boot stuff late (because it has to - * fork idle threads) - but it also needs low mappings for the - * protected-mode entry to work. We zap these entries only after - * the WP-bit has been tested. - */ -#ifndef CONFIG_SMP - zap_low_mappings(); -#endif + entry_trampoline_setup(); + default_ldt_page = virt_to_page(default_ldt); + load_LDT(&init_mm.context); } -kmem_cache_t *pgd_cache; -kmem_cache_t *pmd_cache; +kmem_cache_t *pgd_cache, *pmd_cache, *kpmd_cache; void __init pgtable_cache_init(void) { + void (*ctor)(void *, kmem_cache_t *, unsigned long); + void (*dtor)(void *, kmem_cache_t *, unsigned long); + if (PTRS_PER_PMD > 1) { pmd_cache = kmem_cache_create("pmd", PTRS_PER_PMD*sizeof(pmd_t), @@ -540,13 +527,36 @@ void __init pgtable_cache_init(void) NULL); if (!pmd_cache) panic("pgtable_cache_init(): cannot create pmd cache"); + + if (TASK_SIZE > PAGE_OFFSET) { + kpmd_cache = kmem_cache_create("kpmd", + PTRS_PER_PMD*sizeof(pmd_t), + 0, + SLAB_HWCACHE_ALIGN | SLAB_MUST_HWCACHE_ALIGN, + kpmd_ctor, + NULL); + if (!kpmd_cache) + panic("pgtable_cache_init(): " + "cannot create kpmd cache"); + } } + + if (PTRS_PER_PMD == 1 || TASK_SIZE <= PAGE_OFFSET) + ctor = pgd_ctor; + else + ctor = NULL; + + if (PTRS_PER_PMD == 1 && TASK_SIZE <= PAGE_OFFSET) + dtor = pgd_dtor; + else + dtor = NULL; + pgd_cache = kmem_cache_create("pgd", PTRS_PER_PGD*sizeof(pgd_t), 0, SLAB_HWCACHE_ALIGN | SLAB_MUST_HWCACHE_ALIGN, - pgd_ctor, - PTRS_PER_PMD == 1 ? pgd_dtor : NULL); + ctor, + dtor); if (!pgd_cache) panic("pgtable_cache_init(): Cannot create pgd cache"); } diff -puN arch/i386/mm/pgtable.c~4g-2.6.0-test2-mm2-A5 arch/i386/mm/pgtable.c --- 25/arch/i386/mm/pgtable.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/mm/pgtable.c 2004-01-20 21:32:48.000000000 -0800 @@ -21,6 +21,7 @@ #include #include #include +#include void show_mem(void) { @@ -157,11 +158,20 @@ void pmd_ctor(void *pmd, kmem_cache_t *c memset(pmd, 0, PTRS_PER_PMD*sizeof(pmd_t)); } +void kpmd_ctor(void *__pmd, kmem_cache_t *cache, unsigned long flags) +{ + pmd_t *kpmd, *pmd; + kpmd = pmd_offset(&swapper_pg_dir[PTRS_PER_PGD-1], + (PTRS_PER_PMD - NR_SHARED_PMDS)*PMD_SIZE); + pmd = (pmd_t *)__pmd + (PTRS_PER_PMD - NR_SHARED_PMDS); + + memset(__pmd, 0, (PTRS_PER_PMD - NR_SHARED_PMDS)*sizeof(pmd_t)); + memcpy(pmd, kpmd, NR_SHARED_PMDS*sizeof(pmd_t)); +} + /* - * List of all pgd's needed for non-PAE so it can invalidate entries - * in both cached and uncached pgd's; not needed for PAE since the - * kernel pmd is shared. If PAE were not to share the pmd a similar - * tactic would be needed. This is essentially codepath-based locking + * List of all pgd's needed so it can invalidate entries in both cached + * and uncached pgd's. This is essentially codepath-based locking * against pageattr.c; it is the unique case in which a valid change * of kernel pagetables can't be lazily synchronized by vmalloc faults. * vmalloc faults work because attached pagetables are never freed. @@ -170,30 +180,60 @@ void pmd_ctor(void *pmd, kmem_cache_t *c * could be used. The locking scheme was chosen on the basis of * manfred's recommendations and having no core impact whatsoever. * -- wli + * + * The entire issue goes away when XKVA is configured. */ spinlock_t pgd_lock = SPIN_LOCK_UNLOCKED; LIST_HEAD(pgd_list); -void pgd_ctor(void *pgd, kmem_cache_t *cache, unsigned long unused) +/* + * This is not that hard to figure out. + * (a) PTRS_PER_PMD == 1 means non-PAE. + * (b) PTRS_PER_PMD > 1 means PAE. + * (c) TASK_SIZE > PAGE_OFFSET means XKVA. + * (d) TASK_SIZE <= PAGE_OFFSET means non-XKVA. + * + * Do *NOT* back out the preconstruction like the patch I'm cleaning + * up after this very instant did, or at all, for that matter. + * This is never called when PTRS_PER_PMD > 1 && TASK_SIZE > PAGE_OFFSET. + * -- wli + */ +void pgd_ctor(void *__pgd, kmem_cache_t *cache, unsigned long unused) { + pgd_t *pgd = (pgd_t *)__pgd; unsigned long flags; - if (PTRS_PER_PMD == 1) - spin_lock_irqsave(&pgd_lock, flags); + if (PTRS_PER_PMD == 1) { + if (TASK_SIZE <= PAGE_OFFSET) + spin_lock_irqsave(&pgd_lock, flags); + else + memcpy(&pgd[PTRS_PER_PGD - NR_SHARED_PMDS], + &swapper_pg_dir[PTRS_PER_PGD - NR_SHARED_PMDS], + NR_SHARED_PMDS * sizeof(pgd_t)); + } - memcpy((pgd_t *)pgd + USER_PTRS_PER_PGD, - swapper_pg_dir + USER_PTRS_PER_PGD, - (PTRS_PER_PGD - USER_PTRS_PER_PGD) * sizeof(pgd_t)); + if (TASK_SIZE <= PAGE_OFFSET) + memcpy(pgd + USER_PTRS_PER_PGD, + swapper_pg_dir + USER_PTRS_PER_PGD, + (PTRS_PER_PGD - USER_PTRS_PER_PGD) * sizeof(pgd_t)); if (PTRS_PER_PMD > 1) return; - list_add(&virt_to_page(pgd)->lru, &pgd_list); - spin_unlock_irqrestore(&pgd_lock, flags); - memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t)); + if (TASK_SIZE > PAGE_OFFSET) + memset(pgd, 0, (PTRS_PER_PGD - NR_SHARED_PMDS)*sizeof(pgd_t)); + else { + list_add(&virt_to_page(pgd)->lru, &pgd_list); + spin_unlock_irqrestore(&pgd_lock, flags); + memset(pgd, 0, USER_PTRS_PER_PGD*sizeof(pgd_t)); + } } -/* never called when PTRS_PER_PMD > 1 */ +/* + * Never called when PTRS_PER_PMD > 1 || TASK_SIZE > PAGE_OFFSET + * for with PAE we would list_del() multiple times, and for non-PAE + * with XKVA all the AGP pgd shootdown code is unnecessary. + */ void pgd_dtor(void *pgd, kmem_cache_t *cache, unsigned long unused) { unsigned long flags; /* can be called from interrupt context */ @@ -203,6 +243,12 @@ void pgd_dtor(void *pgd, kmem_cache_t *c spin_unlock_irqrestore(&pgd_lock, flags); } +/* + * See the comments above pgd_ctor() wrt. preconstruction. + * Do *NOT* memcpy() here. If you do, you back out important + * anti- cache pollution code. + * + */ pgd_t *pgd_alloc(struct mm_struct *mm) { int i; @@ -211,15 +257,33 @@ pgd_t *pgd_alloc(struct mm_struct *mm) if (PTRS_PER_PMD == 1 || !pgd) return pgd; + /* + * In the 4G userspace case alias the top 16 MB virtual + * memory range into the user mappings as well (these + * include the trampoline and CPU data structures). + */ for (i = 0; i < USER_PTRS_PER_PGD; ++i) { - pmd_t *pmd = kmem_cache_alloc(pmd_cache, GFP_KERNEL); + kmem_cache_t *cache; + pmd_t *pmd; + + if (TASK_SIZE > PAGE_OFFSET && i == USER_PTRS_PER_PGD - 1) + cache = kpmd_cache; + else + cache = pmd_cache; + + pmd = kmem_cache_alloc(cache, GFP_KERNEL); if (!pmd) goto out_oom; set_pgd(&pgd[i], __pgd(1 + __pa((u64)((u32)pmd)))); } - return pgd; + return pgd; out_oom: + /* + * we don't have to handle the kpmd_cache here, since it's the + * last allocation, and has either nothing to free or when it + * succeeds the whole operation succeeds. + */ for (i--; i >= 0; i--) kmem_cache_free(pmd_cache, (void *)__va(pgd_val(pgd[i])-1)); kmem_cache_free(pgd_cache, pgd); @@ -230,10 +294,29 @@ void pgd_free(pgd_t *pgd) { int i; - /* in the PAE case user pgd entries are overwritten before usage */ - if (PTRS_PER_PMD > 1) - for (i = 0; i < USER_PTRS_PER_PGD; ++i) - kmem_cache_free(pmd_cache, (void *)__va(pgd_val(pgd[i])-1)); /* in the non-PAE case, clear_page_tables() clears user pgd entries */ + if (PTRS_PER_PMD == 1) + goto out_free; + + /* in the PAE case user pgd entries are overwritten before usage */ + for (i = 0; i < USER_PTRS_PER_PGD; ++i) { + kmem_cache_t *cache; + pmd_t *pmd = __va(pgd_val(pgd[i]) - 1); + + /* + * only userspace pmd's are cleared for us + * by mm/memory.c; it's a slab cache invariant + * that we must separate the kernel pmd slab + * all times, else we'll have bad pmd's. + */ + if (TASK_SIZE > PAGE_OFFSET && i == USER_PTRS_PER_PGD - 1) + cache = kpmd_cache; + else + cache = pmd_cache; + + kmem_cache_free(cache, pmd); + } +out_free: kmem_cache_free(pgd_cache, pgd); } + diff -puN drivers/block/scsi_ioctl.c~4g-2.6.0-test2-mm2-A5 drivers/block/scsi_ioctl.c --- 25/drivers/block/scsi_ioctl.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/drivers/block/scsi_ioctl.c 2004-01-20 21:32:48.000000000 -0800 @@ -312,7 +312,7 @@ static int sg_scsi_ioctl(request_queue_t return -EFAULT; if (in_len > PAGE_SIZE || out_len > PAGE_SIZE) return -EINVAL; - if (get_user(opcode, sic->data)) + if (get_user(opcode, (int *)sic->data)) return -EFAULT; bytes = max(in_len, out_len); diff -puN /dev/null include/asm-i386/atomic_kmap.h --- /dev/null 2002-08-30 16:31:37.000000000 -0700 +++ 25-akpm/include/asm-i386/atomic_kmap.h 2004-01-20 21:32:48.000000000 -0800 @@ -0,0 +1,95 @@ +/* + * atomic_kmap.h: temporary virtual kernel memory mappings + * + * Copyright (C) 2003 Ingo Molnar + */ + +#ifndef _ASM_ATOMIC_KMAP_H +#define _ASM_ATOMIC_KMAP_H + +#ifdef __KERNEL__ + +#include +#include + +#ifdef CONFIG_DEBUG_HIGHMEM +#define HIGHMEM_DEBUG 1 +#else +#define HIGHMEM_DEBUG 0 +#endif + +extern pte_t *kmap_pte; +#define kmap_prot PAGE_KERNEL + +#define PKMAP_BASE (0xff000000UL) +#define NR_SHARED_PMDS ((0xffffffff-PKMAP_BASE+1)/PMD_SIZE) + +static inline unsigned long __kmap_atomic_vaddr(enum km_type type) +{ + enum fixed_addresses idx; + + idx = type + KM_TYPE_NR*smp_processor_id(); + return __fix_to_virt(FIX_KMAP_BEGIN + idx); +} + +static inline void *__kmap_atomic_noflush(struct page *page, enum km_type type) +{ + enum fixed_addresses idx; + unsigned long vaddr; + + idx = type + KM_TYPE_NR*smp_processor_id(); + vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx); + /* + * NOTE: entries that rely on some secondary TLB-flush + * effect must not be global: + */ + set_pte(kmap_pte-idx, mk_pte(page, PAGE_KERNEL)); + + return (void*) vaddr; +} + +static inline void *__kmap_atomic(struct page *page, enum km_type type) +{ + enum fixed_addresses idx; + unsigned long vaddr; + + idx = type + KM_TYPE_NR*smp_processor_id(); + vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx); +#if HIGHMEM_DEBUG + BUG_ON(!pte_none(*(kmap_pte-idx))); +#else + /* + * Performance optimization - do not flush if the new + * pte is the same as the old one: + */ + if (pte_val(*(kmap_pte-idx)) == pte_val(mk_pte(page, kmap_prot))) + return (void *) vaddr; +#endif + set_pte(kmap_pte-idx, mk_pte(page, kmap_prot)); + __flush_tlb_one(vaddr); + + return (void*) vaddr; +} + +static inline void __kunmap_atomic(void *kvaddr, enum km_type type) +{ +#if HIGHMEM_DEBUG + unsigned long vaddr = (unsigned long) kvaddr & PAGE_MASK; + enum fixed_addresses idx = type + KM_TYPE_NR*smp_processor_id(); + + BUG_ON(vaddr != __fix_to_virt(FIX_KMAP_BEGIN+idx)); + /* + * force other mappings to Oops if they'll try to access + * this pte without first remap it + */ + pte_clear(kmap_pte-idx); + __flush_tlb_one(vaddr); +#endif +} + +#define __kunmap_atomic_type(type) \ + __kunmap_atomic((void *)__kmap_atomic_vaddr(type), (type)) + +#endif /* __KERNEL__ */ + +#endif /* _ASM_ATOMIC_KMAP_H */ diff -puN include/asm-i386/checksum.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/checksum.h --- 25/include/asm-i386/checksum.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/checksum.h 2004-01-20 21:32:48.000000000 -0800 @@ -25,7 +25,7 @@ asmlinkage unsigned int csum_partial(con * better 64-bit) boundary */ -asmlinkage unsigned int csum_partial_copy_generic( const char *src, char *dst, int len, int sum, +asmlinkage unsigned int direct_csum_partial_copy_generic( const char *src, char *dst, int len, int sum, int *src_err_ptr, int *dst_err_ptr); /* @@ -39,14 +39,19 @@ static __inline__ unsigned int csum_partial_copy_nocheck ( const char *src, char *dst, int len, int sum) { - return csum_partial_copy_generic ( src, dst, len, sum, NULL, NULL); + /* + * The direct function is OK for kernel-space => kernel-space copies: + */ + return direct_csum_partial_copy_generic ( src, dst, len, sum, NULL, NULL); } static __inline__ unsigned int csum_partial_copy_from_user ( const char *src, char *dst, int len, int sum, int *err_ptr) { - return csum_partial_copy_generic ( src, dst, len, sum, err_ptr, NULL); + if (copy_from_user(dst, src, len)) + *err_ptr = -EFAULT; + return csum_partial(dst, len, sum); } /* @@ -172,11 +177,26 @@ static __inline__ unsigned short int csu * Copy and checksum to user */ #define HAVE_CSUM_COPY_USER -static __inline__ unsigned int csum_and_copy_to_user(const char *src, char *dst, +static __inline__ unsigned int direct_csum_and_copy_to_user(const char *src, char *dst, int len, int sum, int *err_ptr) { if (access_ok(VERIFY_WRITE, dst, len)) - return csum_partial_copy_generic(src, dst, len, sum, NULL, err_ptr); + return direct_csum_partial_copy_generic(src, dst, len, sum, NULL, err_ptr); + + if (len) + *err_ptr = -EFAULT; + + return -1; /* invalid checksum */ +} + +static __inline__ unsigned int csum_and_copy_to_user(const char *src, char *dst, + int len, int sum, int *err_ptr) +{ + if (access_ok(VERIFY_WRITE, dst, len)) { + if (copy_to_user(dst, src, len)) + *err_ptr = -EFAULT; + return csum_partial(src, len, sum); + } if (len) *err_ptr = -EFAULT; diff -puN include/asm-i386/desc.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/desc.h --- 25/include/asm-i386/desc.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/desc.h 2004-01-20 21:32:48.000000000 -0800 @@ -21,6 +21,13 @@ struct Xgt_desc_struct { extern struct Xgt_desc_struct idt_descr, cpu_gdt_descr[NR_CPUS]; +extern void trap_init_virtual_IDT(void); +extern void trap_init_virtual_GDT(void); + +asmlinkage int system_call(void); +asmlinkage void lcall7(void); +asmlinkage void lcall27(void); + #define load_TR_desc() __asm__ __volatile__("ltr %%ax"::"a" (GDT_ENTRY_TSS*8)) #define load_LDT_desc() __asm__ __volatile__("lldt %%ax"::"a" (GDT_ENTRY_LDT*8)) @@ -30,6 +37,7 @@ extern struct Xgt_desc_struct idt_descr, */ extern struct desc_struct default_ldt[]; extern void set_intr_gate(unsigned int irq, void * addr); +extern void set_trap_gate(unsigned int n, void *addr); #define _set_tssldt_desc(n,addr,limit,type) \ __asm__ __volatile__ ("movw %w3,0(%2)\n\t" \ @@ -90,31 +98,8 @@ static inline void load_TLS(struct threa #undef C } -static inline void clear_LDT(void) -{ - int cpu = get_cpu(); - - set_ldt_desc(cpu, &default_ldt[0], 5); - load_LDT_desc(); - put_cpu(); -} - -/* - * load one particular LDT into the current CPU - */ -static inline void load_LDT_nolock(mm_context_t *pc, int cpu) -{ - void *segments = pc->ldt; - int count = pc->size; - - if (likely(!count)) { - segments = &default_ldt[0]; - count = 5; - } - - set_ldt_desc(cpu, segments, count); - load_LDT_desc(); -} +extern struct page *default_ldt_page; +extern void load_LDT_nolock(mm_context_t *pc, int cpu); static inline void load_LDT(mm_context_t *pc) { @@ -123,6 +108,6 @@ static inline void load_LDT(mm_context_t put_cpu(); } -#endif /* !__ASSEMBLY__ */ +#endif /* !__ASSEMBLY__ */ #endif diff -puN include/asm-i386/fixmap.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/fixmap.h --- 25/include/asm-i386/fixmap.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/fixmap.h 2004-01-20 21:32:48.000000000 -0800 @@ -18,17 +18,15 @@ #include #include #include -#ifdef CONFIG_HIGHMEM #include #include -#endif /* * Here we define all the compile-time 'special' virtual * addresses. The point is to have a constant address at * compile time, but to set the physical address only - * in the boot process. We allocate these special addresses - * from the end of virtual memory (0xfffff000) backwards. + * in the boot process. We allocate these special addresses + * from the end of virtual memory (0xffffe000) backwards. * Also this lets us do fail-safe vmalloc(), we * can guarantee that these special addresses and * vmalloc()-ed addresses never overlap. @@ -41,11 +39,20 @@ * TLB entries of such buffers will not be flushed across * task switches. */ + +/* + * on UP currently we will have no trace of the fixmap mechanizm, + * no page table allocations, etc. This might change in the + * future, say framebuffers for the console driver(s) could be + * fix-mapped? + */ enum fixed_addresses { FIX_HOLE, FIX_VSYSCALL, #ifdef CONFIG_X86_LOCAL_APIC FIX_APIC_BASE, /* local (CPU) APIC) -- required for SMP or not */ +#else + FIX_VSTACK_HOLE_1, #endif #ifdef CONFIG_X86_IO_APIC FIX_IO_APIC_BASE_0, @@ -57,16 +64,21 @@ enum fixed_addresses { FIX_LI_PCIA, /* Lithium PCI Bridge A */ FIX_LI_PCIB, /* Lithium PCI Bridge B */ #endif -#ifdef CONFIG_X86_F00F_BUG - FIX_F00F_IDT, /* Virtual mapping for IDT */ -#endif + FIX_IDT, + FIX_GDT_1, + FIX_GDT_0, + FIX_TSS_3, + FIX_TSS_2, + FIX_TSS_1, + FIX_TSS_0, + FIX_ENTRY_TRAMPOLINE_1, + FIX_ENTRY_TRAMPOLINE_0, #ifdef CONFIG_X86_CYCLONE_TIMER FIX_CYCLONE_TIMER, /*cyclone timer register*/ + FIX_VSTACK_HOLE_2, #endif -#ifdef CONFIG_HIGHMEM FIX_KMAP_BEGIN, /* reserved pte's for temporary kernel mappings */ FIX_KMAP_END = FIX_KMAP_BEGIN+(KM_TYPE_NR*NR_CPUS)-1, -#endif #ifdef CONFIG_ACPI_BOOT FIX_ACPI_BEGIN, FIX_ACPI_END = FIX_ACPI_BEGIN + FIX_ACPI_PAGES - 1, @@ -95,12 +107,15 @@ extern void __set_fixmap (enum fixed_add __set_fixmap(idx, 0, __pgprot(0)) /* - * used by vmalloc.c. + * used by vmalloc.c and various other places. * * Leave one empty page between vmalloc'ed areas and * the start of the fixmap. + * + * IMPORTANT: dont change FIXADDR_TOP without adjusting KM_VSTACK0 + * and KM_VSTACK1 so that the virtual stack is 8K aligned. */ -#define FIXADDR_TOP (0xfffff000UL) +#define FIXADDR_TOP (0xffffe000UL) #define __FIXADDR_SIZE (__end_of_permanent_fixed_addresses << PAGE_SHIFT) #define FIXADDR_START (FIXADDR_TOP - __FIXADDR_SIZE) diff -puN include/asm-i386/highmem.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/highmem.h --- 25/include/asm-i386/highmem.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/highmem.h 2004-01-20 21:32:48.000000000 -0800 @@ -25,26 +25,19 @@ #include #include #include +#include /* declarations for highmem.c */ extern unsigned long highstart_pfn, highend_pfn; -extern pte_t *kmap_pte; -extern pgprot_t kmap_prot; extern pte_t *pkmap_page_table; - -extern void kmap_init(void); +extern void kmap_init(void) __init; /* * Right now we initialize only a single pte table. It can be extended * easily, subsequent pte tables have to be allocated in one physical * chunk of RAM. */ -#if NR_CPUS <= 32 -#define PKMAP_BASE (0xff800000UL) -#else -#define PKMAP_BASE (0xff600000UL) -#endif #ifdef CONFIG_X86_PAE #define LAST_PKMAP 512 #else diff -puN include/asm-i386/kmap_types.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/kmap_types.h --- 25/include/asm-i386/kmap_types.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/kmap_types.h 2004-01-20 21:32:48.000000000 -0800 @@ -3,30 +3,36 @@ #include -#ifdef CONFIG_DEBUG_HIGHMEM -# define D(n) __KM_FENCE_##n , -#else -# define D(n) -#endif - enum km_type { -D(0) KM_BOUNCE_READ, -D(1) KM_SKB_SUNRPC_DATA, -D(2) KM_SKB_DATA_SOFTIRQ, -D(3) KM_USER0, -D(4) KM_USER1, -D(5) KM_BIO_SRC_IRQ, -D(6) KM_BIO_DST_IRQ, -D(7) KM_PTE0, -D(8) KM_PTE1, -D(9) KM_PTE2, -D(10) KM_IRQ0, -D(11) KM_IRQ1, -D(12) KM_SOFTIRQ0, -D(13) KM_SOFTIRQ1, -D(14) KM_TYPE_NR -}; - -#undef D + /* + * IMPORTANT: don't move these 3 entries, and only add entries in + * pairs: the 4G/4G virtual stack must be 8K aligned on each cpu. + */ + KM_BOUNCE_READ, + KM_VSTACK1, + KM_VSTACK0, + KM_LDT_PAGE15, + KM_LDT_PAGE0 = KM_LDT_PAGE15 + 16-1, + KM_USER_COPY, + KM_VSTACK_HOLE, + KM_SKB_SUNRPC_DATA, + KM_SKB_DATA_SOFTIRQ, + KM_USER0, + KM_USER1, + KM_BIO_SRC_IRQ, + KM_BIO_DST_IRQ, + KM_PTE0, + KM_PTE1, + KM_PTE2, + KM_IRQ0, + KM_IRQ1, + KM_SOFTIRQ0, + KM_SOFTIRQ1, + /* + * Add new entries in pairs: + * the 4G/4G virtual stack must be 8K aligned on each cpu. + */ + KM_TYPE_NR +}; #endif diff -puN include/asm-i386/mmu_context.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/mmu_context.h --- 25/include/asm-i386/mmu_context.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/mmu_context.h 2004-01-20 21:32:48.000000000 -0800 @@ -29,6 +29,10 @@ static inline void switch_mm(struct mm_s { int cpu = smp_processor_id(); +#ifdef CONFIG_X86_SWITCH_PAGETABLES + if (tsk->mm) + tsk->thread_info->user_pgd = (void *)__pa(tsk->mm->pgd); +#endif if (likely(prev != next)) { /* stop flush ipis for the previous mm */ cpu_clear(cpu, prev->cpu_vm_mask); @@ -39,12 +43,14 @@ static inline void switch_mm(struct mm_s cpu_set(cpu, next->cpu_vm_mask); /* Re-load page tables */ +#if !defined(CONFIG_X86_SWITCH_PAGETABLES) load_cr3(next->pgd); +#endif /* * load the LDT, if the LDT is different: */ - if (unlikely(prev->context.ldt != next->context.ldt)) + if (unlikely(prev->context.size + next->context.size)) load_LDT_nolock(&next->context, cpu); } #ifdef CONFIG_SMP @@ -56,7 +62,9 @@ static inline void switch_mm(struct mm_s /* We were in lazy tlb mode and leave_mm disabled * tlb flush IPI delivery. We must reload %cr3. */ +#if !defined(CONFIG_X86_SWITCH_PAGETABLES) load_cr3(next->pgd); +#endif load_LDT_nolock(&next->context, cpu); } } @@ -67,6 +75,6 @@ static inline void switch_mm(struct mm_s asm("movl %0,%%fs ; movl %0,%%gs": :"r" (0)) #define activate_mm(prev, next) \ - switch_mm((prev),(next),NULL) + switch_mm((prev),(next),current) #endif diff -puN include/asm-i386/mmu.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/mmu.h --- 25/include/asm-i386/mmu.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/mmu.h 2004-01-20 21:32:48.000000000 -0800 @@ -8,10 +8,13 @@ * * cpu_vm_mask is used to optimize ldt flushing. */ + +#define MAX_LDT_PAGES 16 + typedef struct { int size; struct semaphore sem; - void *ldt; + struct page *ldt_pages[MAX_LDT_PAGES]; } mm_context_t; #endif diff -puN include/asm-i386/page.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/page.h --- 25/include/asm-i386/page.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/page.h 2004-01-20 21:32:48.000000000 -0800 @@ -1,6 +1,8 @@ #ifndef _I386_PAGE_H #define _I386_PAGE_H +#include + /* PAGE_SHIFT determines the page size */ #define PAGE_SHIFT 12 #define PAGE_SIZE (1UL << PAGE_SHIFT) @@ -9,11 +11,10 @@ #define LARGE_PAGE_MASK (~(LARGE_PAGE_SIZE-1)) #define LARGE_PAGE_SIZE (1UL << PMD_SHIFT) -#ifdef __KERNEL__ -#ifndef __ASSEMBLY__ - #include +#ifdef __KERNEL__ +#ifndef __ASSEMBLY__ #ifdef CONFIG_X86_USE_3DNOW #include @@ -88,8 +89,19 @@ typedef struct { unsigned long pgprot; } * * If you want more physical memory than this then see the CONFIG_HIGHMEM4G * and CONFIG_HIGHMEM64G options in the kernel configuration. + * + * Note: on PAE the kernel must never go below 32 MB, we use the + * first 8 entries of the 2-level boot pgd for PAE magic. */ +#ifdef CONFIG_X86_4G_VM_LAYOUT +#define __PAGE_OFFSET (0x02000000) +#define TASK_SIZE (0xff000000) +#else +#define __PAGE_OFFSET (0xc0000000) +#define TASK_SIZE (0xc0000000) +#endif + /* * This much address space is reserved for vmalloc() and iomap() * as well as fixmap mappings. @@ -114,16 +126,10 @@ static __inline__ int get_order(unsigned #endif /* __ASSEMBLY__ */ -#ifdef __ASSEMBLY__ -#define __PAGE_OFFSET (0xC0000000) -#else -#define __PAGE_OFFSET (0xC0000000UL) -#endif - - #define PAGE_OFFSET ((unsigned long)__PAGE_OFFSET) #define VMALLOC_RESERVE ((unsigned long)__VMALLOC_RESERVE) -#define MAXMEM (-__PAGE_OFFSET-__VMALLOC_RESERVE) +#define __MAXMEM (-__PAGE_OFFSET-__VMALLOC_RESERVE) +#define MAXMEM ((unsigned long)(-PAGE_OFFSET-VMALLOC_RESERVE)) #define __pa(x) ((unsigned long)(x)-PAGE_OFFSET) #define __va(x) ((void *)((unsigned long)(x)+PAGE_OFFSET)) #define pfn_to_kaddr(pfn) __va((pfn) << PAGE_SHIFT) diff -puN include/asm-i386/pgtable.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/pgtable.h --- 25/include/asm-i386/pgtable.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/pgtable.h 2004-01-20 21:32:48.000000000 -0800 @@ -32,16 +32,17 @@ #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page)) extern unsigned long empty_zero_page[1024]; extern pgd_t swapper_pg_dir[1024]; -extern kmem_cache_t *pgd_cache; -extern kmem_cache_t *pmd_cache; +extern kmem_cache_t *pgd_cache, *pmd_cache, *kpmd_cache; extern spinlock_t pgd_lock; extern struct list_head pgd_list; void pmd_ctor(void *, kmem_cache_t *, unsigned long); +void kpmd_ctor(void *, kmem_cache_t *, unsigned long); void pgd_ctor(void *, kmem_cache_t *, unsigned long); void pgd_dtor(void *, kmem_cache_t *, unsigned long); void pgtable_cache_init(void); void paging_init(void); +void setup_identity_mappings(pgd_t *pgd_base, unsigned long start, unsigned long end); #endif /* !__ASSEMBLY__ */ @@ -51,6 +52,11 @@ void paging_init(void); * newer 3-level PAE-mode page tables. */ #ifndef __ASSEMBLY__ + +extern void set_system_gate(unsigned int n, void *addr); +extern void init_entry_mappings(void); +extern void entry_trampoline_setup(void); + #ifdef CONFIG_X86_PAE # include #else @@ -63,7 +69,12 @@ void paging_init(void); #define PGDIR_SIZE (1UL << PGDIR_SHIFT) #define PGDIR_MASK (~(PGDIR_SIZE-1)) -#define USER_PTRS_PER_PGD (TASK_SIZE/PGDIR_SIZE) +#if defined(CONFIG_X86_PAE) && defined(CONFIG_X86_4G_VM_LAYOUT) +# define USER_PTRS_PER_PGD 4 +#else +# define USER_PTRS_PER_PGD ((TASK_SIZE/PGDIR_SIZE) + ((TASK_SIZE % PGDIR_SIZE) + PGDIR_SIZE-1)/PGDIR_SIZE) +#endif + #define FIRST_USER_PGD_NR 0 #define USER_PGD_PTRS (PAGE_OFFSET >> PGDIR_SHIFT) @@ -233,6 +244,7 @@ static inline void ptep_mkdirty(pte_t *p #define mk_pte(page, pgprot) pfn_pte(page_to_pfn(page), (pgprot)) #define mk_pte_huge(entry) ((entry).pte_low |= _PAGE_PRESENT | _PAGE_PSE) +#define mk_pte_phys(physpage, pgprot) pfn_pte((physpage) >> PAGE_SHIFT, pgprot) static inline pte_t pte_modify(pte_t pte, pgprot_t newprot) { diff -puN include/asm-i386/processor.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/processor.h --- 25/include/asm-i386/processor.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/processor.h 2004-01-20 21:32:48.000000000 -0800 @@ -291,11 +291,6 @@ extern unsigned int machine_submodel_id; extern unsigned int BIOS_revision; extern unsigned int mca_pentium_flag; -/* - * User space process size: 3GB (default). - */ -#define TASK_SIZE (PAGE_OFFSET) - /* This decides where the kernel will search for a free chunk of vm * space during mmap's. */ @@ -406,6 +401,7 @@ struct tss_struct { struct thread_struct { /* cached TLS descriptors. */ struct desc_struct tls_array[GDT_ENTRY_TLS_ENTRIES]; + void *stack_page0, *stack_page1; unsigned long esp0; unsigned long sysenter_cs; unsigned long eip; @@ -449,7 +445,8 @@ struct thread_struct { .io_bitmap = { [ 0 ... IO_BITMAP_LONGS] = ~0 }, \ } -static inline void load_esp0(struct tss_struct *tss, struct thread_struct *thread) +static inline void +load_esp0(struct tss_struct *tss, struct thread_struct *thread) { tss->esp0 = thread->esp0; /* This can only happen when SEP is enabled, no need to test "SEP"arately */ @@ -485,6 +482,23 @@ extern void prepare_to_copy(struct task_ */ extern int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags); +#ifdef CONFIG_X86_HIGH_ENTRY +#define virtual_esp0(tsk) \ + ((unsigned long)(tsk)->thread_info->virtual_stack + ((tsk)->thread.esp0 - (unsigned long)(tsk)->thread_info->real_stack)) +#else +# define virtual_esp0(tsk) ((tsk)->thread.esp0) +#endif + +#define load_virtual_esp0(tss, task) \ + do { \ + tss->esp0 = virtual_esp0(task); \ + if (likely(cpu_has_sep) && unlikely(tss->ss1 != task->thread.sysenter_cs)) { \ + tss->ss1 = task->thread.sysenter_cs; \ + wrmsr(MSR_IA32_SYSENTER_CS, \ + task->thread.sysenter_cs, 0); \ + } \ + } while (0) + extern unsigned long thread_saved_pc(struct task_struct *tsk); void show_trace(struct task_struct *task, unsigned long *stack); diff -puN include/asm-i386/string.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/string.h --- 25/include/asm-i386/string.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/string.h 2004-01-20 21:32:48.000000000 -0800 @@ -58,6 +58,29 @@ __asm__ __volatile__( return dest; } +/* + * This is a more generic variant of strncpy_count() suitable for + * implementing string-access routines with all sorts of return + * code semantics. It's used by mm/usercopy.c. + */ +static inline size_t strncpy_count(char * dest,const char *src,size_t count) +{ + __asm__ __volatile__( + + "1:\tdecl %0\n\t" + "js 2f\n\t" + "lodsb\n\t" + "stosb\n\t" + "testb %%al,%%al\n\t" + "jne 1b\n\t" + "2:" + "incl %0" + : "=c" (count) + :"S" (src),"D" (dest),"0" (count) : "memory"); + + return count; +} + static inline char * strcat(char * dest,const char * src) { int d0, d1, d2, d3; diff -puN include/asm-i386/thread_info.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/thread_info.h --- 25/include/asm-i386/thread_info.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/thread_info.h 2004-01-20 21:32:48.000000000 -0800 @@ -33,23 +33,12 @@ struct thread_info { 0-0xBFFFFFFF for user-thead 0-0xFFFFFFFF for kernel-thread */ - struct restart_block restart_block; + void *real_stack, *virtual_stack, *user_pgd; + struct restart_block restart_block; __u8 supervisor_stack[0]; }; -#else /* !__ASSEMBLY__ */ - -/* offsets into the thread_info struct for assembly code access */ -#define TI_TASK 0x00000000 -#define TI_EXEC_DOMAIN 0x00000004 -#define TI_FLAGS 0x00000008 -#define TI_STATUS 0x0000000C -#define TI_CPU 0x00000010 -#define TI_PRE_COUNT 0x00000014 -#define TI_ADDR_LIMIT 0x00000018 -#define TI_RESTART_BLOCK 0x000001C - #endif #define PREEMPT_ACTIVE 0x4000000 @@ -61,7 +50,7 @@ struct thread_info { */ #ifndef __ASSEMBLY__ -#define INIT_THREAD_INFO(tsk) \ +#define INIT_THREAD_INFO(tsk, thread_info) \ { \ .task = &tsk, \ .exec_domain = &default_exec_domain, \ @@ -72,6 +61,7 @@ struct thread_info { .restart_block = { \ .fn = do_no_restart_syscall, \ }, \ + .real_stack = &thread_info, \ } #define init_thread_info (init_thread_union.thread_info) @@ -113,6 +103,7 @@ static inline struct thread_info *curren #define TIF_NEED_RESCHED 3 /* rescheduling necessary */ #define TIF_SINGLESTEP 4 /* restore singlestep on return to user mode */ #define TIF_IRET 5 /* return with iret */ +#define TIF_DB7 6 /* has debug registers */ #define TIF_POLLING_NRFLAG 16 /* true if poll_idle() is polling TIF_NEED_RESCHED */ #define _TIF_SYSCALL_TRACE (1<active_mm) __flush_tlb(); +#endif } static inline void flush_tlb_page(struct vm_area_struct *vma, unsigned long addr) { +#ifndef CONFIG_X86_SWITCH_PAGETABLES if (vma->vm_mm == current->active_mm) __flush_tlb_one(addr); +#endif } static inline void flush_tlb_range(struct vm_area_struct *vma, unsigned long start, unsigned long end) { +#ifndef CONFIG_X86_SWITCH_PAGETABLES if (vma->vm_mm == current->active_mm) __flush_tlb(); +#endif } #else @@ -111,11 +117,10 @@ static inline void flush_tlb_range(struc __flush_tlb() extern void flush_tlb_all(void); -extern void flush_tlb_current_task(void); extern void flush_tlb_mm(struct mm_struct *); extern void flush_tlb_page(struct vm_area_struct *, unsigned long); -#define flush_tlb() flush_tlb_current_task() +#define flush_tlb() flush_tlb_all() static inline void flush_tlb_range(struct vm_area_struct * vma, unsigned long start, unsigned long end) { diff -puN include/asm-i386/uaccess.h~4g-2.6.0-test2-mm2-A5 include/asm-i386/uaccess.h --- 25/include/asm-i386/uaccess.h~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/include/asm-i386/uaccess.h 2004-01-20 21:32:48.000000000 -0800 @@ -26,7 +26,7 @@ #define KERNEL_DS MAKE_MM_SEG(0xFFFFFFFFUL) -#define USER_DS MAKE_MM_SEG(PAGE_OFFSET) +#define USER_DS MAKE_MM_SEG(TASK_SIZE) #define get_ds() (KERNEL_DS) #define get_fs() (current_thread_info()->addr_limit) @@ -149,6 +149,45 @@ extern void __get_user_4(void); :"=a" (ret),"=d" (x) \ :"0" (ptr)) +extern int get_user_size(unsigned int size, void *val, const void *ptr); +extern int put_user_size(unsigned int size, const void *val, void *ptr); +extern int zero_user_size(unsigned int size, void *ptr); +extern int copy_str_fromuser_size(unsigned int size, void *val, const void *ptr); +extern int strlen_fromuser_size(unsigned int size, const void *ptr); + + +# define indirect_get_user(x,ptr) \ +({ int __ret_gu,__val_gu; \ + __typeof__(ptr) __ptr_gu = (ptr); \ + __ret_gu = get_user_size(sizeof(*__ptr_gu), &__val_gu,__ptr_gu) ? -EFAULT : 0;\ + (x) = (__typeof__(*__ptr_gu))__val_gu; \ + __ret_gu; \ +}) +#define indirect_put_user(x,ptr) \ +({ \ + __typeof__(*(ptr)) *__ptr_pu = (ptr), __x_pu = (x); \ + put_user_size(sizeof(*__ptr_pu), &__x_pu, __ptr_pu) ? -EFAULT : 0; \ +}) +#define __indirect_put_user indirect_put_user +#define __indirect_get_user indirect_get_user + +#define indirect_copy_from_user(to,from,n) get_user_size(n,to,from) +#define indirect_copy_to_user(to,from,n) put_user_size(n,from,to) + +#define __indirect_copy_from_user indirect_copy_from_user +#define __indirect_copy_to_user indirect_copy_to_user + +#define indirect_strncpy_from_user(dst, src, count) \ + copy_str_fromuser_size(count, dst, src) + +extern int strlen_fromuser_size(unsigned int size, const void *ptr); +#define indirect_strnlen_user(str, n) strlen_fromuser_size(n, str) +#define indirect_strlen_user(str) indirect_strnlen_user(str, ~0UL >> 1) + +extern int zero_user_size(unsigned int size, void *ptr); + +#define indirect_clear_user(mem, len) zero_user_size(len, mem) +#define __indirect_clear_user clear_user /* Careful: we have to cast the result to the type of the pointer for sign reasons */ /** @@ -168,7 +207,7 @@ extern void __get_user_4(void); * Returns zero on success, or -EFAULT on error. * On error, the variable @x is set to zero. */ -#define get_user(x,ptr) \ +#define direct_get_user(x,ptr) \ ({ int __ret_gu,__val_gu; \ switch(sizeof (*(ptr))) { \ case 1: __get_user_x(1,__ret_gu,__val_gu,ptr); break; \ @@ -198,7 +237,7 @@ extern void __put_user_bad(void); * * Returns zero on success, or -EFAULT on error. */ -#define put_user(x,ptr) \ +#define direct_put_user(x,ptr) \ __put_user_check((__typeof__(*(ptr)))(x),(ptr),sizeof(*(ptr))) @@ -222,7 +261,7 @@ extern void __put_user_bad(void); * Returns zero on success, or -EFAULT on error. * On error, the variable @x is set to zero. */ -#define __get_user(x,ptr) \ +#define __direct_get_user(x,ptr) \ __get_user_nocheck((x),(ptr),sizeof(*(ptr))) @@ -245,7 +284,7 @@ extern void __put_user_bad(void); * * Returns zero on success, or -EFAULT on error. */ -#define __put_user(x,ptr) \ +#define __direct_put_user(x,ptr) \ __put_user_nocheck((__typeof__(*(ptr)))(x),(ptr),sizeof(*(ptr))) #define __put_user_nocheck(x,ptr,size) \ @@ -396,7 +435,7 @@ unsigned long __copy_from_user_ll(void * * On success, this will be zero. */ static inline unsigned long -__copy_to_user(void __user *to, const void *from, unsigned long n) +__direct_copy_to_user(void __user *to, const void *from, unsigned long n) { if (__builtin_constant_p(n)) { unsigned long ret; @@ -434,7 +473,7 @@ __copy_to_user(void __user *to, const vo * data to the requested size using zero bytes. */ static inline unsigned long -__copy_from_user(void *to, const void __user *from, unsigned long n) +__direct_copy_from_user(void *to, const void __user *from, unsigned long n) { if (__builtin_constant_p(n)) { unsigned long ret; @@ -468,11 +507,11 @@ __copy_from_user(void *to, const void __ * On success, this will be zero. */ static inline unsigned long -copy_to_user(void __user *to, const void *from, unsigned long n) +direct_copy_to_user(void __user *to, const void *from, unsigned long n) { might_sleep(); if (access_ok(VERIFY_WRITE, to, n)) - n = __copy_to_user(to, from, n); + n = __direct_copy_to_user(to, from, n); return n; } @@ -493,11 +532,11 @@ copy_to_user(void __user *to, const void * data to the requested size using zero bytes. */ static inline unsigned long -copy_from_user(void *to, const void __user *from, unsigned long n) +direct_copy_from_user(void *to, const void __user *from, unsigned long n) { might_sleep(); if (access_ok(VERIFY_READ, from, n)) - n = __copy_from_user(to, from, n); + n = __direct_copy_from_user(to, from, n); else memset(to, 0, n); return n; @@ -520,10 +559,68 @@ long __strncpy_from_user(char *dst, cons * If there is a limit on the length of a valid string, you may wish to * consider using strnlen_user() instead. */ -#define strlen_user(str) strnlen_user(str, ~0UL >> 1) -long strnlen_user(const char __user *str, long n); -unsigned long clear_user(void __user *mem, unsigned long len); -unsigned long __clear_user(void __user *mem, unsigned long len); +long direct_strncpy_from_user(char *dst, const char *src, long count); +long __direct_strncpy_from_user(char *dst, const char *src, long count); +#define direct_strlen_user(str) direct_strnlen_user(str, ~0UL >> 1) +long direct_strnlen_user(const char *str, long n); +unsigned long direct_clear_user(void *mem, unsigned long len); +unsigned long __direct_clear_user(void *mem, unsigned long len); + +extern int indirect_uaccess; + +#ifdef CONFIG_X86_UACCESS_INDIRECT + +/* + * Return code and zeroing semantics: + + __clear_user 0 <-> bytes not done + clear_user 0 <-> bytes not done + __copy_to_user 0 <-> bytes not done + copy_to_user 0 <-> bytes not done + __copy_from_user 0 <-> bytes not done, zero rest + copy_from_user 0 <-> bytes not done, zero rest + __get_user 0 <-> -EFAULT + get_user 0 <-> -EFAULT + __put_user 0 <-> -EFAULT + put_user 0 <-> -EFAULT + strlen_user strlen + 1 <-> 0 + strnlen_user strlen + 1 (or n+1) <-> 0 + strncpy_from_user strlen (or n) <-> -EFAULT + + */ + +#define __clear_user(mem,len) __indirect_clear_user(mem,len) +#define clear_user(mem,len) indirect_clear_user(mem,len) +#define __copy_to_user(to,from,n) __indirect_copy_to_user(to,from,n) +#define copy_to_user(to,from,n) indirect_copy_to_user(to,from,n) +#define __copy_from_user(to,from,n) __indirect_copy_from_user(to,from,n) +#define copy_from_user(to,from,n) indirect_copy_from_user(to,from,n) +#define __get_user(val,ptr) __indirect_get_user(val,ptr) +#define get_user(val,ptr) indirect_get_user(val,ptr) +#define __put_user(val,ptr) __indirect_put_user(val,ptr) +#define put_user(val,ptr) indirect_put_user(val,ptr) +#define strlen_user(str) indirect_strlen_user(str) +#define strnlen_user(src,count) indirect_strnlen_user(src,count) +#define strncpy_from_user(dst,src,count) \ + indirect_strncpy_from_user(dst,src,count) + +#else + +#define __clear_user __direct_clear_user +#define clear_user direct_clear_user +#define __copy_to_user __direct_copy_to_user +#define copy_to_user direct_copy_to_user +#define __copy_from_user __direct_copy_from_user +#define copy_from_user direct_copy_from_user +#define __get_user __direct_get_user +#define get_user direct_get_user +#define __put_user __direct_put_user +#define put_user direct_put_user +#define strlen_user direct_strlen_user +#define strnlen_user direct_strnlen_user +#define strncpy_from_user direct_strncpy_from_user + +#endif /* CONFIG_X86_UACCESS_INDIRECT */ #endif /* __i386_UACCESS_H */ diff -puN mm/Makefile~4g-2.6.0-test2-mm2-A5 mm/Makefile --- 25/mm/Makefile~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/mm/Makefile 2004-01-20 21:32:48.000000000 -0800 @@ -12,3 +12,6 @@ obj-y := bootmem.o filemap.o mempool.o slab.o swap.o truncate.o vmscan.o $(mmu-y) obj-$(CONFIG_SWAP) += page_io.o swap_state.o swapfile.o + +obj-$(CONFIG_X86_4G) += usercopy.o + diff -puN mm/memory.c~4g-2.6.0-test2-mm2-A5 mm/memory.c --- 25/mm/memory.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/mm/memory.c 2004-01-20 21:32:48.000000000 -0800 @@ -108,7 +108,8 @@ static inline void free_one_pmd(struct m pte_free_tlb(tlb, page); } -static inline void free_one_pgd(struct mmu_gather *tlb, pgd_t * dir) +static inline void free_one_pgd(struct mmu_gather *tlb, pgd_t * dir, + int pgd_idx) { int j; pmd_t * pmd; @@ -122,8 +123,11 @@ static inline void free_one_pgd(struct m } pmd = pmd_offset(dir, 0); pgd_clear(dir); - for (j = 0; j < PTRS_PER_PMD ; j++) + for (j = 0; j < PTRS_PER_PMD ; j++) { + if (pgd_idx * PGDIR_SIZE + j * PMD_SIZE >= TASK_SIZE) + break; free_one_pmd(tlb, pmd+j); + } pmd_free_tlb(tlb, pmd); } @@ -136,11 +140,13 @@ static inline void free_one_pgd(struct m void clear_page_tables(struct mmu_gather *tlb, unsigned long first, int nr) { pgd_t * page_dir = tlb->mm->pgd; + int pgd_idx = first; page_dir += first; do { - free_one_pgd(tlb, page_dir); + free_one_pgd(tlb, page_dir, pgd_idx); page_dir++; + pgd_idx++; } while (--nr); } @@ -438,7 +444,7 @@ zap_pmd_range(struct mmu_gather *tlb, pg unsigned long address, unsigned long size) { pmd_t * pmd; - unsigned long end; + unsigned long end, pgd_boundary; if (pgd_none(*dir)) return; @@ -449,8 +455,9 @@ zap_pmd_range(struct mmu_gather *tlb, pg } pmd = pmd_offset(dir, address); end = address + size; - if (end > ((address + PGDIR_SIZE) & PGDIR_MASK)) - end = ((address + PGDIR_SIZE) & PGDIR_MASK); + pgd_boundary = ((address + PGDIR_SIZE) & PGDIR_MASK); + if (pgd_boundary && (end > pgd_boundary)) + end = pgd_boundary; do { zap_pte_range(tlb, pmd, address, end - address); address = (address + PMD_SIZE) & PMD_MASK; diff -puN mm/slab.c~4g-2.6.0-test2-mm2-A5 mm/slab.c --- 25/mm/slab.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/mm/slab.c 2004-01-20 21:32:48.000000000 -0800 @@ -1226,7 +1226,8 @@ next: cachep = NULL; goto opps; } - slab_size = L1_CACHE_ALIGN(cachep->num*sizeof(kmem_bufctl_t)+sizeof(struct slab)); + slab_size = L1_CACHE_ALIGN(cachep->num*sizeof(kmem_bufctl_t) + + sizeof(struct slab)); /* * If the slab has been placed off-slab, and we have enough space then @@ -1270,10 +1271,13 @@ next: * the cache that's used by kmalloc(24), otherwise * the creation of further caches will BUG(). */ - cachep->array[smp_processor_id()] = &initarray_generic.cache; + cachep->array[smp_processor_id()] = + &initarray_generic.cache; g_cpucache_up = PARTIAL; } else { - cachep->array[smp_processor_id()] = kmalloc(sizeof(struct arraycache_init),GFP_KERNEL); + cachep->array[smp_processor_id()] = + kmalloc(sizeof(struct arraycache_init), + GFP_KERNEL); } BUG_ON(!ac_data(cachep)); ac_data(cachep)->avail = 0; @@ -1287,7 +1291,7 @@ next: } cachep->lists.next_reap = jiffies + REAPTIMEOUT_LIST3 + - ((unsigned long)cachep)%REAPTIMEOUT_LIST3; + ((unsigned long)cachep)%REAPTIMEOUT_LIST3; /* Need the semaphore to access the chain. */ down(&cache_chain_sem); @@ -1300,16 +1304,24 @@ next: list_for_each(p, &cache_chain) { kmem_cache_t *pc = list_entry(p, kmem_cache_t, next); char tmp; - /* This happens when the module gets unloaded and doesn't - destroy its slab cache and noone else reuses the vmalloc - area of the module. Print a warning. */ - if (__get_user(tmp,pc->name)) { - printk("SLAB: cache with size %d has lost its name\n", - pc->objsize); + + /* + * This happens when the module gets unloaded and + * doesn't destroy its slab cache and noone else reuses + * the vmalloc area of the module. Print a warning. + */ +#ifdef CONFIG_X86_UACCESS_INDIRECT + if (__direct_get_user(tmp,pc->name)) { +#else + if (__get_user(tmp,pc->name)) { +#endif + printk("SLAB: cache with size %d has lost its " + "name\n", pc->objsize); continue; } if (!strcmp(pc->name,name)) { - printk("kmem_cache_create: duplicate cache %s\n",name); + printk("kmem_cache_create: duplicate " + "cache %s\n",name); up(&cache_chain_sem); BUG(); } diff -puN /dev/null mm/usercopy.c --- /dev/null 2002-08-30 16:31:37.000000000 -0700 +++ 25-akpm/mm/usercopy.c 2004-01-20 21:32:48.000000000 -0800 @@ -0,0 +1,277 @@ +/* + * linux/mm/usercopy.c + * + * (C) Copyright 2003 Ingo Molnar + * + * Generic implementation of all the user-VM access functions, without + * relying on being able to access the VM directly. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +/* + * Get kernel address of the user page and pin it. + */ +static inline struct page *pin_page(unsigned long addr, int write) +{ + struct mm_struct *mm = current->mm ? : &init_mm; + struct page *page = NULL; + int ret; + + spin_lock(&mm->page_table_lock); + /* + * Do a quick atomic lookup first - this is the fastpath. + */ + page = follow_page(mm, addr, write); + if (likely(page != NULL)) { + if (!PageReserved(page)) + get_page(page); + spin_unlock(&mm->page_table_lock); + return page; + } + + /* + * No luck - bad address or need to fault in the page: + */ + spin_unlock(&mm->page_table_lock); + + /* + * In the context of filemap_copy_from_user(), we are not allowed + * to sleep. We must fail this usercopy attempt and allow + * filemap_copy_from_user() to recover: drop its atomic kmap and use + * a sleeping kmap instead. + */ + if (in_atomic()) + return NULL; + + down_read(&mm->mmap_sem); + ret = get_user_pages(current, mm, addr, 1, write, 0, &page, NULL); + up_read(&mm->mmap_sem); + if (ret <= 0) + return NULL; + return page; +} + +static inline void unpin_page(struct page *page) +{ + put_page(page); +} + +/* + * Access another process' address space. + * Source/target buffer must be kernel space, + * Do not walk the page table directly, use get_user_pages + */ +static int rw_vm(unsigned long addr, void *buf, int len, int write) +{ + if (!len) + return 0; + + /* ignore errors, just check how much was sucessfully transfered */ + while (len) { + struct page *page = NULL; + int bytes, offset; + void *maddr; + + page = pin_page(addr, write); + if (!page) + break; + + bytes = len; + offset = addr & (PAGE_SIZE-1); + if (bytes > PAGE_SIZE-offset) + bytes = PAGE_SIZE-offset; + + maddr = kmap_atomic(page, KM_USER_COPY); + +#define HANDLE_TYPE(type) \ + case sizeof(type): *(type *)(maddr+offset) = *(type *)(buf); break; + + if (write) { + switch (bytes) { + HANDLE_TYPE(char); + HANDLE_TYPE(int); + HANDLE_TYPE(long long); + default: + memcpy(maddr + offset, buf, bytes); + } + } else { +#undef HANDLE_TYPE +#define HANDLE_TYPE(type) \ + case sizeof(type): *(type *)(buf) = *(type *)(maddr+offset); break; + switch (bytes) { + HANDLE_TYPE(char); + HANDLE_TYPE(int); + HANDLE_TYPE(long long); + default: + memcpy(buf, maddr + offset, bytes); + } +#undef HANDLE_TYPE + } + kunmap_atomic(maddr, KM_USER_COPY); + unpin_page(page); + len -= bytes; + buf += bytes; + addr += bytes; + } + + return len; +} + +static int str_vm(unsigned long addr, void *buf0, int len, int copy) +{ + struct mm_struct *mm = current->mm ? : &init_mm; + struct page *page; + void *buf = buf0; + + if (!len) + return len; + + down_read(&mm->mmap_sem); + /* ignore errors, just check how much was sucessfully transfered */ + while (len) { + int bytes, ret, offset, left, copied; + char *maddr; + + ret = get_user_pages(current, mm, addr, 1, copy == 2, 0, &page, NULL); + if (ret <= 0) { + up_read(&mm->mmap_sem); + return -EFAULT; + } + + bytes = len; + offset = addr & (PAGE_SIZE-1); + if (bytes > PAGE_SIZE-offset) + bytes = PAGE_SIZE-offset; + + maddr = kmap_atomic(page, KM_USER_COPY); + if (copy == 2) { + memset(maddr + offset, 0, bytes); + copied = bytes; + left = 0; + } else if (copy == 1) { + left = strncpy_count(buf, maddr + offset, bytes); + copied = bytes - left; + } else { + copied = strnlen(maddr + offset, bytes); + left = bytes - copied; + } + BUG_ON(bytes < 0 || copied < 0); + kunmap_atomic(maddr, KM_USER_COPY); + page_cache_release(page); + len -= copied; + buf += copied; + addr += copied; + if (left) + break; + } + up_read(&mm->mmap_sem); + + return len; +} + +/* + * Copies memory from userspace (ptr) into kernelspace (val). + * + * returns # of bytes not copied. + */ +int get_user_size(unsigned int size, void *val, const void *ptr) +{ + int ret; + + if (unlikely(segment_eq(get_fs(), KERNEL_DS))) + ret = __direct_copy_from_user(val, ptr, size); + else + ret = rw_vm((unsigned long)ptr, val, size, 0); + if (ret) + /* + * Zero the rest: + */ + memset(val + size - ret, 0, ret); + return ret; +} + +/* + * Copies memory from kernelspace (val) into userspace (ptr). + * + * returns # of bytes not copied. + */ +int put_user_size(unsigned int size, const void *val, void *ptr) +{ + if (unlikely(segment_eq(get_fs(), KERNEL_DS))) + return __direct_copy_to_user(ptr, val, size); + else + return rw_vm((unsigned long)ptr, (void *)val, size, 1); +} + +int copy_str_fromuser_size(unsigned int size, void *val, const void *ptr) +{ + int copied, left; + + if (unlikely(segment_eq(get_fs(), KERNEL_DS))) { + left = strncpy_count(val, ptr, size); + copied = size - left; + BUG_ON(copied < 0); + + return copied; + } + left = str_vm((unsigned long)ptr, val, size, 1); + if (left < 0) + return left; + copied = size - left; + BUG_ON(copied < 0); + + return copied; +} + +int strlen_fromuser_size(unsigned int size, const void *ptr) +{ + int copied, left; + + if (unlikely(segment_eq(get_fs(), KERNEL_DS))) { + copied = strnlen(ptr, size) + 1; + BUG_ON(copied < 0); + + return copied; + } + left = str_vm((unsigned long)ptr, NULL, size, 0); + if (left < 0) + return 0; + copied = size - left + 1; + BUG_ON(copied < 0); + + return copied; +} + +int zero_user_size(unsigned int size, void *ptr) +{ + int left; + + if (unlikely(segment_eq(get_fs(), KERNEL_DS))) { + memset(ptr, 0, size); + return 0; + } + left = str_vm((unsigned long)ptr, NULL, size, 2); + if (left < 0) + return size; + return left; +} + +EXPORT_SYMBOL(get_user_size); +EXPORT_SYMBOL(put_user_size); +EXPORT_SYMBOL(zero_user_size); +EXPORT_SYMBOL(copy_str_fromuser_size); +EXPORT_SYMBOL(strlen_fromuser_size); + diff -puN arch/i386/kernel/acpi/boot.c~4g-2.6.0-test2-mm2-A5 arch/i386/kernel/acpi/boot.c --- 25/arch/i386/kernel/acpi/boot.c~4g-2.6.0-test2-mm2-A5 2004-01-20 21:32:48.000000000 -0800 +++ 25-akpm/arch/i386/kernel/acpi/boot.c 2004-01-20 21:32:48.000000000 -0800 @@ -334,7 +334,7 @@ acpi_scan_rsdp ( * RSDP signature. */ for (offset = 0; offset < length; offset += 16) { - if (strncmp((char *) (start + offset), "RSD PTR ", sig_len)) + if (strncmp((char *) __va(start + offset), "RSD PTR ", sig_len)) continue; return (start + offset); } _