OpenCloudOS-Kernel/arch/arm64/kernel/entry.S

951 lines
22 KiB
ArmAsm
Raw Normal View History

/*
* Low-level exception handling code
*
* Copyright (C) 2012 ARM Ltd.
* Authors: Catalin Marinas <catalin.marinas@arm.com>
* Will Deacon <will.deacon@arm.com>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <linux/init.h>
#include <linux/linkage.h>
#include <asm/alternative.h>
#include <asm/assembler.h>
#include <asm/asm-offsets.h>
#include <asm/cpufeature.h>
#include <asm/errno.h>
#include <asm/esr.h>
#include <asm/irq.h>
#include <asm/memory.h>
#include <asm/ptrace.h>
#include <asm/thread_info.h>
#include <asm/asm-uaccess.h>
#include <asm/unistd.h>
/*
* Context tracking subsystem. Used to instrument transitions
* between user and kernel mode.
*/
.macro ct_user_exit, syscall = 0
#ifdef CONFIG_CONTEXT_TRACKING
bl context_tracking_user_exit
.if \syscall == 1
/*
* Save/restore needed during syscalls. Restore syscall arguments from
* the values already saved on stack during kernel_entry.
*/
ldp x0, x1, [sp]
ldp x2, x3, [sp, #S_X2]
ldp x4, x5, [sp, #S_X4]
ldp x6, x7, [sp, #S_X6]
.endif
#endif
.endm
.macro ct_user_enter
#ifdef CONFIG_CONTEXT_TRACKING
bl context_tracking_user_enter
#endif
.endm
/*
* Bad Abort numbers
*-----------------
*/
#define BAD_SYNC 0
#define BAD_IRQ 1
#define BAD_FIQ 2
#define BAD_ERROR 3
.macro kernel_ventry label
.align 7
sub sp, sp, #S_FRAME_SIZE
arm64: add VMAP_STACK overflow detection This patch adds stack overflow detection to arm64, usable when vmap'd stacks are in use. Overflow is detected in a small preamble executed for each exception entry, which checks whether there is enough space on the current stack for the general purpose registers to be saved. If there is not enough space, the overflow handler is invoked on a per-cpu overflow stack. This approach preserves the original exception information in ESR_EL1 (and where appropriate, FAR_EL1). Task and IRQ stacks are aligned to double their size, enabling overflow to be detected with a single bit test. For example, a 16K stack is aligned to 32K, ensuring that bit 14 of the SP must be zero. On an overflow (or underflow), this bit is flipped. Thus, overflow (of less than the size of the stack) can be detected by testing whether this bit is set. The overflow check is performed before any attempt is made to access the stack, avoiding recursive faults (and the loss of exception information these would entail). As logical operations cannot be performed on the SP directly, the SP is temporarily swapped with a general purpose register using arithmetic operations to enable the test to be performed. This gives us a useful error message on stack overflow, as can be trigger with the LKDTM overflow test: [ 305.388749] lkdtm: Performing direct entry OVERFLOW [ 305.395444] Insufficient stack space to handle exception! [ 305.395482] ESR: 0x96000047 -- DABT (current EL) [ 305.399890] FAR: 0xffff00000a5e7f30 [ 305.401315] Task stack: [0xffff00000a5e8000..0xffff00000a5ec000] [ 305.403815] IRQ stack: [0xffff000008000000..0xffff000008004000] [ 305.407035] Overflow stack: [0xffff80003efce4e0..0xffff80003efcf4e0] [ 305.409622] CPU: 0 PID: 1219 Comm: sh Not tainted 4.13.0-rc3-00021-g9636aea #5 [ 305.412785] Hardware name: linux,dummy-virt (DT) [ 305.415756] task: ffff80003d051c00 task.stack: ffff00000a5e8000 [ 305.419221] PC is at recursive_loop+0x10/0x48 [ 305.421637] LR is at recursive_loop+0x38/0x48 [ 305.423768] pc : [<ffff00000859f330>] lr : [<ffff00000859f358>] pstate: 40000145 [ 305.428020] sp : ffff00000a5e7f50 [ 305.430469] x29: ffff00000a5e8350 x28: ffff80003d051c00 [ 305.433191] x27: ffff000008981000 x26: ffff000008f80400 [ 305.439012] x25: ffff00000a5ebeb8 x24: ffff00000a5ebeb8 [ 305.440369] x23: ffff000008f80138 x22: 0000000000000009 [ 305.442241] x21: ffff80003ce65000 x20: ffff000008f80188 [ 305.444552] x19: 0000000000000013 x18: 0000000000000006 [ 305.446032] x17: 0000ffffa2601280 x16: ffff0000081fe0b8 [ 305.448252] x15: ffff000008ff546d x14: 000000000047a4c8 [ 305.450246] x13: ffff000008ff7872 x12: 0000000005f5e0ff [ 305.452953] x11: ffff000008ed2548 x10: 000000000005ee8d [ 305.454824] x9 : ffff000008545380 x8 : ffff00000a5e8770 [ 305.457105] x7 : 1313131313131313 x6 : 00000000000000e1 [ 305.459285] x5 : 0000000000000000 x4 : 0000000000000000 [ 305.461781] x3 : 0000000000000000 x2 : 0000000000000400 [ 305.465119] x1 : 0000000000000013 x0 : 0000000000000012 [ 305.467724] Kernel panic - not syncing: kernel stack overflow [ 305.470561] CPU: 0 PID: 1219 Comm: sh Not tainted 4.13.0-rc3-00021-g9636aea #5 [ 305.473325] Hardware name: linux,dummy-virt (DT) [ 305.475070] Call trace: [ 305.476116] [<ffff000008088ad8>] dump_backtrace+0x0/0x378 [ 305.478991] [<ffff000008088e64>] show_stack+0x14/0x20 [ 305.481237] [<ffff00000895a178>] dump_stack+0x98/0xb8 [ 305.483294] [<ffff0000080c3288>] panic+0x118/0x280 [ 305.485673] [<ffff0000080c2e9c>] nmi_panic+0x6c/0x70 [ 305.486216] [<ffff000008089710>] handle_bad_stack+0x118/0x128 [ 305.486612] Exception stack(0xffff80003efcf3a0 to 0xffff80003efcf4e0) [ 305.487334] f3a0: 0000000000000012 0000000000000013 0000000000000400 0000000000000000 [ 305.488025] f3c0: 0000000000000000 0000000000000000 00000000000000e1 1313131313131313 [ 305.488908] f3e0: ffff00000a5e8770 ffff000008545380 000000000005ee8d ffff000008ed2548 [ 305.489403] f400: 0000000005f5e0ff ffff000008ff7872 000000000047a4c8 ffff000008ff546d [ 305.489759] f420: ffff0000081fe0b8 0000ffffa2601280 0000000000000006 0000000000000013 [ 305.490256] f440: ffff000008f80188 ffff80003ce65000 0000000000000009 ffff000008f80138 [ 305.490683] f460: ffff00000a5ebeb8 ffff00000a5ebeb8 ffff000008f80400 ffff000008981000 [ 305.491051] f480: ffff80003d051c00 ffff00000a5e8350 ffff00000859f358 ffff00000a5e7f50 [ 305.491444] f4a0: ffff00000859f330 0000000040000145 0000000000000000 0000000000000000 [ 305.492008] f4c0: 0001000000000000 0000000000000000 ffff00000a5e8350 ffff00000859f330 [ 305.493063] [<ffff00000808205c>] __bad_stack+0x88/0x8c [ 305.493396] [<ffff00000859f330>] recursive_loop+0x10/0x48 [ 305.493731] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494088] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494425] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494649] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494898] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.495205] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.495453] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.495708] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496000] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496302] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496644] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496894] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.497138] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.497325] [<ffff00000859f3dc>] lkdtm_OVERFLOW+0x14/0x20 [ 305.497506] [<ffff00000859f314>] lkdtm_do_action+0x1c/0x28 [ 305.497786] [<ffff00000859f178>] direct_entry+0xe0/0x170 [ 305.498095] [<ffff000008345568>] full_proxy_write+0x60/0xa8 [ 305.498387] [<ffff0000081fb7f4>] __vfs_write+0x1c/0x128 [ 305.498679] [<ffff0000081fcc68>] vfs_write+0xa0/0x1b0 [ 305.498926] [<ffff0000081fe0fc>] SyS_write+0x44/0xa0 [ 305.499182] Exception stack(0xffff00000a5ebec0 to 0xffff00000a5ec000) [ 305.499429] bec0: 0000000000000001 000000001c4cf5e0 0000000000000009 000000001c4cf5e0 [ 305.499674] bee0: 574f4c465245564f 0000000000000000 0000000000000000 8000000080808080 [ 305.499904] bf00: 0000000000000040 0000000000000038 fefefeff1b4bc2ff 7f7f7f7f7f7fff7f [ 305.500189] bf20: 0101010101010101 0000000000000000 000000000047a4c8 0000000000000038 [ 305.500712] bf40: 0000000000000000 0000ffffa2601280 0000ffffc63f6068 00000000004b5000 [ 305.501241] bf60: 0000000000000001 000000001c4cf5e0 0000000000000009 000000001c4cf5e0 [ 305.501791] bf80: 0000000000000020 0000000000000000 00000000004b5000 000000001c4cc458 [ 305.502314] bfa0: 0000000000000000 0000ffffc63f7950 000000000040a3c4 0000ffffc63f70e0 [ 305.502762] bfc0: 0000ffffa2601268 0000000080000000 0000000000000001 0000000000000040 [ 305.503207] bfe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000 [ 305.503680] [<ffff000008082fb0>] el0_svc_naked+0x24/0x28 [ 305.504720] Kernel Offset: disabled [ 305.505189] CPU features: 0x002082 [ 305.505473] Memory Limit: none [ 305.506181] ---[ end Kernel panic - not syncing: kernel stack overflow This patch was co-authored by Ard Biesheuvel and Mark Rutland. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Will Deacon <will.deacon@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: James Morse <james.morse@arm.com>
2017-07-15 03:30:35 +08:00
#ifdef CONFIG_VMAP_STACK
/*
* Test whether the SP has overflowed, without corrupting a GPR.
* Task and IRQ stacks are aligned to (1 << THREAD_SHIFT).
*/
add sp, sp, x0 // sp' = sp + x0
sub x0, sp, x0 // x0' = sp' - x0 = (sp + x0) - x0 = sp
tbnz x0, #THREAD_SHIFT, 0f
sub x0, sp, x0 // x0'' = sp' - x0' = (sp + x0) - sp = x0
sub sp, sp, x0 // sp'' = sp' - x0 = (sp + x0) - x0 = sp
b \label
0:
/*
* Either we've just detected an overflow, or we've taken an exception
* while on the overflow stack. Either way, we won't return to
* userspace, and can clobber EL0 registers to free up GPRs.
*/
/* Stash the original SP (minus S_FRAME_SIZE) in tpidr_el0. */
msr tpidr_el0, x0
/* Recover the original x0 value and stash it in tpidrro_el0 */
sub x0, sp, x0
msr tpidrro_el0, x0
/* Switch to the overflow stack */
adr_this_cpu sp, overflow_stack + OVERFLOW_STACK_SIZE, x0
/*
* Check whether we were already on the overflow stack. This may happen
* after panic() re-enables interrupts.
*/
mrs x0, tpidr_el0 // sp of interrupted context
sub x0, sp, x0 // delta with top of overflow stack
tst x0, #~(OVERFLOW_STACK_SIZE - 1) // within range?
b.ne __bad_stack // no? -> bad stack pointer
/* We were already on the overflow stack. Restore sp/x0 and carry on. */
sub sp, sp, x0
mrs x0, tpidrro_el0
#endif
b \label
.endm
.macro kernel_entry, el, regsize = 64
.if \regsize == 32
mov w0, w0 // zero upper 32 bits of x0
.endif
stp x0, x1, [sp, #16 * 0]
stp x2, x3, [sp, #16 * 1]
stp x4, x5, [sp, #16 * 2]
stp x6, x7, [sp, #16 * 3]
stp x8, x9, [sp, #16 * 4]
stp x10, x11, [sp, #16 * 5]
stp x12, x13, [sp, #16 * 6]
stp x14, x15, [sp, #16 * 7]
stp x16, x17, [sp, #16 * 8]
stp x18, x19, [sp, #16 * 9]
stp x20, x21, [sp, #16 * 10]
stp x22, x23, [sp, #16 * 11]
stp x24, x25, [sp, #16 * 12]
stp x26, x27, [sp, #16 * 13]
stp x28, x29, [sp, #16 * 14]
.if \el == 0
mrs x21, sp_el0
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr_this_cpu tsk, __entry_task, x20 // Ensure MDSCR_EL1.SS is clear,
ldr x19, [tsk, #TSK_TI_FLAGS] // since we can unmask debug
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
disable_step_tsk x19, x20 // exceptions when scheduling.
mov x29, xzr // fp pointed to user-space
.else
add x21, sp, #S_FRAME_SIZE
get_thread_info tsk
/* Save the task's original addr_limit and set USER_DS (TASK_SIZE_64) */
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x20, [tsk, #TSK_TI_ADDR_LIMIT]
str x20, [sp, #S_ORIG_ADDR_LIMIT]
mov x20, #TASK_SIZE_64
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
str x20, [tsk, #TSK_TI_ADDR_LIMIT]
/* No need to reset PSTATE.UAO, hardware's already set it to 0 for us */
.endif /* \el == 0 */
mrs x22, elr_el1
mrs x23, spsr_el1
stp lr, x21, [sp, #S_LR]
arm64: unwind: reference pt_regs via embedded stack frame As it turns out, the unwind code is slightly broken, and probably has been for a while. The problem is in the dumping of the exception stack, which is intended to dump the contents of the pt_regs struct at each level in the call stack where an exception was taken and routed to a routine marked as __exception (which means its stack frame is right below the pt_regs struct on the stack). 'Right below the pt_regs struct' is ill defined, though: the unwind code assigns 'frame pointer + 0x10' to the .sp member of the stackframe struct at each level, and dump_backtrace() happily dereferences that as the pt_regs pointer when encountering an __exception routine. However, the actual size of the stack frame created by this routine (which could be one of many __exception routines we have in the kernel) is not known, and so frame.sp is pretty useless to figure out where struct pt_regs really is. So it seems the only way to ensure that we can find our struct pt_regs when walking the stack frames is to put it at a known fixed offset of the stack frame pointer that is passed to such __exception routines. The simplest way to do that is to put it inside pt_regs itself, which is the main change implemented by this patch. As a bonus, doing this allows us to get rid of a fair amount of cruft related to walking from one stack to the other, which is especially nice since we intend to introduce yet another stack for overflow handling once we add support for vmapped stacks. It also fixes an inconsistency where we only add a stack frame pointing to ELR_EL1 if we are executing from the IRQ stack but not when we are executing from the task stack. To consistly identify exceptions regs even in the presence of exceptions taken from entry code, we must check whether the next frame was created by entry text, rather than whether the current frame was crated by exception text. To avoid backtracing using PCs that fall in the idmap, or are controlled by userspace, we must explcitly zero the FP and LR in startup paths, and must ensure that the frame embedded in pt_regs is zeroed upon entry from EL0. To avoid these NULL entries showin in the backtrace, unwind_frame() is updated to avoid them. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> [Mark: compare current frame against .entry.text, avoid bogus PCs] Signed-off-by: Mark Rutland <mark.rutland@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: James Morse <james.morse@arm.com> Cc: Will Deacon <will.deacon@arm.com>
2017-07-23 01:45:33 +08:00
/*
* In order to be able to dump the contents of struct pt_regs at the
* time the exception was taken (in case we attempt to walk the call
* stack later), chain it together with the stack frames.
*/
.if \el == 0
stp xzr, xzr, [sp, #S_STACKFRAME]
.else
stp x29, x22, [sp, #S_STACKFRAME]
.endif
add x29, sp, #S_STACKFRAME
#ifdef CONFIG_ARM64_SW_TTBR0_PAN
/*
* Set the TTBR0 PAN bit in SPSR. When the exception is taken from
* EL0, there is no need to check the state of TTBR0_EL1 since
* accesses are always enabled.
* Note that the meaning of this bit differs from the ARMv8.1 PAN
* feature as all TTBR0_EL1 accesses are disabled, not just those to
* user mappings.
*/
alternative_if ARM64_HAS_PAN
b 1f // skip TTBR0 PAN
alternative_else_nop_endif
.if \el != 0
mrs x21, ttbr0_el1
tst x21, #0xffff << 48 // Check for the reserved ASID
orr x23, x23, #PSR_PAN_BIT // Set the emulated PAN in the saved SPSR
b.eq 1f // TTBR0 access already disabled
and x23, x23, #~PSR_PAN_BIT // Clear the emulated PAN in the saved SPSR
.endif
__uaccess_ttbr0_disable x21
1:
#endif
stp x22, x23, [sp, #S_PC]
/* Not in a syscall by default (el0_svc overwrites for real syscall) */
.if \el == 0
mov w21, #NO_SYSCALL
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
str w21, [sp, #S_SYSCALLNO]
.endif
/*
* Set sp_el0 to current thread_info.
*/
.if \el == 0
msr sp_el0, tsk
.endif
/*
* Registers that may be useful after this macro is invoked:
*
* x21 - aborted SP
* x22 - aborted PC
* x23 - aborted PSTATE
*/
.endm
.macro kernel_exit, el
.if \el != 0
/* Restore the task's original addr_limit. */
ldr x20, [sp, #S_ORIG_ADDR_LIMIT]
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
str x20, [tsk, #TSK_TI_ADDR_LIMIT]
/* No need to restore UAO, it will be restored from SPSR_EL1 */
.endif
ldp x21, x22, [sp, #S_PC] // load ELR, SPSR
.if \el == 0
ct_user_enter
.endif
#ifdef CONFIG_ARM64_SW_TTBR0_PAN
/*
* Restore access to TTBR0_EL1. If returning to EL0, no need for SPSR
* PAN bit checking.
*/
alternative_if ARM64_HAS_PAN
b 2f // skip TTBR0 PAN
alternative_else_nop_endif
.if \el != 0
tbnz x22, #22, 1f // Skip re-enabling TTBR0 access if the PSR_PAN_BIT is set
.endif
__uaccess_ttbr0_enable x0
.if \el == 0
/*
* Enable errata workarounds only if returning to user. The only
* workaround currently required for TTBR0_EL1 changes are for the
* Cavium erratum 27456 (broadcast TLBI instructions may cause I-cache
* corruption).
*/
post_ttbr0_update_workaround
.endif
1:
.if \el != 0
and x22, x22, #~PSR_PAN_BIT // ARMv8.0 CPUs do not understand this bit
.endif
2:
#endif
.if \el == 0
ldr x23, [sp, #S_SP] // load return stack pointer
msr sp_el0, x23
#ifdef CONFIG_ARM64_ERRATUM_845719
alternative_if ARM64_WORKAROUND_845719
tbz x22, #4, 1f
#ifdef CONFIG_PID_IN_CONTEXTIDR
mrs x29, contextidr_el1
msr contextidr_el1, x29
#else
msr contextidr_el1, xzr
#endif
1:
alternative_else_nop_endif
#endif
.endif
msr elr_el1, x21 // set up the return data
msr spsr_el1, x22
ldp x0, x1, [sp, #16 * 0]
ldp x2, x3, [sp, #16 * 1]
ldp x4, x5, [sp, #16 * 2]
ldp x6, x7, [sp, #16 * 3]
ldp x8, x9, [sp, #16 * 4]
ldp x10, x11, [sp, #16 * 5]
ldp x12, x13, [sp, #16 * 6]
ldp x14, x15, [sp, #16 * 7]
ldp x16, x17, [sp, #16 * 8]
ldp x18, x19, [sp, #16 * 9]
ldp x20, x21, [sp, #16 * 10]
ldp x22, x23, [sp, #16 * 11]
ldp x24, x25, [sp, #16 * 12]
ldp x26, x27, [sp, #16 * 13]
ldp x28, x29, [sp, #16 * 14]
ldr lr, [sp, #S_LR]
add sp, sp, #S_FRAME_SIZE // restore sp
eret // return to kernel
.endm
.macro irq_stack_entry
mov x19, sp // preserve the original sp
/*
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
* Compare sp with the base of the task stack.
* If the top ~(THREAD_SIZE - 1) bits match, we are on a task stack,
* and should switch to the irq stack.
*/
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x25, [tsk, TSK_STACK]
eor x25, x25, x19
and x25, x25, #~(THREAD_SIZE - 1)
cbnz x25, 9998f
ldr_this_cpu x25, irq_stack_ptr, x26
arm64: kernel: remove {THREAD,IRQ_STACK}_START_SP For historical reasons, we leave the top 16 bytes of our task and IRQ stacks unused, a practice used to ensure that the SP can always be masked to find the base of the current stack (historically, where thread_info could be found). However, this is not necessary, as: * When an exception is taken from a task stack, we decrement the SP by S_FRAME_SIZE and stash the exception registers before we compare the SP against the task stack. In such cases, the SP must be at least S_FRAME_SIZE below the limit, and can be safely masked to determine whether the task stack is in use. * When transitioning to an IRQ stack, we'll place a dummy frame onto the IRQ stack before enabling asynchronous exceptions, or executing code we expect to trigger faults. Thus, if an exception is taken from the IRQ stack, the SP must be at least 16 bytes below the limit. * We no longer mask the SP to find the thread_info, which is now found via sp_el0. Note that historically, the offset was critical to ensure that cpu_switch_to() found the correct stack for new threads that hadn't yet executed ret_from_fork(). Given that, this initial offset serves no purpose, and can be removed. This brings us in-line with other architectures (e.g. x86) which do not rely on this masking. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> [Mark: rebase, kill THREAD_START_SP, commit msg additions] Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Will Deacon <will.deacon@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: James Morse <james.morse@arm.com>
2017-07-21 00:15:45 +08:00
mov x26, #IRQ_STACK_SIZE
add x26, x25, x26
/* switch to the irq stack */
mov sp, x26
9998:
.endm
/*
* x19 should be preserved between irq_stack_entry and
* irq_stack_exit.
*/
.macro irq_stack_exit
mov sp, x19
.endm
/*
* These are the registers used in the syscall handler, and allow us to
* have in theory up to 7 arguments to a function - x0 to x6.
*
* x7 is reserved for the system call number in 32-bit mode.
*/
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
wsc_nr .req w25 // number of system calls
wscno .req w26 // syscall number
xscno .req x26 // syscall number (zero-extended)
stbl .req x27 // syscall table pointer
tsk .req x28 // current thread_info
/*
* Interrupt handling.
*/
.macro irq_handler
ldr_l x1, handle_arch_irq
mov x0, sp
irq_stack_entry
blr x1
irq_stack_exit
.endm
.text
/*
* Exception vectors.
*/
.pushsection ".entry.text", "ax"
.align 11
ENTRY(vectors)
kernel_ventry el1_sync_invalid // Synchronous EL1t
kernel_ventry el1_irq_invalid // IRQ EL1t
kernel_ventry el1_fiq_invalid // FIQ EL1t
kernel_ventry el1_error_invalid // Error EL1t
kernel_ventry el1_sync // Synchronous EL1h
kernel_ventry el1_irq // IRQ EL1h
kernel_ventry el1_fiq_invalid // FIQ EL1h
kernel_ventry el1_error_invalid // Error EL1h
kernel_ventry el0_sync // Synchronous 64-bit EL0
kernel_ventry el0_irq // IRQ 64-bit EL0
kernel_ventry el0_fiq_invalid // FIQ 64-bit EL0
kernel_ventry el0_error_invalid // Error 64-bit EL0
#ifdef CONFIG_COMPAT
kernel_ventry el0_sync_compat // Synchronous 32-bit EL0
kernel_ventry el0_irq_compat // IRQ 32-bit EL0
kernel_ventry el0_fiq_invalid_compat // FIQ 32-bit EL0
kernel_ventry el0_error_invalid_compat // Error 32-bit EL0
#else
kernel_ventry el0_sync_invalid // Synchronous 32-bit EL0
kernel_ventry el0_irq_invalid // IRQ 32-bit EL0
kernel_ventry el0_fiq_invalid // FIQ 32-bit EL0
kernel_ventry el0_error_invalid // Error 32-bit EL0
#endif
END(vectors)
arm64: add VMAP_STACK overflow detection This patch adds stack overflow detection to arm64, usable when vmap'd stacks are in use. Overflow is detected in a small preamble executed for each exception entry, which checks whether there is enough space on the current stack for the general purpose registers to be saved. If there is not enough space, the overflow handler is invoked on a per-cpu overflow stack. This approach preserves the original exception information in ESR_EL1 (and where appropriate, FAR_EL1). Task and IRQ stacks are aligned to double their size, enabling overflow to be detected with a single bit test. For example, a 16K stack is aligned to 32K, ensuring that bit 14 of the SP must be zero. On an overflow (or underflow), this bit is flipped. Thus, overflow (of less than the size of the stack) can be detected by testing whether this bit is set. The overflow check is performed before any attempt is made to access the stack, avoiding recursive faults (and the loss of exception information these would entail). As logical operations cannot be performed on the SP directly, the SP is temporarily swapped with a general purpose register using arithmetic operations to enable the test to be performed. This gives us a useful error message on stack overflow, as can be trigger with the LKDTM overflow test: [ 305.388749] lkdtm: Performing direct entry OVERFLOW [ 305.395444] Insufficient stack space to handle exception! [ 305.395482] ESR: 0x96000047 -- DABT (current EL) [ 305.399890] FAR: 0xffff00000a5e7f30 [ 305.401315] Task stack: [0xffff00000a5e8000..0xffff00000a5ec000] [ 305.403815] IRQ stack: [0xffff000008000000..0xffff000008004000] [ 305.407035] Overflow stack: [0xffff80003efce4e0..0xffff80003efcf4e0] [ 305.409622] CPU: 0 PID: 1219 Comm: sh Not tainted 4.13.0-rc3-00021-g9636aea #5 [ 305.412785] Hardware name: linux,dummy-virt (DT) [ 305.415756] task: ffff80003d051c00 task.stack: ffff00000a5e8000 [ 305.419221] PC is at recursive_loop+0x10/0x48 [ 305.421637] LR is at recursive_loop+0x38/0x48 [ 305.423768] pc : [<ffff00000859f330>] lr : [<ffff00000859f358>] pstate: 40000145 [ 305.428020] sp : ffff00000a5e7f50 [ 305.430469] x29: ffff00000a5e8350 x28: ffff80003d051c00 [ 305.433191] x27: ffff000008981000 x26: ffff000008f80400 [ 305.439012] x25: ffff00000a5ebeb8 x24: ffff00000a5ebeb8 [ 305.440369] x23: ffff000008f80138 x22: 0000000000000009 [ 305.442241] x21: ffff80003ce65000 x20: ffff000008f80188 [ 305.444552] x19: 0000000000000013 x18: 0000000000000006 [ 305.446032] x17: 0000ffffa2601280 x16: ffff0000081fe0b8 [ 305.448252] x15: ffff000008ff546d x14: 000000000047a4c8 [ 305.450246] x13: ffff000008ff7872 x12: 0000000005f5e0ff [ 305.452953] x11: ffff000008ed2548 x10: 000000000005ee8d [ 305.454824] x9 : ffff000008545380 x8 : ffff00000a5e8770 [ 305.457105] x7 : 1313131313131313 x6 : 00000000000000e1 [ 305.459285] x5 : 0000000000000000 x4 : 0000000000000000 [ 305.461781] x3 : 0000000000000000 x2 : 0000000000000400 [ 305.465119] x1 : 0000000000000013 x0 : 0000000000000012 [ 305.467724] Kernel panic - not syncing: kernel stack overflow [ 305.470561] CPU: 0 PID: 1219 Comm: sh Not tainted 4.13.0-rc3-00021-g9636aea #5 [ 305.473325] Hardware name: linux,dummy-virt (DT) [ 305.475070] Call trace: [ 305.476116] [<ffff000008088ad8>] dump_backtrace+0x0/0x378 [ 305.478991] [<ffff000008088e64>] show_stack+0x14/0x20 [ 305.481237] [<ffff00000895a178>] dump_stack+0x98/0xb8 [ 305.483294] [<ffff0000080c3288>] panic+0x118/0x280 [ 305.485673] [<ffff0000080c2e9c>] nmi_panic+0x6c/0x70 [ 305.486216] [<ffff000008089710>] handle_bad_stack+0x118/0x128 [ 305.486612] Exception stack(0xffff80003efcf3a0 to 0xffff80003efcf4e0) [ 305.487334] f3a0: 0000000000000012 0000000000000013 0000000000000400 0000000000000000 [ 305.488025] f3c0: 0000000000000000 0000000000000000 00000000000000e1 1313131313131313 [ 305.488908] f3e0: ffff00000a5e8770 ffff000008545380 000000000005ee8d ffff000008ed2548 [ 305.489403] f400: 0000000005f5e0ff ffff000008ff7872 000000000047a4c8 ffff000008ff546d [ 305.489759] f420: ffff0000081fe0b8 0000ffffa2601280 0000000000000006 0000000000000013 [ 305.490256] f440: ffff000008f80188 ffff80003ce65000 0000000000000009 ffff000008f80138 [ 305.490683] f460: ffff00000a5ebeb8 ffff00000a5ebeb8 ffff000008f80400 ffff000008981000 [ 305.491051] f480: ffff80003d051c00 ffff00000a5e8350 ffff00000859f358 ffff00000a5e7f50 [ 305.491444] f4a0: ffff00000859f330 0000000040000145 0000000000000000 0000000000000000 [ 305.492008] f4c0: 0001000000000000 0000000000000000 ffff00000a5e8350 ffff00000859f330 [ 305.493063] [<ffff00000808205c>] __bad_stack+0x88/0x8c [ 305.493396] [<ffff00000859f330>] recursive_loop+0x10/0x48 [ 305.493731] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494088] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494425] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494649] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.494898] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.495205] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.495453] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.495708] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496000] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496302] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496644] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.496894] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.497138] [<ffff00000859f358>] recursive_loop+0x38/0x48 [ 305.497325] [<ffff00000859f3dc>] lkdtm_OVERFLOW+0x14/0x20 [ 305.497506] [<ffff00000859f314>] lkdtm_do_action+0x1c/0x28 [ 305.497786] [<ffff00000859f178>] direct_entry+0xe0/0x170 [ 305.498095] [<ffff000008345568>] full_proxy_write+0x60/0xa8 [ 305.498387] [<ffff0000081fb7f4>] __vfs_write+0x1c/0x128 [ 305.498679] [<ffff0000081fcc68>] vfs_write+0xa0/0x1b0 [ 305.498926] [<ffff0000081fe0fc>] SyS_write+0x44/0xa0 [ 305.499182] Exception stack(0xffff00000a5ebec0 to 0xffff00000a5ec000) [ 305.499429] bec0: 0000000000000001 000000001c4cf5e0 0000000000000009 000000001c4cf5e0 [ 305.499674] bee0: 574f4c465245564f 0000000000000000 0000000000000000 8000000080808080 [ 305.499904] bf00: 0000000000000040 0000000000000038 fefefeff1b4bc2ff 7f7f7f7f7f7fff7f [ 305.500189] bf20: 0101010101010101 0000000000000000 000000000047a4c8 0000000000000038 [ 305.500712] bf40: 0000000000000000 0000ffffa2601280 0000ffffc63f6068 00000000004b5000 [ 305.501241] bf60: 0000000000000001 000000001c4cf5e0 0000000000000009 000000001c4cf5e0 [ 305.501791] bf80: 0000000000000020 0000000000000000 00000000004b5000 000000001c4cc458 [ 305.502314] bfa0: 0000000000000000 0000ffffc63f7950 000000000040a3c4 0000ffffc63f70e0 [ 305.502762] bfc0: 0000ffffa2601268 0000000080000000 0000000000000001 0000000000000040 [ 305.503207] bfe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000 [ 305.503680] [<ffff000008082fb0>] el0_svc_naked+0x24/0x28 [ 305.504720] Kernel Offset: disabled [ 305.505189] CPU features: 0x002082 [ 305.505473] Memory Limit: none [ 305.506181] ---[ end Kernel panic - not syncing: kernel stack overflow This patch was co-authored by Ard Biesheuvel and Mark Rutland. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Will Deacon <will.deacon@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: James Morse <james.morse@arm.com>
2017-07-15 03:30:35 +08:00
#ifdef CONFIG_VMAP_STACK
/*
* We detected an overflow in kernel_ventry, which switched to the
* overflow stack. Stash the exception regs, and head to our overflow
* handler.
*/
__bad_stack:
/* Restore the original x0 value */
mrs x0, tpidrro_el0
/*
* Store the original GPRs to the new stack. The orginal SP (minus
* S_FRAME_SIZE) was stashed in tpidr_el0 by kernel_ventry.
*/
sub sp, sp, #S_FRAME_SIZE
kernel_entry 1
mrs x0, tpidr_el0
add x0, x0, #S_FRAME_SIZE
str x0, [sp, #S_SP]
/* Stash the regs for handle_bad_stack */
mov x0, sp
/* Time to die */
bl handle_bad_stack
ASM_BUG()
#endif /* CONFIG_VMAP_STACK */
/*
* Invalid mode handlers
*/
.macro inv_entry, el, reason, regsize = 64
kernel_entry \el, \regsize
mov x0, sp
mov x1, #\reason
mrs x2, esr_el1
bl bad_mode
ASM_BUG()
.endm
el0_sync_invalid:
inv_entry 0, BAD_SYNC
ENDPROC(el0_sync_invalid)
el0_irq_invalid:
inv_entry 0, BAD_IRQ
ENDPROC(el0_irq_invalid)
el0_fiq_invalid:
inv_entry 0, BAD_FIQ
ENDPROC(el0_fiq_invalid)
el0_error_invalid:
inv_entry 0, BAD_ERROR
ENDPROC(el0_error_invalid)
#ifdef CONFIG_COMPAT
el0_fiq_invalid_compat:
inv_entry 0, BAD_FIQ, 32
ENDPROC(el0_fiq_invalid_compat)
el0_error_invalid_compat:
inv_entry 0, BAD_ERROR, 32
ENDPROC(el0_error_invalid_compat)
#endif
el1_sync_invalid:
inv_entry 1, BAD_SYNC
ENDPROC(el1_sync_invalid)
el1_irq_invalid:
inv_entry 1, BAD_IRQ
ENDPROC(el1_irq_invalid)
el1_fiq_invalid:
inv_entry 1, BAD_FIQ
ENDPROC(el1_fiq_invalid)
el1_error_invalid:
inv_entry 1, BAD_ERROR
ENDPROC(el1_error_invalid)
/*
* EL1 mode handlers.
*/
.align 6
el1_sync:
kernel_entry 1
mrs x1, esr_el1 // read the syndrome register
lsr x24, x1, #ESR_ELx_EC_SHIFT // exception class
cmp x24, #ESR_ELx_EC_DABT_CUR // data abort in EL1
b.eq el1_da
arm64: Handle el1 synchronous instruction aborts cleanly Executing from a non-executable area gives an ugly message: lkdtm: Performing direct entry EXEC_RODATA lkdtm: attempting ok execution at ffff0000084c0e08 lkdtm: attempting bad execution at ffff000008880700 Bad mode in Synchronous Abort handler detected on CPU2, code 0x8400000e -- IABT (current EL) CPU: 2 PID: 998 Comm: sh Not tainted 4.7.0-rc2+ #13 Hardware name: linux,dummy-virt (DT) task: ffff800077e35780 ti: ffff800077970000 task.ti: ffff800077970000 PC is at lkdtm_rodata_do_nothing+0x0/0x8 LR is at execute_location+0x74/0x88 The 'IABT (current EL)' indicates the error but it's a bit cryptic without knowledge of the ARM ARM. There is also no indication of the specific address which triggered the fault. The increase in kernel page permissions makes hitting this case more likely as well. Handling the case in the vectors gives a much more familiar looking error message: lkdtm: Performing direct entry EXEC_RODATA lkdtm: attempting ok execution at ffff0000084c0840 lkdtm: attempting bad execution at ffff000008880680 Unable to handle kernel paging request at virtual address ffff000008880680 pgd = ffff8000089b2000 [ffff000008880680] *pgd=00000000489b4003, *pud=0000000048904003, *pmd=0000000000000000 Internal error: Oops: 8400000e [#1] PREEMPT SMP Modules linked in: CPU: 1 PID: 997 Comm: sh Not tainted 4.7.0-rc1+ #24 Hardware name: linux,dummy-virt (DT) task: ffff800077f9f080 ti: ffff800008a1c000 task.ti: ffff800008a1c000 PC is at lkdtm_rodata_do_nothing+0x0/0x8 LR is at execute_location+0x74/0x88 Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Laura Abbott <labbott@redhat.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-08-10 09:25:26 +08:00
cmp x24, #ESR_ELx_EC_IABT_CUR // instruction abort in EL1
b.eq el1_ia
cmp x24, #ESR_ELx_EC_SYS64 // configurable trap
b.eq el1_undef
cmp x24, #ESR_ELx_EC_SP_ALIGN // stack alignment exception
b.eq el1_sp_pc
cmp x24, #ESR_ELx_EC_PC_ALIGN // pc alignment exception
b.eq el1_sp_pc
cmp x24, #ESR_ELx_EC_UNKNOWN // unknown exception in EL1
b.eq el1_undef
cmp x24, #ESR_ELx_EC_BREAKPT_CUR // debug exception in EL1
b.ge el1_dbg
b el1_inv
arm64: Handle el1 synchronous instruction aborts cleanly Executing from a non-executable area gives an ugly message: lkdtm: Performing direct entry EXEC_RODATA lkdtm: attempting ok execution at ffff0000084c0e08 lkdtm: attempting bad execution at ffff000008880700 Bad mode in Synchronous Abort handler detected on CPU2, code 0x8400000e -- IABT (current EL) CPU: 2 PID: 998 Comm: sh Not tainted 4.7.0-rc2+ #13 Hardware name: linux,dummy-virt (DT) task: ffff800077e35780 ti: ffff800077970000 task.ti: ffff800077970000 PC is at lkdtm_rodata_do_nothing+0x0/0x8 LR is at execute_location+0x74/0x88 The 'IABT (current EL)' indicates the error but it's a bit cryptic without knowledge of the ARM ARM. There is also no indication of the specific address which triggered the fault. The increase in kernel page permissions makes hitting this case more likely as well. Handling the case in the vectors gives a much more familiar looking error message: lkdtm: Performing direct entry EXEC_RODATA lkdtm: attempting ok execution at ffff0000084c0840 lkdtm: attempting bad execution at ffff000008880680 Unable to handle kernel paging request at virtual address ffff000008880680 pgd = ffff8000089b2000 [ffff000008880680] *pgd=00000000489b4003, *pud=0000000048904003, *pmd=0000000000000000 Internal error: Oops: 8400000e [#1] PREEMPT SMP Modules linked in: CPU: 1 PID: 997 Comm: sh Not tainted 4.7.0-rc1+ #24 Hardware name: linux,dummy-virt (DT) task: ffff800077f9f080 ti: ffff800008a1c000 task.ti: ffff800008a1c000 PC is at lkdtm_rodata_do_nothing+0x0/0x8 LR is at execute_location+0x74/0x88 Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Laura Abbott <labbott@redhat.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-08-10 09:25:26 +08:00
el1_ia:
/*
* Fall through to the Data abort case
*/
el1_da:
/*
* Data abort handling
*/
mrs x3, far_el1
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
// re-enable interrupts if they were enabled in the aborted context
tbnz x23, #7, 1f // PSR_I_BIT
enable_irq
1:
clear_address_tag x0, x3
mov x2, sp // struct pt_regs
bl do_mem_abort
// disable interrupts before pulling preserved data off the stack
disable_irq
kernel_exit 1
el1_sp_pc:
/*
* Stack or PC alignment exception handling
*/
mrs x0, far_el1
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
mov x2, sp
bl do_sp_pc_abort
ASM_BUG()
el1_undef:
/*
* Undefined instruction
*/
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
mov x0, sp
bl do_undefinstr
ASM_BUG()
el1_dbg:
/*
* Debug exception handling
*/
cmp x24, #ESR_ELx_EC_BRK64 // if BRK64
cinc x24, x24, eq // set bit '0'
tbz x24, #0, el1_inv // EL1 only
mrs x0, far_el1
mov x2, sp // struct pt_regs
bl do_debug_exception
kernel_exit 1
el1_inv:
// TODO: add support for undefined instructions in kernel mode
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
mov x0, sp
mov x2, x1
mov x1, #BAD_SYNC
bl bad_mode
ASM_BUG()
ENDPROC(el1_sync)
.align 6
el1_irq:
kernel_entry 1
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
#ifdef CONFIG_TRACE_IRQFLAGS
bl trace_hardirqs_off
#endif
irq_handler
#ifdef CONFIG_PREEMPT
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr w24, [tsk, #TSK_TI_PREEMPT] // get preempt count
cbnz w24, 1f // preempt count != 0
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x0, [tsk, #TSK_TI_FLAGS] // get flags
tbz x0, #TIF_NEED_RESCHED, 1f // needs rescheduling?
bl el1_preempt
1:
#endif
#ifdef CONFIG_TRACE_IRQFLAGS
bl trace_hardirqs_on
#endif
kernel_exit 1
ENDPROC(el1_irq)
#ifdef CONFIG_PREEMPT
el1_preempt:
mov x24, lr
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
1: bl preempt_schedule_irq // irq en/disable is done inside
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x0, [tsk, #TSK_TI_FLAGS] // get new tasks TI_FLAGS
tbnz x0, #TIF_NEED_RESCHED, 1b // needs rescheduling?
ret x24
#endif
/*
* EL0 mode handlers.
*/
.align 6
el0_sync:
kernel_entry 0
mrs x25, esr_el1 // read the syndrome register
lsr x24, x25, #ESR_ELx_EC_SHIFT // exception class
cmp x24, #ESR_ELx_EC_SVC64 // SVC in 64-bit state
b.eq el0_svc
cmp x24, #ESR_ELx_EC_DABT_LOW // data abort in EL0
b.eq el0_da
cmp x24, #ESR_ELx_EC_IABT_LOW // instruction abort in EL0
b.eq el0_ia
cmp x24, #ESR_ELx_EC_FP_ASIMD // FP/ASIMD access
b.eq el0_fpsimd_acc
cmp x24, #ESR_ELx_EC_FP_EXC64 // FP/ASIMD exception
b.eq el0_fpsimd_exc
cmp x24, #ESR_ELx_EC_SYS64 // configurable trap
b.eq el0_sys
cmp x24, #ESR_ELx_EC_SP_ALIGN // stack alignment exception
b.eq el0_sp_pc
cmp x24, #ESR_ELx_EC_PC_ALIGN // pc alignment exception
b.eq el0_sp_pc
cmp x24, #ESR_ELx_EC_UNKNOWN // unknown exception in EL0
b.eq el0_undef
cmp x24, #ESR_ELx_EC_BREAKPT_LOW // debug exception in EL0
b.ge el0_dbg
b el0_inv
#ifdef CONFIG_COMPAT
.align 6
el0_sync_compat:
kernel_entry 0, 32
mrs x25, esr_el1 // read the syndrome register
lsr x24, x25, #ESR_ELx_EC_SHIFT // exception class
cmp x24, #ESR_ELx_EC_SVC32 // SVC in 32-bit state
b.eq el0_svc_compat
cmp x24, #ESR_ELx_EC_DABT_LOW // data abort in EL0
b.eq el0_da
cmp x24, #ESR_ELx_EC_IABT_LOW // instruction abort in EL0
b.eq el0_ia
cmp x24, #ESR_ELx_EC_FP_ASIMD // FP/ASIMD access
b.eq el0_fpsimd_acc
cmp x24, #ESR_ELx_EC_FP_EXC32 // FP/ASIMD exception
b.eq el0_fpsimd_exc
cmp x24, #ESR_ELx_EC_PC_ALIGN // pc alignment exception
b.eq el0_sp_pc
cmp x24, #ESR_ELx_EC_UNKNOWN // unknown exception in EL0
b.eq el0_undef
cmp x24, #ESR_ELx_EC_CP15_32 // CP15 MRC/MCR trap
b.eq el0_undef
cmp x24, #ESR_ELx_EC_CP15_64 // CP15 MRRC/MCRR trap
b.eq el0_undef
cmp x24, #ESR_ELx_EC_CP14_MR // CP14 MRC/MCR trap
b.eq el0_undef
cmp x24, #ESR_ELx_EC_CP14_LS // CP14 LDC/STC trap
b.eq el0_undef
cmp x24, #ESR_ELx_EC_CP14_64 // CP14 MRRC/MCRR trap
b.eq el0_undef
cmp x24, #ESR_ELx_EC_BREAKPT_LOW // debug exception in EL0
b.ge el0_dbg
b el0_inv
el0_svc_compat:
/*
* AArch32 syscall handling
*/
adrp stbl, compat_sys_call_table // load compat syscall table pointer
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
mov wscno, w7 // syscall number in w7 (r7)
mov wsc_nr, #__NR_compat_syscalls
b el0_svc_naked
.align 6
el0_irq_compat:
kernel_entry 0, 32
b el0_irq_naked
#endif
el0_da:
/*
* Data abort handling
*/
mrs x26, far_el1
// enable interrupts before calling the main handler
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg_and_irq
ct_user_exit
clear_address_tag x0, x26
mov x1, x25
mov x2, sp
bl do_mem_abort
b ret_to_user
el0_ia:
/*
* Instruction abort handling
*/
mrs x26, far_el1
// enable interrupts before calling the main handler
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg_and_irq
ct_user_exit
mov x0, x26
mov x1, x25
mov x2, sp
bl do_mem_abort
b ret_to_user
el0_fpsimd_acc:
/*
* Floating Point or Advanced SIMD access
*/
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
ct_user_exit
mov x0, x25
mov x1, sp
bl do_fpsimd_acc
b ret_to_user
el0_fpsimd_exc:
/*
* Floating Point or Advanced SIMD exception
*/
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
ct_user_exit
mov x0, x25
mov x1, sp
bl do_fpsimd_exc
b ret_to_user
el0_sp_pc:
/*
* Stack or PC alignment exception handling
*/
mrs x26, far_el1
// enable interrupts before calling the main handler
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg_and_irq
arm64: entry: fix context tracking for el0_sp_pc Commit 6c81fe7925cc4c42 ("arm64: enable context tracking") did not update el0_sp_pc to use ct_user_exit, but this appears to have been unintentional. In commit 6ab6463aeb5fbc75 ("arm64: adjust el0_sync so that a function can be called") we made x0 available, and in the return to userspace we call ct_user_enter in the kernel_exit macro. Due to this, we currently don't correctly inform RCU of the user->kernel transition, and may erroneously account for time spent in the kernel as if we were in an extended quiescent state when CONFIG_CONTEXT_TRACKING is enabled. As we do record the kernel->user transition, a userspace application making accesses from an unaligned stack pointer can demonstrate the imbalance, provoking the following warning: ------------[ cut here ]------------ WARNING: CPU: 2 PID: 3660 at kernel/context_tracking.c:75 context_tracking_enter+0xd8/0xe4() Modules linked in: CPU: 2 PID: 3660 Comm: a.out Not tainted 4.1.0-rc7+ #8 Hardware name: ARM Juno development board (r0) (DT) Call trace: [<ffffffc000089914>] dump_backtrace+0x0/0x124 [<ffffffc000089a48>] show_stack+0x10/0x1c [<ffffffc0005b3cbc>] dump_stack+0x84/0xc8 [<ffffffc0000b3214>] warn_slowpath_common+0x98/0xd0 [<ffffffc0000b330c>] warn_slowpath_null+0x14/0x20 [<ffffffc00013ada4>] context_tracking_enter+0xd4/0xe4 [<ffffffc0005b534c>] preempt_schedule_irq+0xd4/0x114 [<ffffffc00008561c>] el1_preempt+0x4/0x28 [<ffffffc0001b8040>] exit_files+0x38/0x4c [<ffffffc0000b5b94>] do_exit+0x430/0x978 [<ffffffc0000b614c>] do_group_exit+0x40/0xd4 [<ffffffc0000c0208>] get_signal+0x23c/0x4f4 [<ffffffc0000890b4>] do_signal+0x1ac/0x518 [<ffffffc000089650>] do_notify_resume+0x5c/0x68 ---[ end trace 963c192600337066 ]--- This patch adds the missing ct_user_exit to the el0_sp_pc entry path, correcting the context tracking for this case. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Fixes: 6c81fe7925cc ("arm64: enable context tracking") Cc: <stable@vger.kernel.org> # v3.17+ Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-06-15 23:40:27 +08:00
ct_user_exit
mov x0, x26
mov x1, x25
mov x2, sp
bl do_sp_pc_abort
b ret_to_user
el0_undef:
/*
* Undefined instruction
*/
// enable interrupts before calling the main handler
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg_and_irq
ct_user_exit
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
mov x0, sp
bl do_undefinstr
b ret_to_user
el0_sys:
/*
* System instructions, for trapped cache maintenance instructions
*/
enable_dbg_and_irq
ct_user_exit
mov x0, x25
mov x1, sp
bl do_sysinstr
b ret_to_user
el0_dbg:
/*
* Debug exception handling
*/
tbnz x24, #0, el0_inv // EL0 only
mrs x0, far_el1
mov x1, x25
mov x2, sp
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
bl do_debug_exception
enable_dbg
ct_user_exit
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
b ret_to_user
el0_inv:
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg
ct_user_exit
mov x0, sp
mov x1, #BAD_SYNC
mov x2, x25
bl bad_el0_sync
b ret_to_user
ENDPROC(el0_sync)
.align 6
el0_irq:
kernel_entry 0
el0_irq_naked:
enable_dbg
#ifdef CONFIG_TRACE_IRQFLAGS
bl trace_hardirqs_off
#endif
ct_user_exit
irq_handler
#ifdef CONFIG_TRACE_IRQFLAGS
bl trace_hardirqs_on
#endif
b ret_to_user
ENDPROC(el0_irq)
/*
* This is the fast syscall return path. We do as little as possible here,
* and this includes saving x0 back into the kernel stack.
*/
ret_fast_syscall:
disable_irq // disable interrupts
str x0, [sp, #S_X0] // returned x0
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x1, [tsk, #TSK_TI_FLAGS] // re-check for syscall tracing
and x2, x1, #_TIF_SYSCALL_WORK
cbnz x2, ret_fast_syscall_trace
and x2, x1, #_TIF_WORK_MASK
cbnz x2, work_pending
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_step_tsk x1, x2
kernel_exit 0
ret_fast_syscall_trace:
enable_irq // enable interrupts
b __sys_trace_return_skipped // we already saved x0
/*
* Ok, we need to do extra processing, enter the slow path.
*/
work_pending:
mov x0, sp // 'regs'
bl do_notify_resume
arm64: Add trace_hardirqs_off annotation in ret_to_user When a kernel is built with CONFIG_TRACE_IRQFLAGS the following warning is produced when entering userspace for the first time: WARNING: at /work/Linux/linux-2.6-aarch64/kernel/locking/lockdep.c:3519 Modules linked in: CPU: 1 PID: 1 Comm: systemd Not tainted 4.4.0-rc3+ #639 Hardware name: Juno (DT) task: ffffffc9768a0000 ti: ffffffc9768a8000 task.ti: ffffffc9768a8000 PC is at check_flags.part.22+0x19c/0x1a8 LR is at check_flags.part.22+0x19c/0x1a8 pc : [<ffffffc0000fba6c>] lr : [<ffffffc0000fba6c>] pstate: 600001c5 sp : ffffffc9768abe10 x29: ffffffc9768abe10 x28: ffffffc9768a8000 x27: 0000000000000000 x26: 0000000000000001 x25: 00000000000000a6 x24: ffffffc00064be6c x23: ffffffc0009f249e x22: ffffffc9768a0000 x21: ffffffc97fea5480 x20: 00000000000001c0 x19: ffffffc00169a000 x18: 0000005558cc7b58 x17: 0000007fb78e3180 x16: 0000005558d2e238 x15: ffffffffffffffff x14: 0ffffffffffffffd x13: 0000000000000008 x12: 0101010101010101 x11: 7f7f7f7f7f7f7f7f x10: fefefefefefeff63 x9 : 7f7f7f7f7f7f7f7f x8 : 6e655f7371726964 x7 : 0000000000000001 x6 : ffffffc0001079c4 x5 : 0000000000000000 x4 : 0000000000000001 x3 : ffffffc001698438 x2 : 0000000000000000 x1 : ffffffc9768a0000 x0 : 000000000000002e Call trace: [<ffffffc0000fba6c>] check_flags.part.22+0x19c/0x1a8 [<ffffffc0000fc440>] lock_is_held+0x80/0x98 [<ffffffc00064bafc>] __schedule+0x404/0x730 [<ffffffc00064be6c>] schedule+0x44/0xb8 [<ffffffc000085bb0>] ret_to_user+0x0/0x24 possible reason: unannotated irqs-off. irq event stamp: 502169 hardirqs last enabled at (502169): [<ffffffc000085a98>] el0_irq_naked+0x1c/0x24 hardirqs last disabled at (502167): [<ffffffc0000bb3bc>] __do_softirq+0x17c/0x298 softirqs last enabled at (502168): [<ffffffc0000bb43c>] __do_softirq+0x1fc/0x298 softirqs last disabled at (502143): [<ffffffc0000bb830>] irq_exit+0xa0/0xf0 This happens because we disable interrupts in ret_to_user before calling schedule() in work_resched. This patch adds the necessary trace_hardirqs_off annotation. Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Reported-by: Mark Rutland <mark.rutland@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-12-04 20:42:29 +08:00
#ifdef CONFIG_TRACE_IRQFLAGS
bl trace_hardirqs_on // enabled while in userspace
arm64: Add trace_hardirqs_off annotation in ret_to_user When a kernel is built with CONFIG_TRACE_IRQFLAGS the following warning is produced when entering userspace for the first time: WARNING: at /work/Linux/linux-2.6-aarch64/kernel/locking/lockdep.c:3519 Modules linked in: CPU: 1 PID: 1 Comm: systemd Not tainted 4.4.0-rc3+ #639 Hardware name: Juno (DT) task: ffffffc9768a0000 ti: ffffffc9768a8000 task.ti: ffffffc9768a8000 PC is at check_flags.part.22+0x19c/0x1a8 LR is at check_flags.part.22+0x19c/0x1a8 pc : [<ffffffc0000fba6c>] lr : [<ffffffc0000fba6c>] pstate: 600001c5 sp : ffffffc9768abe10 x29: ffffffc9768abe10 x28: ffffffc9768a8000 x27: 0000000000000000 x26: 0000000000000001 x25: 00000000000000a6 x24: ffffffc00064be6c x23: ffffffc0009f249e x22: ffffffc9768a0000 x21: ffffffc97fea5480 x20: 00000000000001c0 x19: ffffffc00169a000 x18: 0000005558cc7b58 x17: 0000007fb78e3180 x16: 0000005558d2e238 x15: ffffffffffffffff x14: 0ffffffffffffffd x13: 0000000000000008 x12: 0101010101010101 x11: 7f7f7f7f7f7f7f7f x10: fefefefefefeff63 x9 : 7f7f7f7f7f7f7f7f x8 : 6e655f7371726964 x7 : 0000000000000001 x6 : ffffffc0001079c4 x5 : 0000000000000000 x4 : 0000000000000001 x3 : ffffffc001698438 x2 : 0000000000000000 x1 : ffffffc9768a0000 x0 : 000000000000002e Call trace: [<ffffffc0000fba6c>] check_flags.part.22+0x19c/0x1a8 [<ffffffc0000fc440>] lock_is_held+0x80/0x98 [<ffffffc00064bafc>] __schedule+0x404/0x730 [<ffffffc00064be6c>] schedule+0x44/0xb8 [<ffffffc000085bb0>] ret_to_user+0x0/0x24 possible reason: unannotated irqs-off. irq event stamp: 502169 hardirqs last enabled at (502169): [<ffffffc000085a98>] el0_irq_naked+0x1c/0x24 hardirqs last disabled at (502167): [<ffffffc0000bb3bc>] __do_softirq+0x17c/0x298 softirqs last enabled at (502168): [<ffffffc0000bb43c>] __do_softirq+0x1fc/0x298 softirqs last disabled at (502143): [<ffffffc0000bb830>] irq_exit+0xa0/0xf0 This happens because we disable interrupts in ret_to_user before calling schedule() in work_resched. This patch adds the necessary trace_hardirqs_off annotation. Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Reported-by: Mark Rutland <mark.rutland@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-12-04 20:42:29 +08:00
#endif
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x1, [tsk, #TSK_TI_FLAGS] // re-check for single-step
b finish_ret_to_user
/*
* "slow" syscall return path.
*/
ret_to_user:
disable_irq // disable interrupts
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x1, [tsk, #TSK_TI_FLAGS]
and x2, x1, #_TIF_WORK_MASK
cbnz x2, work_pending
finish_ret_to_user:
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_step_tsk x1, x2
kernel_exit 0
ENDPROC(ret_to_user)
/*
* SVC handler.
*/
.align 6
el0_svc:
adrp stbl, sys_call_table // load syscall table pointer
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
mov wscno, w8 // syscall number in w8
mov wsc_nr, #__NR_syscalls
el0_svc_naked: // compat entry point
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
stp x0, xscno, [sp, #S_ORIG_X0] // save the original x0 and syscall number
arm64: debug: avoid accessing mdscr_el1 on fault paths where possible Since mdscr_el1 is part of the debug register group, it is highly likely to be trapped by a hypervisor to prevent virtual machines from debugging (buggering?) each other. Unfortunately, this absolutely destroys our performance, since we access the register on many of our low-level fault handling paths to keep track of the various debug state machines. This patch removes our dependency on mdscr_el1 in the case that debugging is not being used. More specifically we: - Use TIF_SINGLESTEP to indicate that a task is stepping at EL0 and avoid disabling step in the MDSCR when we don't need to. MDSCR_EL1.SS handling is moved to kernel_entry, when trapping from userspace. - Ensure debug exceptions are re-enabled on *all* exception entry paths, even the debug exception handling path (where we re-enable exceptions after invoking the handler). Since we can now rely on MDSCR_EL1.SS being cleared by the entry code, exception handlers can usually enable debug immediately before enabling interrupts. - Remove all debug exception unmasking from ret_to_user and el1_preempt, since we will never get here with debug exceptions masked. This results in a slight change to kernel debug behaviour, where we now step into interrupt handlers and data aborts from EL1 when debugging the kernel, which is actually a useful thing to do. A side-effect of this is that it *does* potentially prevent stepping off {break,watch}points when there is a high-frequency interrupt source (e.g. a timer), so a debugger would need to use either breakpoints or manually disable interrupts to get around this issue. With this patch applied, guest performance is restored under KVM when debug register accesses are trapped (and we get a measurable performance increase on the host on Cortex-A57 too). Cc: Ian Campbell <ian.campbell@citrix.com> Tested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2014-04-30 02:04:06 +08:00
enable_dbg_and_irq
ct_user_exit 1
arm64: split thread_info from task stack This patch moves arm64's struct thread_info from the task stack into task_struct. This protects thread_info from corruption in the case of stack overflows, and makes its address harder to determine if stack addresses are leaked, making a number of attacks more difficult. Precise detection and handling of overflow is left for subsequent patches. Largely, this involves changing code to store the task_struct in sp_el0, and acquire the thread_info from the task struct. Core code now implements current_thread_info(), and as noted in <linux/sched.h> this relies on offsetof(task_struct, thread_info) == 0, enforced by core code. This change means that the 'tsk' register used in entry.S now points to a task_struct, rather than a thread_info as it used to. To make this clear, the TI_* field offsets are renamed to TSK_TI_*, with asm-offsets appropriately updated to account for the structural change. Userspace clobbers sp_el0, and we can no longer restore this from the stack. Instead, the current task is cached in a per-cpu variable that we can safely access from early assembly as interrupts are disabled (and we are thus not preemptible). Both secondary entry and idle are updated to stash the sp and task pointer separately. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Tested-by: Laura Abbott <labbott@redhat.com> Cc: AKASHI Takahiro <takahiro.akashi@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: James Morse <james.morse@arm.com> Cc: Kees Cook <keescook@chromium.org> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-11-04 04:23:13 +08:00
ldr x16, [tsk, #TSK_TI_FLAGS] // check for syscall hooks
tst x16, #_TIF_SYSCALL_WORK
b.ne __sys_trace
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
cmp wscno, wsc_nr // check upper syscall limit
b.hs ni_sys
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
ldr x16, [stbl, xscno, lsl #3] // address in the syscall table
blr x16 // call sys_* routine
b ret_fast_syscall
ni_sys:
mov x0, sp
bl do_ni_syscall
b ret_fast_syscall
ENDPROC(el0_svc)
/*
* This is the really slow path. We're going to be doing context
* switches, and waiting for our parent to respond.
*/
__sys_trace:
cmp wscno, #NO_SYSCALL // user-issued syscall(-1)?
b.ne 1f
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
mov x0, #-ENOSYS // set default errno if so
str x0, [sp, #S_X0]
1: mov x0, sp
bl syscall_trace_enter
cmp w0, #NO_SYSCALL // skip the syscall?
b.eq __sys_trace_return_skipped
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
mov wscno, w0 // syscall number (possibly new)
mov x1, sp // pointer to regs
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
cmp wscno, wsc_nr // check upper syscall limit
b.hs __ni_sys_trace
ldp x0, x1, [sp] // restore the syscall args
ldp x2, x3, [sp, #S_X2]
ldp x4, x5, [sp, #S_X4]
ldp x6, x7, [sp, #S_X6]
arm64: syscallno is secretly an int, make it official The upper 32 bits of the syscallno field in thread_struct are handled inconsistently, being sometimes zero extended and sometimes sign-extended. In fact, only the lower 32 bits seem to have any real significance for the behaviour of the code: it's been OK to handle the upper bits inconsistently because they don't matter. Currently, the only place I can find where those bits are significant is in calling trace_sys_enter(), which may be unintentional: for example, if a compat tracer attempts to cancel a syscall by passing -1 to (COMPAT_)PTRACE_SET_SYSCALL at the syscall-enter-stop, it will be traced as syscall 4294967295 rather than -1 as might be expected (and as occurs for a native tracer doing the same thing). Elsewhere, reads of syscallno cast it to an int or truncate it. There's also a conspicuous amount of code and casting to bodge around the fact that although semantically an int, syscallno is stored as a u64. Let's not pretend any more. In order to preserve the stp x instruction that stores the syscall number in entry.S, this patch special-cases the layout of struct pt_regs for big endian so that the newly 32-bit syscallno field maps onto the low bits of the stored value. This is not beautiful, but benchmarking of the getpid syscall on Juno suggests indicates a minor slowdown if the stp is split into an stp x and stp w. Signed-off-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2017-08-01 22:35:53 +08:00
ldr x16, [stbl, xscno, lsl #3] // address in the syscall table
blr x16 // call sys_* routine
__sys_trace_return:
str x0, [sp, #S_X0] // save returned x0
__sys_trace_return_skipped:
mov x0, sp
bl syscall_trace_exit
b ret_to_user
__ni_sys_trace:
mov x0, sp
bl do_ni_syscall
b __sys_trace_return
.popsection // .entry.text
/*
* Special system call wrappers.
*/
ENTRY(sys_rt_sigreturn_wrapper)
mov x0, sp
b sys_rt_sigreturn
ENDPROC(sys_rt_sigreturn_wrapper)
/*
* Register switch for AArch64. The callee-saved registers need to be saved
* and restored. On entry:
* x0 = previous task_struct (must be preserved across the switch)
* x1 = next task_struct
* Previous and next are guaranteed not to be the same.
*
*/
ENTRY(cpu_switch_to)
mov x10, #THREAD_CPU_CONTEXT
add x8, x0, x10
mov x9, sp
stp x19, x20, [x8], #16 // store callee-saved registers
stp x21, x22, [x8], #16
stp x23, x24, [x8], #16
stp x25, x26, [x8], #16
stp x27, x28, [x8], #16
stp x29, x9, [x8], #16
str lr, [x8]
add x8, x1, x10
ldp x19, x20, [x8], #16 // restore callee-saved registers
ldp x21, x22, [x8], #16
ldp x23, x24, [x8], #16
ldp x25, x26, [x8], #16
ldp x27, x28, [x8], #16
ldp x29, x9, [x8], #16
ldr lr, [x8]
mov sp, x9
msr sp_el0, x1
ret
ENDPROC(cpu_switch_to)
NOKPROBE(cpu_switch_to)
/*
* This is how we return from a fork.
*/
ENTRY(ret_from_fork)
bl schedule_tail
cbz x19, 1f // not a kernel thread
mov x0, x20
blr x19
1: get_thread_info tsk
b ret_to_user
ENDPROC(ret_from_fork)
NOKPROBE(ret_from_fork)