2012-03-05 18:49:27 +07:00
|
|
|
/*
|
|
|
|
* 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>
|
|
|
|
|
2015-06-01 16:47:41 +07:00
|
|
|
#include <asm/alternative.h>
|
2012-03-05 18:49:27 +07:00
|
|
|
#include <asm/assembler.h>
|
|
|
|
#include <asm/asm-offsets.h>
|
2015-03-24 02:07:02 +07:00
|
|
|
#include <asm/cpufeature.h>
|
2012-03-05 18:49:27 +07:00
|
|
|
#include <asm/errno.h>
|
2013-04-08 23:17:03 +07:00
|
|
|
#include <asm/esr.h>
|
2012-03-05 18:49:27 +07:00
|
|
|
#include <asm/thread_info.h>
|
|
|
|
#include <asm/unistd.h>
|
|
|
|
|
2014-05-31 02:34:15 +07:00
|
|
|
/*
|
|
|
|
* 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
|
|
|
|
|
2012-03-05 18:49:27 +07:00
|
|
|
/*
|
|
|
|
* Bad Abort numbers
|
|
|
|
*-----------------
|
|
|
|
*/
|
|
|
|
#define BAD_SYNC 0
|
|
|
|
#define BAD_IRQ 1
|
|
|
|
#define BAD_FIQ 2
|
|
|
|
#define BAD_ERROR 3
|
|
|
|
|
|
|
|
.macro kernel_entry, el, regsize = 64
|
2014-09-29 18:26:41 +07:00
|
|
|
sub sp, sp, #S_FRAME_SIZE
|
2012-03-05 18:49:27 +07:00
|
|
|
.if \regsize == 32
|
|
|
|
mov w0, w0 // zero upper 32 bits of x0
|
|
|
|
.endif
|
2014-09-29 18:26:41 +07:00
|
|
|
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]
|
|
|
|
|
2012-03-05 18:49:27 +07:00
|
|
|
.if \el == 0
|
|
|
|
mrs x21, sp_el0
|
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 01:04:06 +07:00
|
|
|
get_thread_info tsk // Ensure MDSCR_EL1.SS is clear,
|
|
|
|
ldr x19, [tsk, #TI_FLAGS] // since we can unmask debug
|
|
|
|
disable_step_tsk x19, x20 // exceptions when scheduling.
|
2012-03-05 18:49:27 +07:00
|
|
|
.else
|
|
|
|
add x21, sp, #S_FRAME_SIZE
|
|
|
|
.endif
|
|
|
|
mrs x22, elr_el1
|
|
|
|
mrs x23, spsr_el1
|
|
|
|
stp lr, x21, [sp, #S_LR]
|
|
|
|
stp x22, x23, [sp, #S_PC]
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Set syscallno to -1 by default (overridden later if real syscall).
|
|
|
|
*/
|
|
|
|
.if \el == 0
|
|
|
|
mvn x21, xzr
|
|
|
|
str x21, [sp, #S_SYSCALLNO]
|
|
|
|
.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, ret = 0
|
|
|
|
ldp x21, x22, [sp, #S_PC] // load ELR, SPSR
|
|
|
|
.if \el == 0
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_enter
|
2012-03-05 18:49:27 +07:00
|
|
|
ldr x23, [sp, #S_SP] // load return stack pointer
|
2014-09-29 18:26:41 +07:00
|
|
|
msr sp_el0, x23
|
2015-03-24 02:07:02 +07:00
|
|
|
#ifdef CONFIG_ARM64_ERRATUM_845719
|
2015-07-22 18:21:03 +07:00
|
|
|
alternative_if_not ARM64_WORKAROUND_845719
|
|
|
|
nop
|
|
|
|
nop
|
2015-03-24 02:07:02 +07:00
|
|
|
#ifdef CONFIG_PID_IN_CONTEXTIDR
|
2015-07-22 18:21:03 +07:00
|
|
|
nop
|
|
|
|
#endif
|
|
|
|
alternative_else
|
|
|
|
tbz x22, #4, 1f
|
|
|
|
#ifdef CONFIG_PID_IN_CONTEXTIDR
|
|
|
|
mrs x29, contextidr_el1
|
|
|
|
msr contextidr_el1, x29
|
2015-03-24 02:07:02 +07:00
|
|
|
#else
|
2015-07-22 18:21:03 +07:00
|
|
|
msr contextidr_el1, xzr
|
2015-03-24 02:07:02 +07:00
|
|
|
#endif
|
2015-07-22 18:21:03 +07:00
|
|
|
1:
|
|
|
|
alternative_endif
|
2015-03-24 02:07:02 +07:00
|
|
|
#endif
|
2012-03-05 18:49:27 +07:00
|
|
|
.endif
|
2014-09-29 18:26:41 +07:00
|
|
|
msr elr_el1, x21 // set up the return data
|
|
|
|
msr spsr_el1, x22
|
2012-03-05 18:49:27 +07:00
|
|
|
.if \ret
|
|
|
|
ldr x1, [sp, #S_X1] // preserve x0 (syscall return)
|
|
|
|
.else
|
2014-09-29 18:26:41 +07:00
|
|
|
ldp x0, x1, [sp, #16 * 0]
|
2012-03-05 18:49:27 +07:00
|
|
|
.endif
|
2014-09-29 18:26:41 +07:00
|
|
|
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
|
2012-03-05 18:49:27 +07:00
|
|
|
eret // return to kernel
|
|
|
|
.endm
|
|
|
|
|
|
|
|
.macro get_thread_info, rd
|
|
|
|
mov \rd, sp
|
2013-07-24 00:52:31 +07:00
|
|
|
and \rd, \rd, #~(THREAD_SIZE - 1) // top of stack
|
2012-03-05 18:49:27 +07:00
|
|
|
.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.
|
|
|
|
*/
|
|
|
|
sc_nr .req x25 // number of system calls
|
|
|
|
scno .req x26 // syscall number
|
|
|
|
stbl .req x27 // syscall table pointer
|
|
|
|
tsk .req x28 // current thread_info
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Interrupt handling.
|
|
|
|
*/
|
|
|
|
.macro irq_handler
|
2014-11-22 04:50:38 +07:00
|
|
|
adrp x1, handle_arch_irq
|
|
|
|
ldr x1, [x1, #:lo12:handle_arch_irq]
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x0, sp
|
|
|
|
blr x1
|
|
|
|
.endm
|
|
|
|
|
|
|
|
.text
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Exception vectors.
|
|
|
|
*/
|
|
|
|
|
|
|
|
.align 11
|
|
|
|
ENTRY(vectors)
|
|
|
|
ventry el1_sync_invalid // Synchronous EL1t
|
|
|
|
ventry el1_irq_invalid // IRQ EL1t
|
|
|
|
ventry el1_fiq_invalid // FIQ EL1t
|
|
|
|
ventry el1_error_invalid // Error EL1t
|
|
|
|
|
|
|
|
ventry el1_sync // Synchronous EL1h
|
|
|
|
ventry el1_irq // IRQ EL1h
|
|
|
|
ventry el1_fiq_invalid // FIQ EL1h
|
|
|
|
ventry el1_error_invalid // Error EL1h
|
|
|
|
|
|
|
|
ventry el0_sync // Synchronous 64-bit EL0
|
|
|
|
ventry el0_irq // IRQ 64-bit EL0
|
|
|
|
ventry el0_fiq_invalid // FIQ 64-bit EL0
|
|
|
|
ventry el0_error_invalid // Error 64-bit EL0
|
|
|
|
|
|
|
|
#ifdef CONFIG_COMPAT
|
|
|
|
ventry el0_sync_compat // Synchronous 32-bit EL0
|
|
|
|
ventry el0_irq_compat // IRQ 32-bit EL0
|
|
|
|
ventry el0_fiq_invalid_compat // FIQ 32-bit EL0
|
|
|
|
ventry el0_error_invalid_compat // Error 32-bit EL0
|
|
|
|
#else
|
|
|
|
ventry el0_sync_invalid // Synchronous 32-bit EL0
|
|
|
|
ventry el0_irq_invalid // IRQ 32-bit EL0
|
|
|
|
ventry el0_fiq_invalid // FIQ 32-bit EL0
|
|
|
|
ventry el0_error_invalid // Error 32-bit EL0
|
|
|
|
#endif
|
|
|
|
END(vectors)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Invalid mode handlers
|
|
|
|
*/
|
|
|
|
.macro inv_entry, el, reason, regsize = 64
|
|
|
|
kernel_entry el, \regsize
|
|
|
|
mov x0, sp
|
|
|
|
mov x1, #\reason
|
|
|
|
mrs x2, esr_el1
|
|
|
|
b bad_mode
|
|
|
|
.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
|
2014-11-24 19:31:40 +07:00
|
|
|
lsr x24, x1, #ESR_ELx_EC_SHIFT // exception class
|
|
|
|
cmp x24, #ESR_ELx_EC_DABT_CUR // data abort in EL1
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el1_da
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_SYS64 // configurable trap
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el1_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_SP_ALIGN // stack alignment exception
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el1_sp_pc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_PC_ALIGN // pc alignment exception
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el1_sp_pc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_UNKNOWN // unknown exception in EL1
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el1_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_BREAKPT_CUR // debug exception in EL1
|
2012-03-05 18:49:27 +07:00
|
|
|
b.ge el1_dbg
|
|
|
|
b el1_inv
|
|
|
|
el1_da:
|
|
|
|
/*
|
|
|
|
* Data abort 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 01:04:06 +07:00
|
|
|
enable_dbg
|
2012-03-05 18:49:27 +07:00
|
|
|
// re-enable interrupts if they were enabled in the aborted context
|
|
|
|
tbnz x23, #7, 1f // PSR_I_BIT
|
|
|
|
enable_irq
|
|
|
|
1:
|
|
|
|
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 01:04:06 +07:00
|
|
|
enable_dbg
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x2, sp
|
|
|
|
b do_sp_pc_abort
|
|
|
|
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 01:04:06 +07:00
|
|
|
enable_dbg
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x0, sp
|
|
|
|
b do_undefinstr
|
|
|
|
el1_dbg:
|
|
|
|
/*
|
|
|
|
* Debug exception handling
|
|
|
|
*/
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_BRK64 // if BRK64
|
2013-12-04 12:50:20 +07:00
|
|
|
cinc x24, x24, eq // set bit '0'
|
2012-03-05 18:49:27 +07:00
|
|
|
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 01:04:06 +07:00
|
|
|
enable_dbg
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x0, sp
|
2015-07-08 00:00:49 +07:00
|
|
|
mov x2, x1
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x1, #BAD_SYNC
|
|
|
|
b bad_mode
|
|
|
|
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 01:04:06 +07:00
|
|
|
enable_dbg
|
2012-03-05 18:49:27 +07:00
|
|
|
#ifdef CONFIG_TRACE_IRQFLAGS
|
|
|
|
bl trace_hardirqs_off
|
|
|
|
#endif
|
2013-11-13 00:11:53 +07:00
|
|
|
|
2012-03-05 18:49:27 +07:00
|
|
|
irq_handler
|
2013-11-13 00:11:53 +07:00
|
|
|
|
2012-03-05 18:49:27 +07:00
|
|
|
#ifdef CONFIG_PREEMPT
|
2013-11-13 00:11:53 +07:00
|
|
|
get_thread_info tsk
|
2014-01-13 15:57:56 +07:00
|
|
|
ldr w24, [tsk, #TI_PREEMPT] // get preempt count
|
2013-11-05 03:14:58 +07:00
|
|
|
cbnz w24, 1f // preempt count != 0
|
2012-03-05 18:49:27 +07:00
|
|
|
ldr x0, [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 01:04:06 +07:00
|
|
|
1: bl preempt_schedule_irq // irq en/disable is done inside
|
2012-03-05 18:49:27 +07:00
|
|
|
ldr x0, [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
|
2014-11-24 19:31:40 +07:00
|
|
|
lsr x24, x25, #ESR_ELx_EC_SHIFT // exception class
|
|
|
|
cmp x24, #ESR_ELx_EC_SVC64 // SVC in 64-bit state
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_svc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_DABT_LOW // data abort in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_da
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_IABT_LOW // instruction abort in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_ia
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_FP_ASIMD // FP/ASIMD access
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_fpsimd_acc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_FP_EXC64 // FP/ASIMD exception
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_fpsimd_exc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_SYS64 // configurable trap
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_SP_ALIGN // stack alignment exception
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_sp_pc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_PC_ALIGN // pc alignment exception
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_sp_pc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_UNKNOWN // unknown exception in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_BREAKPT_LOW // debug exception in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
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
|
2014-11-24 19:31:40 +07:00
|
|
|
lsr x24, x25, #ESR_ELx_EC_SHIFT // exception class
|
|
|
|
cmp x24, #ESR_ELx_EC_SVC32 // SVC in 32-bit state
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_svc_compat
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_DABT_LOW // data abort in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_da
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_IABT_LOW // instruction abort in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_ia
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_FP_ASIMD // FP/ASIMD access
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_fpsimd_acc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_FP_EXC32 // FP/ASIMD exception
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_fpsimd_exc
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_UNKNOWN // unknown exception in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_CP15_32 // CP15 MRC/MCR trap
|
2013-05-24 18:02:35 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_CP15_64 // CP15 MRRC/MCRR trap
|
2013-05-24 18:02:35 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_CP14_MR // CP14 MRC/MCR trap
|
2013-05-24 18:02:35 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_CP14_LS // CP14 LDC/STC trap
|
2013-05-24 18:02:35 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_CP14_64 // CP14 MRRC/MCRR trap
|
2013-05-24 18:02:35 +07:00
|
|
|
b.eq el0_undef
|
2014-11-24 19:31:40 +07:00
|
|
|
cmp x24, #ESR_ELx_EC_BREAKPT_LOW // debug exception in EL0
|
2012-03-05 18:49:27 +07:00
|
|
|
b.ge el0_dbg
|
|
|
|
b el0_inv
|
|
|
|
el0_svc_compat:
|
|
|
|
/*
|
|
|
|
* AArch32 syscall handling
|
|
|
|
*/
|
2015-01-06 23:42:32 +07:00
|
|
|
adrp stbl, compat_sys_call_table // load compat syscall table pointer
|
2012-03-05 18:49:27 +07:00
|
|
|
uxtw scno, w7 // syscall number in w7 (r7)
|
|
|
|
mov sc_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
|
|
|
|
*/
|
2014-05-31 02:34:14 +07:00
|
|
|
mrs x26, far_el1
|
2012-03-05 18:49:27 +07:00
|
|
|
// 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 01:04:06 +07:00
|
|
|
enable_dbg_and_irq
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_exit
|
2014-05-31 02:34:14 +07:00
|
|
|
bic x0, x26, #(0xff << 56)
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x1, x25
|
|
|
|
mov x2, sp
|
2014-09-29 17:44:01 +07:00
|
|
|
bl do_mem_abort
|
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
el0_ia:
|
|
|
|
/*
|
|
|
|
* Instruction abort handling
|
|
|
|
*/
|
2014-05-31 02:34:14 +07:00
|
|
|
mrs x26, far_el1
|
2012-03-05 18:49:27 +07:00
|
|
|
// 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 01:04:06 +07:00
|
|
|
enable_dbg_and_irq
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_exit
|
2014-05-31 02:34:14 +07:00
|
|
|
mov x0, x26
|
2012-03-05 18:49:27 +07:00
|
|
|
orr x1, x25, #1 << 24 // use reserved ISS bit for instruction aborts
|
|
|
|
mov x2, sp
|
2014-09-29 17:44:01 +07:00
|
|
|
bl do_mem_abort
|
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
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 01:04:06 +07:00
|
|
|
enable_dbg
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_exit
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x0, x25
|
|
|
|
mov x1, sp
|
2014-09-29 17:44:01 +07:00
|
|
|
bl do_fpsimd_acc
|
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
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 01:04:06 +07:00
|
|
|
enable_dbg
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_exit
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x0, x25
|
|
|
|
mov x1, sp
|
2014-09-29 17:44:01 +07:00
|
|
|
bl do_fpsimd_exc
|
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
el0_sp_pc:
|
|
|
|
/*
|
|
|
|
* Stack or PC alignment exception handling
|
|
|
|
*/
|
2014-05-31 02:34:14 +07:00
|
|
|
mrs x26, far_el1
|
2012-03-05 18:49:27 +07:00
|
|
|
// 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 01:04:06 +07:00
|
|
|
enable_dbg_and_irq
|
2015-06-15 22:40:27 +07:00
|
|
|
ct_user_exit
|
2014-05-31 02:34:14 +07:00
|
|
|
mov x0, x26
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x1, x25
|
|
|
|
mov x2, sp
|
2014-09-29 17:44:01 +07:00
|
|
|
bl do_sp_pc_abort
|
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
el0_undef:
|
|
|
|
/*
|
|
|
|
* Undefined instruction
|
|
|
|
*/
|
2013-08-22 17:47:37 +07:00
|
|
|
// 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 01:04:06 +07:00
|
|
|
enable_dbg_and_irq
|
2014-05-31 02:34:15 +07:00
|
|
|
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 01:04:06 +07:00
|
|
|
mov x0, sp
|
2014-09-29 17:44:01 +07:00
|
|
|
bl do_undefinstr
|
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
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 01:04:06 +07:00
|
|
|
bl do_debug_exception
|
|
|
|
enable_dbg
|
2014-05-31 02:34:15 +07:00
|
|
|
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 01:04:06 +07:00
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
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 01:04:06 +07:00
|
|
|
enable_dbg
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_exit
|
2012-03-05 18:49:27 +07:00
|
|
|
mov x0, sp
|
|
|
|
mov x1, #BAD_SYNC
|
2015-07-08 00:00:49 +07:00
|
|
|
mov x2, x25
|
2014-09-29 17:44:01 +07:00
|
|
|
bl bad_mode
|
|
|
|
b ret_to_user
|
2012-03-05 18:49:27 +07:00
|
|
|
ENDPROC(el0_sync)
|
|
|
|
|
|
|
|
.align 6
|
|
|
|
el0_irq:
|
|
|
|
kernel_entry 0
|
|
|
|
el0_irq_naked:
|
|
|
|
enable_dbg
|
|
|
|
#ifdef CONFIG_TRACE_IRQFLAGS
|
|
|
|
bl trace_hardirqs_off
|
|
|
|
#endif
|
2013-11-13 00:11:53 +07:00
|
|
|
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_exit
|
2012-03-05 18:49:27 +07:00
|
|
|
irq_handler
|
2013-11-13 00:11:53 +07:00
|
|
|
|
2012-03-05 18:49:27 +07:00
|
|
|
#ifdef CONFIG_TRACE_IRQFLAGS
|
|
|
|
bl trace_hardirqs_on
|
|
|
|
#endif
|
|
|
|
b ret_to_user
|
|
|
|
ENDPROC(el0_irq)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 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)
|
2015-07-20 21:14:53 +07:00
|
|
|
mov x10, #THREAD_CPU_CONTEXT
|
|
|
|
add x8, x0, x10
|
2012-03-05 18:49:27 +07:00
|
|
|
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]
|
2015-07-20 21:14:53 +07:00
|
|
|
add x8, x1, x10
|
2012-03-05 18:49:27 +07:00
|
|
|
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
|
|
|
|
ret
|
|
|
|
ENDPROC(cpu_switch_to)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 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
|
2015-06-06 04:28:03 +07:00
|
|
|
ldr x1, [tsk, #TI_FLAGS] // re-check for syscall tracing
|
|
|
|
and x2, x1, #_TIF_SYSCALL_WORK
|
|
|
|
cbnz x2, ret_fast_syscall_trace
|
2012-03-05 18:49:27 +07:00
|
|
|
and x2, x1, #_TIF_WORK_MASK
|
|
|
|
cbnz x2, fast_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 01:04:06 +07:00
|
|
|
enable_step_tsk x1, x2
|
2012-03-05 18:49:27 +07:00
|
|
|
kernel_exit 0, ret = 1
|
2015-06-06 04:28:03 +07:00
|
|
|
ret_fast_syscall_trace:
|
|
|
|
enable_irq // enable interrupts
|
|
|
|
b __sys_trace_return
|
2012-03-05 18:49:27 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Ok, we need to do extra processing, enter the slow path.
|
|
|
|
*/
|
|
|
|
fast_work_pending:
|
|
|
|
str x0, [sp, #S_X0] // returned x0
|
|
|
|
work_pending:
|
|
|
|
tbnz x1, #TIF_NEED_RESCHED, work_resched
|
2014-05-08 16:20:23 +07:00
|
|
|
/* TIF_SIGPENDING, TIF_NOTIFY_RESUME or TIF_FOREIGN_FPSTATE case */
|
2012-03-05 18:49:27 +07:00
|
|
|
ldr x2, [sp, #S_PSTATE]
|
|
|
|
mov x0, sp // 'regs'
|
|
|
|
tst x2, #PSR_MODE_MASK // user mode regs?
|
|
|
|
b.ne no_work_pending // returning to kernel
|
2012-10-09 00:04:21 +07:00
|
|
|
enable_irq // enable interrupts for do_notify_resume()
|
2012-03-05 18:49:27 +07:00
|
|
|
bl do_notify_resume
|
|
|
|
b ret_to_user
|
|
|
|
work_resched:
|
|
|
|
bl schedule
|
|
|
|
|
|
|
|
/*
|
|
|
|
* "slow" syscall return path.
|
|
|
|
*/
|
2012-09-10 22:11:46 +07:00
|
|
|
ret_to_user:
|
2012-03-05 18:49:27 +07:00
|
|
|
disable_irq // disable interrupts
|
|
|
|
ldr x1, [tsk, #TI_FLAGS]
|
|
|
|
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 01:04:06 +07:00
|
|
|
enable_step_tsk x1, x2
|
2012-03-05 18:49:27 +07:00
|
|
|
no_work_pending:
|
|
|
|
kernel_exit 0, ret = 0
|
|
|
|
ENDPROC(ret_to_user)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This is how we return from a fork.
|
|
|
|
*/
|
|
|
|
ENTRY(ret_from_fork)
|
|
|
|
bl schedule_tail
|
2012-10-05 18:31:20 +07:00
|
|
|
cbz x19, 1f // not a kernel thread
|
|
|
|
mov x0, x20
|
|
|
|
blr x19
|
|
|
|
1: get_thread_info tsk
|
2012-03-05 18:49:27 +07:00
|
|
|
b ret_to_user
|
|
|
|
ENDPROC(ret_from_fork)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* SVC handler.
|
|
|
|
*/
|
|
|
|
.align 6
|
|
|
|
el0_svc:
|
|
|
|
adrp stbl, sys_call_table // load syscall table pointer
|
|
|
|
uxtw scno, w8 // syscall number in w8
|
|
|
|
mov sc_nr, #__NR_syscalls
|
|
|
|
el0_svc_naked: // compat entry point
|
|
|
|
stp x0, scno, [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 01:04:06 +07:00
|
|
|
enable_dbg_and_irq
|
2014-05-31 02:34:15 +07:00
|
|
|
ct_user_exit 1
|
2012-03-05 18:49:27 +07:00
|
|
|
|
2014-04-30 16:51:29 +07:00
|
|
|
ldr x16, [tsk, #TI_FLAGS] // check for syscall hooks
|
|
|
|
tst x16, #_TIF_SYSCALL_WORK
|
|
|
|
b.ne __sys_trace
|
2012-03-05 18:49:27 +07:00
|
|
|
cmp scno, sc_nr // check upper syscall limit
|
|
|
|
b.hs ni_sys
|
|
|
|
ldr x16, [stbl, scno, lsl #3] // address in the syscall table
|
2014-09-29 17:44:01 +07:00
|
|
|
blr x16 // call sys_* routine
|
|
|
|
b ret_fast_syscall
|
2012-03-05 18:49:27 +07:00
|
|
|
ni_sys:
|
|
|
|
mov x0, sp
|
2014-09-29 17:44:01 +07:00
|
|
|
bl do_ni_syscall
|
|
|
|
b ret_fast_syscall
|
2012-03-05 18:49:27 +07:00
|
|
|
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:
|
2014-11-28 12:26:35 +07:00
|
|
|
mov w0, #-1 // set default errno for
|
|
|
|
cmp scno, x0 // user-issued syscall(-1)
|
|
|
|
b.ne 1f
|
|
|
|
mov x0, #-ENOSYS
|
|
|
|
str x0, [sp, #S_X0]
|
|
|
|
1: mov x0, sp
|
2014-04-30 16:51:30 +07:00
|
|
|
bl syscall_trace_enter
|
2014-11-28 12:26:35 +07:00
|
|
|
cmp w0, #-1 // skip the syscall?
|
|
|
|
b.eq __sys_trace_return_skipped
|
2012-03-05 18:49:27 +07:00
|
|
|
uxtw scno, w0 // syscall number (possibly new)
|
|
|
|
mov x1, sp // pointer to regs
|
|
|
|
cmp scno, sc_nr // check upper syscall limit
|
2014-09-29 17:44:01 +07:00
|
|
|
b.hs __ni_sys_trace
|
2012-03-05 18:49:27 +07:00
|
|
|
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]
|
|
|
|
ldr x16, [stbl, scno, lsl #3] // address in the syscall table
|
2014-09-29 17:44:01 +07:00
|
|
|
blr x16 // call sys_* routine
|
2012-03-05 18:49:27 +07:00
|
|
|
|
|
|
|
__sys_trace_return:
|
2014-11-28 12:26:35 +07:00
|
|
|
str x0, [sp, #S_X0] // save returned x0
|
|
|
|
__sys_trace_return_skipped:
|
2014-04-30 16:51:30 +07:00
|
|
|
mov x0, sp
|
|
|
|
bl syscall_trace_exit
|
2012-03-05 18:49:27 +07:00
|
|
|
b ret_to_user
|
|
|
|
|
2014-09-29 17:44:01 +07:00
|
|
|
__ni_sys_trace:
|
|
|
|
mov x0, sp
|
|
|
|
bl do_ni_syscall
|
|
|
|
b __sys_trace_return
|
|
|
|
|
2012-03-05 18:49:27 +07:00
|
|
|
/*
|
|
|
|
* Special system call wrappers.
|
|
|
|
*/
|
|
|
|
ENTRY(sys_rt_sigreturn_wrapper)
|
|
|
|
mov x0, sp
|
|
|
|
b sys_rt_sigreturn
|
|
|
|
ENDPROC(sys_rt_sigreturn_wrapper)
|