mirror of
https://github.com/torvalds/linux.git
synced 2024-12-22 10:56:40 +00:00
6ebbf2ce43
ARMv6 and greater introduced a new instruction ("bx") which can be used to return from function calls. Recent CPUs perform better when the "bx lr" instruction is used rather than the "mov pc, lr" instruction, and this sequence is strongly recommended to be used by the ARM architecture manual (section A.4.1.1). We provide a new macro "ret" with all its variants for the condition code which will resolve to the appropriate instruction. Rather than doing this piecemeal, and miss some instances, change all the "mov pc" instances to use the new macro, with the exception of the "movs" instruction and the kprobes code. This allows us to detect the "mov pc, lr" case and fix it up - and also gives us the possibility of deploying this for other registers depending on the CPU selection. Reported-by: Will Deacon <will.deacon@arm.com> Tested-by: Stephen Warren <swarren@nvidia.com> # Tegra Jetson TK1 Tested-by: Robert Jarzmik <robert.jarzmik@free.fr> # mioa701_bootresume.S Tested-by: Andrew Lunn <andrew@lunn.ch> # Kirkwood Tested-by: Shawn Guo <shawn.guo@freescale.com> Tested-by: Tony Lindgren <tony@atomide.com> # OMAPs Tested-by: Gregory CLEMENT <gregory.clement@free-electrons.com> # Armada XP, 375, 385 Acked-by: Sekhar Nori <nsekhar@ti.com> # DaVinci Acked-by: Christoffer Dall <christoffer.dall@linaro.org> # kvm/hyp Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com> # PXA3xx Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> # Xen Tested-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> # ARMv7M Tested-by: Simon Horman <horms+renesas@verge.net.au> # Shmobile Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
345 lines
9.1 KiB
ArmAsm
345 lines
9.1 KiB
ArmAsm
#include <linux/init.h>
|
|
#include <linux/linkage.h>
|
|
|
|
#include <asm/assembler.h>
|
|
#include <asm/asm-offsets.h>
|
|
#include <asm/errno.h>
|
|
#include <asm/thread_info.h>
|
|
#include <asm/v7m.h>
|
|
|
|
@ Bad Abort numbers
|
|
@ -----------------
|
|
@
|
|
#define BAD_PREFETCH 0
|
|
#define BAD_DATA 1
|
|
#define BAD_ADDREXCPTN 2
|
|
#define BAD_IRQ 3
|
|
#define BAD_UNDEFINSTR 4
|
|
|
|
@
|
|
@ Most of the stack format comes from struct pt_regs, but with
|
|
@ the addition of 8 bytes for storing syscall args 5 and 6.
|
|
@ This _must_ remain a multiple of 8 for EABI.
|
|
@
|
|
#define S_OFF 8
|
|
|
|
/*
|
|
* The SWI code relies on the fact that R0 is at the bottom of the stack
|
|
* (due to slow/fast restore user regs).
|
|
*/
|
|
#if S_R0 != 0
|
|
#error "Please fix"
|
|
#endif
|
|
|
|
.macro zero_fp
|
|
#ifdef CONFIG_FRAME_POINTER
|
|
mov fp, #0
|
|
#endif
|
|
.endm
|
|
|
|
.macro alignment_trap, rtemp, label
|
|
#ifdef CONFIG_ALIGNMENT_TRAP
|
|
ldr \rtemp, \label
|
|
ldr \rtemp, [\rtemp]
|
|
mcr p15, 0, \rtemp, c1, c0
|
|
#endif
|
|
.endm
|
|
|
|
#ifdef CONFIG_CPU_V7M
|
|
/*
|
|
* ARMv7-M exception entry/exit macros.
|
|
*
|
|
* xPSR, ReturnAddress(), LR (R14), R12, R3, R2, R1, and R0 are
|
|
* automatically saved on the current stack (32 words) before
|
|
* switching to the exception stack (SP_main).
|
|
*
|
|
* If exception is taken while in user mode, SP_main is
|
|
* empty. Otherwise, SP_main is aligned to 64 bit automatically
|
|
* (CCR.STKALIGN set).
|
|
*
|
|
* Linux assumes that the interrupts are disabled when entering an
|
|
* exception handler and it may BUG if this is not the case. Interrupts
|
|
* are disabled during entry and reenabled in the exit macro.
|
|
*
|
|
* v7m_exception_slow_exit is used when returning from SVC or PendSV.
|
|
* When returning to kernel mode, we don't return from exception.
|
|
*/
|
|
.macro v7m_exception_entry
|
|
@ determine the location of the registers saved by the core during
|
|
@ exception entry. Depending on the mode the cpu was in when the
|
|
@ exception happend that is either on the main or the process stack.
|
|
@ Bit 2 of EXC_RETURN stored in the lr register specifies which stack
|
|
@ was used.
|
|
tst lr, #EXC_RET_STACK_MASK
|
|
mrsne r12, psp
|
|
moveq r12, sp
|
|
|
|
@ we cannot rely on r0-r3 and r12 matching the value saved in the
|
|
@ exception frame because of tail-chaining. So these have to be
|
|
@ reloaded.
|
|
ldmia r12!, {r0-r3}
|
|
|
|
@ Linux expects to have irqs off. Do it here before taking stack space
|
|
cpsid i
|
|
|
|
sub sp, #S_FRAME_SIZE-S_IP
|
|
stmdb sp!, {r0-r11}
|
|
|
|
@ load saved r12, lr, return address and xPSR.
|
|
@ r0-r7 are used for signals and never touched from now on. Clobbering
|
|
@ r8-r12 is OK.
|
|
mov r9, r12
|
|
ldmia r9!, {r8, r10-r12}
|
|
|
|
@ calculate the original stack pointer value.
|
|
@ r9 currently points to the memory location just above the auto saved
|
|
@ xPSR.
|
|
@ The cpu might automatically 8-byte align the stack. Bit 9
|
|
@ of the saved xPSR specifies if stack aligning took place. In this case
|
|
@ another 32-bit value is included in the stack.
|
|
|
|
tst r12, V7M_xPSR_FRAMEPTRALIGN
|
|
addne r9, r9, #4
|
|
|
|
@ store saved r12 using str to have a register to hold the base for stm
|
|
str r8, [sp, #S_IP]
|
|
add r8, sp, #S_SP
|
|
@ store r13-r15, xPSR
|
|
stmia r8!, {r9-r12}
|
|
@ store old_r0
|
|
str r0, [r8]
|
|
.endm
|
|
|
|
/*
|
|
* PENDSV and SVCALL are configured to have the same exception
|
|
* priorities. As a kernel thread runs at SVCALL execution priority it
|
|
* can never be preempted and so we will never have to return to a
|
|
* kernel thread here.
|
|
*/
|
|
.macro v7m_exception_slow_exit ret_r0
|
|
cpsid i
|
|
ldr lr, =EXC_RET_THREADMODE_PROCESSSTACK
|
|
|
|
@ read original r12, sp, lr, pc and xPSR
|
|
add r12, sp, #S_IP
|
|
ldmia r12, {r1-r5}
|
|
|
|
@ an exception frame is always 8-byte aligned. To tell the hardware if
|
|
@ the sp to be restored is aligned or not set bit 9 of the saved xPSR
|
|
@ accordingly.
|
|
tst r2, #4
|
|
subne r2, r2, #4
|
|
orrne r5, V7M_xPSR_FRAMEPTRALIGN
|
|
biceq r5, V7M_xPSR_FRAMEPTRALIGN
|
|
|
|
@ ensure bit 0 is cleared in the PC, otherwise behaviour is
|
|
@ unpredictable
|
|
bic r4, #1
|
|
|
|
@ write basic exception frame
|
|
stmdb r2!, {r1, r3-r5}
|
|
ldmia sp, {r1, r3-r5}
|
|
.if \ret_r0
|
|
stmdb r2!, {r0, r3-r5}
|
|
.else
|
|
stmdb r2!, {r1, r3-r5}
|
|
.endif
|
|
|
|
@ restore process sp
|
|
msr psp, r2
|
|
|
|
@ restore original r4-r11
|
|
ldmia sp!, {r0-r11}
|
|
|
|
@ restore main sp
|
|
add sp, sp, #S_FRAME_SIZE-S_IP
|
|
|
|
cpsie i
|
|
bx lr
|
|
.endm
|
|
#endif /* CONFIG_CPU_V7M */
|
|
|
|
@
|
|
@ Store/load the USER SP and LR registers by switching to the SYS
|
|
@ mode. Useful in Thumb-2 mode where "stm/ldm rd, {sp, lr}^" is not
|
|
@ available. Should only be called from SVC mode
|
|
@
|
|
.macro store_user_sp_lr, rd, rtemp, offset = 0
|
|
mrs \rtemp, cpsr
|
|
eor \rtemp, \rtemp, #(SVC_MODE ^ SYSTEM_MODE)
|
|
msr cpsr_c, \rtemp @ switch to the SYS mode
|
|
|
|
str sp, [\rd, #\offset] @ save sp_usr
|
|
str lr, [\rd, #\offset + 4] @ save lr_usr
|
|
|
|
eor \rtemp, \rtemp, #(SVC_MODE ^ SYSTEM_MODE)
|
|
msr cpsr_c, \rtemp @ switch back to the SVC mode
|
|
.endm
|
|
|
|
.macro load_user_sp_lr, rd, rtemp, offset = 0
|
|
mrs \rtemp, cpsr
|
|
eor \rtemp, \rtemp, #(SVC_MODE ^ SYSTEM_MODE)
|
|
msr cpsr_c, \rtemp @ switch to the SYS mode
|
|
|
|
ldr sp, [\rd, #\offset] @ load sp_usr
|
|
ldr lr, [\rd, #\offset + 4] @ load lr_usr
|
|
|
|
eor \rtemp, \rtemp, #(SVC_MODE ^ SYSTEM_MODE)
|
|
msr cpsr_c, \rtemp @ switch back to the SVC mode
|
|
.endm
|
|
|
|
#ifndef CONFIG_THUMB2_KERNEL
|
|
.macro svc_exit, rpsr, irq = 0
|
|
.if \irq != 0
|
|
@ IRQs already off
|
|
#ifdef CONFIG_TRACE_IRQFLAGS
|
|
@ The parent context IRQs must have been enabled to get here in
|
|
@ the first place, so there's no point checking the PSR I bit.
|
|
bl trace_hardirqs_on
|
|
#endif
|
|
.else
|
|
@ IRQs off again before pulling preserved data off the stack
|
|
disable_irq_notrace
|
|
#ifdef CONFIG_TRACE_IRQFLAGS
|
|
tst \rpsr, #PSR_I_BIT
|
|
bleq trace_hardirqs_on
|
|
tst \rpsr, #PSR_I_BIT
|
|
blne trace_hardirqs_off
|
|
#endif
|
|
.endif
|
|
msr spsr_cxsf, \rpsr
|
|
#if defined(CONFIG_CPU_V6)
|
|
ldr r0, [sp]
|
|
strex r1, r2, [sp] @ clear the exclusive monitor
|
|
ldmib sp, {r1 - pc}^ @ load r1 - pc, cpsr
|
|
#elif defined(CONFIG_CPU_32v6K)
|
|
clrex @ clear the exclusive monitor
|
|
ldmia sp, {r0 - pc}^ @ load r0 - pc, cpsr
|
|
#else
|
|
ldmia sp, {r0 - pc}^ @ load r0 - pc, cpsr
|
|
#endif
|
|
.endm
|
|
|
|
.macro restore_user_regs, fast = 0, offset = 0
|
|
ldr r1, [sp, #\offset + S_PSR] @ get calling cpsr
|
|
ldr lr, [sp, #\offset + S_PC]! @ get pc
|
|
msr spsr_cxsf, r1 @ save in spsr_svc
|
|
#if defined(CONFIG_CPU_V6)
|
|
strex r1, r2, [sp] @ clear the exclusive monitor
|
|
#elif defined(CONFIG_CPU_32v6K)
|
|
clrex @ clear the exclusive monitor
|
|
#endif
|
|
.if \fast
|
|
ldmdb sp, {r1 - lr}^ @ get calling r1 - lr
|
|
.else
|
|
ldmdb sp, {r0 - lr}^ @ get calling r0 - lr
|
|
.endif
|
|
mov r0, r0 @ ARMv5T and earlier require a nop
|
|
@ after ldm {}^
|
|
add sp, sp, #S_FRAME_SIZE - S_PC
|
|
movs pc, lr @ return & move spsr_svc into cpsr
|
|
.endm
|
|
|
|
#else /* CONFIG_THUMB2_KERNEL */
|
|
.macro svc_exit, rpsr, irq = 0
|
|
.if \irq != 0
|
|
@ IRQs already off
|
|
#ifdef CONFIG_TRACE_IRQFLAGS
|
|
@ The parent context IRQs must have been enabled to get here in
|
|
@ the first place, so there's no point checking the PSR I bit.
|
|
bl trace_hardirqs_on
|
|
#endif
|
|
.else
|
|
@ IRQs off again before pulling preserved data off the stack
|
|
disable_irq_notrace
|
|
#ifdef CONFIG_TRACE_IRQFLAGS
|
|
tst \rpsr, #PSR_I_BIT
|
|
bleq trace_hardirqs_on
|
|
tst \rpsr, #PSR_I_BIT
|
|
blne trace_hardirqs_off
|
|
#endif
|
|
.endif
|
|
ldr lr, [sp, #S_SP] @ top of the stack
|
|
ldrd r0, r1, [sp, #S_LR] @ calling lr and pc
|
|
clrex @ clear the exclusive monitor
|
|
stmdb lr!, {r0, r1, \rpsr} @ calling lr and rfe context
|
|
ldmia sp, {r0 - r12}
|
|
mov sp, lr
|
|
ldr lr, [sp], #4
|
|
rfeia sp!
|
|
.endm
|
|
|
|
#ifdef CONFIG_CPU_V7M
|
|
/*
|
|
* Note we don't need to do clrex here as clearing the local monitor is
|
|
* part of each exception entry and exit sequence.
|
|
*/
|
|
.macro restore_user_regs, fast = 0, offset = 0
|
|
.if \offset
|
|
add sp, #\offset
|
|
.endif
|
|
v7m_exception_slow_exit ret_r0 = \fast
|
|
.endm
|
|
#else /* ifdef CONFIG_CPU_V7M */
|
|
.macro restore_user_regs, fast = 0, offset = 0
|
|
clrex @ clear the exclusive monitor
|
|
mov r2, sp
|
|
load_user_sp_lr r2, r3, \offset + S_SP @ calling sp, lr
|
|
ldr r1, [sp, #\offset + S_PSR] @ get calling cpsr
|
|
ldr lr, [sp, #\offset + S_PC] @ get pc
|
|
add sp, sp, #\offset + S_SP
|
|
msr spsr_cxsf, r1 @ save in spsr_svc
|
|
.if \fast
|
|
ldmdb sp, {r1 - r12} @ get calling r1 - r12
|
|
.else
|
|
ldmdb sp, {r0 - r12} @ get calling r0 - r12
|
|
.endif
|
|
add sp, sp, #S_FRAME_SIZE - S_SP
|
|
movs pc, lr @ return & move spsr_svc into cpsr
|
|
.endm
|
|
#endif /* ifdef CONFIG_CPU_V7M / else */
|
|
#endif /* !CONFIG_THUMB2_KERNEL */
|
|
|
|
/*
|
|
* Context tracking subsystem. Used to instrument transitions
|
|
* between user and kernel mode.
|
|
*/
|
|
.macro ct_user_exit, save = 1
|
|
#ifdef CONFIG_CONTEXT_TRACKING
|
|
.if \save
|
|
stmdb sp!, {r0-r3, ip, lr}
|
|
bl context_tracking_user_exit
|
|
ldmia sp!, {r0-r3, ip, lr}
|
|
.else
|
|
bl context_tracking_user_exit
|
|
.endif
|
|
#endif
|
|
.endm
|
|
|
|
.macro ct_user_enter, save = 1
|
|
#ifdef CONFIG_CONTEXT_TRACKING
|
|
.if \save
|
|
stmdb sp!, {r0-r3, ip, lr}
|
|
bl context_tracking_user_enter
|
|
ldmia sp!, {r0-r3, ip, lr}
|
|
.else
|
|
bl context_tracking_user_enter
|
|
.endif
|
|
#endif
|
|
.endm
|
|
|
|
/*
|
|
* These are the registers used in the syscall handler, and allow us to
|
|
* have in theory up to 7 arguments to a function - r0 to r6.
|
|
*
|
|
* r7 is reserved for the system call number for thumb mode.
|
|
*
|
|
* Note that tbl == why is intentional.
|
|
*
|
|
* We must set at least "tsk" and "why" when calling ret_with_reschedule.
|
|
*/
|
|
scno .req r7 @ syscall number
|
|
tbl .req r8 @ syscall table pointer
|
|
why .req r8 @ Linux syscall (!= 0)
|
|
tsk .req r9 @ current thread_info
|