linux/arch/xtensa/kernel/coprocessor.S
Max Filippov d6d5f19e21 xtensa: abstract 'entry' and 'retw' in assembly code
Provide abi_entry, abi_entry_default, abi_ret and abi_ret_default macros
that allocate aligned stack frame in windowed and call0 ABIs.
Provide XTENSA_SPILL_STACK_RESERVE macro that specifies required stack
frame size when register spilling is involved.
Replace all uses of 'entry' and 'retw' with the above macros.
This makes most of the xtensa assembly code ready for XEA3 and call0 ABI.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
2019-07-08 10:04:48 -07:00

272 lines
5.6 KiB
ArmAsm

/*
* arch/xtensa/kernel/coprocessor.S
*
* Xtensa processor configuration-specific table of coprocessor and
* other custom register layout information.
*
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file "COPYING" in the main directory of this archive
* for more details.
*
* Copyright (C) 2003 - 2007 Tensilica Inc.
*/
#include <linux/linkage.h>
#include <asm/asm-offsets.h>
#include <asm/processor.h>
#include <asm/coprocessor.h>
#include <asm/thread_info.h>
#include <asm/asm-uaccess.h>
#include <asm/unistd.h>
#include <asm/ptrace.h>
#include <asm/current.h>
#include <asm/pgtable.h>
#include <asm/page.h>
#include <asm/signal.h>
#include <asm/tlbflush.h>
#if XTENSA_HAVE_COPROCESSORS
/*
* Macros for lazy context switch.
*/
#define SAVE_CP_REGS(x) \
.if XTENSA_HAVE_COPROCESSOR(x); \
.align 4; \
.Lsave_cp_regs_cp##x: \
xchal_cp##x##_store a2 a4 a5 a6 a7; \
jx a0; \
.endif
#define SAVE_CP_REGS_TAB(x) \
.if XTENSA_HAVE_COPROCESSOR(x); \
.long .Lsave_cp_regs_cp##x; \
.else; \
.long 0; \
.endif; \
.long THREAD_XTREGS_CP##x
#define LOAD_CP_REGS(x) \
.if XTENSA_HAVE_COPROCESSOR(x); \
.align 4; \
.Lload_cp_regs_cp##x: \
xchal_cp##x##_load a2 a4 a5 a6 a7; \
jx a0; \
.endif
#define LOAD_CP_REGS_TAB(x) \
.if XTENSA_HAVE_COPROCESSOR(x); \
.long .Lload_cp_regs_cp##x; \
.else; \
.long 0; \
.endif; \
.long THREAD_XTREGS_CP##x
SAVE_CP_REGS(0)
SAVE_CP_REGS(1)
SAVE_CP_REGS(2)
SAVE_CP_REGS(3)
SAVE_CP_REGS(4)
SAVE_CP_REGS(5)
SAVE_CP_REGS(6)
SAVE_CP_REGS(7)
LOAD_CP_REGS(0)
LOAD_CP_REGS(1)
LOAD_CP_REGS(2)
LOAD_CP_REGS(3)
LOAD_CP_REGS(4)
LOAD_CP_REGS(5)
LOAD_CP_REGS(6)
LOAD_CP_REGS(7)
.section ".rodata", "a"
.align 4
.Lsave_cp_regs_jump_table:
SAVE_CP_REGS_TAB(0)
SAVE_CP_REGS_TAB(1)
SAVE_CP_REGS_TAB(2)
SAVE_CP_REGS_TAB(3)
SAVE_CP_REGS_TAB(4)
SAVE_CP_REGS_TAB(5)
SAVE_CP_REGS_TAB(6)
SAVE_CP_REGS_TAB(7)
.Lload_cp_regs_jump_table:
LOAD_CP_REGS_TAB(0)
LOAD_CP_REGS_TAB(1)
LOAD_CP_REGS_TAB(2)
LOAD_CP_REGS_TAB(3)
LOAD_CP_REGS_TAB(4)
LOAD_CP_REGS_TAB(5)
LOAD_CP_REGS_TAB(6)
LOAD_CP_REGS_TAB(7)
.previous
/*
* coprocessor_flush(struct thread_info*, index)
* a2 a3
*
* Save coprocessor registers for coprocessor 'index'.
* The register values are saved to or loaded from the coprocessor area
* inside the task_info structure.
*
* Note that this function doesn't update the coprocessor_owner information!
*
*/
ENTRY(coprocessor_flush)
/* reserve 4 bytes on stack to save a0 */
abi_entry(4)
s32i a0, a1, 0
movi a0, .Lsave_cp_regs_jump_table
addx8 a3, a3, a0
l32i a4, a3, 4
l32i a3, a3, 0
add a2, a2, a4
beqz a3, 1f
callx0 a3
1: l32i a0, a1, 0
abi_ret(4)
ENDPROC(coprocessor_flush)
/*
* Entry condition:
*
* a0: trashed, original value saved on stack (PT_AREG0)
* a1: a1
* a2: new stack pointer, original in DEPC
* a3: a3
* depc: a2, original value saved on stack (PT_DEPC)
* excsave_1: dispatch table
*
* PT_DEPC >= VALID_DOUBLE_EXCEPTION_ADDRESS: double exception, DEPC
* < VALID_DOUBLE_EXCEPTION_ADDRESS: regular exception
*/
ENTRY(fast_coprocessor_double)
wsr a0, excsave1
call0 unrecoverable_exception
ENDPROC(fast_coprocessor_double)
ENTRY(fast_coprocessor)
/* Save remaining registers a1-a3 and SAR */
s32i a3, a2, PT_AREG3
rsr a3, sar
s32i a1, a2, PT_AREG1
s32i a3, a2, PT_SAR
mov a1, a2
rsr a2, depc
s32i a2, a1, PT_AREG2
/*
* The hal macros require up to 4 temporary registers. We use a3..a6.
*/
s32i a4, a1, PT_AREG4
s32i a5, a1, PT_AREG5
s32i a6, a1, PT_AREG6
/* Find coprocessor number. Subtract first CP EXCCAUSE from EXCCAUSE */
rsr a3, exccause
addi a3, a3, -EXCCAUSE_COPROCESSOR0_DISABLED
/* Set corresponding CPENABLE bit -> (sar:cp-index, a3: 1<<cp-index)*/
ssl a3 # SAR: 32 - coprocessor_number
movi a2, 1
rsr a0, cpenable
sll a2, a2
or a0, a0, a2
wsr a0, cpenable
rsync
/* Retrieve previous owner. (a3 still holds CP number) */
movi a0, coprocessor_owner # list of owners
addx4 a0, a3, a0 # entry for CP
l32i a4, a0, 0
beqz a4, 1f # skip 'save' if no previous owner
/* Disable coprocessor for previous owner. (a2 = 1 << CP number) */
l32i a5, a4, THREAD_CPENABLE
xor a5, a5, a2 # (1 << cp-id) still in a2
s32i a5, a4, THREAD_CPENABLE
/*
* Get context save area and 'call' save routine.
* (a4 still holds previous owner (thread_info), a3 CP number)
*/
movi a5, .Lsave_cp_regs_jump_table
movi a0, 2f # a0: 'return' address
addx8 a3, a3, a5 # a3: coprocessor number
l32i a2, a3, 4 # a2: xtregs offset
l32i a3, a3, 0 # a3: jump address
add a2, a2, a4
jx a3
/* Note that only a0 and a1 were preserved. */
2: rsr a3, exccause
addi a3, a3, -EXCCAUSE_COPROCESSOR0_DISABLED
movi a0, coprocessor_owner
addx4 a0, a3, a0
/* Set new 'owner' (a0 points to the CP owner, a3 contains the CP nr) */
1: GET_THREAD_INFO (a4, a1)
s32i a4, a0, 0
/* Get context save area and 'call' load routine. */
movi a5, .Lload_cp_regs_jump_table
movi a0, 1f
addx8 a3, a3, a5
l32i a2, a3, 4 # a2: xtregs offset
l32i a3, a3, 0 # a3: jump address
add a2, a2, a4
jx a3
/* Restore all registers and return from exception handler. */
1: l32i a6, a1, PT_AREG6
l32i a5, a1, PT_AREG5
l32i a4, a1, PT_AREG4
l32i a0, a1, PT_SAR
l32i a3, a1, PT_AREG3
l32i a2, a1, PT_AREG2
wsr a0, sar
l32i a0, a1, PT_AREG0
l32i a1, a1, PT_AREG1
rfe
ENDPROC(fast_coprocessor)
.data
ENTRY(coprocessor_owner)
.fill XCHAL_CP_MAX, 4, 0
END(coprocessor_owner)
#endif /* XTENSA_HAVE_COPROCESSORS */