forge/linux-pinenote
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
linux-pinenote/arch/parisc/kernel/perf_asm.S

1693 lines
26 KiB
ArmAsm
Raw Normal View History

Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
/* low-level asm for "intrigue" (PA8500-8700 CPU perf counters)
*
* Copyright (C) 2001 Randolph Chung <tausq at parisc-linux.org>
* Copyright (C) 2001 Hewlett-Packard (Grant Grundler)
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* 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, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include <asm/assembly.h>
arch/parisc/kernel/perf_asm.S: build fix Missing <linux/init.h> header for __HEAD macro. Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2008-05-15 10:57:33 -04:00
#include <linux/init.h>
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
#include <linux/linkage.h>
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
#ifdef CONFIG_64BIT
.level 2.0w
#endif /* CONFIG_64BIT */
#define MTDIAG_1(gr) .word 0x14201840 + gr*0x10000
#define MTDIAG_2(gr) .word 0x14401840 + gr*0x10000
#define MFDIAG_1(gr) .word 0x142008A0 + gr
#define MFDIAG_2(gr) .word 0x144008A0 + gr
#define STDIAG(dr) .word 0x14000AA0 + dr*0x200000
#define SFDIAG(dr) .word 0x14000BA0 + dr*0x200000
#define DR2_SLOW_RET 53
;
; Enable the performance counters
;
; The coprocessor only needs to be enabled when
; starting/stopping the coprocessor with the pmenb/pmdis.
;
Revert "parisc: fix trivial section name warnings" This reverts commit bd3bb8c15b9a80dbddfb7905b237a4a11a4725b4. Signed-off-by: Kyle McMartin <kyle@mcmartin.ca>
2008-05-22 14:36:31 -04:00
.text
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENTRY(perf_intrigue_enable_perf_counters)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
.proc
.callinfo frame=0,NO_CALLS
.entry
ldi 0x20,%r25 ; load up perfmon bit
mfctl ccr,%r26 ; get coprocessor register
or %r25,%r26,%r26 ; set bit
mtctl %r26,ccr ; turn on performance coprocessor
pmenb ; enable performance monitor
ssm 0,0 ; dummy op to ensure completion
sync ; follow ERS
andcm %r26,%r25,%r26 ; clear bit now
mtctl %r26,ccr ; turn off performance coprocessor
nop ; NOPs as specified in ERS
nop
nop
nop
nop
nop
nop
bve (%r2)
nop
.exit
.procend
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENDPROC(perf_intrigue_enable_perf_counters)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENTRY(perf_intrigue_disable_perf_counters)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
.proc
.callinfo frame=0,NO_CALLS
.entry
ldi 0x20,%r25 ; load up perfmon bit
mfctl ccr,%r26 ; get coprocessor register
or %r25,%r26,%r26 ; set bit
mtctl %r26,ccr ; turn on performance coprocessor
pmdis ; disable performance monitor
ssm 0,0 ; dummy op to ensure completion
andcm %r26,%r25,%r26 ; clear bit now
bve (%r2)
mtctl %r26,ccr ; turn off performance coprocessor
.exit
.procend
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENDPROC(perf_intrigue_disable_perf_counters)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
;***********************************************************************
;*
;* Name: perf_rdr_shift_in_W
;*
;* Description:
;* This routine shifts data in from the RDR in arg0 and returns
;* the result in ret0. If the RDR is <= 64 bits in length, it
;* is shifted shifted backup immediately. This is to compensate
;* for RDR10 which has bits that preclude PDC stack operations
;* when they are in the wrong state.
;*
;* Arguments:
;* arg0 : rdr to be read
;* arg1 : bit length of rdr
;*
;* Returns:
;* ret0 = next 64 bits of rdr data from staging register
;*
;* Register usage:
;* arg0 : rdr to be read
;* arg1 : bit length of rdr
;* %r24 - original DR2 value
;* %r1 - scratch
;* %r29 - scratch
;*
;* Returns:
;* ret0 = RDR data (right justified)
;*
;***********************************************************************
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENTRY(perf_rdr_shift_in_W)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
.proc
.callinfo frame=0,NO_CALLS
.entry
;
; read(shift in) the RDR.
;
; NOTE: The PCX-W ERS states that DR2_SLOW_RET must be set before any
; shifting is done, from or to, remote diagnose registers.
;
depdi,z 1,DR2_SLOW_RET,1,%r29
MFDIAG_2 (24)
or %r24,%r29,%r29
MTDIAG_2 (29) ; set DR2_SLOW_RET
nop
nop
nop
nop
;
; Cacheline start (32-byte cacheline)
;
nop
nop
nop
extrd,u arg1,63,6,%r1 ; setup shift amount by bits to move
mtsar %r1
shladd arg0,2,%r0,%r1 ; %r1 = 4 * RDR number
blr %r1,%r0 ; branch to 8-instruction sequence
nop
;
; Cacheline start (32-byte cacheline)
;
;
; RDR 0 sequence
;
SFDIAG (0)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1) ; mtdiag %dr1, %r1
STDIAG (0)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 1 sequence
;
sync
ssm 0,0
SFDIAG (1)
ssm 0,0
MFDIAG_1 (28)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
nop
;
; RDR 2 read sequence
;
SFDIAG (2)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (2)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 3 read sequence
;
b,n perf_rdr_shift_in_W_leave
nop
nop
nop
nop
nop
nop
nop
;
; RDR 4 read sequence
;
sync
ssm 0,0
SFDIAG (4)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 5 read sequence
;
sync
ssm 0,0
SFDIAG (5)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 6 read sequence
;
sync
ssm 0,0
SFDIAG (6)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 7 read sequence
;
b,n perf_rdr_shift_in_W_leave
nop
nop
nop
nop
nop
nop
nop
;
; RDR 8 read sequence
;
b,n perf_rdr_shift_in_W_leave
nop
nop
nop
nop
nop
nop
nop
;
; RDR 9 read sequence
;
b,n perf_rdr_shift_in_W_leave
nop
nop
nop
nop
nop
nop
nop
;
; RDR 10 read sequence
;
SFDIAG (10)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (10)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 11 read sequence
;
SFDIAG (11)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (11)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 12 read sequence
;
b,n perf_rdr_shift_in_W_leave
nop
nop
nop
nop
nop
nop
nop
;
; RDR 13 read sequence
;
sync
ssm 0,0
SFDIAG (13)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 14 read sequence
;
SFDIAG (14)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (14)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 15 read sequence
;
sync
ssm 0,0
SFDIAG (15)
ssm 0,0
MFDIAG_1 (28)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
nop
;
; RDR 16 read sequence
;
sync
ssm 0,0
SFDIAG (16)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 17 read sequence
;
SFDIAG (17)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (17)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 18 read sequence
;
SFDIAG (18)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (18)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 19 read sequence
;
b,n perf_rdr_shift_in_W_leave
nop
nop
nop
nop
nop
nop
nop
;
; RDR 20 read sequence
;
sync
ssm 0,0
SFDIAG (20)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 21 read sequence
;
sync
ssm 0,0
SFDIAG (21)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 22 read sequence
;
sync
ssm 0,0
SFDIAG (22)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 23 read sequence
;
sync
ssm 0,0
SFDIAG (23)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 24 read sequence
;
sync
ssm 0,0
SFDIAG (24)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 25 read sequence
;
sync
ssm 0,0
SFDIAG (25)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 26 read sequence
;
SFDIAG (26)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (26)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 27 read sequence
;
SFDIAG (27)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (27)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 28 read sequence
;
sync
ssm 0,0
SFDIAG (28)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 29 read sequence
;
sync
ssm 0,0
SFDIAG (29)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_W_leave
ssm 0,0
nop
;
; RDR 30 read sequence
;
SFDIAG (30)
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (30)
ssm 0,0
b,n perf_rdr_shift_in_W_leave
;
; RDR 31 read sequence
;
sync
ssm 0,0
SFDIAG (31)
ssm 0,0
MFDIAG_1 (28)
nop
ssm 0,0
nop
;
; Fallthrough
;
perf_rdr_shift_in_W_leave:
bve (%r2)
.exit
MTDIAG_2 (24) ; restore DR2
.procend
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENDPROC(perf_rdr_shift_in_W)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
;***********************************************************************
;*
;* Name: perf_rdr_shift_out_W
;*
;* Description:
;* This routine moves data to the RDR's. The double-word that
;* arg1 points to is loaded and moved into the staging register.
;* Then the STDIAG instruction for the RDR # in arg0 is called
;* to move the data to the RDR.
;*
;* Arguments:
;* arg0 = rdr number
;* arg1 = 64-bit value to write
;* %r24 - DR2 | DR2_SLOW_RET
;* %r23 - original DR2 value
;*
;* Returns:
;* None
;*
;* Register usage:
;*
;***********************************************************************
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENTRY(perf_rdr_shift_out_W)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
.proc
.callinfo frame=0,NO_CALLS
.entry
;
; NOTE: The PCX-W ERS states that DR2_SLOW_RET must be set before any
; shifting is done, from or to, the remote diagnose registers.
;
depdi,z 1,DR2_SLOW_RET,1,%r24
MFDIAG_2 (23)
or %r24,%r23,%r24
MTDIAG_2 (24) ; set DR2_SLOW_RET
MTDIAG_1 (25) ; data to the staging register
shladd arg0,2,%r0,%r1 ; %r1 = 4 * RDR number
blr %r1,%r0 ; branch to 8-instruction sequence
nop
;
; RDR 0 write sequence
;
sync ; RDR 0 write sequence
ssm 0,0
STDIAG (0)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 1 write sequence
;
sync
ssm 0,0
STDIAG (1)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 2 write sequence
;
sync
ssm 0,0
STDIAG (2)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 3 write sequence
;
sync
ssm 0,0
STDIAG (3)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 4 write sequence
;
sync
ssm 0,0
STDIAG (4)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 5 write sequence
;
sync
ssm 0,0
STDIAG (5)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 6 write sequence
;
sync
ssm 0,0
STDIAG (6)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 7 write sequence
;
sync
ssm 0,0
STDIAG (7)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 8 write sequence
;
sync
ssm 0,0
STDIAG (8)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 9 write sequence
;
sync
ssm 0,0
STDIAG (9)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 10 write sequence
;
sync
ssm 0,0
STDIAG (10)
STDIAG (26)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
ssm 0,0
nop
;
; RDR 11 write sequence
;
sync
ssm 0,0
STDIAG (11)
STDIAG (27)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
ssm 0,0
nop
;
; RDR 12 write sequence
;
sync
ssm 0,0
STDIAG (12)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 13 write sequence
;
sync
ssm 0,0
STDIAG (13)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 14 write sequence
;
sync
ssm 0,0
STDIAG (14)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 15 write sequence
;
sync
ssm 0,0
STDIAG (15)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 16 write sequence
;
sync
ssm 0,0
STDIAG (16)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 17 write sequence
;
sync
ssm 0,0
STDIAG (17)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 18 write sequence
;
sync
ssm 0,0
STDIAG (18)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 19 write sequence
;
sync
ssm 0,0
STDIAG (19)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 20 write sequence
;
sync
ssm 0,0
STDIAG (20)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 21 write sequence
;
sync
ssm 0,0
STDIAG (21)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 22 write sequence
;
sync
ssm 0,0
STDIAG (22)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 23 write sequence
;
sync
ssm 0,0
STDIAG (23)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 24 write sequence
;
sync
ssm 0,0
STDIAG (24)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 25 write sequence
;
sync
ssm 0,0
STDIAG (25)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 26 write sequence
;
sync
ssm 0,0
STDIAG (10)
STDIAG (26)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
ssm 0,0
nop
;
; RDR 27 write sequence
;
sync
ssm 0,0
STDIAG (11)
STDIAG (27)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
ssm 0,0
nop
;
; RDR 28 write sequence
;
sync
ssm 0,0
STDIAG (28)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 29 write sequence
;
sync
ssm 0,0
STDIAG (29)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 30 write sequence
;
sync
ssm 0,0
STDIAG (30)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
;
; RDR 31 write sequence
;
sync
ssm 0,0
STDIAG (31)
ssm 0,0
b,n perf_rdr_shift_out_W_leave
nop
ssm 0,0
nop
perf_rdr_shift_out_W_leave:
bve (%r2)
.exit
MTDIAG_2 (23) ; restore DR2
.procend
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENDPROC(perf_rdr_shift_out_W)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
;***********************************************************************
;*
;* Name: rdr_shift_in_U
;*
;* Description:
;* This routine shifts data in from the RDR in arg0 and returns
;* the result in ret0. If the RDR is <= 64 bits in length, it
;* is shifted shifted backup immediately. This is to compensate
;* for RDR10 which has bits that preclude PDC stack operations
;* when they are in the wrong state.
;*
;* Arguments:
;* arg0 : rdr to be read
;* arg1 : bit length of rdr
;*
;* Returns:
;* ret0 = next 64 bits of rdr data from staging register
;*
;* Register usage:
;* arg0 : rdr to be read
;* arg1 : bit length of rdr
;* %r24 - original DR2 value
;* %r23 - DR2 | DR2_SLOW_RET
;* %r1 - scratch
;*
;***********************************************************************
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENTRY(perf_rdr_shift_in_U)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
.proc
.callinfo frame=0,NO_CALLS
.entry
; read(shift in) the RDR.
;
; NOTE: The PCX-U ERS states that DR2_SLOW_RET must be set before any
; shifting is done, from or to, remote diagnose registers.
depdi,z 1,DR2_SLOW_RET,1,%r29
MFDIAG_2 (24)
or %r24,%r29,%r29
MTDIAG_2 (29) ; set DR2_SLOW_RET
nop
nop
nop
nop
;
; Start of next 32-byte cacheline
;
nop
nop
nop
extrd,u arg1,63,6,%r1
mtsar %r1
shladd arg0,2,%r0,%r1 ; %r1 = 4 * RDR number
blr %r1,%r0 ; branch to 8-instruction sequence
nop
;
; Start of next 32-byte cacheline
;
SFDIAG (0) ; RDR 0 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (0)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (1) ; RDR 1 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (1)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
sync ; RDR 2 read sequence
ssm 0,0
SFDIAG (4)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 3 read sequence
ssm 0,0
SFDIAG (3)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 4 read sequence
ssm 0,0
SFDIAG (4)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 5 read sequence
ssm 0,0
SFDIAG (5)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 6 read sequence
ssm 0,0
SFDIAG (6)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 7 read sequence
ssm 0,0
SFDIAG (7)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
b,n perf_rdr_shift_in_U_leave
nop
nop
nop
nop
nop
nop
nop
SFDIAG (9) ; RDR 9 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (9)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (10) ; RDR 10 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (10)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (11) ; RDR 11 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (11)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (12) ; RDR 12 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (12)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (13) ; RDR 13 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (13)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (14) ; RDR 14 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (14)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (15) ; RDR 15 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (15)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
sync ; RDR 16 read sequence
ssm 0,0
SFDIAG (16)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
SFDIAG (17) ; RDR 17 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (17)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (18) ; RDR 18 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (18)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
b,n perf_rdr_shift_in_U_leave
nop
nop
nop
nop
nop
nop
nop
sync ; RDR 20 read sequence
ssm 0,0
SFDIAG (20)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 21 read sequence
ssm 0,0
SFDIAG (21)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 22 read sequence
ssm 0,0
SFDIAG (22)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 23 read sequence
ssm 0,0
SFDIAG (23)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 24 read sequence
ssm 0,0
SFDIAG (24)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
sync ; RDR 25 read sequence
ssm 0,0
SFDIAG (25)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
SFDIAG (26) ; RDR 26 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (26)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (27) ; RDR 27 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (27)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
sync ; RDR 28 read sequence
ssm 0,0
SFDIAG (28)
ssm 0,0
MFDIAG_1 (28)
b,n perf_rdr_shift_in_U_leave
ssm 0,0
nop
b,n perf_rdr_shift_in_U_leave
nop
nop
nop
nop
nop
nop
nop
SFDIAG (30) ; RDR 30 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (30)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
SFDIAG (31) ; RDR 31 read sequence
ssm 0,0
MFDIAG_1 (28)
shrpd ret0,%r0,%sar,%r1
MTDIAG_1 (1)
STDIAG (31)
ssm 0,0
b,n perf_rdr_shift_in_U_leave
nop
perf_rdr_shift_in_U_leave:
bve (%r2)
.exit
MTDIAG_2 (24) ; restore DR2
.procend
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENDPROC(perf_rdr_shift_in_U)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
;***********************************************************************
;*
;* Name: rdr_shift_out_U
;*
;* Description:
;* This routine moves data to the RDR's. The double-word that
;* arg1 points to is loaded and moved into the staging register.
;* Then the STDIAG instruction for the RDR # in arg0 is called
;* to move the data to the RDR.
;*
;* Arguments:
;* arg0 = rdr target
;* arg1 = buffer pointer
;*
;* Returns:
;* None
;*
;* Register usage:
;* arg0 = rdr target
;* arg1 = buffer pointer
;* %r24 - DR2 | DR2_SLOW_RET
;* %r23 - original DR2 value
;*
;***********************************************************************
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENTRY(perf_rdr_shift_out_U)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
.proc
.callinfo frame=0,NO_CALLS
.entry
;
; NOTE: The PCX-U ERS states that DR2_SLOW_RET must be set before any
; shifting is done, from or to, the remote diagnose registers.
;
depdi,z 1,DR2_SLOW_RET,1,%r24
MFDIAG_2 (23)
or %r24,%r23,%r24
MTDIAG_2 (24) ; set DR2_SLOW_RET
MTDIAG_1 (25) ; data to the staging register
shladd arg0,2,%r0,%r1 ; %r1 = 4 * RDR number
blr %r1,%r0 ; branch to 8-instruction sequence
nop
;
; 32-byte cachline aligned
;
sync ; RDR 0 write sequence
ssm 0,0
STDIAG (0)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 1 write sequence
ssm 0,0
STDIAG (1)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 2 write sequence
ssm 0,0
STDIAG (2)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 3 write sequence
ssm 0,0
STDIAG (3)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 4 write sequence
ssm 0,0
STDIAG (4)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 5 write sequence
ssm 0,0
STDIAG (5)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 6 write sequence
ssm 0,0
STDIAG (6)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 7 write sequence
ssm 0,0
STDIAG (7)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 8 write sequence
ssm 0,0
STDIAG (8)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 9 write sequence
ssm 0,0
STDIAG (9)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 10 write sequence
ssm 0,0
STDIAG (10)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 11 write sequence
ssm 0,0
STDIAG (11)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 12 write sequence
ssm 0,0
STDIAG (12)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 13 write sequence
ssm 0,0
STDIAG (13)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 14 write sequence
ssm 0,0
STDIAG (14)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 15 write sequence
ssm 0,0
STDIAG (15)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 16 write sequence
ssm 0,0
STDIAG (16)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 17 write sequence
ssm 0,0
STDIAG (17)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 18 write sequence
ssm 0,0
STDIAG (18)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 19 write sequence
ssm 0,0
STDIAG (19)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 20 write sequence
ssm 0,0
STDIAG (20)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 21 write sequence
ssm 0,0
STDIAG (21)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 22 write sequence
ssm 0,0
STDIAG (22)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 23 write sequence
ssm 0,0
STDIAG (23)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 24 write sequence
ssm 0,0
STDIAG (24)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 25 write sequence
ssm 0,0
STDIAG (25)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 26 write sequence
ssm 0,0
STDIAG (26)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 27 write sequence
ssm 0,0
STDIAG (27)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 28 write sequence
ssm 0,0
STDIAG (28)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 29 write sequence
ssm 0,0
STDIAG (29)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 30 write sequence
ssm 0,0
STDIAG (30)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
sync ; RDR 31 write sequence
ssm 0,0
STDIAG (31)
ssm 0,0
b,n perf_rdr_shift_out_U_leave
nop
ssm 0,0
nop
perf_rdr_shift_out_U_leave:
bve (%r2)
.exit
MTDIAG_2 (23) ; restore DR2
.procend
[PARISC] more ENTRY(), ENDPROC(), END() conversions Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
2007-01-24 22:36:32 +01:00
ENDPROC(perf_rdr_shift_out_U)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-16 15:20:36 -07:00
Reference in a new issue Copy permalink