2005-04-16 16:20:36 -06:00
|
|
|
/* $Id: trampoline.S,v 1.26 2002/02/09 19:49:30 davem Exp $
|
|
|
|
* trampoline.S: Jump start slave processors on sparc64.
|
|
|
|
*
|
|
|
|
* Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <asm/head.h>
|
|
|
|
#include <asm/asi.h>
|
|
|
|
#include <asm/lsu.h>
|
|
|
|
#include <asm/dcr.h>
|
|
|
|
#include <asm/dcu.h>
|
|
|
|
#include <asm/pstate.h>
|
|
|
|
#include <asm/page.h>
|
|
|
|
#include <asm/pgtable.h>
|
|
|
|
#include <asm/spitfire.h>
|
|
|
|
#include <asm/processor.h>
|
|
|
|
#include <asm/thread_info.h>
|
|
|
|
#include <asm/mmu.h>
|
|
|
|
|
|
|
|
.data
|
|
|
|
.align 8
|
|
|
|
call_method:
|
|
|
|
.asciz "call-method"
|
|
|
|
.align 8
|
|
|
|
itlb_load:
|
|
|
|
.asciz "SUNW,itlb-load"
|
|
|
|
.align 8
|
|
|
|
dtlb_load:
|
|
|
|
.asciz "SUNW,dtlb-load"
|
|
|
|
|
|
|
|
.text
|
|
|
|
.align 8
|
|
|
|
.globl sparc64_cpu_startup, sparc64_cpu_startup_end
|
|
|
|
sparc64_cpu_startup:
|
|
|
|
flushw
|
|
|
|
|
|
|
|
BRANCH_IF_CHEETAH_BASE(g1,g5,cheetah_startup)
|
|
|
|
BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g5,cheetah_plus_startup)
|
|
|
|
|
|
|
|
ba,pt %xcc, spitfire_startup
|
|
|
|
nop
|
|
|
|
|
|
|
|
cheetah_plus_startup:
|
|
|
|
/* Preserve OBP chosen DCU and DCR register settings. */
|
|
|
|
ba,pt %xcc, cheetah_generic_startup
|
|
|
|
nop
|
|
|
|
|
|
|
|
cheetah_startup:
|
|
|
|
mov DCR_BPE | DCR_RPE | DCR_SI | DCR_IFPOE | DCR_MS, %g1
|
|
|
|
wr %g1, %asr18
|
|
|
|
|
|
|
|
sethi %uhi(DCU_ME|DCU_RE|DCU_HPE|DCU_SPE|DCU_SL|DCU_WE), %g5
|
|
|
|
or %g5, %ulo(DCU_ME|DCU_RE|DCU_HPE|DCU_SPE|DCU_SL|DCU_WE), %g5
|
|
|
|
sllx %g5, 32, %g5
|
|
|
|
or %g5, DCU_DM | DCU_IM | DCU_DC | DCU_IC, %g5
|
|
|
|
stxa %g5, [%g0] ASI_DCU_CONTROL_REG
|
|
|
|
membar #Sync
|
|
|
|
|
|
|
|
cheetah_generic_startup:
|
|
|
|
mov TSB_EXTENSION_P, %g3
|
|
|
|
stxa %g0, [%g3] ASI_DMMU
|
|
|
|
stxa %g0, [%g3] ASI_IMMU
|
|
|
|
membar #Sync
|
|
|
|
|
|
|
|
mov TSB_EXTENSION_S, %g3
|
|
|
|
stxa %g0, [%g3] ASI_DMMU
|
|
|
|
membar #Sync
|
|
|
|
|
|
|
|
mov TSB_EXTENSION_N, %g3
|
|
|
|
stxa %g0, [%g3] ASI_DMMU
|
|
|
|
stxa %g0, [%g3] ASI_IMMU
|
|
|
|
membar #Sync
|
|
|
|
|
|
|
|
/* Disable STICK_INT interrupts. */
|
|
|
|
sethi %hi(0x80000000), %g5
|
|
|
|
sllx %g5, 32, %g5
|
|
|
|
wr %g5, %asr25
|
|
|
|
|
|
|
|
ba,pt %xcc, startup_continue
|
|
|
|
nop
|
|
|
|
|
|
|
|
spitfire_startup:
|
|
|
|
mov (LSU_CONTROL_IC | LSU_CONTROL_DC | LSU_CONTROL_IM | LSU_CONTROL_DM), %g1
|
|
|
|
stxa %g1, [%g0] ASI_LSU_CONTROL
|
|
|
|
membar #Sync
|
|
|
|
|
|
|
|
startup_continue:
|
|
|
|
wrpr %g0, 15, %pil
|
|
|
|
|
|
|
|
sethi %hi(0x80000000), %g2
|
|
|
|
sllx %g2, 32, %g2
|
|
|
|
wr %g2, 0, %tick_cmpr
|
|
|
|
|
|
|
|
/* Call OBP by hand to lock KERNBASE into i/d tlbs.
|
|
|
|
* We lock 2 consequetive entries if we are 'bigkernel'.
|
|
|
|
*/
|
|
|
|
mov %o0, %l0
|
|
|
|
|
|
|
|
sethi %hi(prom_entry_lock), %g2
|
|
|
|
1: ldstub [%g2 + %lo(prom_entry_lock)], %g1
|
[SPARC64]: Avoid membar instructions in delay slots.
In particular, avoid membar instructions in the delay
slot of a jmpl instruction.
UltraSPARC-I, II, IIi, and IIe have a bug, documented in
the UltraSPARC-IIi User's Manual, Appendix K, Erratum 51
The long and short of it is that if the IMU unit misses
on a branch or jmpl, and there is a store buffer synchronizing
membar in the delay slot, the chip can stop fetching instructions.
If interrupts are enabled or some other trap is enabled, the
chip will unwedge itself, but performance will suffer.
We already had a workaround for this bug in a few spots, but
it's better to have the entire tree sanitized for this rule.
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-27 16:42:04 -06:00
|
|
|
membar #StoreLoad | #StoreStore
|
2005-04-16 16:20:36 -06:00
|
|
|
brnz,pn %g1, 1b
|
[SPARC64]: Avoid membar instructions in delay slots.
In particular, avoid membar instructions in the delay
slot of a jmpl instruction.
UltraSPARC-I, II, IIi, and IIe have a bug, documented in
the UltraSPARC-IIi User's Manual, Appendix K, Erratum 51
The long and short of it is that if the IMU unit misses
on a branch or jmpl, and there is a store buffer synchronizing
membar in the delay slot, the chip can stop fetching instructions.
If interrupts are enabled or some other trap is enabled, the
chip will unwedge itself, but performance will suffer.
We already had a workaround for this bug in a few spots, but
it's better to have the entire tree sanitized for this rule.
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-27 16:42:04 -06:00
|
|
|
nop
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
sethi %hi(p1275buf), %g2
|
|
|
|
or %g2, %lo(p1275buf), %g2
|
|
|
|
ldx [%g2 + 0x10], %l2
|
|
|
|
mov %sp, %l1
|
|
|
|
add %l2, -(192 + 128), %sp
|
|
|
|
flushw
|
|
|
|
|
|
|
|
sethi %hi(call_method), %g2
|
|
|
|
or %g2, %lo(call_method), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x00]
|
|
|
|
mov 5, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x08]
|
|
|
|
mov 1, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x10]
|
|
|
|
sethi %hi(itlb_load), %g2
|
|
|
|
or %g2, %lo(itlb_load), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x18]
|
|
|
|
sethi %hi(mmu_ihandle_cache), %g2
|
|
|
|
lduw [%g2 + %lo(mmu_ihandle_cache)], %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x20]
|
|
|
|
sethi %hi(KERNBASE), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x28]
|
|
|
|
sethi %hi(kern_locked_tte_data), %g2
|
|
|
|
ldx [%g2 + %lo(kern_locked_tte_data)], %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x30]
|
|
|
|
|
|
|
|
mov 15, %g2
|
|
|
|
BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
|
|
|
|
|
|
|
|
mov 63, %g2
|
|
|
|
1:
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x38]
|
|
|
|
sethi %hi(p1275buf), %g2
|
|
|
|
or %g2, %lo(p1275buf), %g2
|
|
|
|
ldx [%g2 + 0x08], %o1
|
|
|
|
call %o1
|
|
|
|
add %sp, (2047 + 128), %o0
|
|
|
|
|
|
|
|
sethi %hi(bigkernel), %g2
|
|
|
|
lduw [%g2 + %lo(bigkernel)], %g2
|
|
|
|
cmp %g2, 0
|
|
|
|
be,pt %icc, do_dtlb
|
|
|
|
nop
|
|
|
|
|
|
|
|
sethi %hi(call_method), %g2
|
|
|
|
or %g2, %lo(call_method), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x00]
|
|
|
|
mov 5, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x08]
|
|
|
|
mov 1, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x10]
|
|
|
|
sethi %hi(itlb_load), %g2
|
|
|
|
or %g2, %lo(itlb_load), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x18]
|
|
|
|
sethi %hi(mmu_ihandle_cache), %g2
|
|
|
|
lduw [%g2 + %lo(mmu_ihandle_cache)], %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x20]
|
|
|
|
sethi %hi(KERNBASE + 0x400000), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x28]
|
|
|
|
sethi %hi(kern_locked_tte_data), %g2
|
|
|
|
ldx [%g2 + %lo(kern_locked_tte_data)], %g2
|
|
|
|
sethi %hi(0x400000), %g1
|
|
|
|
add %g2, %g1, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x30]
|
|
|
|
|
|
|
|
mov 14, %g2
|
|
|
|
BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
|
|
|
|
|
|
|
|
mov 62, %g2
|
|
|
|
1:
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x38]
|
|
|
|
sethi %hi(p1275buf), %g2
|
|
|
|
or %g2, %lo(p1275buf), %g2
|
|
|
|
ldx [%g2 + 0x08], %o1
|
|
|
|
call %o1
|
|
|
|
add %sp, (2047 + 128), %o0
|
|
|
|
|
|
|
|
do_dtlb:
|
|
|
|
sethi %hi(call_method), %g2
|
|
|
|
or %g2, %lo(call_method), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x00]
|
|
|
|
mov 5, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x08]
|
|
|
|
mov 1, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x10]
|
|
|
|
sethi %hi(dtlb_load), %g2
|
|
|
|
or %g2, %lo(dtlb_load), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x18]
|
|
|
|
sethi %hi(mmu_ihandle_cache), %g2
|
|
|
|
lduw [%g2 + %lo(mmu_ihandle_cache)], %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x20]
|
|
|
|
sethi %hi(KERNBASE), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x28]
|
|
|
|
sethi %hi(kern_locked_tte_data), %g2
|
|
|
|
ldx [%g2 + %lo(kern_locked_tte_data)], %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x30]
|
|
|
|
|
|
|
|
mov 15, %g2
|
|
|
|
BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
|
|
|
|
|
|
|
|
mov 63, %g2
|
|
|
|
1:
|
|
|
|
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x38]
|
|
|
|
sethi %hi(p1275buf), %g2
|
|
|
|
or %g2, %lo(p1275buf), %g2
|
|
|
|
ldx [%g2 + 0x08], %o1
|
|
|
|
call %o1
|
|
|
|
add %sp, (2047 + 128), %o0
|
|
|
|
|
|
|
|
sethi %hi(bigkernel), %g2
|
|
|
|
lduw [%g2 + %lo(bigkernel)], %g2
|
|
|
|
cmp %g2, 0
|
|
|
|
be,pt %icc, do_unlock
|
|
|
|
nop
|
|
|
|
|
|
|
|
sethi %hi(call_method), %g2
|
|
|
|
or %g2, %lo(call_method), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x00]
|
|
|
|
mov 5, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x08]
|
|
|
|
mov 1, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x10]
|
|
|
|
sethi %hi(dtlb_load), %g2
|
|
|
|
or %g2, %lo(dtlb_load), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x18]
|
|
|
|
sethi %hi(mmu_ihandle_cache), %g2
|
|
|
|
lduw [%g2 + %lo(mmu_ihandle_cache)], %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x20]
|
|
|
|
sethi %hi(KERNBASE + 0x400000), %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x28]
|
|
|
|
sethi %hi(kern_locked_tte_data), %g2
|
|
|
|
ldx [%g2 + %lo(kern_locked_tte_data)], %g2
|
|
|
|
sethi %hi(0x400000), %g1
|
|
|
|
add %g2, %g1, %g2
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x30]
|
|
|
|
|
|
|
|
mov 14, %g2
|
|
|
|
BRANCH_IF_ANY_CHEETAH(g1,g5,1f)
|
|
|
|
|
|
|
|
mov 62, %g2
|
|
|
|
1:
|
|
|
|
|
|
|
|
stx %g2, [%sp + 2047 + 128 + 0x38]
|
|
|
|
sethi %hi(p1275buf), %g2
|
|
|
|
or %g2, %lo(p1275buf), %g2
|
|
|
|
ldx [%g2 + 0x08], %o1
|
|
|
|
call %o1
|
|
|
|
add %sp, (2047 + 128), %o0
|
|
|
|
|
|
|
|
do_unlock:
|
|
|
|
sethi %hi(prom_entry_lock), %g2
|
|
|
|
stb %g0, [%g2 + %lo(prom_entry_lock)]
|
|
|
|
membar #StoreStore | #StoreLoad
|
|
|
|
|
|
|
|
mov %l1, %sp
|
|
|
|
flushw
|
|
|
|
|
|
|
|
mov %l0, %o0
|
|
|
|
|
|
|
|
wrpr %g0, (PSTATE_PRIV | PSTATE_PEF), %pstate
|
|
|
|
wr %g0, 0, %fprs
|
|
|
|
|
|
|
|
/* XXX Buggy PROM... */
|
|
|
|
srl %o0, 0, %o0
|
|
|
|
ldx [%o0], %g6
|
|
|
|
|
|
|
|
wr %g0, ASI_P, %asi
|
|
|
|
|
|
|
|
mov PRIMARY_CONTEXT, %g7
|
|
|
|
stxa %g0, [%g7] ASI_DMMU
|
|
|
|
membar #Sync
|
|
|
|
mov SECONDARY_CONTEXT, %g7
|
|
|
|
stxa %g0, [%g7] ASI_DMMU
|
|
|
|
membar #Sync
|
|
|
|
|
|
|
|
mov 1, %g5
|
|
|
|
sllx %g5, THREAD_SHIFT, %g5
|
|
|
|
sub %g5, (STACKFRAME_SZ + STACK_BIAS), %g5
|
|
|
|
add %g6, %g5, %sp
|
|
|
|
mov 0, %fp
|
|
|
|
|
|
|
|
wrpr %g0, 0, %wstate
|
|
|
|
wrpr %g0, 0, %tl
|
|
|
|
|
|
|
|
/* Setup the trap globals, then we can resurface. */
|
|
|
|
rdpr %pstate, %o1
|
|
|
|
mov %g6, %o2
|
|
|
|
wrpr %o1, PSTATE_AG, %pstate
|
|
|
|
sethi %hi(sparc64_ttable_tl0), %g5
|
|
|
|
wrpr %g5, %tba
|
|
|
|
mov %o2, %g6
|
|
|
|
|
|
|
|
wrpr %o1, PSTATE_MG, %pstate
|
|
|
|
#define KERN_HIGHBITS ((_PAGE_VALID|_PAGE_SZ4MB)^0xfffff80000000000)
|
|
|
|
#define KERN_LOWBITS (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_W)
|
|
|
|
|
|
|
|
mov TSB_REG, %g1
|
|
|
|
stxa %g0, [%g1] ASI_DMMU
|
|
|
|
membar #Sync
|
|
|
|
mov TLB_SFSR, %g1
|
|
|
|
sethi %uhi(KERN_HIGHBITS), %g2
|
|
|
|
or %g2, %ulo(KERN_HIGHBITS), %g2
|
|
|
|
sllx %g2, 32, %g2
|
|
|
|
or %g2, KERN_LOWBITS, %g2
|
|
|
|
|
|
|
|
BRANCH_IF_ANY_CHEETAH(g3,g7,9f)
|
|
|
|
|
|
|
|
ba,pt %xcc, 1f
|
|
|
|
nop
|
|
|
|
|
|
|
|
9:
|
|
|
|
sethi %uhi(VPTE_BASE_CHEETAH), %g3
|
|
|
|
or %g3, %ulo(VPTE_BASE_CHEETAH), %g3
|
|
|
|
ba,pt %xcc, 2f
|
|
|
|
sllx %g3, 32, %g3
|
|
|
|
1:
|
|
|
|
sethi %uhi(VPTE_BASE_SPITFIRE), %g3
|
|
|
|
or %g3, %ulo(VPTE_BASE_SPITFIRE), %g3
|
|
|
|
sllx %g3, 32, %g3
|
|
|
|
|
|
|
|
2:
|
|
|
|
clr %g7
|
|
|
|
#undef KERN_HIGHBITS
|
|
|
|
#undef KERN_LOWBITS
|
|
|
|
|
|
|
|
wrpr %o1, 0x0, %pstate
|
|
|
|
ldx [%g6 + TI_TASK], %g4
|
|
|
|
|
|
|
|
wrpr %g0, 0, %wstate
|
|
|
|
|
|
|
|
call init_irqwork_curcpu
|
|
|
|
nop
|
|
|
|
|
|
|
|
BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g2,g3,1f)
|
|
|
|
ba,pt %xcc, 2f
|
|
|
|
nop
|
|
|
|
|
|
|
|
1: /* Start using proper page size encodings in ctx register. */
|
|
|
|
sethi %uhi(CTX_CHEETAH_PLUS_NUC), %g3
|
|
|
|
mov PRIMARY_CONTEXT, %g1
|
|
|
|
sllx %g3, 32, %g3
|
|
|
|
sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2
|
|
|
|
or %g3, %g2, %g3
|
|
|
|
stxa %g3, [%g1] ASI_DMMU
|
|
|
|
membar #Sync
|
|
|
|
|
|
|
|
2:
|
|
|
|
rdpr %pstate, %o1
|
|
|
|
or %o1, PSTATE_IE, %o1
|
|
|
|
wrpr %o1, 0, %pstate
|
|
|
|
|
|
|
|
call prom_set_trap_table
|
|
|
|
sethi %hi(sparc64_ttable_tl0), %o0
|
|
|
|
|
|
|
|
call smp_callin
|
|
|
|
nop
|
|
|
|
call cpu_idle
|
|
|
|
mov 0, %o0
|
|
|
|
call cpu_panic
|
|
|
|
nop
|
|
|
|
1: b,a,pt %xcc, 1b
|
|
|
|
|
|
|
|
.align 8
|
|
|
|
sparc64_cpu_startup_end:
|