linux_dsm_epyc7002/arch/sparc64/kernel/trampoline.S
David S. Miller 56fb4df6da [SPARC64]: Elminate all usage of hard-coded trap globals.
UltraSPARC has special sets of global registers which are switched to
for certain trap types.  There is one set for MMU related traps, one
set of Interrupt Vector processing, and another set (called the
Alternate globals) for all other trap types.

For what seems like forever we've hard coded the values in some of
these trap registers.  Some examples include:

1) Interrupt Vector global %g6 holds current processors interrupt
   work struct where received interrupts are managed for IRQ handler
   dispatch.

2) MMU global %g7 holds the base of the page tables of the currently
   active address space.

3) Alternate global %g6 held the current_thread_info() value.

Such hardcoding has resulted in some serious issues in many areas.
There are some code sequences where having another register available
would help clean up the implementation.  Taking traps such as
cross-calls from the OBP firmware requires some trick code sequences
wherein we have to save away and restore all of the special sets of
global registers when we enter/exit OBP.

We were also using the IMMU TSB register on SMP to hold the per-cpu
area base address, which doesn't work any longer now that we actually
use the TSB facility of the cpu.

The implementation is pretty straight forward.  One tricky bit is
getting the current processor ID as that is different on different cpu
variants.  We use a stub with a fancy calling convention which we
patch at boot time.  The calling convention is that the stub is
branched to and the (PC - 4) to return to is in register %g1.  The cpu
number is left in %g6.  This stub can be invoked by using the
__GET_CPUID macro.

We use an array of per-cpu trap state to store the current thread and
physical address of the current address space's page tables.  The
TRAP_LOAD_THREAD_REG loads %g6 with the current thread from this
table, it uses __GET_CPUID and also clobbers %g1.

TRAP_LOAD_IRQ_WORK is used by the interrupt vector processing to load
the current processor's IRQ software state into %g6.  It also uses
__GET_CPUID and clobbers %g1.

Finally, TRAP_LOAD_PGD_PHYS loads the physical address base of the
current address space's page tables into %g7, it clobbers %g1 and uses
__GET_CPUID.

Many refinements are possible, as well as some tuning, with this stuff
in place.

Signed-off-by: David S. Miller <davem@davemloft.net>
2006-03-20 01:11:16 -08:00

327 lines
7.2 KiB
ArmAsm

/* $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
membar #StoreLoad | #StoreStore
brnz,pn %g1, 1b
nop
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(prom_mmu_ihandle_cache), %g2
lduw [%g2 + %lo(prom_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(prom_mmu_ihandle_cache), %g2
lduw [%g2 + %lo(prom_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(prom_mmu_ihandle_cache), %g2
lduw [%g2 + %lo(prom_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(prom_mmu_ihandle_cache), %g2
lduw [%g2 + %lo(prom_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
/* Load TBA, then we can resurface. */
sethi %hi(sparc64_ttable_tl0), %g5
wrpr %g5, %tba
ldx [%g6 + TI_TASK], %g4
wrpr %g0, 0, %wstate
call init_irqwork_curcpu
nop
call init_cur_cpu_trap
nop
/* Start using proper page size encodings in ctx register. */
sethi %hi(sparc64_kern_pri_context), %g3
ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2
mov PRIMARY_CONTEXT, %g1
stxa %g2, [%g1] ASI_DMMU
membar #Sync
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: