kernel_optimize_test/arch/sparc64/kernel/trampoline.S
David S. Miller 64658743fd [SPARC64]: Remove most limitations to kernel image size.
Currently kernel images are limited to 8MB in size, and this causes
problems especially when enabling features that take up a lot of
kernel image space such as lockdep.

The code now will align the kernel image size up to 4MB and map that
many locked TLB entries.  So, the only practical limitation is the
number of available locked TLB entries which is 16 on Cheetah and 64
on pre-Cheetah sparc64 cpus.  Niagara cpus don't actually have hw
locked TLB entry support.  Rather, the hypervisor transparently
provides support for "locked" TLB entries since it runs with physical
addressing and does the initial TLB miss processing.

Fully utilizing this change requires some help from SILO, a patch for
which will be submitted to the maintainer.  Essentially, SILO will
only currently map up to 8MB for the kernel image and that needs to be
increased.

Note that neither this patch nor the SILO bits will help with network
booting.  The openfirmware code will only map up to a certain amount
of kernel image during a network boot and there isn't much we can to
about that other than to implemented a layered network booting
facility.  Solaris has this, and calls it "wanboot" and we may
implement something similar at some point.

Signed-off-by: David S. Miller <davem@davemloft.net>
2008-03-21 17:01:38 -07:00

412 lines
9.1 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 <linux/init.h>
#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>
#include <asm/hypervisor.h>
#include <asm/cpudata.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"
/* XXX __cpuinit this thing XXX */
#define TRAMP_STACK_SIZE 1024
.align 16
tramp_stack:
.skip TRAMP_STACK_SIZE
__CPUINIT
.align 8
.globl sparc64_cpu_startup, sparc64_cpu_startup_end
sparc64_cpu_startup:
BRANCH_IF_SUN4V(g1, niagara_startup)
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
/* fallthru */
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
/* fallthru */
niagara_startup:
/* 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:
mov %o0, %l0
BRANCH_IF_SUN4V(g1, niagara_lock_tlb)
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 'num_kernel_image_mappings' consequetive entries.
*/
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
add %l2, -(192 + 128), %sp
flushw
/* Setup the loop variables:
* %l3: VADDR base
* %l4: TTE base
* %l5: Loop iterator, iterates from 0 to 'num_kernel_image_mappings'
* %l6: Number of TTE entries to map
* %l7: Highest TTE entry number, we count down
*/
sethi %hi(KERNBASE), %l3
sethi %hi(kern_locked_tte_data), %l4
ldx [%l4 + %lo(kern_locked_tte_data)], %l4
clr %l5
sethi %hi(num_kernel_image_mappings), %l6
lduw [%l6 + %lo(num_kernel_image_mappings)], %l6
add %l6, 1, %l6
mov 15, %l7
BRANCH_IF_ANY_CHEETAH(g1,g5,2f)
mov 63, %l7
2:
3:
/* Lock into I-MMU */
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]
/* Each TTE maps 4MB, convert index to offset. */
sllx %l5, 22, %g1
add %l3, %g1, %g2
stx %g2, [%sp + 2047 + 128 + 0x28] ! VADDR
add %l4, %g1, %g2
stx %g2, [%sp + 2047 + 128 + 0x30] ! TTE
/* TTE index is highest minus loop index. */
sub %l7, %l5, %g2
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
/* Lock into D-MMU */
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]
/* Each TTE maps 4MB, convert index to offset. */
sllx %l5, 22, %g1
add %l3, %g1, %g2
stx %g2, [%sp + 2047 + 128 + 0x28] ! VADDR
add %l4, %g1, %g2
stx %g2, [%sp + 2047 + 128 + 0x30] ! TTE
/* TTE index is highest minus loop index. */
sub %l7, %l5, %g2
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
add %l5, 1, %l5
cmp %l5, %l6
bne,pt %xcc, 3b
nop
sethi %hi(prom_entry_lock), %g2
stb %g0, [%g2 + %lo(prom_entry_lock)]
membar #StoreStore | #StoreLoad
ba,pt %xcc, after_lock_tlb
nop
niagara_lock_tlb:
sethi %hi(KERNBASE), %l3
sethi %hi(kern_locked_tte_data), %l4
ldx [%l4 + %lo(kern_locked_tte_data)], %l4
clr %l5
sethi %hi(num_kernel_image_mappings), %l6
lduw [%l6 + %lo(num_kernel_image_mappings)], %l6
add %l6, 1, %l6
1:
mov HV_FAST_MMU_MAP_PERM_ADDR, %o5
sllx %l5, 22, %g2
add %l3, %g2, %o0
clr %o1
add %l4, %g2, %o2
mov HV_MMU_IMMU, %o3
ta HV_FAST_TRAP
mov HV_FAST_MMU_MAP_PERM_ADDR, %o5
sllx %l5, 22, %g2
add %l3, %g2, %o0
clr %o1
add %l4, %g2, %o2
mov HV_MMU_DMMU, %o3
ta HV_FAST_TRAP
add %l5, 1, %l5
cmp %l5, %l6
bne,pt %xcc, 1b
nop
after_lock_tlb:
wrpr %g0, (PSTATE_PRIV | PSTATE_PEF), %pstate
wr %g0, 0, %fprs
wr %g0, ASI_P, %asi
mov PRIMARY_CONTEXT, %g7
661: stxa %g0, [%g7] ASI_DMMU
.section .sun4v_1insn_patch, "ax"
.word 661b
stxa %g0, [%g7] ASI_MMU
.previous
membar #Sync
mov SECONDARY_CONTEXT, %g7
661: stxa %g0, [%g7] ASI_DMMU
.section .sun4v_1insn_patch, "ax"
.word 661b
stxa %g0, [%g7] ASI_MMU
.previous
membar #Sync
/* Everything we do here, until we properly take over the
* trap table, must be done with extreme care. We cannot
* make any references to %g6 (current thread pointer),
* %g4 (current task pointer), or %g5 (base of current cpu's
* per-cpu area) until we properly take over the trap table
* from the firmware and hypervisor.
*
* Get onto temporary stack which is in the locked kernel image.
*/
sethi %hi(tramp_stack), %g1
or %g1, %lo(tramp_stack), %g1
add %g1, TRAMP_STACK_SIZE, %g1
sub %g1, STACKFRAME_SZ + STACK_BIAS + 256, %sp
mov 0, %fp
/* Put garbage in these registers to trap any access to them. */
set 0xdeadbeef, %g4
set 0xdeadbeef, %g5
set 0xdeadbeef, %g6
call init_irqwork_curcpu
nop
sethi %hi(tlb_type), %g3
lduw [%g3 + %lo(tlb_type)], %g2
cmp %g2, 3
bne,pt %icc, 1f
nop
call hard_smp_processor_id
nop
call sun4v_register_mondo_queues
nop
1: call init_cur_cpu_trap
ldx [%l0], %o0
/* 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
661: stxa %g2, [%g1] ASI_DMMU
.section .sun4v_1insn_patch, "ax"
.word 661b
stxa %g2, [%g1] ASI_MMU
.previous
membar #Sync
wrpr %g0, 0, %wstate
/* As a hack, put &init_thread_union into %g6.
* prom_world() loads from here to restore the %asi
* register.
*/
sethi %hi(init_thread_union), %g6
or %g6, %lo(init_thread_union), %g6
sethi %hi(is_sun4v), %o0
lduw [%o0 + %lo(is_sun4v)], %o0
brz,pt %o0, 1f
nop
TRAP_LOAD_TRAP_BLOCK(%g2, %g3)
add %g2, TRAP_PER_CPU_FAULT_INFO, %g2
stxa %g2, [%g0] ASI_SCRATCHPAD
/* Compute physical address:
*
* paddr = kern_base + (mmfsa_vaddr - KERNBASE)
*/
sethi %hi(KERNBASE), %g3
sub %g2, %g3, %g2
sethi %hi(kern_base), %g3
ldx [%g3 + %lo(kern_base)], %g3
add %g2, %g3, %o1
sethi %hi(sparc64_ttable_tl0), %o0
set prom_set_trap_table_name, %g2
stx %g2, [%sp + 2047 + 128 + 0x00]
mov 2, %g2
stx %g2, [%sp + 2047 + 128 + 0x08]
mov 0, %g2
stx %g2, [%sp + 2047 + 128 + 0x10]
stx %o0, [%sp + 2047 + 128 + 0x18]
stx %o1, [%sp + 2047 + 128 + 0x20]
sethi %hi(p1275buf), %g2
or %g2, %lo(p1275buf), %g2
ldx [%g2 + 0x08], %o1
call %o1
add %sp, (2047 + 128), %o0
ba,pt %xcc, 2f
nop
1: sethi %hi(sparc64_ttable_tl0), %o0
set prom_set_trap_table_name, %g2
stx %g2, [%sp + 2047 + 128 + 0x00]
mov 1, %g2
stx %g2, [%sp + 2047 + 128 + 0x08]
mov 0, %g2
stx %g2, [%sp + 2047 + 128 + 0x10]
stx %o0, [%sp + 2047 + 128 + 0x18]
sethi %hi(p1275buf), %g2
or %g2, %lo(p1275buf), %g2
ldx [%g2 + 0x08], %o1
call %o1
add %sp, (2047 + 128), %o0
2: ldx [%l0], %g6
ldx [%g6 + TI_TASK], %g4
mov 1, %g5
sllx %g5, THREAD_SHIFT, %g5
sub %g5, (STACKFRAME_SZ + STACK_BIAS), %g5
add %g6, %g5, %sp
mov 0, %fp
rdpr %pstate, %o1
or %o1, PSTATE_IE, %o1
wrpr %o1, 0, %pstate
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: