AuxXxilium/linux_dsm_epyc7002
1
0
Fork 0
mirror of https://github.com/AuxXxilium/linux_dsm_epyc7002.git synced 2024-12-21 15:48:56 +07:00
Code Issues Actions Packages Projects Releases Wiki Activity
d11c2a0de2
linux_dsm_epyc7002/arch/sparc/include/asm/ttable.h

678 lines
20 KiB
C
Raw Normal View History

sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
#ifndef _SPARC64_TTABLE_H
#define _SPARC64_TTABLE_H
#include <asm/utrap.h>
sparc64: Make %pil level 15 a pseudo-NMI. So that we can profile code even in a local_irq_disable() section, only write 14 (instead of 15) into the %pil register to disable IRQs. This allows PIL level 15 to serve as a pseudo NMI. Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-24 12:55:29 +07:00
#include <asm/pil.h>
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
#ifdef __ASSEMBLY__
#include <asm/thread_info.h>
#endif
#define BOOT_KERNEL b sparc64_boot; nop; nop; nop; nop; nop; nop; nop;
/* We need a "cleaned" instruction... */
#define CLEAN_WINDOW \
rdpr %cleanwin, %l0; add %l0, 1, %l0; \
wrpr %l0, 0x0, %cleanwin; \
clr %o0; clr %o1; clr %o2; clr %o3; \
clr %o4; clr %o5; clr %o6; clr %o7; \
clr %l0; clr %l1; clr %l2; clr %l3; \
clr %l4; clr %l5; clr %l6; clr %l7; \
retry; \
nop;nop;nop;nop;nop;nop;nop;nop;nop;nop;nop;nop;
#define TRAP(routine) \
sethi %hi(109f), %g7; \
ba,pt %xcc, etrap; \
109: or %g7, %lo(109b), %g7; \
call routine; \
add %sp, PTREGS_OFF, %o0; \
ba,pt %xcc, rtrap; \
nop; \
nop;
#define TRAP_7INSNS(routine) \
sethi %hi(109f), %g7; \
ba,pt %xcc, etrap; \
109: or %g7, %lo(109b), %g7; \
call routine; \
add %sp, PTREGS_OFF, %o0; \
ba,pt %xcc, rtrap; \
nop;
#define TRAP_SAVEFPU(routine) \
sethi %hi(109f), %g7; \
ba,pt %xcc, do_fptrap; \
109: or %g7, %lo(109b), %g7; \
call routine; \
add %sp, PTREGS_OFF, %o0; \
ba,pt %xcc, rtrap; \
nop; \
nop;
#define TRAP_NOSAVE(routine) \
ba,pt %xcc, routine; \
nop; \
nop; nop; nop; nop; nop; nop;
#define TRAP_NOSAVE_7INSNS(routine) \
ba,pt %xcc, routine; \
nop; \
nop; nop; nop; nop; nop;
#define TRAPTL1(routine) \
sethi %hi(109f), %g7; \
ba,pt %xcc, etraptl1; \
109: or %g7, %lo(109b), %g7; \
call routine; \
add %sp, PTREGS_OFF, %o0; \
ba,pt %xcc, rtrap; \
nop; \
nop;
#define TRAP_ARG(routine, arg) \
sethi %hi(109f), %g7; \
ba,pt %xcc, etrap; \
109: or %g7, %lo(109b), %g7; \
add %sp, PTREGS_OFF, %o0; \
call routine; \
mov arg, %o1; \
ba,pt %xcc, rtrap; \
nop;
#define TRAPTL1_ARG(routine, arg) \
sethi %hi(109f), %g7; \
ba,pt %xcc, etraptl1; \
109: or %g7, %lo(109b), %g7; \
add %sp, PTREGS_OFF, %o0; \
call routine; \
mov arg, %o1; \
ba,pt %xcc, rtrap; \
nop;
#define SYSCALL_TRAP(routine, systbl) \
rdpr %pil, %g2; \
mov TSTATE_SYSCALL, %g3; \
sethi %hi(109f), %g7; \
ba,pt %xcc, etrap_syscall; \
109: or %g7, %lo(109b), %g7; \
sethi %hi(systbl), %l7; \
ba,pt %xcc, routine; \
or %l7, %lo(systbl), %l7;
#define TRAP_UTRAP(handler,lvl) \
mov handler, %g3; \
ba,pt %xcc, utrap_trap; \
mov lvl, %g4; \
nop; \
nop; \
nop; \
nop; \
nop;
#ifdef CONFIG_COMPAT
#define LINUX_32BIT_SYSCALL_TRAP SYSCALL_TRAP(linux_sparc_syscall32, sys_call_table32)
#else
#define LINUX_32BIT_SYSCALL_TRAP BTRAP(0x110)
#endif
#define LINUX_64BIT_SYSCALL_TRAP SYSCALL_TRAP(linux_sparc_syscall, sys_call_table64)
#define GETCC_TRAP TRAP(getcc)
#define SETCC_TRAP TRAP(setcc)
#define BREAKPOINT_TRAP TRAP(breakpoint_trap)
#ifdef CONFIG_TRACE_IRQFLAGS
#define TRAP_IRQ(routine, level) \
rdpr %pil, %g2; \
sparc64: Make %pil level 15 a pseudo-NMI. So that we can profile code even in a local_irq_disable() section, only write 14 (instead of 15) into the %pil register to disable IRQs. This allows PIL level 15 to serve as a pseudo NMI. Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-24 12:55:29 +07:00
wrpr %g0, PIL_NORMAL_MAX, %pil; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
sethi %hi(1f-4), %g7; \
ba,pt %xcc, etrap_irq; \
or %g7, %lo(1f-4), %g7; \
nop; \
nop; \
nop; \
.subsection 2; \
1: call trace_hardirqs_off; \
nop; \
mov level, %o0; \
call routine; \
add %sp, PTREGS_OFF, %o1; \
ba,a,pt %xcc, rtrap_irq; \
.previous;
#else
#define TRAP_IRQ(routine, level) \
rdpr %pil, %g2; \
sparc64: Make %pil level 15 a pseudo-NMI. So that we can profile code even in a local_irq_disable() section, only write 14 (instead of 15) into the %pil register to disable IRQs. This allows PIL level 15 to serve as a pseudo NMI. Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-24 12:55:29 +07:00
wrpr %g0, PIL_NORMAL_MAX, %pil; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
ba,pt %xcc, etrap_irq; \
rd %pc, %g7; \
mov level, %o0; \
call routine; \
add %sp, PTREGS_OFF, %o1; \
ba,a,pt %xcc, rtrap_irq;
#endif
sparc64: Make %pil level 15 a pseudo-NMI. So that we can profile code even in a local_irq_disable() section, only write 14 (instead of 15) into the %pil register to disable IRQs. This allows PIL level 15 to serve as a pseudo NMI. Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-24 12:55:29 +07:00
#define TRAP_NMI_IRQ(routine, level) \
rdpr %pil, %g2; \
wrpr %g0, PIL_NMI, %pil; \
ba,pt %xcc, etrap_irq; \
rd %pc, %g7; \
mov level, %o0; \
call routine; \
add %sp, PTREGS_OFF, %o1; \
sparc64: Make special trap return path for TRAP_NMI(). We don't want the rtrap path to try and run softirqs or anything like that when returning from a PIL==15 NMI. Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-26 13:24:59 +07:00
ba,a,pt %xcc, rtrap_nmi;
sparc64: Make %pil level 15 a pseudo-NMI. So that we can profile code even in a local_irq_disable() section, only write 14 (instead of 15) into the %pil register to disable IRQs. This allows PIL level 15 to serve as a pseudo NMI. Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-24 12:55:29 +07:00
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
#define TRAP_IVEC TRAP_NOSAVE(do_ivec)
#define BTRAP(lvl) TRAP_ARG(bad_trap, lvl)
#define BTRAPTL1(lvl) TRAPTL1_ARG(bad_trap_tl1, lvl)
#define FLUSH_WINDOW_TRAP \
ba,pt %xcc, etrap; \
rd %pc, %g7; \
flushw; \
ldx [%sp + PTREGS_OFF + PT_V9_TNPC], %l1; \
add %l1, 4, %l2; \
stx %l1, [%sp + PTREGS_OFF + PT_V9_TPC]; \
ba,pt %xcc, rtrap; \
stx %l2, [%sp + PTREGS_OFF + PT_V9_TNPC];
#ifdef CONFIG_KPROBES
#define KPROBES_TRAP(lvl) TRAP_IRQ(kprobe_trap, lvl)
#else
#define KPROBES_TRAP(lvl) TRAP_ARG(bad_trap, lvl)
#endif
#ifdef CONFIG_KGDB
#define KGDB_TRAP(lvl) TRAP_IRQ(kgdb_trap, lvl)
#else
#define KGDB_TRAP(lvl) TRAP_ARG(bad_trap, lvl)
#endif
#define SUN4V_ITSB_MISS \
ldxa [%g0] ASI_SCRATCHPAD, %g2; \
ldx [%g2 + HV_FAULT_I_ADDR_OFFSET], %g4; \
ldx [%g2 + HV_FAULT_I_CTX_OFFSET], %g5; \
srlx %g4, 22, %g6; \
ba,pt %xcc, sun4v_itsb_miss; \
nop; \
nop; \
nop;
#define SUN4V_DTSB_MISS \
ldxa [%g0] ASI_SCRATCHPAD, %g2; \
ldx [%g2 + HV_FAULT_D_ADDR_OFFSET], %g4; \
ldx [%g2 + HV_FAULT_D_CTX_OFFSET], %g5; \
srlx %g4, 22, %g6; \
ba,pt %xcc, sun4v_dtsb_miss; \
nop; \
nop; \
nop;
/* Before touching these macros, you owe it to yourself to go and
* see how arch/sparc64/kernel/winfixup.S works... -DaveM
*
* For the user cases we used to use the %asi register, but
* it turns out that the "wr xxx, %asi" costs ~5 cycles, so
* now we use immediate ASI loads and stores instead. Kudos
* to Greg Onufer for pointing out this performance anomaly.
*
* Further note that we cannot use the g2, g4, g5, and g7 alternate
* globals in the spill routines, check out the save instruction in
* arch/sparc64/kernel/etrap.S to see what I mean about g2, and
* g4/g5 are the globals which are preserved by etrap processing
* for the caller of it. The g7 register is the return pc for
* etrap. Finally, g6 is the current thread register so we cannot
* us it in the spill handlers either. Most of these rules do not
* apply to fill processing, only g6 is not usable.
*/
/* Normal kernel spill */
#define SPILL_0_NORMAL \
stx %l0, [%sp + STACK_BIAS + 0x00]; \
stx %l1, [%sp + STACK_BIAS + 0x08]; \
stx %l2, [%sp + STACK_BIAS + 0x10]; \
stx %l3, [%sp + STACK_BIAS + 0x18]; \
stx %l4, [%sp + STACK_BIAS + 0x20]; \
stx %l5, [%sp + STACK_BIAS + 0x28]; \
stx %l6, [%sp + STACK_BIAS + 0x30]; \
stx %l7, [%sp + STACK_BIAS + 0x38]; \
stx %i0, [%sp + STACK_BIAS + 0x40]; \
stx %i1, [%sp + STACK_BIAS + 0x48]; \
stx %i2, [%sp + STACK_BIAS + 0x50]; \
stx %i3, [%sp + STACK_BIAS + 0x58]; \
stx %i4, [%sp + STACK_BIAS + 0x60]; \
stx %i5, [%sp + STACK_BIAS + 0x68]; \
stx %i6, [%sp + STACK_BIAS + 0x70]; \
stx %i7, [%sp + STACK_BIAS + 0x78]; \
saved; retry; nop; nop; nop; nop; nop; nop; \
nop; nop; nop; nop; nop; nop; nop; nop;
#define SPILL_0_NORMAL_ETRAP \
etrap_kernel_spill: \
stx %l0, [%sp + STACK_BIAS + 0x00]; \
stx %l1, [%sp + STACK_BIAS + 0x08]; \
stx %l2, [%sp + STACK_BIAS + 0x10]; \
stx %l3, [%sp + STACK_BIAS + 0x18]; \
stx %l4, [%sp + STACK_BIAS + 0x20]; \
stx %l5, [%sp + STACK_BIAS + 0x28]; \
stx %l6, [%sp + STACK_BIAS + 0x30]; \
stx %l7, [%sp + STACK_BIAS + 0x38]; \
stx %i0, [%sp + STACK_BIAS + 0x40]; \
stx %i1, [%sp + STACK_BIAS + 0x48]; \
stx %i2, [%sp + STACK_BIAS + 0x50]; \
stx %i3, [%sp + STACK_BIAS + 0x58]; \
stx %i4, [%sp + STACK_BIAS + 0x60]; \
stx %i5, [%sp + STACK_BIAS + 0x68]; \
stx %i6, [%sp + STACK_BIAS + 0x70]; \
stx %i7, [%sp + STACK_BIAS + 0x78]; \
saved; \
sub %g1, 2, %g1; \
ba,pt %xcc, etrap_save; \
wrpr %g1, %cwp; \
nop; nop; nop; nop; nop; nop; nop; nop; \
nop; nop; nop; nop;
/* Normal 64bit spill */
#define SPILL_1_GENERIC(ASI) \
add %sp, STACK_BIAS + 0x00, %g1; \
stxa %l0, [%g1 + %g0] ASI; \
mov 0x08, %g3; \
stxa %l1, [%g1 + %g3] ASI; \
add %g1, 0x10, %g1; \
stxa %l2, [%g1 + %g0] ASI; \
stxa %l3, [%g1 + %g3] ASI; \
add %g1, 0x10, %g1; \
stxa %l4, [%g1 + %g0] ASI; \
stxa %l5, [%g1 + %g3] ASI; \
add %g1, 0x10, %g1; \
stxa %l6, [%g1 + %g0] ASI; \
stxa %l7, [%g1 + %g3] ASI; \
add %g1, 0x10, %g1; \
stxa %i0, [%g1 + %g0] ASI; \
stxa %i1, [%g1 + %g3] ASI; \
add %g1, 0x10, %g1; \
stxa %i2, [%g1 + %g0] ASI; \
stxa %i3, [%g1 + %g3] ASI; \
add %g1, 0x10, %g1; \
stxa %i4, [%g1 + %g0] ASI; \
stxa %i5, [%g1 + %g3] ASI; \
add %g1, 0x10, %g1; \
stxa %i6, [%g1 + %g0] ASI; \
stxa %i7, [%g1 + %g3] ASI; \
saved; \
retry; nop; nop; \
b,a,pt %xcc, spill_fixup_dax; \
b,a,pt %xcc, spill_fixup_mna; \
b,a,pt %xcc, spill_fixup;
#define SPILL_1_GENERIC_ETRAP \
etrap_user_spill_64bit: \
stxa %l0, [%sp + STACK_BIAS + 0x00] %asi; \
stxa %l1, [%sp + STACK_BIAS + 0x08] %asi; \
stxa %l2, [%sp + STACK_BIAS + 0x10] %asi; \
stxa %l3, [%sp + STACK_BIAS + 0x18] %asi; \
stxa %l4, [%sp + STACK_BIAS + 0x20] %asi; \
stxa %l5, [%sp + STACK_BIAS + 0x28] %asi; \
stxa %l6, [%sp + STACK_BIAS + 0x30] %asi; \
stxa %l7, [%sp + STACK_BIAS + 0x38] %asi; \
stxa %i0, [%sp + STACK_BIAS + 0x40] %asi; \
stxa %i1, [%sp + STACK_BIAS + 0x48] %asi; \
stxa %i2, [%sp + STACK_BIAS + 0x50] %asi; \
stxa %i3, [%sp + STACK_BIAS + 0x58] %asi; \
stxa %i4, [%sp + STACK_BIAS + 0x60] %asi; \
stxa %i5, [%sp + STACK_BIAS + 0x68] %asi; \
stxa %i6, [%sp + STACK_BIAS + 0x70] %asi; \
stxa %i7, [%sp + STACK_BIAS + 0x78] %asi; \
saved; \
sub %g1, 2, %g1; \
ba,pt %xcc, etrap_save; \
wrpr %g1, %cwp; \
nop; nop; nop; nop; nop; \
nop; nop; nop; nop; \
ba,a,pt %xcc, etrap_spill_fixup_64bit; \
ba,a,pt %xcc, etrap_spill_fixup_64bit; \
ba,a,pt %xcc, etrap_spill_fixup_64bit;
#define SPILL_1_GENERIC_ETRAP_FIXUP \
etrap_spill_fixup_64bit: \
ldub [%g6 + TI_WSAVED], %g1; \
sll %g1, 3, %g3; \
add %g6, %g3, %g3; \
stx %sp, [%g3 + TI_RWIN_SPTRS]; \
sll %g1, 7, %g3; \
add %g6, %g3, %g3; \
stx %l0, [%g3 + TI_REG_WINDOW + 0x00]; \
stx %l1, [%g3 + TI_REG_WINDOW + 0x08]; \
stx %l2, [%g3 + TI_REG_WINDOW + 0x10]; \
stx %l3, [%g3 + TI_REG_WINDOW + 0x18]; \
stx %l4, [%g3 + TI_REG_WINDOW + 0x20]; \
stx %l5, [%g3 + TI_REG_WINDOW + 0x28]; \
stx %l6, [%g3 + TI_REG_WINDOW + 0x30]; \
stx %l7, [%g3 + TI_REG_WINDOW + 0x38]; \
stx %i0, [%g3 + TI_REG_WINDOW + 0x40]; \
stx %i1, [%g3 + TI_REG_WINDOW + 0x48]; \
stx %i2, [%g3 + TI_REG_WINDOW + 0x50]; \
stx %i3, [%g3 + TI_REG_WINDOW + 0x58]; \
stx %i4, [%g3 + TI_REG_WINDOW + 0x60]; \
stx %i5, [%g3 + TI_REG_WINDOW + 0x68]; \
stx %i6, [%g3 + TI_REG_WINDOW + 0x70]; \
stx %i7, [%g3 + TI_REG_WINDOW + 0x78]; \
add %g1, 1, %g1; \
stb %g1, [%g6 + TI_WSAVED]; \
saved; \
rdpr %cwp, %g1; \
sub %g1, 2, %g1; \
ba,pt %xcc, etrap_save; \
wrpr %g1, %cwp; \
nop; nop; nop
/* Normal 32bit spill */
#define SPILL_2_GENERIC(ASI) \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
and %sp, 1, %g3; \
brnz,pn %g3, (. - (128 + 4)); \
srl %sp, 0, %sp; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
stwa %l0, [%sp + %g0] ASI; \
mov 0x04, %g3; \
stwa %l1, [%sp + %g3] ASI; \
add %sp, 0x08, %g1; \
stwa %l2, [%g1 + %g0] ASI; \
stwa %l3, [%g1 + %g3] ASI; \
add %g1, 0x08, %g1; \
stwa %l4, [%g1 + %g0] ASI; \
stwa %l5, [%g1 + %g3] ASI; \
add %g1, 0x08, %g1; \
stwa %l6, [%g1 + %g0] ASI; \
stwa %l7, [%g1 + %g3] ASI; \
add %g1, 0x08, %g1; \
stwa %i0, [%g1 + %g0] ASI; \
stwa %i1, [%g1 + %g3] ASI; \
add %g1, 0x08, %g1; \
stwa %i2, [%g1 + %g0] ASI; \
stwa %i3, [%g1 + %g3] ASI; \
add %g1, 0x08, %g1; \
stwa %i4, [%g1 + %g0] ASI; \
stwa %i5, [%g1 + %g3] ASI; \
add %g1, 0x08, %g1; \
stwa %i6, [%g1 + %g0] ASI; \
stwa %i7, [%g1 + %g3] ASI; \
saved; \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
retry; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
b,a,pt %xcc, spill_fixup_dax; \
b,a,pt %xcc, spill_fixup_mna; \
b,a,pt %xcc, spill_fixup;
#define SPILL_2_GENERIC_ETRAP \
etrap_user_spill_32bit: \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
and %sp, 1, %g3; \
brnz,pn %g3, etrap_user_spill_64bit; \
srl %sp, 0, %sp; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
stwa %l0, [%sp + 0x00] %asi; \
stwa %l1, [%sp + 0x04] %asi; \
stwa %l2, [%sp + 0x08] %asi; \
stwa %l3, [%sp + 0x0c] %asi; \
stwa %l4, [%sp + 0x10] %asi; \
stwa %l5, [%sp + 0x14] %asi; \
stwa %l6, [%sp + 0x18] %asi; \
stwa %l7, [%sp + 0x1c] %asi; \
stwa %i0, [%sp + 0x20] %asi; \
stwa %i1, [%sp + 0x24] %asi; \
stwa %i2, [%sp + 0x28] %asi; \
stwa %i3, [%sp + 0x2c] %asi; \
stwa %i4, [%sp + 0x30] %asi; \
stwa %i5, [%sp + 0x34] %asi; \
stwa %i6, [%sp + 0x38] %asi; \
stwa %i7, [%sp + 0x3c] %asi; \
saved; \
sub %g1, 2, %g1; \
ba,pt %xcc, etrap_save; \
wrpr %g1, %cwp; \
nop; nop; nop; nop; \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
nop; nop; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
ba,a,pt %xcc, etrap_spill_fixup_32bit; \
ba,a,pt %xcc, etrap_spill_fixup_32bit; \
ba,a,pt %xcc, etrap_spill_fixup_32bit;
#define SPILL_2_GENERIC_ETRAP_FIXUP \
etrap_spill_fixup_32bit: \
ldub [%g6 + TI_WSAVED], %g1; \
sll %g1, 3, %g3; \
add %g6, %g3, %g3; \
stx %sp, [%g3 + TI_RWIN_SPTRS]; \
sll %g1, 7, %g3; \
add %g6, %g3, %g3; \
stw %l0, [%g3 + TI_REG_WINDOW + 0x00]; \
stw %l1, [%g3 + TI_REG_WINDOW + 0x04]; \
stw %l2, [%g3 + TI_REG_WINDOW + 0x08]; \
stw %l3, [%g3 + TI_REG_WINDOW + 0x0c]; \
stw %l4, [%g3 + TI_REG_WINDOW + 0x10]; \
stw %l5, [%g3 + TI_REG_WINDOW + 0x14]; \
stw %l6, [%g3 + TI_REG_WINDOW + 0x18]; \
stw %l7, [%g3 + TI_REG_WINDOW + 0x1c]; \
stw %i0, [%g3 + TI_REG_WINDOW + 0x20]; \
stw %i1, [%g3 + TI_REG_WINDOW + 0x24]; \
stw %i2, [%g3 + TI_REG_WINDOW + 0x28]; \
stw %i3, [%g3 + TI_REG_WINDOW + 0x2c]; \
stw %i4, [%g3 + TI_REG_WINDOW + 0x30]; \
stw %i5, [%g3 + TI_REG_WINDOW + 0x34]; \
stw %i6, [%g3 + TI_REG_WINDOW + 0x38]; \
stw %i7, [%g3 + TI_REG_WINDOW + 0x3c]; \
add %g1, 1, %g1; \
stb %g1, [%g6 + TI_WSAVED]; \
saved; \
rdpr %cwp, %g1; \
sub %g1, 2, %g1; \
ba,pt %xcc, etrap_save; \
wrpr %g1, %cwp; \
nop; nop; nop
#define SPILL_1_NORMAL SPILL_1_GENERIC(ASI_AIUP)
#define SPILL_2_NORMAL SPILL_2_GENERIC(ASI_AIUP)
#define SPILL_3_NORMAL SPILL_0_NORMAL
#define SPILL_4_NORMAL SPILL_0_NORMAL
#define SPILL_5_NORMAL SPILL_0_NORMAL
#define SPILL_6_NORMAL SPILL_0_NORMAL
#define SPILL_7_NORMAL SPILL_0_NORMAL
#define SPILL_0_OTHER SPILL_0_NORMAL
#define SPILL_1_OTHER SPILL_1_GENERIC(ASI_AIUS)
#define SPILL_2_OTHER SPILL_2_GENERIC(ASI_AIUS)
#define SPILL_3_OTHER SPILL_3_NORMAL
#define SPILL_4_OTHER SPILL_4_NORMAL
#define SPILL_5_OTHER SPILL_5_NORMAL
#define SPILL_6_OTHER SPILL_6_NORMAL
#define SPILL_7_OTHER SPILL_7_NORMAL
/* Normal kernel fill */
#define FILL_0_NORMAL \
ldx [%sp + STACK_BIAS + 0x00], %l0; \
ldx [%sp + STACK_BIAS + 0x08], %l1; \
ldx [%sp + STACK_BIAS + 0x10], %l2; \
ldx [%sp + STACK_BIAS + 0x18], %l3; \
ldx [%sp + STACK_BIAS + 0x20], %l4; \
ldx [%sp + STACK_BIAS + 0x28], %l5; \
ldx [%sp + STACK_BIAS + 0x30], %l6; \
ldx [%sp + STACK_BIAS + 0x38], %l7; \
ldx [%sp + STACK_BIAS + 0x40], %i0; \
ldx [%sp + STACK_BIAS + 0x48], %i1; \
ldx [%sp + STACK_BIAS + 0x50], %i2; \
ldx [%sp + STACK_BIAS + 0x58], %i3; \
ldx [%sp + STACK_BIAS + 0x60], %i4; \
ldx [%sp + STACK_BIAS + 0x68], %i5; \
ldx [%sp + STACK_BIAS + 0x70], %i6; \
ldx [%sp + STACK_BIAS + 0x78], %i7; \
restored; retry; nop; nop; nop; nop; nop; nop; \
nop; nop; nop; nop; nop; nop; nop; nop;
#define FILL_0_NORMAL_RTRAP \
kern_rtt_fill: \
rdpr %cwp, %g1; \
sub %g1, 1, %g1; \
wrpr %g1, %cwp; \
ldx [%sp + STACK_BIAS + 0x00], %l0; \
ldx [%sp + STACK_BIAS + 0x08], %l1; \
ldx [%sp + STACK_BIAS + 0x10], %l2; \
ldx [%sp + STACK_BIAS + 0x18], %l3; \
ldx [%sp + STACK_BIAS + 0x20], %l4; \
ldx [%sp + STACK_BIAS + 0x28], %l5; \
ldx [%sp + STACK_BIAS + 0x30], %l6; \
ldx [%sp + STACK_BIAS + 0x38], %l7; \
ldx [%sp + STACK_BIAS + 0x40], %i0; \
ldx [%sp + STACK_BIAS + 0x48], %i1; \
ldx [%sp + STACK_BIAS + 0x50], %i2; \
ldx [%sp + STACK_BIAS + 0x58], %i3; \
ldx [%sp + STACK_BIAS + 0x60], %i4; \
ldx [%sp + STACK_BIAS + 0x68], %i5; \
ldx [%sp + STACK_BIAS + 0x70], %i6; \
ldx [%sp + STACK_BIAS + 0x78], %i7; \
restored; \
add %g1, 1, %g1; \
ba,pt %xcc, kern_rtt_restore; \
wrpr %g1, %cwp; \
nop; nop; nop; nop; nop; \
nop; nop; nop; nop;
/* Normal 64bit fill */
#define FILL_1_GENERIC(ASI) \
add %sp, STACK_BIAS + 0x00, %g1; \
ldxa [%g1 + %g0] ASI, %l0; \
mov 0x08, %g2; \
mov 0x10, %g3; \
ldxa [%g1 + %g2] ASI, %l1; \
mov 0x18, %g5; \
ldxa [%g1 + %g3] ASI, %l2; \
ldxa [%g1 + %g5] ASI, %l3; \
add %g1, 0x20, %g1; \
ldxa [%g1 + %g0] ASI, %l4; \
ldxa [%g1 + %g2] ASI, %l5; \
ldxa [%g1 + %g3] ASI, %l6; \
ldxa [%g1 + %g5] ASI, %l7; \
add %g1, 0x20, %g1; \
ldxa [%g1 + %g0] ASI, %i0; \
ldxa [%g1 + %g2] ASI, %i1; \
ldxa [%g1 + %g3] ASI, %i2; \
ldxa [%g1 + %g5] ASI, %i3; \
add %g1, 0x20, %g1; \
ldxa [%g1 + %g0] ASI, %i4; \
ldxa [%g1 + %g2] ASI, %i5; \
ldxa [%g1 + %g3] ASI, %i6; \
ldxa [%g1 + %g5] ASI, %i7; \
restored; \
retry; nop; nop; nop; nop; \
b,a,pt %xcc, fill_fixup_dax; \
b,a,pt %xcc, fill_fixup_mna; \
b,a,pt %xcc, fill_fixup;
#define FILL_1_GENERIC_RTRAP \
user_rtt_fill_64bit: \
ldxa [%sp + STACK_BIAS + 0x00] %asi, %l0; \
ldxa [%sp + STACK_BIAS + 0x08] %asi, %l1; \
ldxa [%sp + STACK_BIAS + 0x10] %asi, %l2; \
ldxa [%sp + STACK_BIAS + 0x18] %asi, %l3; \
ldxa [%sp + STACK_BIAS + 0x20] %asi, %l4; \
ldxa [%sp + STACK_BIAS + 0x28] %asi, %l5; \
ldxa [%sp + STACK_BIAS + 0x30] %asi, %l6; \
ldxa [%sp + STACK_BIAS + 0x38] %asi, %l7; \
ldxa [%sp + STACK_BIAS + 0x40] %asi, %i0; \
ldxa [%sp + STACK_BIAS + 0x48] %asi, %i1; \
ldxa [%sp + STACK_BIAS + 0x50] %asi, %i2; \
ldxa [%sp + STACK_BIAS + 0x58] %asi, %i3; \
ldxa [%sp + STACK_BIAS + 0x60] %asi, %i4; \
ldxa [%sp + STACK_BIAS + 0x68] %asi, %i5; \
ldxa [%sp + STACK_BIAS + 0x70] %asi, %i6; \
ldxa [%sp + STACK_BIAS + 0x78] %asi, %i7; \
ba,pt %xcc, user_rtt_pre_restore; \
restored; \
nop; nop; nop; nop; nop; nop; \
nop; nop; nop; nop; nop; \
ba,a,pt %xcc, user_rtt_fill_fixup; \
ba,a,pt %xcc, user_rtt_fill_fixup; \
ba,a,pt %xcc, user_rtt_fill_fixup;
/* Normal 32bit fill */
#define FILL_2_GENERIC(ASI) \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
and %sp, 1, %g3; \
brnz,pn %g3, (. - (128 + 4)); \
srl %sp, 0, %sp; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
lduwa [%sp + %g0] ASI, %l0; \
mov 0x04, %g2; \
mov 0x08, %g3; \
lduwa [%sp + %g2] ASI, %l1; \
mov 0x0c, %g5; \
lduwa [%sp + %g3] ASI, %l2; \
lduwa [%sp + %g5] ASI, %l3; \
add %sp, 0x10, %g1; \
lduwa [%g1 + %g0] ASI, %l4; \
lduwa [%g1 + %g2] ASI, %l5; \
lduwa [%g1 + %g3] ASI, %l6; \
lduwa [%g1 + %g5] ASI, %l7; \
add %g1, 0x10, %g1; \
lduwa [%g1 + %g0] ASI, %i0; \
lduwa [%g1 + %g2] ASI, %i1; \
lduwa [%g1 + %g3] ASI, %i2; \
lduwa [%g1 + %g5] ASI, %i3; \
add %g1, 0x10, %g1; \
lduwa [%g1 + %g0] ASI, %i4; \
lduwa [%g1 + %g2] ASI, %i5; \
lduwa [%g1 + %g3] ASI, %i6; \
lduwa [%g1 + %g5] ASI, %i7; \
restored; \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
retry; nop; nop; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
b,a,pt %xcc, fill_fixup_dax; \
b,a,pt %xcc, fill_fixup_mna; \
b,a,pt %xcc, fill_fixup;
#define FILL_2_GENERIC_RTRAP \
user_rtt_fill_32bit: \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
and %sp, 1, %g3; \
brnz,pn %g3, user_rtt_fill_64bit; \
srl %sp, 0, %sp; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
lduwa [%sp + 0x00] %asi, %l0; \
lduwa [%sp + 0x04] %asi, %l1; \
lduwa [%sp + 0x08] %asi, %l2; \
lduwa [%sp + 0x0c] %asi, %l3; \
lduwa [%sp + 0x10] %asi, %l4; \
lduwa [%sp + 0x14] %asi, %l5; \
lduwa [%sp + 0x18] %asi, %l6; \
lduwa [%sp + 0x1c] %asi, %l7; \
lduwa [%sp + 0x20] %asi, %i0; \
lduwa [%sp + 0x24] %asi, %i1; \
lduwa [%sp + 0x28] %asi, %i2; \
lduwa [%sp + 0x2c] %asi, %i3; \
lduwa [%sp + 0x30] %asi, %i4; \
lduwa [%sp + 0x34] %asi, %i5; \
lduwa [%sp + 0x38] %asi, %i6; \
lduwa [%sp + 0x3c] %asi, %i7; \
ba,pt %xcc, user_rtt_pre_restore; \
restored; \
nop; nop; nop; nop; nop; \
sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller <davem@davemloft.net>
2012-10-27 05:18:37 +07:00
nop; nop; nop; \
sparc: copy sparc64 specific files to asm-sparc Used the following script to copy the files: cd include set -e SPARC64=`ls asm-sparc64` for FILE in ${SPARC64}; do if [ -f asm-sparc/$FILE ]; then echo $FILE exist in asm-sparc else git mv asm-sparc64/$FILE asm-sparc/$FILE printf "#include <asm-sparc/$FILE>\n" > asm-sparc64/$FILE git add asm-sparc64/$FILE fi done Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
2008-06-20 01:26:19 +07:00
ba,a,pt %xcc, user_rtt_fill_fixup; \
ba,a,pt %xcc, user_rtt_fill_fixup; \
ba,a,pt %xcc, user_rtt_fill_fixup;
#define FILL_1_NORMAL FILL_1_GENERIC(ASI_AIUP)
#define FILL_2_NORMAL FILL_2_GENERIC(ASI_AIUP)
#define FILL_3_NORMAL FILL_0_NORMAL
#define FILL_4_NORMAL FILL_0_NORMAL
#define FILL_5_NORMAL FILL_0_NORMAL
#define FILL_6_NORMAL FILL_0_NORMAL
#define FILL_7_NORMAL FILL_0_NORMAL
#define FILL_0_OTHER FILL_0_NORMAL
#define FILL_1_OTHER FILL_1_GENERIC(ASI_AIUS)
#define FILL_2_OTHER FILL_2_GENERIC(ASI_AIUS)
#define FILL_3_OTHER FILL_3_NORMAL
#define FILL_4_OTHER FILL_4_NORMAL
#define FILL_5_OTHER FILL_5_NORMAL
#define FILL_6_OTHER FILL_6_NORMAL
#define FILL_7_OTHER FILL_7_NORMAL
#endif /* !(_SPARC64_TTABLE_H) */
Reference in New Issue Copy Permalink