mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-25 10:49:25 +07:00
517ffce4e1
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>
159 lines
3.8 KiB
ArmAsm
159 lines
3.8 KiB
ArmAsm
/* winfixup.S: Handle cases where user stack pointer is found to be bogus.
|
|
*
|
|
* Copyright (C) 1997, 2006 David S. Miller (davem@davemloft.net)
|
|
*/
|
|
|
|
#include <asm/asi.h>
|
|
#include <asm/head.h>
|
|
#include <asm/page.h>
|
|
#include <asm/ptrace.h>
|
|
#include <asm/processor.h>
|
|
#include <asm/spitfire.h>
|
|
#include <asm/thread_info.h>
|
|
|
|
.text
|
|
|
|
/* It used to be the case that these register window fault
|
|
* handlers could run via the save and restore instructions
|
|
* done by the trap entry and exit code. They now do the
|
|
* window spill/fill by hand, so that case no longer can occur.
|
|
*/
|
|
|
|
.align 32
|
|
fill_fixup:
|
|
TRAP_LOAD_THREAD_REG(%g6, %g1)
|
|
rdpr %tstate, %g1
|
|
and %g1, TSTATE_CWP, %g1
|
|
or %g4, FAULT_CODE_WINFIXUP, %g4
|
|
stb %g4, [%g6 + TI_FAULT_CODE]
|
|
stx %g5, [%g6 + TI_FAULT_ADDR]
|
|
wrpr %g1, %cwp
|
|
ba,pt %xcc, etrap
|
|
rd %pc, %g7
|
|
call do_sparc64_fault
|
|
add %sp, PTREGS_OFF, %o0
|
|
ba,pt %xcc, rtrap
|
|
nop
|
|
|
|
/* Be very careful about usage of the trap globals here.
|
|
* You cannot touch %g5 as that has the fault information.
|
|
*/
|
|
spill_fixup:
|
|
spill_fixup_mna:
|
|
spill_fixup_dax:
|
|
TRAP_LOAD_THREAD_REG(%g6, %g1)
|
|
ldx [%g6 + TI_FLAGS], %g1
|
|
andcc %sp, 0x1, %g0
|
|
movne %icc, 0, %g1
|
|
andcc %g1, _TIF_32BIT, %g0
|
|
ldub [%g6 + TI_WSAVED], %g1
|
|
sll %g1, 3, %g3
|
|
add %g6, %g3, %g3
|
|
stx %sp, [%g3 + TI_RWIN_SPTRS]
|
|
sll %g1, 7, %g3
|
|
bne,pt %xcc, 1f
|
|
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]
|
|
ba,pt %xcc, 2f
|
|
stx %i7, [%g3 + TI_REG_WINDOW + 0x78]
|
|
1: 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]
|
|
2: add %g1, 1, %g1
|
|
stb %g1, [%g6 + TI_WSAVED]
|
|
rdpr %tstate, %g1
|
|
andcc %g1, TSTATE_PRIV, %g0
|
|
saved
|
|
be,pn %xcc, 1f
|
|
and %g1, TSTATE_CWP, %g1
|
|
retry
|
|
1: mov FAULT_CODE_WRITE | FAULT_CODE_DTLB | FAULT_CODE_WINFIXUP, %g4
|
|
stb %g4, [%g6 + TI_FAULT_CODE]
|
|
stx %g5, [%g6 + TI_FAULT_ADDR]
|
|
wrpr %g1, %cwp
|
|
ba,pt %xcc, etrap
|
|
rd %pc, %g7
|
|
call do_sparc64_fault
|
|
add %sp, PTREGS_OFF, %o0
|
|
ba,a,pt %xcc, rtrap
|
|
|
|
winfix_mna:
|
|
andn %g3, 0x7f, %g3
|
|
add %g3, 0x78, %g3
|
|
wrpr %g3, %tnpc
|
|
done
|
|
|
|
fill_fixup_mna:
|
|
rdpr %tstate, %g1
|
|
and %g1, TSTATE_CWP, %g1
|
|
wrpr %g1, %cwp
|
|
ba,pt %xcc, etrap
|
|
rd %pc, %g7
|
|
sethi %hi(tlb_type), %g1
|
|
lduw [%g1 + %lo(tlb_type)], %g1
|
|
cmp %g1, 3
|
|
bne,pt %icc, 1f
|
|
add %sp, PTREGS_OFF, %o0
|
|
mov %l4, %o2
|
|
call sun4v_do_mna
|
|
mov %l5, %o1
|
|
ba,a,pt %xcc, rtrap
|
|
1: mov %l4, %o1
|
|
mov %l5, %o2
|
|
call mem_address_unaligned
|
|
nop
|
|
ba,a,pt %xcc, rtrap
|
|
|
|
winfix_dax:
|
|
andn %g3, 0x7f, %g3
|
|
add %g3, 0x74, %g3
|
|
wrpr %g3, %tnpc
|
|
done
|
|
|
|
fill_fixup_dax:
|
|
rdpr %tstate, %g1
|
|
and %g1, TSTATE_CWP, %g1
|
|
wrpr %g1, %cwp
|
|
ba,pt %xcc, etrap
|
|
rd %pc, %g7
|
|
sethi %hi(tlb_type), %g1
|
|
mov %l4, %o1
|
|
lduw [%g1 + %lo(tlb_type)], %g1
|
|
mov %l5, %o2
|
|
cmp %g1, 3
|
|
bne,pt %icc, 1f
|
|
add %sp, PTREGS_OFF, %o0
|
|
call sun4v_data_access_exception
|
|
nop
|
|
ba,a,pt %xcc, rtrap
|
|
1: call spitfire_data_access_exception
|
|
nop
|
|
ba,a,pt %xcc, rtrap
|