mirrors/cosmopolitan
1
0
Fork 0
mirror of https://github.com/jart/cosmopolitan.git synced 2024-05-18 03:22:40 +00:00
Code Issues Projects Releases Wiki Activity
cosmopolitan/libc/nexgen32e/sha1ni.S
Justine Tunney 957c61cbbf
Release Cosmopolitan v3.3 
This change upgrades to GCC 12.3 and GNU binutils 2.42. The GNU linker
appears to have changed things so that only a single de-duplicated str
table is present in the binary, and it gets placed wherever the linker
wants, regardless of what the linker script says. To cope with that we
need to stop using .ident to embed licenses. As such, this change does
significant work to revamp how third party licenses are defined in the
codebase, using `.section .notice,"aR",@progbits`.

This new GCC 12.3 toolchain has support for GNU indirect functions. It
lets us support __target_clones__ for the first time. This is used for
optimizing the performance of libc string functions such as strlen and
friends so far on x86, by ensuring AVX systems favor a second codepath
that uses VEX encoding. It shaves some latency off certain operations.
It's a useful feature to have for scientific computing for the reasons
explained by the test/libcxx/openmp_test.cc example which compiles for
fifteen different microarchitectures. Thanks to the upgrades, it's now
also possible to use newer instruction sets, such as AVX512FP16, VNNI.

Cosmo now uses the %gs register on x86 by default for TLS. Doing it is
helpful for any program that links `cosmo_dlopen()`. Such programs had
to recompile their binaries at startup to change the TLS instructions.
That's not great, since it means every page in the executable needs to
be faulted. The work of rewriting TLS-related x86 opcodes, is moved to
fixupobj.com instead. This is great news for MacOS x86 users, since we
previously needed to morph the binary every time for that platform but
now that's no longer necessary. The only platforms where we need fixup
of TLS x86 opcodes at runtime are now Windows, OpenBSD, and NetBSD. On
Windows we morph TLS to point deeper into the TIB, based on a TlsAlloc
assignment, and on OpenBSD/NetBSD we morph %gs back into %fs since the
kernels do not allow us to specify a value for the %gs register.

OpenBSD users are now required to use APE Loader to run Cosmo binaries
and assimilation is no longer possible. OpenBSD kernel needs to change
to allow programs to specify a value for the %gs register, or it needs
to stop marking executable pages loaded by the kernel as mimmutable().

This release fixes __constructor__, .ctor, .init_array, and lastly the
.preinit_array so they behave the exact same way as glibc.

We no longer use hex constants to define math.h symbols like M_PI.
2024-02-20 13:27:59 -08:00

289 lines
8.2 KiB
ArmAsm
Raw Permalink Blame History

/*-*- mode:unix-assembly; indent-tabs-mode:t; tab-width:8; coding:utf-8 -*-│
│ vi: set noet ft=asm ts=8 sw=8 fenc=utf-8 :vi │
╞══════════════════════════════════════════════════════════════════════════════╡
│ │
│ Copyright 2015 Intel Corporation │
│ │
│ Redistribution and use in source and binary forms, with or without │
│ modification, are permitted provided that the following conditions │
│ are met: │
│ │
│ * Redistributions of source code must retain the above copyright │
│ notice, this list of conditions and the following disclaimer. │
│ * Redistributions in binary form must reproduce the above copyright │
│ notice, this list of conditions and the following disclaimer in │
│ the documentation and/or other materials provided with the │
│ distribution. │
│ * Neither the name of Intel Corporation nor the names of its │
│ contributors may be used to endorse or promote products derived │
│ from this software without specific prior written permission. │
│ │
│ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS │
│ "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT │
│ LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR │
│ A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT │
│ OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, │
│ SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT │
│ LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, │
│ DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY │
│ THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT │
│ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE │
│ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. │
│ │
╚─────────────────────────────────────────────────────────────────────────────*/
#include "libc/macros.internal.h"
.section .notice,"aR",@progbits
.asciz "\n\n\
Intel SHA-NI (BSD-3 License)\n\
Copyright 2015 Intel Corporation\n\
Sean Gulley <sean.m.gulley@intel.com>\n\
Tim Chen <tim.c.chen@linux.intel.com>"
.text
.balign 32
#define FRAME_SIZE 32
#define DIGEST_PTR %rdi
#define DATA_PTR %rsi
#define NUM_BLKS %rdx
#define ABCD %xmm0
#define E0 %xmm1 /* Need two E's b/c they ping pong */
#define E1 %xmm2
#define MSG0 %xmm3
#define MSG1 %xmm4
#define MSG2 %xmm5
#define MSG3 %xmm6
#define SHUF_MASK %xmm7
// Performs Intel® SHA-NI™ optimized SHA-1 update.
//
// The function takes a pointer to the current hash values, a
// pointer to the input data, and a number of 64 byte blocks to
// process. Once all blocks have been processed, the digest pointer
// is updated with the resulting hash value. The function only
// processes complete blocks, there is no functionality to store
// partial blocks. All message padding and hash value
// initialization must be done outside the update function.
//
// The indented lines in the loop are instructions related to
// rounds processing. The non-indented lines are instructions
// related to the message schedule.
//
// void sha1_transform_ni(uint32_t digest[static 5],
// const void *data,
// uint32_t numBlocks);
//
// @param %rdi points to output digest
// @param %rsi points to input data
// @param %rdx is number of 64-byte blocks to process
// @see X86_HAVE(SHA)
.ftrace1
sha1_transform_ni:
.ftrace2
push %rbp
mov %rsp,%rbp
sub $FRAME_SIZE,%rsp
shl $6,NUM_BLKS # convert to bytes
jz .Ldone_hash
add DATA_PTR,NUM_BLKS # pointer to end of data
// load initial hash values
movdqa UPPER_WORD_MASK(%rip),E1
pinsrd $3,1*16(DIGEST_PTR),E0
movdqu 0*16(DIGEST_PTR),ABCD
pand E1,E0
pshufd $0x1B,ABCD,ABCD
movdqa PSHUFFLE_BYTE_FLIP_MASK(%rip),SHUF_MASK
.Lloop0:
// Save hash values for addition after rounds
movdqa E0,(0*16)(%rsp)
movdqa ABCD,(1*16)(%rsp)
// Rounds 0-3
movdqu 0*16(DATA_PTR),MSG0
pshufb SHUF_MASK,MSG0
paddd MSG0,E0
movdqa ABCD,E1
sha1rnds4 $0,E0,ABCD
// Rounds 4-7
movdqu 1*16(DATA_PTR),MSG1
pshufb SHUF_MASK,MSG1
sha1nexte MSG1,E1
movdqa ABCD,E0
sha1rnds4 $0,E1,ABCD
sha1msg1 MSG1,MSG0
// Rounds 8-11
movdqu 2*16(DATA_PTR),MSG2
pshufb SHUF_MASK,MSG2
sha1nexte MSG2,E0
movdqa ABCD,E1
sha1rnds4 $0,E0,ABCD
sha1msg1 MSG2,MSG1
pxor MSG2,MSG0
// Rounds 12-15
movdqu 3*16(DATA_PTR),MSG3
pshufb SHUF_MASK,MSG3
sha1nexte MSG3,E1
movdqa ABCD,E0
sha1msg2 MSG3,MSG0
sha1rnds4 $0,E1,ABCD
sha1msg1 MSG3,MSG2
pxor MSG3,MSG1
// Rounds 16-19
sha1nexte MSG0,E0
movdqa ABCD,E1
sha1msg2 MSG0,MSG1
sha1rnds4 $0,E0,ABCD
sha1msg1 MSG0,MSG3
pxor MSG0,MSG2
// Rounds 20-23
sha1nexte MSG1,E1
movdqa ABCD,E0
sha1msg2 MSG1,MSG2
sha1rnds4 $1,E1,ABCD
sha1msg1 MSG1,MSG0
pxor MSG1,MSG3
// Rounds 24-27
sha1nexte MSG2,E0
movdqa ABCD,E1
sha1msg2 MSG2,MSG3
sha1rnds4 $1,E0,ABCD
sha1msg1 MSG2,MSG1
pxor MSG2,MSG0
// Rounds 28-31
sha1nexte MSG3,E1
movdqa ABCD,E0
sha1msg2 MSG3,MSG0
sha1rnds4 $1,E1,ABCD
sha1msg1 MSG3,MSG2
pxor MSG3,MSG1
// Rounds 32-35
sha1nexte MSG0,E0
movdqa ABCD,E1
sha1msg2 MSG0,MSG1
sha1rnds4 $1,E0,ABCD
sha1msg1 MSG0,MSG3
pxor MSG0,MSG2
// Rounds 36-39
sha1nexte MSG1,E1
movdqa ABCD,E0
sha1msg2 MSG1,MSG2
sha1rnds4 $1,E1,ABCD
sha1msg1 MSG1,MSG0
pxor MSG1,MSG3
// Rounds 40-43
sha1nexte MSG2,E0
movdqa ABCD,E1
sha1msg2 MSG2,MSG3
sha1rnds4 $2,E0,ABCD
sha1msg1 MSG2,MSG1
pxor MSG2,MSG0
// Rounds 44-47
sha1nexte MSG3,E1
movdqa ABCD,E0
sha1msg2 MSG3,MSG0
sha1rnds4 $2,E1,ABCD
sha1msg1 MSG3,MSG2
pxor MSG3,MSG1
// Rounds 48-51
sha1nexte MSG0,E0
movdqa ABCD,E1
sha1msg2 MSG0,MSG1
sha1rnds4 $2,E0,ABCD
sha1msg1 MSG0,MSG3
pxor MSG0,MSG2
// Rounds 52-55
sha1nexte MSG1,E1
movdqa ABCD,E0
sha1msg2 MSG1,MSG2
sha1rnds4 $2,E1,ABCD
sha1msg1 MSG1,MSG0
pxor MSG1,MSG3
// Rounds 56-59
sha1nexte MSG2,E0
movdqa ABCD,E1
sha1msg2 MSG2,MSG3
sha1rnds4 $2,E0,ABCD
sha1msg1 MSG2,MSG1
pxor MSG2,MSG0
// Rounds 60-63
sha1nexte MSG3,E1
movdqa ABCD,E0
sha1msg2 MSG3,MSG0
sha1rnds4 $3,E1,ABCD
sha1msg1 MSG3,MSG2
pxor MSG3,MSG1
// Rounds 64-67
sha1nexte MSG0,E0
movdqa ABCD,E1
sha1msg2 MSG0,MSG1
sha1rnds4 $3,E0,ABCD
sha1msg1 MSG0,MSG3
pxor MSG0,MSG2
// Rounds 68-71
sha1nexte MSG1,E1
movdqa ABCD,E0
sha1msg2 MSG1,MSG2
sha1rnds4 $3,E1,ABCD
pxor MSG1,MSG3
// Rounds 72-75
sha1nexte MSG2,E0
movdqa ABCD,E1
sha1msg2 MSG2,MSG3
sha1rnds4 $3,E0,ABCD
// Rounds 76-79
sha1nexte MSG3,E1
movdqa ABCD,E0
sha1rnds4 $3,E1,ABCD
// Add current hash values with previously saved
sha1nexte (0*16)(%rsp),E0
paddd (1*16)(%rsp),ABCD
// Increment data pointer and loop if more to process
add $64,DATA_PTR
cmp NUM_BLKS,DATA_PTR
jne .Lloop0
// Write hash values back in the correct order
pshufd $0x1B,ABCD,ABCD
movdqu ABCD,0*16(DIGEST_PTR)
pextrd $3,E0,1*16(DIGEST_PTR)
.Ldone_hash:
leave
ret
.endfn sha1_transform_ni,globl
.section .rodata.cst16.PSHUFFLE_BYTE_FLIP_MASK, "aM", @progbits, 16
.balign 16
PSHUFFLE_BYTE_FLIP_MASK:
.octa 0x000102030405060708090a0b0c0d0e0f
.section .rodata.cst16.UPPER_WORD_MASK, "aM", @progbits, 16
.balign 16
UPPER_WORD_MASK:
.octa 0xFFFFFFFF000000000000000000000000
Reference in a new issue View git blame Copy permalink