From: "Jan Beulich" <JBeulich@suse.com>
To: xen-devel <xen-devel@lists.xenproject.org>
Cc: George Dunlap <George.Dunlap@eu.citrix.com>,
Andrew Cooper <andrew.cooper3@citrix.com>,
Wei Liu <wei.liu2@citrix.com>
Subject: [PATCH v3 05/34] x86emul: support basic AVX512 moves
Date: Tue, 18 Sep 2018 05:56:36 -0600 [thread overview]
Message-ID: <5BA0E7F402000078001E9537@prv1-mh.provo.novell.com> (raw)
In-Reply-To: <5BA0E58F02000078001E94CE@prv1-mh.provo.novell.com>
Note: SDM Vol 2 rev 067 is not really consistent about EVEX.L'L for LIG
insns - the only place where this is made explicit is a table in
the section titled "Vector Length Orthogonality": While they
tolerate 0, 1, and 2, a value of 3 uniformly leads to #UD.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
---
v3: Restrict k-reg reading to insns with memory operand. Shrink scope of
"disp8scale".
v2: Move "full" into more narrow scope.
--- a/tools/tests/x86_emulator/test_x86_emulator.c
+++ b/tools/tests/x86_emulator/test_x86_emulator.c
@@ -1985,6 +1985,53 @@ int main(int argc, char **argv)
else
printf("skipped\n");
+ printf("%-40s", "Testing {evex} vmovq %xmm1,32(%edx)...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovq_to_mem);
+
+ asm volatile ( "pcmpgtb %%xmm1, %%xmm1\n"
+ put_insn(evex_vmovq_to_mem, "%{evex%} vmovq %%xmm1, 32(%0)")
+ :: "d" (NULL) );
+
+ memset(res, 0xdb, 64);
+ set_insn(evex_vmovq_to_mem);
+ regs.ecx = 0;
+ regs.edx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(evex_vmovq_to_mem) ||
+ *((uint64_t *)res + 4) ||
+ memcmp(res, res + 10, 24) ||
+ memcmp(res, res + 6, 8) )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
+ printf("%-40s", "Testing {evex} vmovq 32(%edx),%xmm0...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovq_from_mem);
+
+ asm volatile ( "pcmpeqb %%xmm0, %%xmm0\n"
+ put_insn(evex_vmovq_from_mem, "%{evex%} vmovq 32(%0), %%xmm0")
+ :: "d" (NULL) );
+
+ set_insn(evex_vmovq_from_mem);
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(evex_vmovq_from_mem) )
+ goto fail;
+ asm ( "vmovq %1, %%xmm1\n\t"
+ "vpcmpeqq %%zmm0, %%zmm1, %%k0\n"
+ "kmovw %%k0, %0" : "=r" (rc) : "m" (res[8]) );
+ if ( rc != 0xff )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
printf("%-40s", "Testing movdqu %xmm2,(%ecx)...");
if ( stack_exec && cpu_has_sse2 )
{
@@ -2085,6 +2132,118 @@ int main(int argc, char **argv)
else
printf("skipped\n");
+ printf("%-40s", "Testing vmovdqu32 %zmm2,(%ecx){%k1}...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(vmovdqu32_to_mem);
+
+ memset(res, 0x55, 128);
+
+ asm volatile ( "vpcmpeqd %%ymm2, %%ymm2, %%ymm2\n\t"
+ "kmovw %1,%%k1\n"
+ put_insn(vmovdqu32_to_mem,
+ "vmovdqu32 %%zmm2, (%0)%{%%k1%}")
+ :: "c" (NULL), "rm" (res[0]) );
+ set_insn(vmovdqu32_to_mem);
+
+ regs.ecx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( (rc != X86EMUL_OKAY) || memcmp(res + 16, res + 24, 32) ||
+ !check_eip(vmovdqu32_to_mem) )
+ goto fail;
+
+ res[16] = ~0; res[18] = ~0; res[20] = ~0; res[22] = ~0;
+ res[24] = 0; res[26] = 0; res[28] = 0; res[30] = 0;
+ if ( memcmp(res, res + 16, 64) )
+ goto fail;
+
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
+ printf("%-40s", "Testing vmovdqu32 64(%edx),%zmm2{%k2}...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(vmovdqu32_from_mem);
+
+ asm volatile ( "knotw %%k1, %%k2\n"
+ put_insn(vmovdqu32_from_mem,
+ "vmovdqu32 64(%0), %%zmm2%{%%k2%}")
+ :: "d" (NULL) );
+
+ set_insn(vmovdqu32_from_mem);
+ regs.ecx = 0;
+ regs.edx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(vmovdqu32_from_mem) )
+ goto fail;
+ asm ( "vpcmpeqd %1, %%zmm2, %%k0\n\t"
+ "kmovw %%k0, %0" : "=r" (rc) : "m" (res[0]) );
+ if ( rc != 0xffff )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
+ printf("%-40s", "Testing vmovdqu16 %zmm3,(%ecx){%k1}...");
+ if ( stack_exec && cpu_has_avx512bw )
+ {
+ decl_insn(vmovdqu16_to_mem);
+
+ memset(res, 0x55, 128);
+
+ asm volatile ( "vpcmpeqw %%ymm3, %%ymm3, %%ymm3\n\t"
+ "kmovd %1,%%k1\n"
+ put_insn(vmovdqu16_to_mem,
+ "vmovdqu16 %%zmm3, (%0)%{%%k1%}")
+ :: "c" (NULL), "rm" (res[0]) );
+ set_insn(vmovdqu16_to_mem);
+
+ regs.ecx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( (rc != X86EMUL_OKAY) || memcmp(res + 16, res + 24, 32) ||
+ !check_eip(vmovdqu16_to_mem) )
+ goto fail;
+
+ for ( i = 16; i < 24; ++i )
+ res[i] |= 0x0000ffff;
+ for ( ; i < 32; ++i )
+ res[i] &= 0xffff0000;
+ if ( memcmp(res, res + 16, 64) )
+ goto fail;
+
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
+ printf("%-40s", "Testing vmovdqu16 64(%edx),%zmm3{%k2}...");
+ if ( stack_exec && cpu_has_avx512bw )
+ {
+ decl_insn(vmovdqu16_from_mem);
+
+ asm volatile ( "knotd %%k1, %%k2\n"
+ put_insn(vmovdqu16_from_mem,
+ "vmovdqu16 64(%0), %%zmm3%{%%k2%}")
+ :: "d" (NULL) );
+
+ set_insn(vmovdqu16_from_mem);
+ regs.ecx = 0;
+ regs.edx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(vmovdqu16_from_mem) )
+ goto fail;
+ asm ( "vpcmpeqw %1, %%zmm3, %%k0\n\t"
+ "kmovd %%k0, %0" : "=r" (rc) : "m" (res[0]) );
+ if ( rc != 0xffffffff )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
printf("%-40s", "Testing movsd %xmm5,(%ecx)...");
memset(res, 0x77, 64);
memset(res + 10, 0x66, 8);
@@ -2186,6 +2345,71 @@ int main(int argc, char **argv)
else
printf("skipped\n");
+ printf("%-40s", "Testing vmovsd %xmm5,16(%ecx){%k3}...");
+ memset(res, 0x88, 128);
+ memset(res + 20, 0x77, 8);
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(vmovsd_masked_to_mem);
+
+ asm volatile ( "vbroadcastsd %0, %%ymm5\n\t"
+ "kxorw %%k3, %%k3, %%k3\n"
+ put_insn(vmovsd_masked_to_mem,
+ "vmovsd %%xmm5, 16(%1)%{%%k3%}")
+ :: "m" (res[20]), "c" (NULL) );
+
+ set_insn(vmovsd_masked_to_mem);
+ regs.ecx = 0;
+ regs.edx = 0;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( (rc != X86EMUL_OKAY) || !check_eip(vmovsd_masked_to_mem) )
+ goto fail;
+
+ asm volatile ( "kmovw %0, %%k3\n" :: "m" (res[20]) );
+
+ set_insn(vmovsd_masked_to_mem);
+ regs.ecx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( (rc != X86EMUL_OKAY) || !check_eip(vmovsd_masked_to_mem) ||
+ memcmp(res, res + 16, 64) )
+ goto fail;
+
+ printf("okay\n");
+ }
+ else
+ {
+ printf("skipped\n");
+ memset(res + 4, 0x77, 8);
+ }
+
+ printf("%-40s", "Testing vmovaps (%edx),%zmm7{%k3}{z}...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(vmovaps_masked_from_mem);
+
+ asm volatile ( "vpcmpeqd %%xmm7, %%xmm7, %%xmm7\n\t"
+ "vbroadcastss %%xmm7, %%zmm7\n"
+ put_insn(vmovaps_masked_from_mem,
+ "vmovaps (%0), %%zmm7%{%%k3%}%{z%}")
+ :: "d" (NULL) );
+
+ set_insn(vmovaps_masked_from_mem);
+ regs.edx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(vmovaps_masked_from_mem) )
+ goto fail;
+ asm ( "vcmpeqps %1, %%zmm7, %%k0\n\t"
+ "vxorps %%xmm0, %%xmm0, %%xmm0\n\t"
+ "vcmpeqps %%zmm0, %%zmm7, %%k1\n\t"
+ "kxorw %%k1, %%k0, %%k0\n\t"
+ "kmovw %%k0, %0" : "=r" (rc) : "m" (res[16]) );
+ if ( rc != 0xffff )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
printf("%-40s", "Testing movd %mm3,32(%ecx)...");
if ( stack_exec && cpu_has_mmx )
{
@@ -2341,6 +2565,55 @@ int main(int argc, char **argv)
else
printf("skipped\n");
+ printf("%-40s", "Testing {evex} vmovd %xmm3,32(%ecx)...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovd_to_mem);
+
+ asm volatile ( "pcmpeqb %%xmm3, %%xmm3\n"
+ put_insn(evex_vmovd_to_mem,
+ "%{evex%} vmovd %%xmm3, 32(%0)")
+ :: "c" (NULL) );
+
+ memset(res, 0xbd, 64);
+ set_insn(evex_vmovd_to_mem);
+ regs.ecx = (unsigned long)res;
+ regs.edx = 0;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(evex_vmovd_to_mem) ||
+ res[8] + 1 ||
+ memcmp(res, res + 9, 28) ||
+ memcmp(res, res + 6, 8) )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
+ printf("%-40s", "Testing {evex} vmovd 32(%ecx),%xmm4...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovd_from_mem);
+
+ asm volatile ( "pcmpeqb %%xmm4, %%xmm4\n"
+ put_insn(evex_vmovd_from_mem,
+ "%{evex%} vmovd 32(%0), %%xmm4")
+ :: "c" (NULL) );
+
+ set_insn(evex_vmovd_from_mem);
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(evex_vmovd_from_mem) )
+ goto fail;
+ asm ( "vmovd %1, %%xmm0\n\t"
+ "vpcmpeqd %%zmm4, %%zmm0, %%k0\n\t"
+ "kmovw %%k0, %0" : "=r" (rc) : "m" (res[8]) );
+ if ( rc != 0xffff )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
printf("%-40s", "Testing movd %mm3,%ebx...");
if ( stack_exec && cpu_has_mmx )
{
@@ -2507,6 +2780,57 @@ int main(int argc, char **argv)
else
printf("skipped\n");
+ printf("%-40s", "Testing {evex} vmovd %xmm2,%ebx...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovd_to_reg);
+
+ /* See comment next to movd above. */
+ asm volatile ( "pcmpeqb %%xmm2, %%xmm2\n"
+ put_insn(evex_vmovd_to_reg,
+ "%{evex%} vmovd %%xmm2, %%ebx")
+ :: );
+
+ set_insn(evex_vmovd_to_reg);
+#ifdef __x86_64__
+ regs.rbx = 0xbdbdbdbdbdbdbdbdUL;
+#else
+ regs.ebx = 0xbdbdbdbdUL;
+#endif
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( (rc != X86EMUL_OKAY) || !check_eip(evex_vmovd_to_reg) ||
+ regs.ebx != 0xffffffff )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
+ printf("%-40s", "Testing {evex} vmovd %ebx,%xmm1...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovd_from_reg);
+
+ /* See comment next to movd above. */
+ asm volatile ( "pcmpgtb %%xmm1, %%xmm1\n"
+ put_insn(evex_vmovd_from_reg,
+ "%{evex%} vmovd %%ebx, %%xmm1")
+ :: );
+
+ set_insn(evex_vmovd_from_reg);
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( (rc != X86EMUL_OKAY) || !check_eip(evex_vmovd_from_reg) )
+ goto fail;
+ asm ( "vmovd %1, %%xmm0\n\t"
+ "vpcmpeqd %%zmm1, %%zmm0, %%k0\n\t"
+ "kmovw %%k0, %0" : "=r" (rc) : "m" (res[8]) );
+ if ( rc != 0xffff )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
#ifdef __x86_64__
printf("%-40s", "Testing movq %mm3,32(%ecx)...");
if ( stack_exec && cpu_has_mmx )
@@ -2584,6 +2908,36 @@ int main(int argc, char **argv)
else
printf("skipped\n");
+ printf("%-40s", "Testing {evex} vmovq %xmm11,32(%ecx)...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovq_to_mem2);
+
+ asm volatile ( "pcmpeqb %%xmm11, %%xmm11\n"
+#if 0 /* This may not work, as the assembler might pick opcode D6. */
+ put_insn(evex_vmovq_to_mem2,
+ "{evex} vmovq %%xmm11, 32(%0)")
+#else
+ put_insn(evex_vmovq_to_mem2,
+ ".byte 0x62, 0xf1, 0xfd, 0x08, 0x7e, 0x49, 0x04")
+#endif
+ :: "c" (NULL) );
+
+ memset(res, 0xbd, 64);
+ set_insn(evex_vmovq_to_mem2);
+ regs.ecx = (unsigned long)res;
+ regs.edx = 0;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(evex_vmovq_to_mem2) ||
+ *((long *)res + 4) + 1 ||
+ memcmp(res, res + 10, 24) ||
+ memcmp(res, res + 6, 8) )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
+
printf("%-40s", "Testing movq %mm3,%rbx...");
if ( stack_exec && cpu_has_mmx )
{
@@ -2643,6 +2997,28 @@ int main(int argc, char **argv)
}
else
printf("skipped\n");
+
+ printf("%-40s", "Testing vmovq %xmm22,%rbx...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovq_to_reg);
+
+ /* See comment next to movd above. */
+ asm volatile ( "pcmpeqq %%xmm2, %%xmm2\n\t"
+ "vmovq %%xmm2, %%xmm22\n"
+ put_insn(evex_vmovq_to_reg, "vmovq %%xmm22, %%rbx")
+ :: );
+
+ set_insn(evex_vmovq_to_reg);
+ regs.rbx = 0xbdbdbdbdbdbdbdbdUL;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(evex_vmovq_to_reg) ||
+ regs.rbx + 1 )
+ goto fail;
+ printf("okay\n");
+ }
+ else
+ printf("skipped\n");
#endif
printf("%-40s", "Testing maskmovq %mm4,%mm4...");
@@ -2812,6 +3188,32 @@ int main(int argc, char **argv)
goto fail;
printf("okay\n");
}
+ else
+ printf("skipped\n");
+
+ printf("%-40s", "Testing vmovntdqa 64(%ecx),%zmm4...");
+ if ( stack_exec && cpu_has_avx512f )
+ {
+ decl_insn(evex_vmovntdqa);
+
+ asm volatile ( "vpxor %%xmm4, %%xmm4, %%xmm4\n"
+ put_insn(evex_vmovntdqa, "vmovntdqa 64(%0), %%zmm4")
+ :: "c" (NULL) );
+
+ set_insn(evex_vmovntdqa);
+ memset(res, 0x55, 192);
+ memset(res + 16, 0xff, 64);
+ regs.ecx = (unsigned long)res;
+ rc = x86_emulate(&ctxt, &emulops);
+ if ( rc != X86EMUL_OKAY || !check_eip(evex_vmovntdqa) )
+ goto fail;
+ asm ( "vpbroadcastd %1, %%zmm2\n\t"
+ "vpcmpeqd %%zmm4, %%zmm2, %%k0\n\t"
+ "kmovw %%k0, %0" : "=r" (rc) : "0" (~0) );
+ if ( rc != 0xffff )
+ goto fail;
+ printf("okay\n");
+ }
else
printf("skipped\n");
--- a/tools/tests/x86_emulator/x86-emulate.c
+++ b/tools/tests/x86_emulator/x86-emulate.c
@@ -210,6 +210,7 @@ int emul_test_get_fpu(
if ( cpu_has_avx )
break;
case X86EMUL_FPU_opmask:
+ case X86EMUL_FPU_zmm:
if ( cpu_has_avx512f )
break;
default:
--- a/tools/tests/x86_emulator/x86-emulate.h
+++ b/tools/tests/x86_emulator/x86-emulate.h
@@ -266,6 +266,16 @@ static inline uint64_t xgetbv(uint32_t x
(res.b & (1U << 30)) != 0; \
})
+#define cpu_has_avx512vl ({ \
+ struct cpuid_leaf res; \
+ emul_test_cpuid(1, 0, &res, NULL); \
+ if ( !(res.c & (1U << 27)) || ((xgetbv(0) & 0xe6) != 0xe6) ) \
+ res.b = 0; \
+ else \
+ emul_test_cpuid(7, 0, &res, NULL); \
+ (res.b & (1U << 31)) != 0; \
+})
+
int emul_test_cpuid(
uint32_t leaf,
uint32_t subleaf,
--- a/xen/arch/x86/x86_emulate/x86_emulate.c
+++ b/xen/arch/x86/x86_emulate/x86_emulate.c
@@ -243,9 +243,25 @@ enum simd_opsize {
};
typedef uint8_t simd_opsize_t;
+enum disp8scale {
+ /* Values 0 ... 4 are explicit sizes. */
+ d8s_bw = 5,
+ d8s_dq,
+ /*
+ * All further values must strictly be last and in the order
+ * given so that arithmetic on the values works.
+ */
+ d8s_vl,
+ d8s_vl_by_2,
+ d8s_vl_by_4,
+ d8s_vl_by_8,
+};
+typedef uint8_t disp8scale_t;
+
static const struct twobyte_table {
opcode_desc_t desc;
- simd_opsize_t size;
+ simd_opsize_t size:4;
+ disp8scale_t d8s:4;
} twobyte_table[256] = {
[0x00] = { ModRM },
[0x01] = { ImplicitOps|ModRM },
@@ -260,8 +276,8 @@ static const struct twobyte_table {
[0x0d] = { ImplicitOps|ModRM },
[0x0e] = { ImplicitOps },
[0x0f] = { ModRM|SrcImmByte },
- [0x10] = { DstImplicit|SrcMem|ModRM|Mov, simd_any_fp },
- [0x11] = { DstMem|SrcImplicit|ModRM|Mov, simd_any_fp },
+ [0x10] = { DstImplicit|SrcMem|ModRM|Mov, simd_any_fp, d8s_vl },
+ [0x11] = { DstMem|SrcImplicit|ModRM|Mov, simd_any_fp, d8s_vl },
[0x12] = { DstImplicit|SrcMem|ModRM|Mov, simd_other },
[0x13] = { DstMem|SrcImplicit|ModRM|Mov, simd_other },
[0x14 ... 0x15] = { DstImplicit|SrcMem|ModRM, simd_packed_fp },
@@ -270,10 +286,10 @@ static const struct twobyte_table {
[0x18 ... 0x1f] = { ImplicitOps|ModRM },
[0x20 ... 0x21] = { DstMem|SrcImplicit|ModRM },
[0x22 ... 0x23] = { DstImplicit|SrcMem|ModRM },
- [0x28] = { DstImplicit|SrcMem|ModRM|Mov, simd_packed_fp },
- [0x29] = { DstMem|SrcImplicit|ModRM|Mov, simd_packed_fp },
+ [0x28] = { DstImplicit|SrcMem|ModRM|Mov, simd_packed_fp, d8s_vl },
+ [0x29] = { DstMem|SrcImplicit|ModRM|Mov, simd_packed_fp, d8s_vl },
[0x2a] = { DstImplicit|SrcMem|ModRM|Mov, simd_other },
- [0x2b] = { DstMem|SrcImplicit|ModRM|Mov, simd_any_fp },
+ [0x2b] = { DstMem|SrcImplicit|ModRM|Mov, simd_any_fp, d8s_vl },
[0x2c ... 0x2d] = { DstImplicit|SrcMem|ModRM|Mov, simd_other },
[0x2e ... 0x2f] = { ImplicitOps|ModRM|TwoOp },
[0x30 ... 0x35] = { ImplicitOps },
@@ -292,8 +308,8 @@ static const struct twobyte_table {
[0x63 ... 0x67] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
[0x68 ... 0x6a] = { DstImplicit|SrcMem|ModRM, simd_other },
[0x6b ... 0x6d] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
- [0x6e] = { DstImplicit|SrcMem|ModRM|Mov },
- [0x6f] = { DstImplicit|SrcMem|ModRM|Mov, simd_packed_int },
+ [0x6e] = { DstImplicit|SrcMem|ModRM|Mov, simd_none, d8s_dq },
+ [0x6f] = { DstImplicit|SrcMem|ModRM|Mov, simd_packed_int, d8s_vl },
[0x70] = { SrcImmByte|ModRM|TwoOp, simd_other },
[0x71 ... 0x73] = { DstImplicit|SrcImmByte|ModRM },
[0x74 ... 0x76] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
@@ -301,8 +317,8 @@ static const struct twobyte_table {
[0x78] = { ImplicitOps|ModRM },
[0x79] = { DstReg|SrcMem|ModRM, simd_packed_int },
[0x7c ... 0x7d] = { DstImplicit|SrcMem|ModRM, simd_other },
- [0x7e] = { DstMem|SrcImplicit|ModRM|Mov },
- [0x7f] = { DstMem|SrcImplicit|ModRM|Mov, simd_packed_int },
+ [0x7e] = { DstMem|SrcImplicit|ModRM|Mov, simd_none, d8s_dq },
+ [0x7f] = { DstMem|SrcImplicit|ModRM|Mov, simd_packed_int, d8s_vl },
[0x80 ... 0x8f] = { DstImplicit|SrcImm },
[0x90 ... 0x9f] = { ByteOp|DstMem|SrcNone|ModRM|Mov },
[0xa0 ... 0xa1] = { ImplicitOps|Mov },
@@ -344,14 +360,14 @@ static const struct twobyte_table {
[0xd0] = { DstImplicit|SrcMem|ModRM, simd_other },
[0xd1 ... 0xd3] = { DstImplicit|SrcMem|ModRM, simd_other },
[0xd4 ... 0xd5] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
- [0xd6] = { DstMem|SrcImplicit|ModRM|Mov, simd_other },
+ [0xd6] = { DstMem|SrcImplicit|ModRM|Mov, simd_other, 3 },
[0xd7] = { DstReg|SrcImplicit|ModRM|Mov },
[0xd8 ... 0xdf] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
[0xe0] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
[0xe1 ... 0xe2] = { DstImplicit|SrcMem|ModRM, simd_other },
[0xe3 ... 0xe5] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
[0xe6] = { DstImplicit|SrcMem|ModRM|Mov, simd_other },
- [0xe7] = { DstMem|SrcImplicit|ModRM|Mov, simd_packed_int },
+ [0xe7] = { DstMem|SrcImplicit|ModRM|Mov, simd_packed_int, d8s_vl },
[0xe8 ... 0xef] = { DstImplicit|SrcMem|ModRM, simd_packed_int },
[0xf0] = { DstImplicit|SrcMem|ModRM|Mov, simd_other },
[0xf1 ... 0xf3] = { DstImplicit|SrcMem|ModRM, simd_other },
@@ -406,6 +422,7 @@ static const struct ext0f38_table {
uint8_t to_mem:1;
uint8_t two_op:1;
uint8_t vsib:1;
+ disp8scale_t d8s:4;
} ext0f38_table[256] = {
[0x00 ... 0x0b] = { .simd_size = simd_packed_int },
[0x0c ... 0x0f] = { .simd_size = simd_packed_fp },
@@ -418,7 +435,7 @@ static const struct ext0f38_table {
[0x1c ... 0x1e] = { .simd_size = simd_packed_int, .two_op = 1 },
[0x20 ... 0x25] = { .simd_size = simd_other, .two_op = 1 },
[0x28 ... 0x29] = { .simd_size = simd_packed_int },
- [0x2a] = { .simd_size = simd_packed_int, .two_op = 1 },
+ [0x2a] = { .simd_size = simd_packed_int, .two_op = 1, .d8s = d8s_vl },
[0x2b] = { .simd_size = simd_packed_int },
[0x2c ... 0x2d] = { .simd_size = simd_other },
[0x2e ... 0x2f] = { .simd_size = simd_other, .to_mem = 1 },
@@ -656,6 +673,22 @@ union evex {
};
};
+#define EVEX_PFX_BYTES 4
+#define init_evex(stub) ({ \
+ uint8_t *buf_ = get_stub(stub); \
+ buf_[0] = 0x62; \
+ buf_ + EVEX_PFX_BYTES; \
+})
+
+#define copy_EVEX(ptr, evex) ({ \
+ if ( !mode_64bit() ) \
+ (evex).reg |= 8; \
+ (ptr)[1 - EVEX_PFX_BYTES] = (evex).raw[0]; \
+ (ptr)[2 - EVEX_PFX_BYTES] = (evex).raw[1]; \
+ (ptr)[3 - EVEX_PFX_BYTES] = (evex).raw[2]; \
+ container_of((ptr) + 1 - EVEX_PFX_BYTES, typeof(evex), raw[0]); \
+})
+
#define rep_prefix() (vex.pfx >= vex_f3)
#define repe_prefix() (vex.pfx == vex_f3)
#define repne_prefix() (vex.pfx == vex_f2)
@@ -768,6 +801,7 @@ typedef union {
uint64_t mmx;
uint64_t __attribute__ ((aligned(16))) xmm[2];
uint64_t __attribute__ ((aligned(32))) ymm[4];
+ uint64_t __attribute__ ((aligned(64))) zmm[8];
} mmval_t;
/*
@@ -1183,6 +1217,11 @@ static int _get_fpu(
switch ( type )
{
+ case X86EMUL_FPU_zmm:
+ if ( !(xcr0 & X86_XCR0_ZMM) || !(xcr0 & X86_XCR0_HI_ZMM) ||
+ !(xcr0 & X86_XCR0_OPMASK) )
+ return X86EMUL_UNHANDLEABLE;
+ /* fall through */
case X86EMUL_FPU_ymm:
if ( !(xcr0 & X86_XCR0_SSE) || !(xcr0 & X86_XCR0_YMM) )
return X86EMUL_UNHANDLEABLE;
@@ -1777,6 +1816,7 @@ static bool vcpu_has(
#define vcpu_has_clwb() vcpu_has( 7, EBX, 24, ctxt, ops)
#define vcpu_has_sha() vcpu_has( 7, EBX, 29, ctxt, ops)
#define vcpu_has_avx512bw() vcpu_has( 7, EBX, 30, ctxt, ops)
+#define vcpu_has_avx512vl() vcpu_has( 7, EBX, 31, ctxt, ops)
#define vcpu_has_rdpid() vcpu_has( 7, ECX, 22, ctxt, ops)
#define vcpu_has_clzero() vcpu_has(0x80000008, EBX, 0, ctxt, ops)
@@ -2150,6 +2190,62 @@ static unsigned long *decode_vex_gpr(
return decode_gpr(regs, ~vex_reg & (mode_64bit() ? 0xf : 7));
}
+static unsigned int decode_disp8scale(enum disp8scale scale,
+ const struct x86_emulate_state *state)
+{
+ switch ( scale )
+ {
+ case d8s_bw:
+ return state->evex.w;
+
+ default:
+ if ( scale < d8s_vl )
+ return scale;
+ if ( state->evex.br )
+ {
+ case d8s_dq:
+ return 2 + state->evex.w;
+ }
+ break;
+ }
+
+ switch ( state->simd_size )
+ {
+ case simd_any_fp:
+ case simd_single_fp:
+ if ( !(state->evex.pfx & VEX_PREFIX_SCALAR_MASK) )
+ break;
+ /* fall through */
+ case simd_scalar_opc:
+ case simd_scalar_vexw:
+ return 2 + state->evex.w;
+
+ case simd_128:
+ /* These should have an explicit size specified. */
+ ASSERT_UNREACHABLE();
+ return 4;
+
+ default:
+ break;
+ }
+
+ return 4 + state->evex.lr - (scale - d8s_vl);
+}
+
+#define avx512_vlen_check(lig) do { \
+ switch ( evex.lr ) \
+ { \
+ default: \
+ generate_exception(EXC_UD); \
+ case 2: \
+ break; \
+ case 0: case 1: \
+ if (!(lig)) \
+ host_and_vcpu_must_have(avx512vl); \
+ break; \
+ } \
+} while ( false )
+
static bool is_aligned(enum x86_segment seg, unsigned long offs,
unsigned int size, struct x86_emulate_ctxt *ctxt,
const struct x86_emulate_ops *ops)
@@ -2399,6 +2495,7 @@ x86_decode_twobyte(
if ( vex.pfx == vex_f3 ) /* movq xmm/m64,xmm */
{
case X86EMUL_OPC_VEX_F3(0, 0x7e): /* vmovq xmm/m64,xmm */
+ case X86EMUL_OPC_EVEX_F3(0, 0x7e): /* vmovq xmm/m64,xmm */
state->desc = DstImplicit | SrcMem | TwoOp;
state->simd_size = simd_other;
/* Avoid the state->desc clobbering of TwoOp below. */
@@ -2834,6 +2931,8 @@ x86_decode(
if ( d & ModRM )
{
+ unsigned int disp8scale = 0;
+
d &= ~ModRM;
#undef ModRM /* Only its aliases are valid to use from here on. */
modrm_reg = ((rex_prefix & 4) << 1) | ((modrm & 0x38) >> 3);
@@ -2876,6 +2975,9 @@ x86_decode(
break;
case ext_0f:
+ if ( evex_encoded() )
+ disp8scale = decode_disp8scale(twobyte_table[b].d8s, state);
+
switch ( b )
{
case 0x20: /* mov cr,reg */
@@ -2900,6 +3002,8 @@ x86_decode(
if ( ext0f38_table[b].vsib )
d |= vSIB;
state->simd_size = ext0f38_table[b].simd_size;
+ if ( evex_encoded() )
+ disp8scale = decode_disp8scale(ext0f38_table[b].d8s, state);
break;
case ext_8f09:
@@ -2968,7 +3072,7 @@ x86_decode(
ea.mem.off = insn_fetch_type(int16_t);
break;
case 1:
- ea.mem.off += insn_fetch_type(int8_t);
+ ea.mem.off += insn_fetch_type(int8_t) << disp8scale;
break;
case 2:
ea.mem.off += insn_fetch_type(int16_t);
@@ -3027,7 +3131,7 @@ x86_decode(
pc_rel = mode_64bit();
break;
case 1:
- ea.mem.off += insn_fetch_type(int8_t);
+ ea.mem.off += insn_fetch_type(int8_t) << disp8scale;
break;
case 2:
ea.mem.off += insn_fetch_type(int32_t);
@@ -3228,10 +3332,11 @@ x86_emulate(
struct x86_emulate_state state;
int rc;
uint8_t b, d, *opc = NULL;
- unsigned int first_byte = 0, insn_bytes = 0;
+ unsigned int first_byte = 0, elem_bytes, insn_bytes = 0;
+ uint64_t op_mask = ~0ULL;
bool singlestep = (_regs.eflags & X86_EFLAGS_TF) &&
!is_branch_step(ctxt, ops);
- bool sfence = false;
+ bool sfence = false, fault_suppression = false;
struct operand src = { .reg = PTR_POISON };
struct operand dst = { .reg = PTR_POISON };
unsigned long cr4;
@@ -3272,6 +3377,7 @@ x86_emulate(
b = ctxt->opcode;
d = state.desc;
#define state (&state)
+ elem_bytes = 4 << evex.w;
generate_exception_if(state->not_64bit && mode_64bit(), EXC_UD);
@@ -3346,6 +3452,28 @@ x86_emulate(
break;
}
+ /* With a memory operand, fetch the mask register in use (if any). */
+ if ( ea.type == OP_MEM && evex.opmsk )
+ {
+ uint8_t *stb = get_stub(stub);
+
+ /* KMOV{W,Q} %k<n>, (%rax) */
+ stb[0] = 0xc4;
+ stb[1] = 0xe1;
+ stb[2] = cpu_has_avx512bw ? 0xf8 : 0x78;
+ stb[3] = 0x91;
+ stb[4] = evex.opmsk << 3;
+ insn_bytes = 5;
+ stb[5] = 0xc3;
+
+ invoke_stub("", "", "+m" (op_mask) : "a" (&op_mask));
+
+ insn_bytes = 0;
+ put_stub(stub);
+
+ fault_suppression = true;
+ }
+
/* Decode (but don't fetch) the destination operand: register or memory. */
switch ( d & DstMask )
{
@@ -5708,6 +5836,41 @@ x86_emulate(
insn_bytes = PFX_BYTES + 2;
break;
+ CASE_SIMD_PACKED_FP(_EVEX, 0x0f, 0x2b): /* vmovntp{s,d} [xyz]mm,mem */
+ generate_exception_if(ea.type != OP_MEM || evex.opmsk, EXC_UD);
+ sfence = true;
+ fault_suppression = false;
+ /* fall through */
+ CASE_SIMD_PACKED_FP(_EVEX, 0x0f, 0x10): /* vmovup{s,d} [xyz]mm/mem,[xyz]mm{k} */
+ CASE_SIMD_SCALAR_FP(_EVEX, 0x0f, 0x10): /* vmovs{s,d} mem,xmm{k} */
+ /* vmovs{s,d} xmm,xmm,xmm{k} */
+ CASE_SIMD_PACKED_FP(_EVEX, 0x0f, 0x11): /* vmovup{s,d} [xyz]mm,[xyz]mm/mem{k} */
+ CASE_SIMD_SCALAR_FP(_EVEX, 0x0f, 0x11): /* vmovs{s,d} xmm,mem{k} */
+ /* vmovs{s,d} xmm,xmm,xmm{k} */
+ CASE_SIMD_PACKED_FP(_EVEX, 0x0f, 0x28): /* vmovap{s,d} [xyz]mm/mem,[xyz]mm{k} */
+ CASE_SIMD_PACKED_FP(_EVEX, 0x0f, 0x29): /* vmovap{s,d} [xyz]mm,[xyz]mm/mem{k} */
+ /* vmovs{s,d} to/from memory have only two operands. */
+ if ( (b & ~1) == 0x10 && ea.type == OP_MEM )
+ d |= TwoOp;
+ generate_exception_if(evex.br, EXC_UD);
+ generate_exception_if(evex.w != (evex.pfx & VEX_PREFIX_DOUBLE_MASK),
+ EXC_UD);
+ host_and_vcpu_must_have(avx512f);
+ avx512_vlen_check(evex.pfx & VEX_PREFIX_SCALAR_MASK);
+ simd_zmm:
+ get_fpu(X86EMUL_FPU_zmm);
+ opc = init_evex(stub);
+ opc[0] = b;
+ opc[1] = modrm;
+ if ( ea.type == OP_MEM )
+ {
+ /* convert memory operand to (%rAX) */
+ evex.b = 1;
+ opc[1] &= 0x38;
+ }
+ insn_bytes = EVEX_PFX_BYTES + 2;
+ break;
+
case X86EMUL_OPC_66(0x0f, 0x12): /* movlpd m64,xmm */
case X86EMUL_OPC_VEX_66(0x0f, 0x12): /* vmovlpd m64,xmm,xmm */
CASE_SIMD_PACKED_FP(, 0x0f, 0x13): /* movlp{s,d} xmm,m64 */
@@ -6348,6 +6511,41 @@ x86_emulate(
ASSERT(!state->simd_size);
break;
+ case X86EMUL_OPC_EVEX_66(0x0f, 0x6e): /* vmov{d,q} r/m,xmm */
+ case X86EMUL_OPC_EVEX_66(0x0f, 0x7e): /* vmov{d,q} xmm,r/m */
+ generate_exception_if((evex.lr || evex.opmsk || evex.br ||
+ evex.reg != 0xf || !evex.RX),
+ EXC_UD);
+ host_and_vcpu_must_have(avx512f);
+ get_fpu(X86EMUL_FPU_zmm);
+
+ opc = init_evex(stub);
+ opc[0] = b;
+ /* Convert memory/GPR operand to (%rAX). */
+ evex.b = 1;
+ if ( !mode_64bit() )
+ evex.w = 0;
+ opc[1] = modrm & 0x38;
+ insn_bytes = EVEX_PFX_BYTES + 2;
+ opc[2] = 0xc3;
+
+ copy_EVEX(opc, evex);
+ invoke_stub("", "", "+m" (src.val) : "a" (&src.val));
+ dst.val = src.val;
+
+ put_stub(stub);
+ ASSERT(!state->simd_size);
+ break;
+
+ case X86EMUL_OPC_EVEX_F3(0x0f, 0x7e): /* vmovq xmm/m64,xmm */
+ case X86EMUL_OPC_EVEX_66(0x0f, 0xd6): /* vmovq xmm,xmm/m64 */
+ generate_exception_if(evex.lr || !evex.w || evex.opmsk || evex.br,
+ EXC_UD);
+ host_and_vcpu_must_have(avx512f);
+ d |= TwoOp;
+ op_bytes = 8;
+ goto simd_zmm;
+
case X86EMUL_OPC_66(0x0f, 0xe7): /* movntdq xmm,m128 */
case X86EMUL_OPC_VEX_66(0x0f, 0xe7): /* vmovntdq {x,y}mm,mem */
generate_exception_if(ea.type != OP_MEM, EXC_UD);
@@ -6368,6 +6566,30 @@ x86_emulate(
goto simd_0f_avx;
goto simd_0f_sse2;
+ case X86EMUL_OPC_EVEX_66(0x0f, 0xe7): /* vmovntdq [xyz]mm,mem */
+ generate_exception_if(ea.type != OP_MEM || evex.opmsk || evex.w,
+ EXC_UD);
+ sfence = true;
+ fault_suppression = false;
+ /* fall through */
+ case X86EMUL_OPC_EVEX_66(0x0f, 0x6f): /* vmovdqa{32,64} [xyz]mm/mem,[xyz]mm{k} */
+ case X86EMUL_OPC_EVEX_F3(0x0f, 0x6f): /* vmovdqu{32,64} [xyz]mm/mem,[xyz]mm{k} */
+ case X86EMUL_OPC_EVEX_66(0x0f, 0x7f): /* vmovdqa{32,64} [xyz]mm,[xyz]mm/mem{k} */
+ case X86EMUL_OPC_EVEX_F3(0x0f, 0x7f): /* vmovdqu{32,64} [xyz]mm,[xyz]mm/mem{k} */
+ vmovdqa:
+ generate_exception_if(evex.br, EXC_UD);
+ host_and_vcpu_must_have(avx512f);
+ avx512_vlen_check(false);
+ d |= TwoOp;
+ op_bytes = 16 << evex.lr;
+ goto simd_zmm;
+
+ case X86EMUL_OPC_EVEX_F2(0x0f, 0x6f): /* vmovdqu{8,16} [xyz]mm/mem,[xyz]mm{k} */
+ case X86EMUL_OPC_EVEX_F2(0x0f, 0x7f): /* vmovdqu{8,16} [xyz]mm,[xyz]mm/mem{k} */
+ host_and_vcpu_must_have(avx512bw);
+ elem_bytes = 1 << evex.w;
+ goto vmovdqa;
+
case X86EMUL_OPC_VEX_66(0x0f, 0xd6): /* vmovq xmm,xmm/m64 */
generate_exception_if(vex.l, EXC_UD);
d |= TwoOp;
@@ -7734,6 +7956,15 @@ x86_emulate(
}
goto movdqa;
+ case X86EMUL_OPC_EVEX_66(0x0f38, 0x2a): /* vmovntdqa mem,[xyz]mm */
+ generate_exception_if(ea.type != OP_MEM || evex.opmsk || evex.w,
+ EXC_UD);
+ /* Ignore the non-temporal hint for now, using vmovdqa32 instead. */
+ asm volatile ( "mfence" ::: "memory" );
+ b = 0x6f;
+ evex.opcx = vex_0f;
+ goto vmovdqa;
+
case X86EMUL_OPC_VEX_66(0x0f38, 0x2c): /* vmaskmovps mem,{x,y}mm,{x,y}mm */
case X86EMUL_OPC_VEX_66(0x0f38, 0x2d): /* vmaskmovpd mem,{x,y}mm,{x,y}mm */
case X86EMUL_OPC_VEX_66(0x0f38, 0x2e): /* vmaskmovps {x,y}mm,{x,y}mm,mem */
@@ -8787,17 +9018,27 @@ x86_emulate(
else if ( state->simd_size )
{
generate_exception_if(!op_bytes, EXC_UD);
- generate_exception_if(vex.opcx && (d & TwoOp) && vex.reg != 0xf,
+ generate_exception_if((vex.opcx && (d & TwoOp) &&
+ (vex.reg != 0xf || (evex_encoded() && !evex.RX))),
EXC_UD);
if ( !opc )
BUG();
- opc[insn_bytes - PFX_BYTES] = 0xc3;
- copy_REX_VEX(opc, rex_prefix, vex);
+ if ( evex_encoded() )
+ {
+ opc[insn_bytes - EVEX_PFX_BYTES] = 0xc3;
+ copy_EVEX(opc, evex);
+ }
+ else
+ {
+ opc[insn_bytes - PFX_BYTES] = 0xc3;
+ copy_REX_VEX(opc, rex_prefix, vex);
+ }
if ( ea.type == OP_MEM )
{
uint32_t mxcsr = 0;
+ uint64_t full = 0;
if ( op_bytes < 16 ||
(vex.opcx
@@ -8819,6 +9060,44 @@ x86_emulate(
!is_aligned(ea.mem.seg, ea.mem.off, op_bytes,
ctxt, ops),
EXC_GP, 0);
+
+ if ( evex.br )
+ {
+ ASSERT((d & DstMask) != DstMem);
+ op_bytes = elem_bytes;
+ }
+ if ( evex.opmsk )
+ {
+ ASSERT(!(op_bytes % elem_bytes));
+ full = ~0ULL >> (64 - op_bytes / elem_bytes);
+ op_mask &= full;
+ }
+ if ( fault_suppression )
+ {
+ if ( !op_mask )
+ goto simd_no_mem;
+ if ( !evex.br )
+ {
+ first_byte = __builtin_ctzll(op_mask);
+ op_mask >>= first_byte;
+ full >>= first_byte;
+ first_byte *= elem_bytes;
+ op_bytes = (64 - __builtin_clzll(op_mask)) * elem_bytes;
+ }
+ }
+ /*
+ * Independent of fault suppression we may need to read (parts of)
+ * the memory operand for the purpose of merging without splitting
+ * the write below into multiple ones. Note that the EVEX.Z check
+ * here isn't strictly needed, due to there not currently being
+ * any instructions allowing zeroing-merging on memory writes (and
+ * we raise #UD during DstMem processing far above in this case),
+ * yet conceptually the read is then unnecessary.
+ */
+ if ( evex.opmsk && !evex.z && (d & DstMask) == DstMem &&
+ op_mask != full )
+ d = (d & ~SrcMask) | SrcMem;
+
switch ( d & SrcMask )
{
case SrcMem:
@@ -8865,7 +9144,10 @@ x86_emulate(
}
}
else
+ {
+ simd_no_mem:
dst.type = OP_NONE;
+ }
/* {,v}maskmov{q,dqu}, as an exception, uses rDI. */
if ( likely((ctxt->opcode & ~(X86EMUL_OPC_PFX_MASK |
--- a/xen/arch/x86/x86_emulate/x86_emulate.h
+++ b/xen/arch/x86/x86_emulate/x86_emulate.h
@@ -171,6 +171,7 @@ enum x86_emulate_fpu_type {
X86EMUL_FPU_xmm, /* SSE instruction set (%xmm0-%xmm7/15) */
X86EMUL_FPU_ymm, /* AVX/XOP instruction set (%ymm0-%ymm7/15) */
X86EMUL_FPU_opmask, /* AVX512 opmask instruction set (%k0-%k7) */
+ X86EMUL_FPU_zmm, /* AVX512 instruction set (%zmm0-%zmm7/31) */
/* This sentinel will never be passed to ->get_fpu(). */
X86EMUL_FPU_none
};
--- a/xen/include/asm-x86/cpufeature.h
+++ b/xen/include/asm-x86/cpufeature.h
@@ -105,6 +105,7 @@
#define cpu_has_smap boot_cpu_has(X86_FEATURE_SMAP)
#define cpu_has_sha boot_cpu_has(X86_FEATURE_SHA)
#define cpu_has_avx512bw boot_cpu_has(X86_FEATURE_AVX512BW)
+#define cpu_has_avx512vl boot_cpu_has(X86_FEATURE_AVX512VL)
/* CPUID level 0x80000007.edx */
#define cpu_has_itsc boot_cpu_has(X86_FEATURE_ITSC)
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
next prev parent reply other threads:[~2018-09-18 11:56 UTC|newest]
Thread overview: 465+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-08-09 8:15 [PATCH 0/6] x86emul: fixes, improvements, and beginnings of AVX512 support Jan Beulich
2018-08-09 8:23 ` [PATCH 1/6] x86emul: fix FMA scalar operand sizes Jan Beulich
2018-08-09 8:24 ` [PATCH 2/6] x86emul: extend MASKMOV{Q,DQU} tests Jan Beulich
2018-08-09 8:24 ` [PATCH 3/6] x86emul: support AVX512 opmask insns Jan Beulich
2018-08-09 8:25 ` [PATCH 4/6] x86emul: clean up AVX2 insn use in test harness Jan Beulich
2018-08-09 8:25 ` [PATCH 5/6] x86emul: correct EVEX decoding Jan Beulich
2018-08-09 8:26 ` [PATCH 6/6] x86emul: generalize vector length handling for AVX512/EVEX Jan Beulich
2018-08-29 14:20 ` [PATCH v2 0/6] x86emul: fixes, improvements, and beginnings of AVX512 support Jan Beulich
2018-08-29 14:23 ` [PATCH v2 1/6] x86emul: fix FMA scalar operand sizes Jan Beulich
2018-09-03 16:43 ` Andrew Cooper
2018-09-04 7:52 ` Jan Beulich
2018-08-29 14:23 ` [PATCH v2 2/6] x86emul: extend MASKMOV{Q,DQU} tests Jan Beulich
2018-09-03 16:44 ` [PATCH v2 2/6] x86emul: extend MASKMOV{Q, DQU} tests Andrew Cooper
2018-08-29 14:24 ` [PATCH v2 3/6] x86emul: support AVX512 opmask insns Jan Beulich
2018-09-03 17:57 ` Andrew Cooper
2018-09-04 7:58 ` Jan Beulich
2018-08-29 14:24 ` [PATCH v2 4/6] x86emul: clean up AVX2 insn use in test harness Jan Beulich
2018-09-03 18:04 ` Andrew Cooper
2018-08-29 14:25 ` [PATCH v2 5/6] x86emul: correct EVEX decoding Jan Beulich
2018-09-04 10:48 ` Andrew Cooper
2018-09-04 12:48 ` Jan Beulich
2018-08-29 14:25 ` [PATCH v2 6/6] x86emul: generalize vector length handling for AVX512/EVEX Jan Beulich
2018-09-04 11:02 ` Andrew Cooper
2018-09-04 12:50 ` Jan Beulich
2018-09-18 11:46 ` [PATCH v3 00/34] x86emul: fixes, improvements, and beginnings of AVX512 support Jan Beulich
2018-09-18 11:53 ` [PATCH v3 01/34] x86emul: support AVX512 opmask insns Jan Beulich
2018-10-25 18:32 ` Andrew Cooper
2018-10-26 9:03 ` Jan Beulich
2018-10-26 11:29 ` Andrew Cooper
2018-10-26 11:59 ` Jan Beulich
2018-10-26 12:19 ` Andrew Cooper
2018-10-26 12:34 ` Jan Beulich
2018-09-18 11:53 ` [PATCH v3 02/34] x86/HVM: grow MMIO cache data size to 64 bytes Jan Beulich
2018-09-18 16:05 ` Paul Durrant
2018-10-25 18:36 ` Andrew Cooper
2018-10-26 9:04 ` Jan Beulich
2018-10-26 11:29 ` Andrew Cooper
2018-09-18 11:55 ` [PATCH v3 03/34] x86emul: correct EVEX decoding Jan Beulich
2018-09-18 11:55 ` [PATCH v3 04/34] x86emul: generalize vector length handling for AVX512/EVEX Jan Beulich
2018-09-18 11:56 ` Jan Beulich [this message]
2018-09-18 11:57 ` [PATCH v3 06/34] x86emul: test for correct EVEX Disp8 scaling Jan Beulich
2018-09-18 11:57 ` [PATCH v3 07/34] x86emul: use AVX512 logic for emulating V{, P}MASKMOV* Jan Beulich
2018-09-18 11:58 ` [PATCH v3 08/34] x86emul: support AVX512F legacy-equivalent arithmetic FP insns Jan Beulich
2018-09-18 11:59 ` [PATCH v3 09/34] x86emul: support AVX512DQ logic " Jan Beulich
2018-09-18 11:59 ` [PATCH v3 10/34] x86emul: support AVX512F "normal" FP compare insns Jan Beulich
2018-09-18 12:00 ` [PATCH v3 11/34] x86emul: support AVX512F misc legacy-equivalent FP insns Jan Beulich
2018-09-18 12:00 ` [PATCH v3 12/34] x86emul: support AVX512F fused-multiply-add insns Jan Beulich
2018-09-18 12:01 ` [PATCH v3 13/34] x86emul: support AVX512F legacy-equivalent logic insns Jan Beulich
2018-09-18 12:02 ` [PATCH v3 14/34] x86emul: support AVX512{F, DQ} FP broadcast insns Jan Beulich
2018-09-18 12:03 ` [PATCH v3 15/34] x86emul: support AVX512F v{, u}comis{d, s} insns Jan Beulich
2018-09-18 12:03 ` [PATCH v3 16/34] x86emul/test: introduce eq() Jan Beulich
2018-09-18 12:04 ` [PATCH v3 17/34] x86emul: support AVX512{F, BW} packed integer compare insns Jan Beulich
2018-09-18 12:05 ` [PATCH v3 18/34] x86emul: support AVX512{F, BW} packed integer arithmetic insns Jan Beulich
2018-09-18 12:05 ` [PATCH v3 19/34] x86emul: use simd_128 also for legacy vector shift insns Jan Beulich
2018-09-18 12:05 ` [PATCH v3 20/34] x86emul: support AVX512{F, BW} shift/rotate insns Jan Beulich
2018-09-18 12:06 ` [PATCH v3 21/34] x86emul: support AVX512{F, BW, DQ} extract insns Jan Beulich
2018-09-18 12:07 ` [PATCH v3 22/34] x86emul: support AVX512{F, BW, DQ} insert insns Jan Beulich
2018-09-18 12:07 ` [PATCH v3 23/34] x86emul: basic AVX512F testing Jan Beulich
2018-09-18 12:08 ` [PATCH v3 24/34] x86emul: support AVX512{F, BW, DQ} integer broadcast insns Jan Beulich
2018-09-18 12:09 ` [PATCH v3 25/34] x86emul: basic AVX512VL testing Jan Beulich
2018-09-18 12:09 ` [PATCH v3 26/34] x86emul: support AVX512{F, BW} zero- and sign-extending moves Jan Beulich
2018-09-18 12:09 ` [PATCH v3 27/34] x86emul: support AVX512{F, BW} down conversion moves Jan Beulich
2018-09-18 12:10 ` [PATCH v3 28/34] x86emul: support AVX512{F, BW} integer unpack insns Jan Beulich
2018-09-18 12:11 ` [PATCH v3 29/34] x86emul: support AVX512{F, BW, _VBMI} full permute insns Jan Beulich
2018-09-18 12:11 ` [PATCH v3 29/34] x86emul: support AVX512{F, BW} integer shuffle insns Jan Beulich
2018-09-18 12:12 ` [PATCH v3 30/34] x86emul: support AVX512{BW, DQ} mask move insns Jan Beulich
2018-09-18 12:14 ` [PATCH v3 32/34] x86emul: basic AVX512BW testing Jan Beulich
2018-09-18 12:14 ` [PATCH v3 33/34] x86emul: basic AVX512DQ testing Jan Beulich
2018-09-18 12:14 ` [PATCH v3 34/34] x86emul: also allow running the 32-bit harness on a 64-bit distro Jan Beulich
2018-09-25 13:14 ` [PATCH v4 00/44] x86emul: fixes, improvements, and beginnings of AVX512 support Jan Beulich
2018-09-25 13:25 ` [PATCH v4 01/44] x86emul: support AVX512 opmask insns Jan Beulich
2018-09-26 6:06 ` Jan Beulich
2018-09-25 13:26 ` [PATCH v4 02/44] x86/HVM: grow MMIO cache data size to 64 bytes Jan Beulich
2018-09-25 13:27 ` [PATCH v4 03/44] x86emul: correct EVEX decoding Jan Beulich
2018-10-26 14:33 ` Andrew Cooper
2018-09-25 13:28 ` [PATCH v4 04/44] x86emul: generalize vector length handling for AVX512/EVEX Jan Beulich
2018-10-26 16:10 ` Andrew Cooper
2018-09-25 13:28 ` [PATCH v4 05/44] x86emul: support basic AVX512 moves Jan Beulich
2018-11-13 17:12 ` Andrew Cooper
2018-11-14 14:35 ` Jan Beulich
2018-11-14 16:26 ` Andrew Cooper
2018-11-15 9:54 ` Jan Beulich
2018-09-25 13:29 ` [PATCH v4 06/44] x86emul: test for correct EVEX Disp8 scaling Jan Beulich
2018-11-12 17:42 ` Andrew Cooper
2018-11-13 11:12 ` Jan Beulich
2018-11-13 15:45 ` Andrew Cooper
2018-11-14 14:17 ` Jan Beulich
2018-11-14 14:42 ` Andrew Cooper
2018-11-14 14:58 ` Jan Beulich
2018-09-25 13:29 ` [PATCH v4 07/44] x86emul: also allow running the 32-bit harness on a 64-bit distro Jan Beulich
2018-11-12 17:50 ` Andrew Cooper
2018-11-13 11:42 ` Jan Beulich
2018-11-13 15:58 ` Andrew Cooper
2018-11-14 8:42 ` Jan Beulich
2018-09-25 13:30 ` [PATCH v4 08/44] x86emul: use AVX512 logic for emulating V{, P}MASKMOV* Jan Beulich
2018-11-13 18:12 ` Andrew Cooper
2018-09-25 13:31 ` [PATCH v4 09/44] x86emul: support AVX512F legacy-equivalent arithmetic FP insns Jan Beulich
2018-11-13 18:21 ` Andrew Cooper
2018-09-25 13:32 ` [PATCH v4 10/44] x86emul: support AVX512DQ logic " Jan Beulich
2018-11-13 18:56 ` Andrew Cooper
2018-09-25 13:32 ` [PATCH v4 11/44] x86emul: support AVX512F "normal" FP compare insns Jan Beulich
2018-11-13 19:04 ` Andrew Cooper
2018-11-14 14:41 ` Jan Beulich
2018-11-14 14:45 ` Andrew Cooper
2018-09-25 13:33 ` [PATCH v4 12/44] x86emul: support AVX512F misc legacy-equivalent FP insns Jan Beulich
2018-11-13 19:17 ` Andrew Cooper
2018-09-25 13:33 ` [PATCH v4 13/44] x86emul: support AVX512F fused-multiply-add insns Jan Beulich
2018-11-13 19:28 ` Andrew Cooper
2018-09-25 13:34 ` [PATCH v4 14/44] x86emul: support AVX512F legacy-equivalent logic insns Jan Beulich
2018-11-13 19:30 ` Andrew Cooper
2018-09-25 13:35 ` [PATCH v4 15/44] x86emul: support AVX512{F, DQ} FP broadcast insns Jan Beulich
2018-11-13 19:37 ` Andrew Cooper
2018-09-25 13:35 ` [PATCH v4 16/44] x86emul: support AVX512F v{, u}comis{d, s} insns Jan Beulich
2018-11-13 19:39 ` Andrew Cooper
2018-09-25 13:36 ` [PATCH v4 17/44] x86emul/test: introduce eq() Jan Beulich
2018-10-26 11:31 ` Andrew Cooper
2018-09-25 13:37 ` [PATCH v4 18/44] x86emul: support AVX512{F, BW} packed integer compare insns Jan Beulich
2018-09-25 13:37 ` [PATCH v4 19/44] x86emul: support AVX512{F, BW} packed integer arithmetic insns Jan Beulich
2018-09-25 13:38 ` [PATCH v4 20/44] x86emul: use simd_128 also for legacy vector shift insns Jan Beulich
2018-09-25 13:39 ` [PATCH v4 21/44] x86emul: support AVX512{F, BW} shift/rotate insns Jan Beulich
2018-09-25 13:40 ` [PATCH v4 22/44] x86emul: support AVX512{F, BW, DQ} extract insns Jan Beulich
2018-09-25 13:40 ` [PATCH v4 23/44] x86emul: support AVX512{F, BW, DQ} insert insns Jan Beulich
2018-09-25 13:41 ` [PATCH v4 24/44] x86emul: basic AVX512F testing Jan Beulich
2018-09-25 13:41 ` [PATCH v4 25/44] x86emul: support AVX512{F, BW, DQ} integer broadcast insns Jan Beulich
2018-09-25 13:42 ` [PATCH v4 26/44] x86emul: basic AVX512VL testing Jan Beulich
2018-09-25 13:43 ` [PATCH v4 27/44] x86emul: support AVX512{F, BW} zero- and sign-extending moves Jan Beulich
2018-09-25 13:43 ` [PATCH v4 28/44] x86emul: support AVX512{F, BW} down conversion moves Jan Beulich
2018-09-25 13:44 ` [PATCH v4 29/44] x86emul: support AVX512{F, BW} integer unpack insns Jan Beulich
2018-09-25 13:44 ` [PATCH v4 30/44] x86emul: support AVX512{F, BW, _VBMI} full permute insns Jan Beulich
2018-09-25 13:46 ` [PATCH v4 31/44] x86emul: support AVX512{F, BW} integer shuffle insns Jan Beulich
2018-09-25 13:46 ` [PATCH v4 32/44] x86emul: support AVX512{BW, DQ} mask move insns Jan Beulich
2018-09-25 13:47 ` [PATCH v4 33/44] x86emul: basic AVX512BW testing Jan Beulich
2018-09-25 13:48 ` [PATCH v4 34/44] x86emul: basic AVX512DQ testing Jan Beulich
2018-09-25 13:48 ` [PATCH v4 35/44] x86emul: support AVX512F move high/low insns Jan Beulich
2018-09-25 13:49 ` [PATCH v4 36/44] x86emul: support AVX512F move duplicate insns Jan Beulich
2018-09-25 13:49 ` [PATCH v4 37/44] x86emul: support AVX512{F, BW, VBMI} permute insns Jan Beulich
2018-09-25 13:50 ` [PATCH v4 38/44] x86emul: support AVX512BW pack insns Jan Beulich
2018-09-25 13:51 ` [PATCH v4 39/44] x86emul: support AVX512F floating-point conversion insns Jan Beulich
2018-09-25 13:52 ` [PATCH v4 40/44] x86emul: support AVX512F legacy-equivalent packed int/FP " Jan Beulich
2018-09-25 13:53 ` [PATCH v4 41/44] x86emul: support AVX512F legacy-equivalent scalar " Jan Beulich
2018-09-25 13:53 ` [PATCH v4 42/44] x86emul: support AVX512DQ packed quad-int/FP " Jan Beulich
2018-09-25 13:54 ` [PATCH v4 43/44] x86emul: support AVX512{F, DQ} uint-to-FP " Jan Beulich
2018-09-25 13:55 ` [PATCH v4 44/44] x86emul: support AVX512{F, DQ} FP-to-uint " Jan Beulich
2018-11-19 10:00 ` [PATCH v5 00/47] x86emul: fair parts of AVX512 support Jan Beulich
2018-11-19 10:13 ` [PATCH v5 01/47] x86emul: introduce IMPOSSIBLE() Jan Beulich
2018-11-19 18:11 ` Andrew Cooper
2018-11-20 8:12 ` Jan Beulich
2018-11-19 10:13 ` [PATCH v5 02/47] x86emul: support basic AVX512 moves Jan Beulich
2018-11-19 18:35 ` Andrew Cooper
2018-11-19 10:14 ` [PATCH v5 03/47] x86emul: test for correct EVEX Disp8 scaling Jan Beulich
2018-11-19 18:35 ` Andrew Cooper
2018-11-19 10:14 ` [PATCH v5 04/47] x86emul: also allow running the 32-bit harness on a 64-bit distro Jan Beulich
2018-11-19 18:40 ` Andrew Cooper
2018-11-19 10:15 ` [PATCH v5 05/47] x86emul: use AVX512 logic for emulating V{, P}MASKMOV* Jan Beulich
2018-11-19 10:16 ` [PATCH v5 06/47] x86emul: support AVX512F legacy-equivalent arithmetic FP insns Jan Beulich
2018-11-19 10:16 ` [PATCH v5 07/47] x86emul: support AVX512DQ logic " Jan Beulich
2018-11-19 10:17 ` [PATCH v5 08/47] x86emul: support basic AVX512F FP compare insns Jan Beulich
2018-11-19 10:17 ` [PATCH v5 09/47] x86emul: support AVX512F misc legacy-equivalent FP insns Jan Beulich
2018-11-19 10:18 ` [PATCH v5 10/47] x86emul: support AVX512F fused-multiply-add insns Jan Beulich
2018-11-19 10:18 ` [PATCH v5 11/47] x86emul: support AVX512F legacy-equivalent logic insns Jan Beulich
2018-11-19 10:19 ` [PATCH v5 12/47] x86emul: support AVX512{F, DQ} FP broadcast insns Jan Beulich
2018-11-19 18:44 ` Andrew Cooper
2018-11-19 10:20 ` [PATCH v5 13/47] x86emul: support AVX512F v{, u}comis{d, s} insns Jan Beulich
2018-11-19 10:21 ` [PATCH v5 14/47] x86emul: support AVX512{F, BW} packed integer compare insns Jan Beulich
2018-11-19 19:09 ` Andrew Cooper
2018-11-19 10:21 ` [PATCH v5 15/47] x86emul: support AVX512{F, BW} packed integer arithmetic insns Jan Beulich
2018-11-19 19:18 ` Andrew Cooper
2018-11-19 10:22 ` [PATCH v5 16/47] x86emul: use simd_128 also for legacy vector shift insns Jan Beulich
2018-11-19 19:21 ` Andrew Cooper
2018-11-19 10:22 ` [PATCH v5 17/47] x86emul: support AVX512{F, BW} shift/rotate insns Jan Beulich
2018-11-19 10:23 ` [PATCH v5 18/47] x86emul: support AVX512{F, BW, DQ} extract insns Jan Beulich
2018-11-19 10:23 ` [PATCH v5 19/47] x86emul: support AVX512{F, BW, DQ} insert insns Jan Beulich
2018-11-19 10:24 ` [PATCH v5 20/47] x86emul: basic AVX512F testing Jan Beulich
2018-11-19 10:25 ` [PATCH v5 21/47] x86emul: support AVX512{F, BW, DQ} integer broadcast insns Jan Beulich
2018-11-19 10:25 ` [PATCH v5 22/47] x86emul: basic AVX512VL testing Jan Beulich
2018-11-19 10:26 ` [PATCH v5 23/47] x86emul: support AVX512{F, BW} zero- and sign-extending moves Jan Beulich
2018-11-19 10:26 ` [PATCH v5 24/47] x86emul: support AVX512{F, BW} down conversion moves Jan Beulich
2018-11-19 10:28 ` [PATCH v5 25/47] x86emul: support AVX512{F, BW} integer unpack insns Jan Beulich
2018-11-19 10:28 ` [PATCH v5 26/47] x86emul: support AVX512{F, BW, _VBMI} full permute insns Jan Beulich
2018-11-19 10:29 ` [PATCH v5 27/47] x86emul: support AVX512{F, BW} integer shuffle insns Jan Beulich
2018-11-19 10:30 ` [PATCH v5 28/47] x86emul: support AVX512{BW, DQ} mask move insns Jan Beulich
2018-11-19 10:30 ` [PATCH v5 29/47] x86emul: basic AVX512BW testing Jan Beulich
2018-11-19 10:31 ` [PATCH v5 30/47] x86emul: basic AVX512DQ testing Jan Beulich
2018-11-19 10:31 ` [PATCH v5 31/47] x86emul: support AVX512F move high/low insns Jan Beulich
2018-11-19 10:31 ` [PATCH v5 32/47] x86emul: support AVX512F move duplicate insns Jan Beulich
2018-11-19 10:32 ` [PATCH v5 33/47] x86emul: support AVX512{F, BW, VBMI} permute insns Jan Beulich
2018-11-19 10:33 ` [PATCH v5 34/47] x86emul: support AVX512BW pack insns Jan Beulich
2018-11-19 10:33 ` [PATCH v5 35/47] x86emul: support AVX512F floating-point conversion insns Jan Beulich
2018-11-19 10:34 ` [PATCH v5 36/47] x86emul: support AVX512F legacy-equivalent packed int/FP " Jan Beulich
2018-11-19 10:35 ` [PATCH v5 37/47] x86emul: support AVX512F legacy-equivalent scalar " Jan Beulich
2018-11-19 10:36 ` [PATCH v5 38/47] x86emul: support AVX512DQ packed quad-int/FP " Jan Beulich
2018-11-19 10:37 ` [PATCH v5 39/47] x86emul: support AVX512{F, DQ} uint-to-FP " Jan Beulich
2018-11-19 10:37 ` [PATCH v5 40/47] x86emul: support AVX512{F, DQ} FP-to-uint " Jan Beulich
2018-11-19 10:38 ` [PATCH v5 41/47] x86emul: support remaining AVX512F legacy-equivalent insns Jan Beulich
2018-11-19 10:38 ` [PATCH v5 42/47] x86emul: support remaining AVX512BW " Jan Beulich
2018-11-19 10:39 ` [PATCH v5 43/47] x86emul: support AVX512{F, ER} reciprocal insns Jan Beulich
2018-11-19 10:39 ` [PATCH v5 44/47] x86emul: support AVX512F floating point manipulation insns Jan Beulich
2018-11-19 10:40 ` [PATCH v5 45/47] x86emul: support AVX512DQ " Jan Beulich
2018-11-19 10:40 ` [PATCH v5 46/47] x86emul: support AVX512{F, _VBMI2} compress/expand insns Jan Beulich
2018-11-19 10:41 ` [PATCH v5 47/47] x86emul: support remaining misc AVX512{F, BW} insns Jan Beulich
2018-12-06 9:45 ` [PATCH v6 00/42] x86emul: fair parts of AVX512 support Jan Beulich
2018-12-06 9:51 ` [PATCH v6 01/42] x86emul: support AVX512{F, BW} shift/rotate insns Jan Beulich
2018-12-06 9:51 ` [PATCH v6 02/42] x86emul: support AVX512{F, BW, DQ} extract insns Jan Beulich
2018-12-06 9:51 ` [PATCH v6 03/42] x86emul: support AVX512{F, BW, DQ} insert insns Jan Beulich
2018-12-06 9:52 ` [PATCH v6 04/42] x86emul: basic AVX512F testing Jan Beulich
2018-12-06 9:53 ` [PATCH v6 05/42] x86emul: support AVX512{F, BW, DQ} integer broadcast insns Jan Beulich
2018-12-06 9:53 ` [PATCH v6 06/42] x86emul: basic AVX512VL testing Jan Beulich
2018-12-06 9:53 ` [PATCH v6 07/42] x86emul: support AVX512{F, BW} zero- and sign-extending moves Jan Beulich
2018-12-06 9:54 ` [PATCH v6 08/42] x86emul: support AVX512{F, BW} down conversion moves Jan Beulich
2018-12-06 9:54 ` [PATCH v6 09/42] x86emul: support AVX512{F, BW} integer unpack insns Jan Beulich
2018-12-06 9:55 ` [PATCH v6 10/42] x86emul: support AVX512{F, BW, _VBMI} full permute insns Jan Beulich
2018-12-06 9:55 ` [PATCH v6 11/42] x86emul: support AVX512{F, BW} integer shuffle insns Jan Beulich
2018-12-06 9:55 ` [PATCH v6 12/42] x86emul: support AVX512{BW, DQ} mask move insns Jan Beulich
2018-12-06 9:56 ` [PATCH v6 13/42] x86emul: basic AVX512BW testing Jan Beulich
2018-12-06 9:57 ` [PATCH v6 14/42] x86emul: basic AVX512DQ testing Jan Beulich
2018-12-06 9:57 ` [PATCH v6 15/42] x86emul: support AVX512F move high/low insns Jan Beulich
2018-12-06 9:58 ` [PATCH v6 16/42] x86emul: support AVX512F move duplicate insns Jan Beulich
2018-12-06 9:59 ` [PATCH v6 17/42] x86emul: support AVX512{F, BW, _VBMI} permute insns Jan Beulich
2018-12-06 9:59 ` [PATCH v6 18/42] x86emul: support AVX512BW pack insns Jan Beulich
2018-12-06 10:00 ` [PATCH v6 19/42] x86emul: support AVX512F floating-point conversion insns Jan Beulich
2018-12-06 10:00 ` [PATCH v6 20/42] x86emul: support AVX512F legacy-equivalent packed int/FP " Jan Beulich
2018-12-06 10:01 ` [PATCH v6 21/42] x86emul: support AVX512F legacy-equivalent scalar " Jan Beulich
2018-12-06 10:01 ` [PATCH v6 22/42] x86emul: support AVX512DQ packed quad-int/FP " Jan Beulich
2018-12-06 10:02 ` [PATCH v6 23/42] x86emul: support AVX512{F, DQ} uint-to-FP " Jan Beulich
2018-12-06 10:02 ` [PATCH v6 24/42] x86emul: support AVX512{F, DQ} FP-to-uint " Jan Beulich
2018-12-06 10:04 ` [PATCH v6 25/42] x86emul: support remaining AVX512F legacy-equivalent insns Jan Beulich
2018-12-06 10:04 ` [PATCH v6 26/42] x86emul: support remaining AVX512BW " Jan Beulich
2018-12-06 10:04 ` [PATCH v6 27/42] x86emul: support AVX512{F, ER} reciprocal insns Jan Beulich
2018-12-06 10:05 ` [PATCH v6 28/42] x86emul: support AVX512F floating point manipulation insns Jan Beulich
2018-12-06 10:05 ` [PATCH v6 29/42] x86emul: support AVX512DQ " Jan Beulich
2018-12-06 10:06 ` [PATCH v6 30/42] x86emul: support AVX512{F, _VBMI2} compress/expand insns Jan Beulich
2018-12-06 10:07 ` [PATCH v6 31/42] x86emul: support remaining misc AVX512{F, BW} insns Jan Beulich
2018-12-06 10:07 ` [PATCH v6 32/42] x86emul: support AVX512F gather insns Jan Beulich
2018-12-06 10:08 ` [PATCH v6 33/42] x86emul: add high register S/G test cases Jan Beulich
2018-12-06 10:08 ` [PATCH v6 34/42] x86emul: support AVX512F scatter insns Jan Beulich
2018-12-06 10:09 ` [PATCH v6 35/42] x86emul: support AVX512PF insns Jan Beulich
2018-12-06 10:09 ` [PATCH v6 36/42] x86emul: support AVX512CD insns Jan Beulich
2018-12-06 10:10 ` [PATCH v6 37/42] x86emul: complete support of AVX512_VBMI insns Jan Beulich
2018-12-06 10:10 ` [PATCH v6 38/42] x86emul: support of AVX512* population count insns Jan Beulich
2018-12-06 10:11 ` [PATCH v6 39/42] x86emul: support of AVX512_IFMA insns Jan Beulich
2018-12-06 10:11 ` [PATCH v6 40/42] x86emul: support remaining AVX512_VBMI2 insns Jan Beulich
2018-12-06 10:12 ` [PATCH v6 41/42] x86emul: support AVX512_4FMAPS insns Jan Beulich
2018-12-06 10:12 ` [PATCH v6 42/42] x86emul: support AVX512_4VNNIW insns Jan Beulich
2018-12-19 14:17 ` [PATCH v7 00/49] x86emul: remaining AVX512 support Jan Beulich
2018-12-19 14:34 ` Jan Beulich
2018-12-19 14:35 ` [PATCH v7 01/49] x86emul: rename evex.br to evex.brs Jan Beulich
2018-12-19 15:14 ` Andrew Cooper
2018-12-19 14:36 ` [PATCH v7 02/49] x86emul: support AVX512{F, BW} shift/rotate insns Jan Beulich
2018-12-19 16:00 ` Andrew Cooper
2018-12-19 14:36 ` [PATCH v7 03/49] x86emul: support AVX512{F, BW, DQ} extract insns Jan Beulich
2018-12-19 18:20 ` Andrew Cooper
2018-12-20 7:49 ` Jan Beulich
2018-12-19 14:37 ` [PATCH v7 04/49] x86emul: support AVX512{F, BW, DQ} insert insns Jan Beulich
2019-03-14 15:35 ` Andrew Cooper
2018-12-19 14:37 ` [PATCH v7 05/49] x86emul: basic AVX512F testing Jan Beulich
2019-03-14 15:42 ` Andrew Cooper
2018-12-19 14:38 ` [PATCH v7 06/49] x86emul: support AVX512{F, BW, DQ} integer broadcast insns Jan Beulich
2019-03-14 16:38 ` Andrew Cooper
2019-03-14 17:15 ` Jan Beulich
2019-03-15 16:39 ` Andrew Cooper
2019-03-18 9:45 ` Jan Beulich
2018-12-19 14:38 ` [PATCH v7 07/49] x86emul: basic AVX512VL testing Jan Beulich
2019-03-14 16:39 ` Andrew Cooper
2018-12-19 14:41 ` [PATCH v7 08/49] x86emul: support AVX512{F, BW} zero- and sign-extending moves Jan Beulich
2018-12-19 14:41 ` [PATCH v7 09/49] x86emul: support AVX512{F, BW} down conversion moves Jan Beulich
2018-12-19 14:42 ` [PATCH v7 10/49] x86emul: support AVX512{F, BW} integer unpack insns Jan Beulich
2018-12-19 14:42 ` [PATCH v7 11/49] x86emul: support AVX512{F, BW, _VBMI} full permute insns Jan Beulich
2018-12-19 14:43 ` [PATCH v7 12/49] x86emul: support AVX512{F, BW} integer shuffle insns Jan Beulich
2018-12-19 14:43 ` [PATCH v7 13/49] x86emul: support AVX512{BW, DQ} mask move insns Jan Beulich
2018-12-19 14:44 ` [PATCH v7 14/49] x86emul: basic AVX512BW testing Jan Beulich
2018-12-19 14:46 ` [PATCH v7 15/49] x86emul: basic AVX512DQ testing Jan Beulich
2018-12-19 14:46 ` [PATCH v7 16/49] x86emul: support AVX512F move high/low insns Jan Beulich
2018-12-19 14:47 ` [PATCH v7 17/49] x86emul: support AVX512F move duplicate insns Jan Beulich
2018-12-19 14:47 ` [PATCH v7 18/49] x86emul: support AVX512{F, BW, _VBMI} permute insns Jan Beulich
2018-12-19 14:48 ` [PATCH v7 19/49] x86emul: support AVX512BW pack insns Jan Beulich
2018-12-19 14:48 ` [PATCH v7 20/49] x86emul: support AVX512F floating-point conversion insns Jan Beulich
2018-12-19 14:48 ` [PATCH v7 21/49] x86emul: support AVX512F legacy-equivalent packed int/FP " Jan Beulich
2018-12-19 14:51 ` [PATCH v7 22/49] x86emul: support AVX512F legacy-equivalent scalar " Jan Beulich
2018-12-19 14:51 ` [PATCH v7 23/49] x86emul: support AVX512DQ packed quad-int/FP " Jan Beulich
2018-12-19 14:52 ` [PATCH v7 24/49] x86emul: support AVX512{F, DQ} uint-to-FP " Jan Beulich
2018-12-19 14:52 ` [PATCH v7 25/49] x86emul: support AVX512{F, DQ} FP-to-uint " Jan Beulich
2018-12-19 14:53 ` [PATCH v7 26/49] x86emul: support remaining AVX512F legacy-equivalent insns Jan Beulich
2018-12-19 14:53 ` [PATCH v7 27/49] x86emul: support remaining AVX512BW " Jan Beulich
2018-12-19 14:54 ` [PATCH v7 28/49] x86emul: support AVX512{F, ER} reciprocal insns Jan Beulich
2018-12-19 14:55 ` [PATCH v7 29/49] x86emul: support AVX512F floating point manipulation insns Jan Beulich
2018-12-19 14:55 ` [PATCH v7 30/49] x86emul: support AVX512DQ " Jan Beulich
2018-12-19 14:56 ` [PATCH v7 31/49] x86emul: support AVX512{F, _VBMI2} compress/expand insns Jan Beulich
2018-12-19 14:56 ` [PATCH v7 32/49] x86emul: support remaining misc AVX512{F, BW} insns Jan Beulich
2018-12-19 14:57 ` [PATCH v7 33/49] x86emul: support AVX512F gather insns Jan Beulich
2018-12-19 14:57 ` [PATCH v7 34/49] x86emul: add high register S/G test cases Jan Beulich
2018-12-19 14:58 ` [PATCH v7 35/49] x86emul: support AVX512F scatter insns Jan Beulich
2018-12-19 14:59 ` [PATCH v7 36/49] x86emul: support AVX512PF insns Jan Beulich
2018-12-19 14:59 ` [PATCH v7 37/49] x86emul: support AVX512CD insns Jan Beulich
2018-12-19 15:00 ` [PATCH v7 38/49] x86emul: complete support of AVX512_VBMI insns Jan Beulich
2018-12-19 15:00 ` [PATCH v7 39/49] x86emul: support of AVX512* population count insns Jan Beulich
2018-12-19 15:01 ` [PATCH v7 40/49] x86emul: support of AVX512_IFMA insns Jan Beulich
2018-12-19 15:01 ` [PATCH v7 42/49] x86emul: support remaining AVX512_VBMI2 insns Jan Beulich
2018-12-19 15:02 ` [PATCH v7 42/49] x86emul: support AVX512_4FMAPS insns Jan Beulich
2018-12-19 15:05 ` [PATCH v7 43/49] x86emul: support AVX512_4VNNIW insns Jan Beulich
2018-12-19 15:06 ` [PATCH v7 44/49] x86emul: support AVX512_VNNI insns Jan Beulich
2018-12-19 15:06 ` [PATCH v7 45/49] x86emul: support VPCLMULQDQ insns Jan Beulich
2018-12-19 15:07 ` [PATCH v7 46/49] x86emul: support VAES insns Jan Beulich
2018-12-19 15:07 ` [PATCH v7 47/49] x86emul: support GFNI insns Jan Beulich
2018-12-19 15:07 ` [PATCH v7 48/49] x86emul: restore ordering within main switch statement Jan Beulich
2018-12-19 15:08 ` [PATCH v7 49/49] tools: re-sync CPUID leaf 7 tables Jan Beulich
2019-03-14 11:07 ` Andrew Cooper
2019-03-15 10:30 ` [PATCH v8 00/50] x86emul: remaining AVX512 support Jan Beulich
2019-03-15 10:36 ` [PATCH v8 01/50] x86emul: no need to set fault_suppression to false for VMOVNT* Jan Beulich
2019-03-15 16:52 ` Andrew Cooper
2019-03-15 10:36 ` [PATCH v8 02/50] x86emul: support AVX512{F, BW, DQ} extract insns Jan Beulich
2019-03-15 17:51 ` Andrew Cooper
2019-03-15 10:37 ` [PATCH v8 03/50] x86emul: support AVX512{F, BW, DQ} insert insns Jan Beulich
2019-03-15 10:38 ` [PATCH v8 04/50] x86emul: basic AVX512F testing Jan Beulich
2019-03-15 10:39 ` [PATCH v8 05/50] x86emul: support AVX512{F, BW, DQ} integer broadcast insns Jan Beulich
2019-03-15 10:39 ` [PATCH v8 06/50] x86emul: basic AVX512VL testing Jan Beulich
2019-03-15 10:40 ` [PATCH v8 07/50] x86emul: support AVX512{F, BW} zero- and sign-extending moves Jan Beulich
2019-03-15 18:02 ` Andrew Cooper
2019-03-15 10:40 ` [PATCH v8 08/50] x86emul: support AVX512{F, BW} down conversion moves Jan Beulich
2019-03-15 18:10 ` Andrew Cooper
2019-03-15 10:41 ` [PATCH v8 09/50] x86emul: support AVX512{F, BW} integer unpack insns Jan Beulich
2019-03-15 18:21 ` Andrew Cooper
2019-03-18 9:55 ` Jan Beulich
2019-05-20 12:11 ` Andrew Cooper
2019-05-20 12:11 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:41 ` [PATCH v8 10/50] x86emul: support AVX512{F, BW, _VBMI} full permute insns Jan Beulich
2019-05-17 16:50 ` Andrew Cooper
2019-05-17 16:50 ` [Xen-devel] " Andrew Cooper
2019-05-20 6:55 ` Jan Beulich
2019-05-20 6:55 ` [Xen-devel] " Jan Beulich
2019-05-20 12:10 ` Andrew Cooper
2019-05-20 12:10 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:43 ` [PATCH v8 11/50] x86emul: support AVX512{F, BW} integer shuffle insns Jan Beulich
2019-05-17 17:01 ` Andrew Cooper
2019-05-17 17:01 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:43 ` [PATCH v8 12/50] x86emul: support AVX512{BW, DQ} mask move insns Jan Beulich
2019-05-17 17:02 ` Andrew Cooper
2019-05-17 17:02 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:43 ` [PATCH v8 13/50] x86emul: basic AVX512BW testing Jan Beulich
2019-05-17 17:03 ` Andrew Cooper
2019-05-17 17:03 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:44 ` [PATCH v8 14/50] x86emul: basic AVX512DQ testing Jan Beulich
2019-05-17 17:03 ` Andrew Cooper
2019-05-17 17:03 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:44 ` [PATCH v8 15/50] x86emul: support AVX512F move high/low insns Jan Beulich
2019-05-21 10:59 ` Andrew Cooper
2019-05-21 10:59 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:45 ` [PATCH v8 16/50] x86emul: support AVX512F move duplicate insns Jan Beulich
2019-05-21 11:09 ` Andrew Cooper
2019-05-21 11:09 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:46 ` [PATCH v8 17/50] x86emul: support AVX512{F, BW, _VBMI} permute insns Jan Beulich
2019-05-21 11:24 ` Andrew Cooper
2019-05-21 11:24 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:46 ` [PATCH v8 18/50] x86emul: support AVX512BW pack insns Jan Beulich
2019-05-21 11:26 ` Andrew Cooper
2019-05-21 11:26 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:47 ` [PATCH v8 19/50] x86emul: support AVX512F floating-point conversion insns Jan Beulich
2019-05-21 11:33 ` Andrew Cooper
2019-05-21 11:33 ` [Xen-devel] " Andrew Cooper
2019-05-21 15:46 ` Jan Beulich
2019-05-21 15:46 ` [Xen-devel] " Jan Beulich
2019-05-23 16:08 ` Andrew Cooper
2019-05-23 16:08 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:47 ` [PATCH v8 20/50] x86emul: support AVX512F legacy-equivalent packed int/FP " Jan Beulich
2019-05-21 11:37 ` Andrew Cooper
2019-05-21 11:37 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:52 ` [PATCH v8 21/50] x86emul: support AVX512F legacy-equivalent scalar " Jan Beulich
2019-05-21 11:44 ` Andrew Cooper
2019-05-21 11:44 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:52 ` [PATCH v8 22/50] x86emul: support AVX512DQ packed quad-int/FP " Jan Beulich
2019-05-21 11:53 ` Andrew Cooper
2019-05-21 11:53 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:53 ` [PATCH v8 23/50] x86emul: support AVX512{F, DQ} uint-to-FP " Jan Beulich
2019-05-21 11:58 ` Andrew Cooper
2019-05-21 11:58 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:54 ` [PATCH v8 24/50] x86emul: support AVX512{F, DQ} FP-to-uint " Jan Beulich
2019-05-21 12:09 ` Andrew Cooper
2019-05-21 12:09 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:54 ` [PATCH v8 25/50] x86emul: support remaining AVX512F legacy-equivalent insns Jan Beulich
2019-05-21 13:06 ` Andrew Cooper
2019-05-21 13:06 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:54 ` [PATCH v8 26/50] x86emul: support remaining AVX512BW " Jan Beulich
2019-05-21 13:08 ` Andrew Cooper
2019-05-21 13:08 ` [Xen-devel] " Andrew Cooper
2019-05-21 13:34 ` Jan Beulich
2019-05-21 13:34 ` [Xen-devel] " Jan Beulich
2019-05-23 16:10 ` Andrew Cooper
2019-05-23 16:10 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:55 ` [PATCH v8 27/50] x86emul: support AVX512{F, ER} reciprocal insns Jan Beulich
2019-05-23 16:15 ` Andrew Cooper
2019-05-23 16:15 ` [Xen-devel] " Andrew Cooper
2019-05-24 6:43 ` Jan Beulich
2019-05-24 6:43 ` [Xen-devel] " Jan Beulich
2019-05-24 20:48 ` Andrew Cooper
2019-05-24 20:48 ` [Xen-devel] " Andrew Cooper
2019-05-27 8:02 ` Jan Beulich
2019-05-27 8:02 ` [Xen-devel] " Jan Beulich
2019-05-29 10:00 ` Andrew Cooper
2019-05-29 10:00 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:56 ` [PATCH v8 28/50] x86emul: support AVX512F floating point manipulation insns Jan Beulich
2019-05-29 12:51 ` Andrew Cooper
2019-05-29 12:51 ` [Xen-devel] " Andrew Cooper
2019-05-29 13:15 ` Jan Beulich
2019-05-29 13:15 ` [Xen-devel] " Jan Beulich
2019-06-10 14:01 ` Andrew Cooper
2019-06-10 14:03 ` Andrew Cooper
2019-03-15 10:56 ` [PATCH v8 29/50] x86emul: support AVX512DQ " Jan Beulich
2019-06-10 14:06 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:56 ` [PATCH v8 30/50] x86emul: support AVX512{F, _VBMI2} compress/expand insns Jan Beulich
2019-06-10 14:51 ` [Xen-devel] " Andrew Cooper
2019-06-11 10:20 ` Jan Beulich
2019-06-18 16:24 ` Andrew Cooper
2019-06-19 6:38 ` Jan Beulich
2019-03-15 10:58 ` [PATCH v8 31/50] x86emul: support remaining misc AVX512{F, BW} insns Jan Beulich
2019-06-18 16:42 ` [Xen-devel] " Andrew Cooper
2019-06-19 6:44 ` Jan Beulich
2019-03-15 10:58 ` [PATCH v8 32/50] x86emul: support AVX512F gather insns Jan Beulich
2019-06-19 12:05 ` [Xen-devel] " Andrew Cooper
2019-06-19 12:43 ` Jan Beulich
2019-03-15 10:59 ` [PATCH v8 33/50] x86emul: add high register S/G test cases Jan Beulich
2019-06-19 12:07 ` [Xen-devel] " Andrew Cooper
2019-03-15 10:59 ` [PATCH v8 34/50] x86emul: support AVX512F scatter insns Jan Beulich
2019-03-15 11:00 ` [PATCH v8 35/50] x86emul: support AVX512PF insns Jan Beulich
2019-03-15 11:00 ` [PATCH v8 36/50] x86emul: support AVX512CD insns Jan Beulich
2019-06-19 12:13 ` [Xen-devel] " Andrew Cooper
2019-03-15 11:01 ` [PATCH v8 37/50] x86emul: complete support of AVX512_VBMI insns Jan Beulich
2019-06-19 12:16 ` [Xen-devel] " Andrew Cooper
2019-03-15 11:01 ` [PATCH v8 38/50] x86emul: support of AVX512* population count insns Jan Beulich
2019-06-19 12:22 ` [Xen-devel] " Andrew Cooper
2019-06-19 12:48 ` Jan Beulich
2019-03-15 11:02 ` [PATCH v8 39/50] x86emul: support of AVX512_IFMA insns Jan Beulich
2019-06-19 12:23 ` [Xen-devel] " Andrew Cooper
2019-03-15 11:02 ` [PATCH v8 40/50] x86emul: support remaining AVX512_VBMI2 insns Jan Beulich
2019-06-19 12:25 ` [Xen-devel] " Andrew Cooper
2019-03-15 11:04 ` [PATCH v8 41/50] x86emul: support AVX512_4FMAPS insns Jan Beulich
2019-06-19 14:58 ` [Xen-devel] " Andrew Cooper
2019-06-21 6:50 ` Jan Beulich
2019-03-15 11:04 ` [PATCH v8 42/50] x86emul: support AVX512_4VNNIW insns Jan Beulich
2019-06-19 14:58 ` [Xen-devel] " Andrew Cooper
2019-03-15 11:04 ` [PATCH v8 43/50] x86emul: support AVX512_VNNI insns Jan Beulich
2019-06-19 15:01 ` [Xen-devel] " Andrew Cooper
2019-06-21 6:55 ` Jan Beulich
2019-03-15 11:05 ` [PATCH v8 44/50] x86emul: support VPCLMULQDQ insns Jan Beulich
2019-06-21 12:52 ` [Xen-devel] " Andrew Cooper
2019-06-21 13:44 ` Jan Beulich
2019-03-15 11:06 ` [PATCH v8 45/50] x86emul: support VAES insns Jan Beulich
2019-06-21 12:57 ` [Xen-devel] " Andrew Cooper
2019-03-15 11:06 ` [PATCH v8 46/50] x86emul: support GFNI insns Jan Beulich
2019-06-21 13:19 ` [Xen-devel] " Andrew Cooper
2019-06-21 13:33 ` Andrew Cooper
2019-06-21 14:00 ` Jan Beulich
2019-06-21 14:20 ` Andrew Cooper
2019-06-21 15:02 ` Jan Beulich
2019-06-25 6:48 ` Jan Beulich
2019-03-15 11:07 ` [PATCH v8 47/50] x86emul: restore ordering within main switch statement Jan Beulich
2019-06-21 13:20 ` [Xen-devel] " Andrew Cooper
2019-03-15 11:07 ` [PATCH v8 48/50] x86emul: add an AES/VAES test case to the harness Jan Beulich
2019-06-21 13:36 ` [Xen-devel] " Andrew Cooper
2019-06-21 14:04 ` Jan Beulich
2019-06-21 14:20 ` Andrew Cooper
2019-03-15 11:08 ` [PATCH v8 49/50] x86emul: add a SHA " Jan Beulich
2019-06-21 13:51 ` [Xen-devel] " Andrew Cooper
2019-06-21 14:10 ` Jan Beulich
2019-06-21 14:23 ` Andrew Cooper
2019-03-15 11:08 ` [PATCH v8 50/50] x86emul: add a PCLMUL/VPCLMUL " Jan Beulich
2019-06-21 13:58 ` [Xen-devel] " Andrew Cooper
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5BA0E7F402000078001E9537@prv1-mh.provo.novell.com \
--to=jbeulich@suse.com \
--cc=George.Dunlap@eu.citrix.com \
--cc=andrew.cooper3@citrix.com \
--cc=wei.liu2@citrix.com \
--cc=xen-devel@lists.xenproject.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).