From: Paolo Bonzini <pbonzini@redhat.com>
To: qemu-devel@nongnu.org
Cc: richard.henderson@linaro.org
Subject: [PATCH 13/14] target/i386: use higher-precision arithmetic to compute CF
Date: Mon, 28 Oct 2024 16:18:50 +0100 [thread overview]
Message-ID: <20241028151851.376355-14-pbonzini@redhat.com> (raw)
In-Reply-To: <20241028151851.376355-1-pbonzini@redhat.com>
If the operands of the arithmetic instruction fit within a half-register,
it's easiest to use a comparison instruction to compute the carry.
`
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
target/i386/tcg/cc_helper_template.h.inc | 37 ++++++++++++++++++++++++
1 file changed, 37 insertions(+)
diff --git a/target/i386/tcg/cc_helper_template.h.inc b/target/i386/tcg/cc_helper_template.h.inc
index 8af8b8539f9..f29a6dfb77c 100644
--- a/target/i386/tcg/cc_helper_template.h.inc
+++ b/target/i386/tcg/cc_helper_template.h.inc
@@ -22,12 +22,17 @@
#if DATA_BITS == 8
#define SUFFIX b
#define DATA_TYPE uint8_t
+#define WIDER_TYPE uint32_t
#elif DATA_BITS == 16
#define SUFFIX w
#define DATA_TYPE uint16_t
+#define WIDER_TYPE uint32_t
#elif DATA_BITS == 32
#define SUFFIX l
#define DATA_TYPE uint32_t
+#if HOST_LONG_BITS <= 64
+#define WIDER_TYPE uint64_t
+#endif
#elif DATA_BITS == 64
#define SUFFIX q
#define DATA_TYPE uint64_t
@@ -62,9 +67,18 @@ static uint32_t glue(compute_all_adc, SUFFIX)(DATA_TYPE dst, DATA_TYPE src1,
DATA_TYPE src3)
{
uint32_t cf, pf, af, zf, sf, of;
+
+#ifdef WIDER_TYPE
+ WIDER_TYPE src13 = (WIDER_TYPE) src1 + (WIDER_TYPE) src3;
+ DATA_TYPE src2 = dst - src13;
+
+ cf = dst < src13;
+#else
DATA_TYPE src2 = dst - src1 - src3;
cf = (src3 ? dst <= src1 : dst < src1);
+#endif
+
pf = compute_pf(dst);
af = (dst ^ src1 ^ src2) & 0x10;
zf = (dst == 0) << 6;
@@ -76,7 +90,13 @@ static uint32_t glue(compute_all_adc, SUFFIX)(DATA_TYPE dst, DATA_TYPE src1,
static int glue(compute_c_adc, SUFFIX)(DATA_TYPE dst, DATA_TYPE src1,
DATA_TYPE src3)
{
+#ifdef WIDER_TYPE
+ WIDER_TYPE src13 = (WIDER_TYPE) src1 + (WIDER_TYPE) src3;
+
+ return dst < src13;
+#else
return src3 ? dst <= src1 : dst < src1;
+#endif
}
static uint32_t glue(compute_all_sub, SUFFIX)(DATA_TYPE dst, DATA_TYPE src2)
@@ -104,9 +124,18 @@ static uint32_t glue(compute_all_sbb, SUFFIX)(DATA_TYPE dst, DATA_TYPE src2,
DATA_TYPE src3)
{
uint32_t cf, pf, af, zf, sf, of;
+
+#ifdef WIDER_TYPE
+ WIDER_TYPE src23 = (WIDER_TYPE) src2 + (WIDER_TYPE) src3;
+ DATA_TYPE src1 = dst + src23;
+
+ cf = src1 < src23;
+#else
DATA_TYPE src1 = dst + src2 + src3;
cf = (src3 ? src1 <= src2 : src1 < src2);
+#endif
+
pf = compute_pf(dst);
af = (dst ^ src1 ^ src2) & 0x10;
zf = (dst == 0) << 6;
@@ -118,9 +147,16 @@ static uint32_t glue(compute_all_sbb, SUFFIX)(DATA_TYPE dst, DATA_TYPE src2,
static int glue(compute_c_sbb, SUFFIX)(DATA_TYPE dst, DATA_TYPE src2,
DATA_TYPE src3)
{
+#ifdef WIDER_TYPE
+ WIDER_TYPE src23 = (WIDER_TYPE) src2 + (WIDER_TYPE) src3;
+ DATA_TYPE src1 = dst + src23;
+
+ return src1 < src23;
+#else
DATA_TYPE src1 = dst + src2 + src3;
return (src3 ? src1 <= src2 : src1 < src2);
+#endif
}
static uint32_t glue(compute_all_logic, SUFFIX)(DATA_TYPE dst, DATA_TYPE src1)
@@ -258,3 +294,4 @@ static int glue(compute_c_blsi, SUFFIX)(DATA_TYPE dst, DATA_TYPE src1)
#undef DATA_TYPE
#undef DATA_MASK
#undef SUFFIX
+#undef WIDER_TYPE
--
2.47.0
next prev parent reply other threads:[~2024-10-28 15:21 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-10-28 15:18 [PATCH v2 00/14] target/i386: miscellaneous flags improvements Paolo Bonzini
2024-10-28 15:18 ` [PATCH 01/14] target/i386: use tcg_gen_ext_tl when applicable Paolo Bonzini
2024-10-28 15:18 ` [PATCH 02/14] target/i386: Tidy cc_op_str usage Paolo Bonzini
2024-10-28 15:18 ` [PATCH 03/14] target/i386: remove CC_OP_CLR Paolo Bonzini
2024-10-28 15:18 ` [PATCH 04/14] target/i386: Rearrange CCOp Paolo Bonzini
2024-10-28 15:18 ` [PATCH 05/14] target/i386: Introduce cc_op_size Paolo Bonzini
2024-10-28 15:18 ` [PATCH 06/14] target/i386: Wrap cc_op_live with a validity check Paolo Bonzini
2024-10-28 15:18 ` [PATCH 07/14] target/i386: optimize computation of ZF from CC_OP_DYNAMIC Paolo Bonzini
2024-10-29 15:43 ` Richard Henderson
2024-10-28 15:18 ` [PATCH 08/14] target/i386: optimize TEST+Jxx sequences Paolo Bonzini
2024-10-28 15:18 ` [PATCH 09/14] target/i386: add a few more trivial CCPrepare cases Paolo Bonzini
2024-10-28 15:18 ` [PATCH 10/14] target/i386: add a note about gen_jcc1 Paolo Bonzini
2024-10-29 15:44 ` Richard Henderson
2024-10-28 15:18 ` [PATCH 11/14] target/i386: make flag variables unsigned Paolo Bonzini
2024-10-28 15:18 ` [PATCH 12/14] target/i386: use compiler builtin to compute PF Paolo Bonzini
2024-10-29 15:45 ` Richard Henderson
2024-10-28 15:18 ` Paolo Bonzini [this message]
2024-10-29 15:46 ` [PATCH 13/14] target/i386: use higher-precision arithmetic to compute CF Richard Henderson
2024-10-28 15:18 ` [PATCH 14/14] target/i386: use + to put flags together Paolo Bonzini
2024-10-29 15:46 ` Richard Henderson
-- strict thread matches above, loose matches on Subject: below --
2024-10-20 15:53 [PATCH 00/14] target/i386: miscellaneous flags improvements Paolo Bonzini
2024-10-20 15:53 ` [PATCH 13/14] target/i386: use higher-precision arithmetic to compute CF Paolo Bonzini
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=20241028151851.376355-14-pbonzini@redhat.com \
--to=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=richard.henderson@linaro.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).