[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [PATCH 51/81] tcg/aarch64: Fix tcg_out_movi
From: |
Michael Roth |
Subject: |
[Qemu-devel] [PATCH 51/81] tcg/aarch64: Fix tcg_out_movi |
Date: |
Mon, 20 Mar 2017 18:08:15 -0500 |
From: Richard Henderson <address@hidden>
There were some patterns, like 0x0000_ffff_ffff_00ff, for which we
would select to begin a multi-insn sequence with MOVN, but would
fail to set the 0x0000 lane back from 0xffff.
Signed-off-by: Richard Henderson <address@hidden>
Message-Id: <address@hidden>
(cherry picked from commit 8cf9a3d3f7a4b95f33e0bda5416b9c93ec887dd3)
Signed-off-by: Michael Roth <address@hidden>
---
tcg/aarch64/tcg-target.inc.c | 57 +++++++++++++++++++-------------------------
1 file changed, 24 insertions(+), 33 deletions(-)
diff --git a/tcg/aarch64/tcg-target.inc.c b/tcg/aarch64/tcg-target.inc.c
index 6c68681..2d7cc35 100644
--- a/tcg/aarch64/tcg-target.inc.c
+++ b/tcg/aarch64/tcg-target.inc.c
@@ -581,11 +581,9 @@ static void tcg_out_logicali(TCGContext *s, AArch64Insn
insn, TCGType ext,
static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg rd,
tcg_target_long value)
{
- AArch64Insn insn;
int i, wantinv, shift;
tcg_target_long svalue = value;
tcg_target_long ivalue = ~value;
- tcg_target_long imask;
/* For 32-bit values, discard potential garbage in value. For 64-bit
values within [2**31, 2**32-1], we can create smaller sequences by
@@ -631,42 +629,35 @@ static void tcg_out_movi(TCGContext *s, TCGType type,
TCGReg rd,
/* Would it take fewer insns to begin with MOVN? For the value and its
inverse, count the number of 16-bit lanes that are 0. */
- for (i = wantinv = imask = 0; i < 64; i += 16) {
+ for (i = wantinv = 0; i < 64; i += 16) {
tcg_target_long mask = 0xffffull << i;
- if ((value & mask) == 0) {
- wantinv -= 1;
- }
- if ((ivalue & mask) == 0) {
- wantinv += 1;
- imask |= mask;
- }
+ wantinv -= ((value & mask) == 0);
+ wantinv += ((ivalue & mask) == 0);
}
- /* If we had more 0xffff than 0x0000, invert VALUE and use MOVN. */
- insn = I3405_MOVZ;
- if (wantinv > 0) {
- value = ivalue;
- insn = I3405_MOVN;
- }
-
- /* Find the lowest lane that is not 0x0000. */
- shift = ctz64(value) & (63 & -16);
- tcg_out_insn_3405(s, insn, type, rd, value >> shift, shift);
-
- if (wantinv > 0) {
- /* Re-invert the value, so MOVK sees non-inverted bits. */
- value = ~value;
- /* Clear out all the 0xffff lanes. */
- value ^= imask;
- }
- /* Clear out the lane that we just set. */
- value &= ~(0xffffUL << shift);
-
- /* Iterate until all lanes have been set, and thus cleared from VALUE. */
- while (value) {
+ if (wantinv <= 0) {
+ /* Find the lowest lane that is not 0x0000. */
shift = ctz64(value) & (63 & -16);
- tcg_out_insn(s, 3405, MOVK, type, rd, value >> shift, shift);
+ tcg_out_insn(s, 3405, MOVZ, type, rd, value >> shift, shift);
+ /* Clear out the lane that we just set. */
value &= ~(0xffffUL << shift);
+ /* Iterate until all non-zero lanes have been processed. */
+ while (value) {
+ shift = ctz64(value) & (63 & -16);
+ tcg_out_insn(s, 3405, MOVK, type, rd, value >> shift, shift);
+ value &= ~(0xffffUL << shift);
+ }
+ } else {
+ /* Like above, but with the inverted value and MOVN to start. */
+ shift = ctz64(ivalue) & (63 & -16);
+ tcg_out_insn(s, 3405, MOVN, type, rd, ivalue >> shift, shift);
+ ivalue &= ~(0xffffUL << shift);
+ while (ivalue) {
+ shift = ctz64(ivalue) & (63 & -16);
+ /* Provide MOVK with the non-inverted value. */
+ tcg_out_insn(s, 3405, MOVK, type, rd, ~(ivalue >> shift), shift);
+ ivalue &= ~(0xffffUL << shift);
+ }
}
}
--
2.7.4
- [Qemu-devel] [PATCH 57/81] cpu-exec: fix icount out-of-bounds access, (continued)
- [Qemu-devel] [PATCH 57/81] cpu-exec: fix icount out-of-bounds access, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 54/81] block/iscsi: avoid data corruption with cache=writeback, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 58/81] ahci: advertise HOST_CAP_64, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 62/81] block/nfs: fix naming of runtime opts, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 65/81] target-ppc, tcg: fix usermode segfault with pthread_create(), Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 67/81] target/sparc: Restore ldstub of odd asis, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 68/81] apic: reset apic_delivered global variable on machine reset, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 41/81] ui/gtk: fix crash at startup when no console is available, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 44/81] qemu-thread: fix qemu_thread_set_name() race in qemu_thread_create(), Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 49/81] char: fix ctrl-a b not working, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 51/81] tcg/aarch64: Fix tcg_out_movi,
Michael Roth <=
- [Qemu-devel] [PATCH 52/81] ui: use evdev keymap when running under wayland, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 56/81] cirrus: fix oob access issue (CVE-2017-2615), Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 70/81] qga: ignore EBUSY when freezing a filesystem, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 75/81] NetRxPkt: Do not try to pull more data than present, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 06/81] 9pfs: local: open/opendir: don't follow symlinks, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 59/81] target/s390x: use "qemu" cpu model in user mode, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 74/81] NetRxPkt: Fix memory corruption on VLAN header stripping, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 69/81] target-i386: correctly propagate retaddr into SVM helpers, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 08/81] 9pfs: local: llistxattr: don't follow symlinks, Michael Roth, 2017/03/20
- [Qemu-devel] [PATCH 61/81] block/nfs: fix NULL pointer dereference in URI parsing, Michael Roth, 2017/03/20