From: Chinmay Rath <rathc@linux.ibm.com>
To: qemu-devel@nongnu.org, qemu-ppc@nongnu.org,
harshpb@linux.ibm.com, milesg@linux.ibm.com
Cc: npiggin@gmail.com, richard.henderson@linaro.org,
vishalc@linux.ibm.com, tshah@linux.ibm.com,
shivangu@linux.ibm.com, ojaswin@linux.ibm.com,
aboorvad@linux.ibm.com, amachhiw@linux.ibm.com, sv@linux.ibm.com,
shivani@linux.ibm.com, mkchauras@gmail.com, uverma@linux.ibm.com,
nikhilks@linux.ibm.com, Chinmay Rath <rathc@linux.ibm.com>
Subject: [PATCH 17/29] target/ppc: Move wait instruction to decodetree
Date: Thu, 20 Aug 2026 19:30:46 +0530 [thread overview]
Message-ID: <20260820140114.327023-18-rathc@linux.ibm.com> (raw)
In-Reply-To: <20260820140114.327023-1-rathc@linux.ibm.com>
From: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Implement wait instruction using decodetree with
proper handling of WC and PL fields across different ISA
versions. Since the opcode changed between ISA 2.x and 3.x
we have defined 2 different decodetree instructions that
are handled by the same helper.
The changes have been tested by comparing the tcg generated
before and after the change for wait instruction. The test
was repeated for 2 types of machines:
Power10 - which should use the newer wait implementation
e500mc - which should use older PPC_WAIT implementation
Both generated the same tcg ops with and without the change.
Further, with the changes we were able to boot into Fedora
distro as well.
Signed-off-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Signed-off-by: Chinmay Rath <rathc@linux.ibm.com>
---
target/ppc/insn32.decode | 15 +++
target/ppc/translate.c | 91 -----------------
target/ppc/translate/storage-ctrl-impl.c.inc | 100 +++++++++++++++++++
3 files changed, 115 insertions(+), 91 deletions(-)
diff --git a/target/ppc/insn32.decode b/target/ppc/insn32.decode
index 67fb181542..3b8387b60c 100644
--- a/target/ppc/insn32.decode
+++ b/target/ppc/insn32.decode
@@ -1480,6 +1480,21 @@ ISYNC 010011 ----- ----- ----- 0010010110 -
@XL_bfa ...... bf:3 .. bfa:3 .. ..... ..... ..... .
MCRF 010011 ... -- ... -- ----- 00000 00000 - @XL_bfa
+# Wait Instructions
+
+# We can use the same X_wait argument set for both WAIT and older ISA2.x
+# compliant WAIT_2_x. Depending on the version, we may or may not use the
+# arguments. For more information, check do_wait() helper.
+
+&X_wait wc pl
+@X_wait ...... ..- wc:2 --- pl:2 ----- .......... - &X_wait
+
+# According to ISA v3.1, bit 6 and 7 are implementation dependent and "unless
+# the intention is to use the implementation-dependent field, these bits must be
+# coded zero". Hence, we code them to 0.
+WAIT 011111 00-.. ---.. ----- 0000011110 - @X_wait
+WAIT_ISA_2_X 011111 ---.. ---.. ----- 0000111110 - @X_wait
+
# Branch History Rolling Buffer (BHRB) Instructions
&XFX_bhrbe rt bhrbe
diff --git a/target/ppc/translate.c b/target/ppc/translate.c
index f7c7f53d82..2d5157080c 100644
--- a/target/ppc/translate.c
+++ b/target/ppc/translate.c
@@ -2727,95 +2727,6 @@ static inline void gen_check_tlb_flush(DisasContext *ctx, bool global)
static inline void gen_check_tlb_flush(DisasContext *ctx, bool global) { }
#endif
-/* wait */
-static void gen_wait(DisasContext *ctx)
-{
- uint32_t wc;
-
- if (ctx->insns_flags & PPC_WAIT) {
- /* v2.03-v2.07 define an older incompatible 'wait' encoding. */
-
- if (ctx->insns_flags2 & PPC2_PM_ISA206) {
- /* v2.06 introduced the WC field. WC > 0 may be treated as no-op. */
- wc = WC(ctx->opcode);
- } else {
- wc = 0;
- }
-
- } else if (ctx->insns_flags2 & PPC2_ISA300) {
- /* v3.0 defines a new 'wait' encoding. */
- wc = WC(ctx->opcode);
- if (ctx->insns_flags2 & PPC2_ISA310) {
- uint32_t pl = PL(ctx->opcode);
-
- /* WC 1,2 may be treated as no-op. WC 3 is reserved. */
- if (wc == 3) {
- gen_invalid(ctx);
- return;
- }
-
- /* PL 1-3 are reserved. If WC=2 then the insn is treated as noop. */
- if (pl > 0 && wc != 2) {
- gen_invalid(ctx);
- return;
- }
-
- } else { /* ISA300 */
- /* WC 1-3 are reserved */
- if (wc > 0) {
- gen_invalid(ctx);
- return;
- }
- }
-
- } else {
- warn_report("wait instruction decoded with wrong ISA flags.");
- gen_invalid(ctx);
- return;
- }
-
- /*
- * wait without WC field or with WC=0 waits for an exception / interrupt
- * to occur.
- */
- if (wc == 0) {
- TCGv_i32 t0 = tcg_constant_i32(1);
- tcg_gen_st_i32(t0, tcg_env,
- -offsetof(PowerPCCPU, env) + offsetof(CPUState, halted));
- /* Stop translation, as the CPU is supposed to sleep from now */
- gen_exception_nip(ctx, EXCP_HLT, ctx->base.pc_next);
- }
-
- /*
- * Other wait types must not just wait until an exception occurs because
- * ignoring their other wake-up conditions could cause a hang.
- *
- * For v2.06 and 2.07, wc=1,2,3 are architected but may be implemented as
- * no-ops.
- *
- * wc=1 and wc=3 explicitly allow the instruction to be treated as a no-op.
- *
- * wc=2 waits for an implementation-specific condition, such could be
- * always true, so it can be implemented as a no-op.
- *
- * For v3.1, wc=1,2 are architected but may be implemented as no-ops.
- *
- * wc=1 (waitrsv) waits for an exception or a reservation to be lost.
- * Reservation-loss may have implementation-specific conditions, so it
- * can be implemented as a no-op.
- *
- * wc=2 waits for an exception or an amount of time to pass. This
- * amount is implementation-specific so it can be implemented as a
- * no-op.
- *
- * ISA v3.1 allows for execution to resume "in the rare case of
- * an implementation-dependent event", so in any case software must
- * not depend on the architected resumption condition to become
- * true, so no-op implementations should be architecturally correct
- * (if suboptimal).
- */
-}
-
#if defined(TARGET_PPC64)
static void gen_doze(DisasContext *ctx)
{
@@ -5197,8 +5108,6 @@ GEN_HANDLER(lswx, 0x1F, 0x15, 0x10, 0x00000001, PPC_STRING),
GEN_HANDLER(stswi, 0x1F, 0x15, 0x16, 0x00000001, PPC_STRING),
GEN_HANDLER(stswx, 0x1F, 0x15, 0x14, 0x00000001, PPC_STRING),
/* ISA v3.0 changed the extended opcode from 62 to 30 */
-GEN_HANDLER(wait, 0x1F, 0x1E, 0x01, 0x039FF801, PPC_WAIT),
-GEN_HANDLER_E(wait, 0x1F, 0x1E, 0x00, 0x039CF801, PPC_NONE, PPC2_ISA300),
GEN_HANDLER(rfi, 0x13, 0x12, 0x01, 0x03FF8001, PPC_FLOW),
#if defined(TARGET_PPC64)
GEN_HANDLER(rfid, 0x13, 0x12, 0x00, 0x03FF8001, PPC_64B),
diff --git a/target/ppc/translate/storage-ctrl-impl.c.inc b/target/ppc/translate/storage-ctrl-impl.c.inc
index 943229c438..21c7999d5d 100644
--- a/target/ppc/translate/storage-ctrl-impl.c.inc
+++ b/target/ppc/translate/storage-ctrl-impl.c.inc
@@ -270,3 +270,103 @@ static bool do_tlbie(DisasContext *ctx, arg_X_tlbie *a, bool local)
TRANS_FLAGS(MEM_TLBIE, TLBIE, do_tlbie, false)
TRANS_FLAGS(MEM_TLBIE, TLBIEL, do_tlbie, true)
+
+/*
+ * Decodetree populates a->wc and a->pl based on ISA v3.1, however they may
+ * or may not be used based on the ISA:
+ * For ISA < 2.06 - both wc and pl are ignored
+ * For ISA == 2.06, 2.07, 3.0 - wc is used and pl is ignored
+ * For ISA 3.1 - both wc and pl are used
+ */
+static bool do_wait(DisasContext *ctx, arg_X_wait *a)
+{
+ uint32_t wc;
+
+ if (ctx->insns_flags & PPC_WAIT) {
+ /* v2.03-v2.07 define an older incompatible 'wait' encoding. */
+
+ if (ctx->insns_flags2 & PPC2_PM_ISA206) {
+ /* v2.06 introduced the WC field. WC > 0 may be treated as no-op. */
+ wc = a->wc;
+ } else {
+ wc = 0;
+ }
+
+ } else if (ctx->insns_flags2 & PPC2_ISA300) {
+ /* v3.0 defines a new 'wait' encoding. */
+ wc = a->wc;
+ if (ctx->insns_flags2 & PPC2_ISA310) {
+ uint32_t pl = a->pl;
+
+ /* WC 1,2 may be treated as no-op. WC 3 is reserved. */
+ if (wc == 3) {
+ gen_invalid(ctx);
+ return true;
+ }
+
+ /* PL 1-3 are reserved. If WC=2 then the insn is treated as noop. */
+ if (pl > 0 && wc != 2) {
+ gen_invalid(ctx);
+ return true;
+ }
+
+ } else { /* ISA300 */
+ /* WC 1-3 are reserved */
+ if (wc > 0) {
+ gen_invalid(ctx);
+ return true;
+ }
+ }
+
+ } else {
+ warn_report("wait instruction decoded with wrong ISA flags.");
+ gen_invalid(ctx);
+ return true;
+ }
+
+ /*
+ * wait without WC field or with WC=0 waits for an exception / interrupt
+ * to occur.
+ */
+ if (wc == 0) {
+ TCGv_i32 t0 = tcg_constant_i32(1);
+ tcg_gen_st_i32(t0, tcg_env,
+ -offsetof(PowerPCCPU, env) + offsetof(CPUState, halted));
+ /* Stop translation, as the CPU is supposed to sleep from now */
+ gen_exception_nip(ctx, EXCP_HLT, ctx->base.pc_next);
+ }
+
+ /*
+ * Other wait types must not just wait until an exception occurs because
+ * ignoring their other wake-up conditions could cause a hang.
+ *
+ * For v2.06 and 2.07, wc=1,2,3 are architected but may be implemented as
+ * no-ops.
+ *
+ * wc=1 and wc=3 explicitly allow the instruction to be treated as a no-op.
+ *
+ * wc=2 waits for an implementation-specific condition, such could be
+ * always true, so it can be implemented as a no-op.
+ *
+ * For v3.1, wc=1,2 are architected but may be implemented as no-ops.
+ *
+ * wc=1 (waitrsv) waits for an exception or a reservation to be lost.
+ * Reservation-loss may have implementation-specific conditions, so it
+ * can be implemented as a no-op.
+ *
+ * wc=2 waits for an exception or an amount of time to pass. This
+ * amount is implementation-specific so it can be implemented as a
+ * no-op.
+ *
+ * ISA v3.1 allows for execution to resume "in the rare case of
+ * an implementation-dependent event", so in any case software must
+ * not depend on the architected resumption condition to become
+ * true, so no-op implementations should be architecturally correct
+ * (if suboptimal).
+ */
+ return true;
+}
+
+TRANS_FLAGS(WAIT, WAIT_ISA_2_X, do_wait)
+TRANS_FLAGS2(ISA300, WAIT, do_wait)
+
--
2.55.0
next prev parent reply other threads:[~2026-08-20 14:06 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-20 14:00 [PATCH 00/29] target/ppc: Moving instructions to decodetree specification Chinmay Rath
2026-08-20 14:00 ` [PATCH 01/29] target/ppc: Migrate extswsli to decodetree Chinmay Rath
2026-08-20 14:00 ` [PATCH 02/29] target/ppc: Migrate atomic loads " Chinmay Rath
2026-08-20 14:00 ` [PATCH 03/29] target/ppc: Convert cache instructions " Chinmay Rath
2026-08-20 14:00 ` [PATCH 04/29] target/ppc: Move vector merge " Chinmay Rath
2026-08-20 14:00 ` [PATCH 05/29] target/ppc: Move vector pack " Chinmay Rath
2026-08-20 14:00 ` [PATCH 06/29] target/ppc: Move st{b, h, w, d, q}cx " Chinmay Rath
2026-08-20 14:00 ` [PATCH 07/29] target/ppc: convert slw, srw instruction via decode spec Chinmay Rath
2026-08-20 14:00 ` [PATCH 08/29] target/ppc: convert sraw[i] " Chinmay Rath
2026-08-20 14:00 ` [PATCH 09/29] target/ppc: Convert mcrf to decode tree Chinmay Rath
2026-08-20 14:00 ` [PATCH 10/29] target/ppc: Move fixed-point Shift insns to decodetree Chinmay Rath
2026-08-20 14:00 ` [PATCH 11/29] target/ppc: Move fixed-point byte-reversal store " Chinmay Rath
2026-08-20 14:00 ` [PATCH 12/29] target/ppc: Move GPR atomic load/store instructions " Chinmay Rath
2026-08-20 14:00 ` [PATCH 13/29] target/ppc: Move isync instruction " Chinmay Rath
2026-08-20 14:00 ` [PATCH 14/29] target/ppc: Convert b{a, l, la} to decode tree Chinmay Rath
2026-08-20 14:00 ` [PATCH 15/29] target/ppc: move various conditional branch insns to decodetree Chinmay Rath
2026-08-20 14:00 ` [PATCH 16/29] target/ppc: Fix TRANS* macro variadic arguments handling Chinmay Rath
2026-08-20 14:00 ` Chinmay Rath [this message]
2026-08-20 14:00 ` [PATCH 18/29] target/ppc: Move sleep & friends to decodetree Chinmay Rath
2026-08-20 14:00 ` [PATCH 19/29] target/ppc: Refactor sleep and its variants to use a common helper Chinmay Rath
2026-08-20 15:15 ` Miles Glenn
2026-08-20 14:00 ` [PATCH 20/29] target/ppc: Move Condition Register access instructions to decodetree Chinmay Rath
2026-08-20 14:00 ` [PATCH 21/29] target/ppc: Move Condition Register logical " Chinmay Rath
2026-08-20 14:00 ` [PATCH 22/29] target/ppc: make do_ea_calc_ra available for 32 bit builds Chinmay Rath
2026-08-20 14:00 ` [PATCH 23/29] target/ppc: Move Fixed-Point Load/Store String instructions to decodetree Chinmay Rath
2026-08-20 14:00 ` [PATCH 24/29] target/ppc: Move VMX integer arithmetic and BCD " Chinmay Rath
2026-08-20 14:00 ` [PATCH 25/29] target/ppc: Move rlwimi, rlwinm " Chinmay Rath
2026-08-20 14:00 ` [PATCH 26/29] target/ppc: Move lmw, stmw " Chinmay Rath
2026-08-20 14:00 ` [PATCH 27/29] target/ppc: Move mfmsr, mtmsr[d] " Chinmay Rath
2026-08-20 14:00 ` [PATCH 28/29] target/ppc: Move byte-reverse " Chinmay Rath
2026-08-20 14:00 ` [PATCH 29/29] target/ppc: Move system call and rfi " Chinmay Rath
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=20260820140114.327023-18-rathc@linux.ibm.com \
--to=rathc@linux.ibm.com \
--cc=aboorvad@linux.ibm.com \
--cc=amachhiw@linux.ibm.com \
--cc=harshpb@linux.ibm.com \
--cc=milesg@linux.ibm.com \
--cc=mkchauras@gmail.com \
--cc=nikhilks@linux.ibm.com \
--cc=npiggin@gmail.com \
--cc=ojaswin@linux.ibm.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
--cc=richard.henderson@linaro.org \
--cc=shivangu@linux.ibm.com \
--cc=shivani@linux.ibm.com \
--cc=sv@linux.ibm.com \
--cc=tshah@linux.ibm.com \
--cc=uverma@linux.ibm.com \
--cc=vishalc@linux.ibm.com \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.