public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r14-10099] i386: Avoid =&r, r, r andn double-word alternative for ia32 [PR114810]
@ 2024-04-23 21:31 Jakub Jelinek
0 siblings, 0 replies; only message in thread
From: Jakub Jelinek @ 2024-04-23 21:31 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:628c2221d38715a64f828e3635317293d150e001
commit r14-10099-g628c2221d38715a64f828e3635317293d150e001
Author: Jakub Jelinek <jakub@redhat.com>
Date: Tue Apr 23 23:30:27 2024 +0200
i386: Avoid =&r,r,r andn double-word alternative for ia32 [PR114810]
As discussed in the PR, on ia32 with its 8 GPRs, where 1 is always fixed
and other 2 often are as well having an alternative which needs 3
double-word registers is just too much for RA.
The following patch splits that alternative into two, one with o is used
even on ia32, but one with the 3x r is used just for -m64/-mx32.
Tried to reduce the testcase further, but it wasn't easily possible.
2024-04-23 Jakub Jelinek <jakub@redhat.com>
PR target/114810
* config/i386/i386.md (*andn<dwi>3_doubleword_bmi): Split the =&r,r,ro
alternative into =&r,r,r enabled only for x64 and =&r,r,o.
* g++.target/i386/pr114810.C: New test.
Diff:
---
gcc/config/i386/i386.md | 9 +-
gcc/testsuite/g++.target/i386/pr114810.C | 861 +++++++++++++++++++++++++++++++
2 files changed, 866 insertions(+), 4 deletions(-)
diff --git a/gcc/config/i386/i386.md b/gcc/config/i386/i386.md
index adab1ef9e04..80e64c603eb 100644
--- a/gcc/config/i386/i386.md
+++ b/gcc/config/i386/i386.md
@@ -12482,10 +12482,10 @@
})
(define_insn_and_split "*andn<dwi>3_doubleword_bmi"
- [(set (match_operand:<DWI> 0 "register_operand" "=&r,r,r")
+ [(set (match_operand:<DWI> 0 "register_operand" "=&r,&r,r,r")
(and:<DWI>
- (not:<DWI> (match_operand:<DWI> 1 "register_operand" "r,0,r"))
- (match_operand:<DWI> 2 "nonimmediate_operand" "ro,ro,0")))
+ (not:<DWI> (match_operand:<DWI> 1 "register_operand" "r,r,0,r"))
+ (match_operand:<DWI> 2 "nonimmediate_operand" "r,o,ro,0")))
(clobber (reg:CC FLAGS_REG))]
"TARGET_BMI"
"#"
@@ -12496,7 +12496,8 @@
(parallel [(set (match_dup 3)
(and:DWIH (not:DWIH (match_dup 4)) (match_dup 5)))
(clobber (reg:CC FLAGS_REG))])]
- "split_double_mode (<DWI>mode, &operands[0], 3, &operands[0], &operands[3]);")
+ "split_double_mode (<DWI>mode, &operands[0], 3, &operands[0], &operands[3]);"
+ [(set_attr "isa" "x64,*,*,*")])
(define_insn_and_split "*andn<mode>3_doubleword"
[(set (match_operand:DWI 0 "register_operand")
diff --git a/gcc/testsuite/g++.target/i386/pr114810.C b/gcc/testsuite/g++.target/i386/pr114810.C
new file mode 100644
index 00000000000..a7b3643ca21
--- /dev/null
+++ b/gcc/testsuite/g++.target/i386/pr114810.C
@@ -0,0 +1,861 @@
+// PR target/114810
+// { dg-do compile { target { { { *-*-linux* } && ia32 } && c++17 } } }
+// { dg-options "-mstackrealign -O2 -mbmi -fno-exceptions -fno-plt -march=x86-64 -w" }
+// { dg-additional-options "-fpie" { target pie } }
+
+enum E1 { a, dp, b, jm, c, dq, d, mj, e, dr, f, jn, h, dt, j, nt, l, du, m, jo, n, dv, o, mk, p, dw, q, jp, s, dx, t, ol, u, dy, v, jq, w };
+enum dz { x, ml, y };
+struct ea { short g; } z, jr;
+long long aa;
+struct eb { ea ab; ea dp[]; };
+enum ac { };
+typedef enum { } nu;
+struct ad { ac k; };
+unsigned ec (long);
+struct ae;
+int js (ae);
+unsigned af ();
+struct ed;
+template < int ag > struct ee { using ah = ed[ag]; };
+template < int ag > struct array { typename ee < ag >::ah ai; ed & operator[] (int aj) { return ai[aj]; } };
+struct { void dp (...); } ak;
+void ef (int);
+template < typename al > struct jt { al & operator[] (short); };
+struct am { void operator= (bool); };
+struct an { am operator[] (unsigned); };
+template < typename, unsigned, unsigned >using eg = an;
+struct ao;
+struct ae { ae (ao *); };
+struct mm { mm (); mm (int); };
+enum ap { };
+enum eh { };
+bool aq, ju, ar, ei, nv, as, ej, at;
+struct jv
+{
+ jv (eh au):dp (au) {}
+ jv ();
+ operator eh ();
+ unsigned av ()
+ {
+ aq = dp & 7;
+ return dp * (aq ? : 4);
+ }
+ unsigned ek ()
+ {
+ int aw;
+ bool mn = dp & 7;
+ aw = dp * (mn ? : 4);
+ return aw + 3 >> 2;
+ }
+ eh dp;
+} ax, el, ay, jw, az, em, ba, om;
+struct ed
+{
+ ed ():bb (), dp () {}
+ int bc () { return bb; }
+ jv en () { return (eh) dp; }
+ unsigned ek ()
+ {
+ jv bd;
+ bd = (eh) dp;
+ return bd.ek ();
+ }
+ ap jx ();
+ unsigned bb:24;
+ int dp:8;
+};
+struct be { short dp = 0; } bf, eo;
+struct bg
+{
+ bg ();
+ bg (ed r)
+ {
+ dp.bh = r;
+ if (r.bc ())
+ mo = true;
+ else
+ bi = true;
+ }
+ static bg ep (int);
+ bg (be);
+ struct { ed bh; } dp;
+ union { char mo:1; char bi:1; short bj = 0; };
+} jy, bk, eq, bl, mp, bm, er;
+struct bn
+{
+ explicit bn (ed bo):bh (bo) {}
+ ed dp ();
+ ed bh;
+ be es;
+ char bj = 0;
+};
+struct bp
+{
+ eg < int, 6, 4 > dp;
+};
+jt < bg > bq;
+jt < bn > definitions;
+struct ao
+{
+ bp & br ();
+};
+enum jz:short;
+template < typename > using bs = ae;
+ao *et ();
+short bt, nw;
+struct bu
+{
+ int dp;
+};
+dz bv;
+unsigned eu;
+struct bw
+{
+ ac k;
+ unsigned dp;
+} *bx;
+bool ka ();
+struct by
+{
+ bool dp;
+};
+typedef enum
+{ bz, ev } ca;
+typedef enum
+{
+ mq, cb, ew, cc, kb, cd, ex, ce
+} on;
+typedef struct cf
+{
+ on jx;
+ char dp;
+ char cg;
+} kc;
+struct ch
+{
+ kc *dp;
+};
+typedef enum
+{
+ ci, ey, cj, mr, ck, ez, cl, kd, cm, fa, cn, nx, co, fb, cp, ke, cq, fc,
+ cr, ms, cs, fd, ct, kf, cu, fe, cv, os, cw, ff, cx, kg, cy, fg, cz, mt, da
+} fh;
+typedef struct { cf db; fh dp; kc dc; ch kh[]; } dd;
+nu fi ();
+typedef enum
+{ de, ny } fj;
+typedef struct { fj jx; } df;
+typedef struct { by dp; } dg;
+dg *ki;
+struct dh
+{
+ ad *dp;
+ eb *args;
+ bw *di;
+ bu *block;
+ struct
+ {
+ struct
+ {
+ bool fk;
+ } dj;
+ bool dp;
+ bool dk;
+ } mu;
+ ed dl[84];
+};
+ed fl (dh *, ea);
+enum dm
+{ };
+dm kj ();
+enum
+{ dn };
+struct fm
+{
+ operator ed ();
+ operator bg ();
+ bn dc ();
+ ao *operator-> ();
+};
+struct oo
+{
+ oo (ed);
+ oo (bg);
+ oo (fm);
+};
+struct fn
+{
+ enum dp
+ {
+ kk, fo, mv, fp, kl, fq
+ };
+ jv lm;
+ fn (bw *, bu *);
+ fn fr ();
+ ed bo (jv);
+ bn dc ();
+ bn dc (be);
+ ed km (oo);
+ fm copy (bn, oo);
+ fm fs ();
+ fm nz (bn, oo, oo, oo);
+ void ft (oo);
+ void ft (E1, bn, oo);
+ fm ft (E1, bn, oo, oo);
+ void ft (E1, bn, oo, oo, oo);
+ void ft (E1, bn, bn, oo);
+ void kn (E1, bn, oo);
+ fm kn (dp, bn, oo);
+ fm kn (dp, bn, bn, oo);
+ fm fu (E1, bn, oo, oo);
+ fm fu (E1, bn, bn, oo, oo);
+ fm fu (dp, bn, bn, oo, oo);
+ fm mw (E1, bn, int);
+ fm fv (E1, bn, oo, oo);
+ fm fv (dp, bn, oo, oo);
+ void dp (E1, bn, mm, bool = false, bool = false, bool = false);
+ void dp (E1, bn, oo, oo, mm = false, bool = false, bool = false);
+ fm ds (E1, bn, oo, short, char = 0, bool = false);
+ fm ds (E1, bn, oo, oo, short = 0, char = 0, bool = false);
+ fm ko (E1, bn, oo, oo);
+ fm ko (E1, bn, oo, oo, oo);
+ fm fw (E1, bn, oo, oo);
+ fm dp (E1, bn, oo, oo, oo);
+ fm oy (E1, bn, oo, short, char, char, bool, bool);
+ fm fx (E1, bn, oo, oo, oo);
+};
+int kp, fy, mx, fz, kq, ga, oa, gb;
+int *kr, *gc, *my;
+long gd;
+ca ks;
+namespace
+{
+ void ge (dh *, char *, unsigned, cf *, char *);
+ ed kt (dh *, kc *);
+ ed gf (dh *, ed, bg = bg (), bg = bg ());
+ void mz (bool);
+ ed gg (dh *, fn, ed, ed);
+ ed ku (dh *, fn, ed, unsigned, bool);
+ ed gh (dh *, ed);
+ void op (dh *, ed, int, ed);
+ ed op (dh *, ed, int, jv);
+ void gi (dh *, ed, unsigned);
+ void kv (dh *, ed, bg, ed);
+ ed gj (dh *, ed, ed = ed ());
+ void na (dh *, int *);
+ void gk (dh *, ed, ed);
+ ed kw (dh *);
+ bool gl (dh *);
+ void ob (dh *, dd *);
+ void gm (dh * ctx, dd * db)
+ {
+ fn bld (ctx->di, ctx->block);
+ ed dst = kt (ctx, &db->dc), dp = kt (ctx, db->kh[0].dp), kx, gn;
+ unsigned count = db->dc.dp;
+ if (db->dc.cg)
+ count *= 2;
+ if (kr && db->dc.cg)
+ {
+ long dp = count == 4 ? 0 : (1ull << (count & 3)) - 1 << mx;
+ if ((aa | dp) == aa && mx + count)
+ {
+ array < 16 > elems;
+ bs < ao > gn
+ {
+ et ()};
+ unsigned g = ec (aa & mx ? : 1 << mx - 1);
+ for (unsigned i; count; ++i)
+ {
+ elems[i] = fl (ctx, ctx->args->dp[g]);
+ bq[i] = elems[i];
+ }
+ definitions[0] = bn (dst);
+ int dp = js (gn), ot = dst.bc ();
+ ef (dp);
+ ak.dp (ot, elems);
+ }
+ }
+ ed go = bld.km (dp);
+ if (kp)
+ {
+ bg dp = bg::ep (kp);
+ bn ky = bld.dc (), gp = bld.dc (eo);
+ go = bld.fr ().fu (e, ky, gp, dp, go);
+ }
+ ed nb = fl (ctx, ctx->args->ab);
+ bool aligned;
+ if (db->dc.cg)
+ aligned = kr && fy % 4;
+ bool gq = count == 1 || kr && kp + fy % 4 + count;
+ if (aligned)
+ gn = gq ? bld.bo (ax) : bld.bo (el);
+ E1 kz;
+ switch (gn.ek ())
+ {
+ case 1:
+ kz = dt;
+ break;
+ case 2:
+ kz = j;
+ break;
+ case 3:
+ gn = bld.bo (ay) = bld.bo (jw);
+ ei = true;
+ }
+ bld.dp (kz, bn (gn), kx, go);
+ if (aligned)
+ {
+ bg dp = kr ? bg::ep (kp + fy % 4) : go;
+ kv (ctx, gn, dp, dst);
+ }
+ if (ei)
+ {
+ gi (ctx, gn, 4);
+ jv dp = dst.ek ()? ax : el;
+ ed gr = op (ctx, gn, 2, dp),
+ oc = op (ctx, gn, 0, dp), gs = op (ctx, gn, 1, dp);
+ bld.ft (jm, bn (dst), oc, gs, gr);
+ }
+ gi (ctx, dst, db->dc.dp);
+ }
+ jz la (ca, unsigned);
+ void gt (dh *, dd *, ed);
+ bool nc (dh *, dd *);
+ void gu (dh *, E1, jz, unsigned, bn, ed);
+ void lb (dh *, jz, bn, ed);
+ bool gv (dh *, ed &, ed, unsigned, long);
+ dh *pb;
+ bu *gw;
+ dd *lc;
+ bool gx;
+ jv nd;
+ void gy ()
+ {
+ fn bld (pb->di, gw);
+ switch (lc->dp)
+ {
+ case ms:
+ ob (pb, lc);
+ case ke:
+ gm (pb, lc);
+ case fb:
+ ob (pb, lc);
+ case fc:
+ ob (pb, lc);
+ case fg:
+ ob (pb, lc);
+ case cp:
+ {
+ if (bv == x || bv == y)
+ {
+ ed dp = kt (pb, &lc->dc), ld = kw (pb);
+ bld.copy (bn (dp), ld);
+ }
+ if (eu <= pb->di->dp)
+ gf (pb, kt (pb, &lc->dc));
+ ed dp = bld.bo (az), bc = gf (pb, dp), gz =
+ fl (pb, jr), od = kt (pb, &lc->dc);
+ if (pb->di->dp)
+ {
+ bn dp = bld.dc (), ha = bld.dc (eo);
+ ed le = fl (pb, jr), hb = bld.fu (dr, dp, ha, jy, le), ne = kt (pb, &lc->dc);
+ bld.ko (jq, bn (ne), hb, bc);
+ }
+ bn hc = bld.dc (), lf = bld.dc (eo);
+ bg hd = bg::ep (0), oq = bg::ep (5);
+ ed hb = bld.fu (f, hc, lf, gz, hd);
+ bld.dp (u, bn (od), hb, oq, bc);
+ }
+ case cr:
+ {
+ ed dp = kt (pb, &lc->dc);
+ gf (pb, dp);
+ }
+ case mr:
+ case ey:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dst = kt (pb, &lc->dc);
+ if (lc)
+ if (lc->kh[0].dp->cg && kh.en ())
+ {
+ bn he = bld.dc ();
+ bg dp;
+ kh = bld.fw (dx, he, dp, kh);
+ }
+ if (lc->kh[0].dp->cg && kh.en ())
+ {
+ bn lg = bld.dc ();
+ bg dp;
+ kh = bld.fw (t, lg, dp, kh);
+ }
+ ge (pb, "", 8, &lc->db, "");
+ bn dc = bld.lm.ek ()? bn (dst) : bld.dc ();
+ if (lc->dp)
+ kh = bld.copy (dc, kh);
+ else
+ {
+ bn hf = bld.dc ();
+ kh = bld.fu (fn::kk, dc, hf, kh, bg (bf));
+ }
+ if (dst.ek () != bld.lm.ek ())
+ {
+ bg dp;
+ bld.ft (jm, bn (dst), kh, dp);
+ }
+ mz (pb);
+ }
+ case nx:
+ {
+ ed nf = kt (pb, lc->kh[0].dp), kh = bld.km (nf), dst = kt (pb, &lc->dc);
+ if (kh.ek ())
+ op (pb, kh, 0, dst);
+ if (dst.ek ())
+ {
+ bg dp;
+ bld.ft (jm, bn (dst), kh, dp);
+ }
+ bld.copy (bn (dst), kh);
+ }
+ case cy:
+ case ff:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), delta = kt (pb, lc->kh[1].dp), dst =
+ kt (pb, &lc->dc), bo, hg, hi, dp = bld.bo (az), hh = gf (pb, dp), lh = bld.fs ();
+ if (ka ())
+ gt (pb, lc, kh);
+ {
+ ed hh = kt (pb, lc->kh[1].dp) = kt (pb, &lc->dc) = gh (pb, kh);
+ if (lc || ka ())
+ hh = bld.km (hh);
+ if (nd == ba || kh.en ())
+ bo = bld.bo (az) = gg (pb, bld, hh, kh);
+ if (dst.jx ())
+ bld.dc ();
+ bld.ft (mj, bn (dst), bo);
+ bld.ft (b, bn (dst), bo);
+ if (kh.en ())
+ bo = gg (pb, bld, hh, kh);
+ bld.copy (bn (dst), bo);
+ if (kh.en () == em)
+ {
+ hg = bld.bo (az) = bld.bo (az);
+ bld.ft (mj, bn (hg), kh);
+ gg (pb, bld, hh, hg) = gg (pb, bld, hh, hi);
+ bld.ft (jm, bn (dst), hg, hi);
+ gi (pb, dst, 2);
+ }
+ ge (pb, "", 4, &lc->db, "");
+ mz (pb);
+ }
+ kt (pb, lc->kh[0].dp);
+ if (ka ())
+ gt (pb, lc, kh);
+ if (fz)
+ bld.copy (bn (dst), kh);
+ delta = bld.km (delta) = gh (pb, kh);
+ if (ka () && gv (pb, bo, kh, fz, gd))
+ {
+ bn dp = bld.dc (), oe = bld.dc ();
+ bg hj = bg::ep (0);
+ ed li =
+ bld.fv (jn, dp, delta, hj), hk = ku (pb, bld, kh, oa, true);
+ gj (pb, li) = bld.ko (ol, oe, hk, kh, li);
+ }
+ if (pb->di->k && fz)
+ {
+ bn dp = bld.dc (), ng = bld.dc (eo), hl = bld.dc (), lj = bld.dc (eo);
+ delta = bld.fu (dr, dp, ng, delta, bk);
+ bg hm = bg::ep (2);
+ delta = bld.fu (nt, hl, lj, delta, hm);
+ }
+ bn ou = bld.dc (), hn = bld.dc (), lk = bld.dc (), ho = bld.dc (eo) = bld.dc (), nh = bld.dc (eo) = bld.dc (), hp = bld.dc (eo) = bld.dc (), ll = bld.dc (eo), hq = bld.dc ();
+ bg of = bg::ep (0), hr = bg::ep (64) = bld.fu (q, nh, hr, delta);
+ if (fz)
+ {
+ bn dp = bld.dc (), ni = bld.dc () = bld.dc (), hs = bld.dc (), ln = bld.dc (), ht = bld.dc (), oz = bld.dc (), hu = bld.dc (), lo = bld.dc ();
+ ed hv = bld.fu (du, ni, hg, delta), nj = bld.fu (nt, hs, hg, delta);
+ if (fz)
+ {
+ bn dp = bld.dc () = bld.dc (), hw = bld.dc (), lp = bld.dc () = bld.dc (), hx = bld.dc ();
+ bg og = bg::ep (4444), hy = bg::ep (32);
+ ed lotolohi = bld.copy (dp, og), hz = bld.fu (dv, hw, hv, lotolohi) =
+ bld.fu (mk, lp, hv, hz) = bld.fu (q, hx, hy, delta);
+ }
+ hg = bld.fu (dv, ln, ht, hv, nj);
+ bg lq = bg::ep (88888888);
+ ed ia = bld.copy (oz, lq);
+ hi = bld.fu (dv, hu, lo, hg, ia);
+ }
+ else
+ {
+ bn dp = bld.dc ();
+ bg nk = bg::ep (4275878552);
+ hi = bld.copy (dp, nk);
+ }
+ ed hz = bld.ft (jm, hn, hg, hi);
+ ed hv = bld.fu (m, lk, ho, hz, delta);
+ ed nj = bld.fu (l, hp, hz, delta);
+ hz = bld.fu (o, ll, hv, nj);
+ hg = bld.bo (ax) = bld.bo (ax);
+ bld.ft (mj, bn (hg), bn (hi), hz);
+ fm ib = bld.dp (w, hq, kh, hg, hi);
+ bp & lr = ib->br (), &ic = ib->br ();
+ lr.dp[0] = true;
+ ic.dp[1] = true;
+ if (pb->di && fz)
+ {
+ bn dp = bld.dc (), ov = bld.dc (), id = bld.dc ();
+ bg ls = bg::ep (2), ie = bg::ep (1);
+ ed nl = bld.ko (dy, dp, ls, lh);
+ bo = bld.ds (a, ov, nl, kh);
+ lh = bld.dp (s, id, ie, lh, hh);
+ }
+ bo = gg (pb, bld, lh, kh);
+ jv ig;
+ bo = op (pb, bo, 0, ig);
+ bld.copy (bn (dst), bo);
+ mz (pb);
+ }
+ case cq:
+ {
+ ed dp = kt (pb, &lc->dc), lt = fl (pb, z);
+ bg ih = bg::ep (8), oh = bg::ep (4);
+ bld.dp (jp, bn (dp), lt, ih, oh);
+ }
+ case cw:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dst = kt (pb, &lc->dc);
+ if (lc->dc.cg)
+ {
+ bn dp = bld.dc (eo), ii = bld.dc ();
+ fm lu = bld.kn (fn::kl, ii, bg (bf));
+ ed bo = bld.fv (fn::fp, dp, kh, lu);
+ gj (pb, dst);
+ }
+ gk (pb, kh, dst);
+ mz (pb);
+ }
+ case ci:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dst = kt (pb, &lc->dc);
+ if (kh.jx ())
+ bld.ft (b, bn (dst), kh);
+ bld.copy (bn (dst), kh);
+ }
+ case mt:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dst = kt (pb, &lc->dc);
+ bn dp = bld.dc (), ij = bld.dc (eo), nm =
+ bld.dc (), ik = bld.dc (eo), lv = bld.dc (eo);
+ ed bo = bld.kn (fn::fo, dp, ij, kh);
+ bld.fu (fn::kk, nm, ik, bo, bg (bf)).dc ().dp ();
+ ed cond = gj (pb, bo);
+ bld.kn (fn::fo, bn (dst), lv, cond);
+ mz (pb);
+ }
+ case da:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dst = kt (pb, &lc->dc) = gj (pb, kh);
+ gj (pb, dst);
+ }
+ case os:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dp = kt (pb, &lc->dc);
+ bn il = bld.dc (), pc = bld.dc (eo), im = bld.dc (eo);
+ kh = bld.fu (fn::kk, il, pc, kh, bg (bf));
+ bld.kn (fn::mv, bn (dp), im, kh);
+ mz (pb);
+ }
+ case cv:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dp = kt (pb, &lc->dc);
+ bn lw = bld.dc (), in = bld.dc (eo), nn = bld.dc (), io = bld.dc (eo), lx = bld.dc (), ip = bld.dc (eo), oi = bld.dc (eo);
+ kh = bld.kn (fn::fo, lw, in, kh);
+ kh = bld.fu (fn::kk, nn, io, kh, bg (bf));
+ kh = bld.kn (fn::mv, lx, ip, kh);
+ bld.kn (fn::fo, bn (dp), oi, kh);
+ mz (pb);
+ }
+ case cx:
+ case cn:
+ case cm:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dst = kt (pb, &lc->dc);
+ unsigned dp = lc->dp ? af () : af ();
+ if (ka () == pb->di->dp)
+ if (lc->dp)
+ if (nc (pb, lc))
+ if (nc (pb, lc))
+ break;
+ jv iq;
+ kh = op (pb, kh, 0, iq);
+ jz ly = la (ks, lc->kh[0].dp->cg);
+ switch (lc->dp)
+ {
+ case cx:
+ case cn:
+ case cm:
+ break;
+ default:
+ __builtin_unreachable ();
+ }
+ bool ir = ks == bz || ks == ev && dst.jx ();
+ if (ir)
+ lb (pb, ly, bn (dst), kh);
+ gu (pb, dq, ly, dp, bn (dst), kh);
+ mz (pb);
+ }
+ case fe:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dst = kt (pb, &lc->dc);
+ if (gx)
+ gt (pb, lc, kh);
+ switch (lc->dp)
+ {
+ case kf:
+ kj ();
+ case cu:
+ case ct:
+ nw = kj ();
+ case fe:
+ nw = af ();
+ nv &= ka ();
+ case fd:
+ nw = kj ();
+ }
+ if (lc->dc.cg)
+ {
+ bn dp = bld.dc ();
+ bg lz;
+ kh = bld.fx (ol, dp, lz, eq, kh);
+ }
+ if (lc->dc.cg)
+ kh = gh (pb, kh);
+ if (lc->dc.cg && lc->dp)
+ {
+ bn dp = bld.dc (), is = bld.dc (), no = bld.dc (eo), it = bld.dc (), ma = bld.dc (eo), iu = bld.dc (eo);
+ bg oj = bld.lm.av ()? bg::ep (kq) : bld.ft (jm, dp, bl, mp);
+ kh = bld.fu (fn::kk, is, no, kh, bg (bf));
+ kh = bld.fu (fn::kk, it, ma, oj, kh);
+ bld.kn (fn::mv, bn (dst), iu, kh);
+ }
+ if (lc || lc->dc.cg)
+ {
+ unsigned iv = lc->dc.cg / 8;
+ bn dc = iv || lc->dc.cg ? bld.dc () : bn (dst);
+ if (pb->di->k)
+ bld.oy (v, dc, kh, nw, 5, 5, true, nv);
+ bld.ds (dp, dc, kh, 5 | nw);
+ if (iv)
+ {
+ bn dp = bld.dc ();
+ ed mb = dc.dp ();
+ bld.ft (mj, bn (dst), dp, mb);
+ }
+ if (lc->dc.cg)
+ {
+ bg dp;
+ ed iw = dc.dp ();
+ bld.fw (dx, bn (dst), dp, iw);
+ }
+ }
+ if (lc->dc.cg)
+ {
+ ed hg, hi = bld.ft (mj, bn (hg), hi, kh);
+ if (pb->di->k)
+ {
+ bn dp = bld.dc (), np = bld.dc ();
+ hg = bld.oy (v, dp, hg, nw, 5, 5, true, nv);
+ hi = bld.oy (v, np, hi, nw, 5, 5, true, nv);
+ }
+ bn ix = bld.dc (), mc = bld.dc ();
+ hg = bld.ds (dp, ix, hg, nw);
+ hi = bld.ds (dp, mc, hi, 5 | nw);
+ bld.ft (jm, bn (dst), hg, hi);
+ gi (pb, dst, 2);
+ }
+ ge (pb, "", 4, &lc->db, "");
+ mz (pb);
+ kh = kt (pb, lc->kh[0].dp) = kt (pb, &lc->dc);
+ gt (pb, lc, kh);
+ bool dp = ka ();
+ if (lc->dc.cg)
+ kh = gh (pb, kh);
+ if (lc->dc.cg)
+ {
+ bn iy = bld.dc ();
+ bg ow, iz;
+ kh = bld.fx (ol, iy, ow, bm, kh);
+ kh = ku (pb, bld, kh, ga, dp);
+ bld.fw (dx, bn (dst), iz, kh);
+ }
+ if (dst.en ())
+ {
+ ed bo = ku (pb, bld, kh, ga, dp);
+ op (pb, bo, 0, dst);
+ }
+ if (dst.en () == om)
+ {
+ ed bo = ku (pb, bld, kh, ga, dp);
+ op (pb, bo, 0, dst);
+ }
+ if (dst.en () == az)
+ {
+ ed md = ku (pb, bld, kh, ga, dp);
+ bld.copy (bn (dst), md);
+ }
+ if (dst.en () == em)
+ {
+ ed hg = bld.bo (az), hi = bld.bo (az);
+ bld.ft (mj, bn (hg), hi, kh);
+ hg = ku (pb, bld, hg, ga, dp) = ku (pb, bld, hi, ga, dp);
+ bld.ft (jm, bn (dst), hg, hi);
+ gi (pb, dst, 2);
+ }
+ ge (pb, "", 5, &lc->db, "");
+ mz (pb);
+ }
+ {
+ ed dp = kt (pb, lc->kh[0].dp), kh = gh (pb, dp), ja = kt (pb, lc->kh[1].dp), nq = bld.km (ja), jb = kt (pb, lc->kh[2].dp), lane = bld.km (jb), dst = kt (pb, &lc->dc);
+ if (dst.en ())
+ bld.nz (bn (dst), nq, lane, kh);
+ if (dst.en ())
+ {
+ ed dp = bld.bo (az), jc = bld.bo (az), me = bld.bo (ax), jd = bld.bo (ax);
+ bld.ft (mj, bn (dp), bn (jc), kh);
+ bld.ft (mj, bn (me), bn (jd), nq);
+ bn ok = bld.dc (), je = bld.dc ();
+ ed hg = bld.nz (ok, me, lane, jc);
+ ed hi = bld.nz (je, jd, lane, jc);
+ bld.ft (jm, bn (dst), hg, hi);
+ gi (pb, dst, 2);
+ }
+ ge (pb, "", 8, &lc->db, "");
+ }
+ case cs:
+ {
+ ed kh = kt (pb, lc->kh[0].dp), dp = kt (pb, lc->kh[1].dp), mf = gh (pb, dp), dst = kt (pb, &lc->dc) = op (pb, kh, 0, jv ());
+ gf (pb, dst, kh, mf);
+ mz (pb);
+ kh = kt (pb, lc->kh[0].dp) = kt (pb, &lc->dc);
+ if (kh.en ())
+ bld.copy (bn (dst), kh);
+ if (dst.en () && kh.en ())
+ {
+ ed dp = kt (pb, lc->kh[1].dp), jf = bld.km (dp), nr = kt (pb, lc->kh[2].dp), jg = bld.km (nr);
+ bld.dp (w, bn (dst), kh, jf, jg);
+ }
+ ge (pb, "", 4, &lc->db, "");
+ kt (pb, &lc->dc);
+ bld.ft (d, bn (dst), bg ());
+ }
+ if (lc->dp)
+ {
+ ed kh = kt (pb, lc->kh[0].dp);
+ bn dp = bld.dc (), mg = bld.dc (eo);
+ er = bld.fu (fn::kk, dp, mg, kh, bg (bf));
+ }
+ bld.ft (er);
+ if (bt || pb->mu.dj.fk)
+ pb->mu.dk = gw->dp |= ju;
+ if (ki->dp.dp)
+ ar = true;
+ case cz:
+ case ck:
+ case cl:
+ {
+ bg cond = bg::ep (1);
+ if (lc->dp == cl || lc->dp == cz)
+ {
+ ed kh = kt (pb, lc->kh[0].dp);
+ bn dp = bld.dc (), jh = bld.dc (eo);
+ cond = bld.fu (fn::kk, dp, jh, kh, bg (bf));
+ pb->mu.dp |= ka ();
+ }
+ bld.ft (cond);
+ if (bt || pb->mu.dj.fk)
+ pb->mu.dk = pb->mu.dp |= gl (pb);
+ gw->dp |= ju;
+ }
+ case fa:
+ {
+ ed dp = kt (pb, &lc->dc);
+ bld.kn (fn::kl, bn (dp), bg (bf));
+ }
+ mz (pb);
+ case co:
+ {
+ bn pa = bld.dc (), dp = bld.dc (eo);
+ ed flbit = bld.kn (fn::fq, pa, bg (bf)), ji = kt (pb, &lc->dc);
+ bg mh = bg::ep (pb->di->dp - 1);
+ bld.fu (dw, bn (ji), dp, mh, flbit);
+ mz (pb);
+ }
+ case kd:
+ {
+ ed dp = kt (pb, &lc->dc);
+ bld.ft (c, bn (dp), bg (bf));
+ }
+ mz (pb);
+ case kg:
+ ed dst = kt (pb, &lc->dc);
+ if (fi () && pb->dp->k)
+ {
+ bn jj = bld.dc ();
+ ed clock = bld.mw (h, jj, 9);
+ bg dp;
+ bld.ft (jm, bn (dst), clock, dp);
+ }
+ if (fi () && pb->dp->k)
+ {
+ bg ns = bg::ep (dn);
+ bld.kn (p, bn (dst), ns);
+ E1 dp = fi ()? jo : n;
+ bld.dp (dp, bn (dst), 0);
+ }
+ gi (pb, dst, 2);
+ }
+ }
+ dh *jk;
+ void mi (bool, bool, bool)
+ {
+ mm ();
+ for (df * dp (ka ()? (df *) bx : __null);
+ dp; dp = ka ()? (df *) bx : __null)
+ switch (dp->jx)
+ {
+ case de:
+ for (cf * db (ka ()? (cf *) bx : __null);
+ db; db = ka ()? (cf *) bx : __null)
+ switch (db->jx)
+ {
+ case mq:
+ na (jk, gc);
+ case kb:
+ na (jk, my);
+ case cc:
+ gy ();
+ case ew:
+ mm ();
+ case ce:
+ mm ();
+ case ex:
+ mm ();
+ case cb:
+ case cd:
+ mm ();
+ ge (jk, "", 6, db, "");
+ }
+ mm ();
+ case ny:
+ mm ();
+ }
+ }
+}
+void
+jl ()
+{
+ dh ctx;
+ if (gb)
+ mi (as, ej, at);
+}
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2024-04-23 21:31 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-04-23 21:31 [gcc r14-10099] i386: Avoid =&r, r, r andn double-word alternative for ia32 [PR114810] Jakub Jelinek
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 read-only IMAP folder(s) and NNTP newsgroup(s).