public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "steven at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug tree-optimization/23455] tree load PRE is not working properly Date: Mon, 02 Jan 2006 21:27:00 -0000 [thread overview] Message-ID: <20060102212721.15468.qmail@sourceware.org> (raw) In-Reply-To: <bug-23455-7849@http.gcc.gnu.org/bugzilla/> ------- Comment #6 from steven at gcc dot gnu dot org 2006-01-02 21:27 ------- Ehm, wouldn't "unsigned char *" alias everything? GCSE doesn't do load PRE for me on the original test case, either. And, as long as "outbuf[outcnt] = bi_buf;" has a V_MAY_DEF for outcnt, you're not going to see any load PRE happening here. With a changed test case, GCSE _does_ perform load PRE: unsigned long outcnt; extern void flush_outbuf(void); void bi_windup(unsigned int *outbuf, unsigned int bi_buf) { unsigned long t1 = outcnt; outbuf[t1] = bi_buf; unsigned long t2 = outcnt; if (t2 == 16384) flush_outbuf(); unsigned long t3 = outcnt; outbuf[t3] = bi_buf; } This gives me the following asm on x86-64: bi_windup: movq %rbx, -16(%rsp) movq %rbp, -8(%rsp) subq $24, %rsp movq outcnt(%rip), %rax movq %rdi, %rbp movl %esi, %ebx cmpq $16384, %rax movl %esi, (%rbp,%rax,4) je .L6 .L2: movl %ebx, (%rbp,%rax,4) movq 8(%rsp), %rbx movq 16(%rsp), %rbp addq $24, %rsp ret .p2align 4,,7 .L6: call flush_outbuf movq outcnt(%rip), %rax jmp .L2 Looking at the .pre tree dump, there isn't anything obvious in the way of doing load PRE here AFAICT: bi_windup (outbufD.1866, bi_bufD.1867) { unsigned intD.3 storetmp.21D.1918; long unsigned intD.4 t3D.1872; long unsigned intD.4 t2D.1871; long unsigned intD.4 t1D.1870; unsigned intD.3 * D.1878; unsigned intD.3 * D.1877; long unsigned intD.4 D.1876; unsigned intD.3 * D.1875; unsigned intD.3 * D.1874; long unsigned intD.4 D.1873; # BLOCK 2 freq:10000 # PRED: ENTRY [100.0%] (fallthru,exec) # VUSE <outcntD.1863_1>; t1D.1870_2 = outcntD.1863; D.1873_3 = t1D.1870_2 * 4; D.1874_4 = (unsigned intD.3 *) D.1873_3; D.1875_6 = D.1874_4 + outbufD.1866_5; # TMT.4D.1900_16 = V_MAY_DEF <TMT.4D.1900_15>; *D.1875_6 = bi_bufD.1867_7; if (t1D.1870_2 == 16384) goto <L0>; else goto <L2>; # SUCC: 3 [26.2%] (true,exec) 5 [73.8%] (false,exec) # BLOCK 5 freq:7378 # PRED: 2 [73.8%] (false,exec) <L2>:; goto <bb 4> (<L1>); # SUCC: 4 [100.0%] (fallthru) # BLOCK 3 freq:2622 # PRED: 2 [26.2%] (true,exec) <L0>:; # outcntD.1863_18 = V_MAY_DEF <outcntD.1863_1>; # TMT.4D.1900_19 = V_MAY_DEF <TMT.4D.1900_16>; flush_outbuf (); # SUCC: 4 [100.0%] (fallthru,exec) # BLOCK 4 freq:10000 # PRED: 5 [100.0%] (fallthru) 3 [100.0%] (fallthru,exec) # TMT.4D.1900_14 = PHI <TMT.4D.1900_16(5), TMT.4D.1900_19(3)>; # outcntD.1863_13 = PHI <outcntD.1863_1(5), outcntD.1863_18(3)>; <L1>:; # VUSE <outcntD.1863_13>; t3D.1872_9 = outcntD.1863; D.1876_10 = t3D.1872_9 * 4; D.1877_11 = (unsigned intD.3 *) D.1876_10; D.1878_12 = D.1877_11 + outbufD.1866_5; # TMT.4D.1900_17 = V_MAY_DEF <TMT.4D.1900_14>; *D.1878_12 = bi_bufD.1867_7; return; # SUCC: EXIT [100.0%] } -- steven at gcc dot gnu dot org changed: What |Removed |Added ---------------------------------------------------------------------------- Keywords| |TREE Summary|load PRE is missing |tree load PRE is not working | |properly http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23455
next prev parent reply other threads:[~2006-01-02 21:27 UTC|newest] Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top [not found] <bug-23455-7849@http.gcc.gnu.org/bugzilla/> 2006-01-02 19:31 ` [Bug tree-optimization/23455] load PRE is missing bonzini at gnu dot org 2006-01-02 21:27 ` steven at gcc dot gnu dot org [this message] 2006-01-02 21:30 ` [Bug tree-optimization/23455] tree load PRE is not working properly steven at gcc dot gnu dot org 2006-01-03 8:04 ` bonzini at gnu dot org 2006-01-03 18:29 ` dberlin at gcc dot gnu dot org 2006-01-04 22:35 ` steven at gcc dot gnu dot org 2006-01-07 0:25 ` steven at gcc dot gnu dot org 2008-04-07 1:41 ` [Bug tree-optimization/23455] tree load PRE is not working for global variables pinskia at gcc dot gnu dot org 2008-04-07 1:45 ` pinskia at gcc dot gnu dot org 2008-04-07 1:47 ` pinskia at gcc dot gnu dot org 2008-07-06 8:53 ` steven at gcc dot gnu dot org 2008-07-08 16:12 ` dberlin at gcc dot gnu dot org 2008-07-08 20:28 ` rguenth at gcc dot gnu dot org
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=20060102212721.15468.qmail@sourceware.org \ --to=gcc-bugzilla@gcc.gnu.org \ --cc=gcc-bugs@gcc.gnu.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: linkBe 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 read-only IMAP folder(s) and NNTP newsgroup(s).