* [SH][committed] Fix PR 65979 / PR 66611
@ 2015-06-26 2:36 Oleg Endo
2015-06-27 1:33 ` Oleg Endo
0 siblings, 1 reply; 2+ messages in thread
From: Oleg Endo @ 2015-06-26 2:36 UTC (permalink / raw)
To: gcc-patches
[-- Attachment #1: Type: text/plain, Size: 422 bytes --]
Hi,
This fixes the problem of emitting an insn from within a peephole pattern which can't be recognized due to mismatching operand constraints. Tested by Kaz on sh4-linux. Committed to trunk as r224987. Backport to GCC 5 will follow.
Cheers,
Oleg
gcc/ChangeLog:
PR target/65979
PR target/66611
* config/sh/sh.md (tstsi_t peephole2): Use insn_invalid_p to check if
the replacement insn will work.
[-- Attachment #2: sh_pr65979_pr66611.patch --]
[-- Type: application/octet-stream, Size: 1664 bytes --]
Index: gcc/config/sh/sh.md
===================================================================
--- gcc/config/sh/sh.md (revision 224565)
+++ gcc/config/sh/sh.md (working copy)
@@ -14727,8 +14727,19 @@
if (REGNO (operands[1]) == REGNO (operands[2]))
operands[2] = gen_rtx_REG (SImode, REGNO (operands[0]));
- sh_check_add_incdec_notes (emit_insn (gen_rtx_SET (operands[2],
- operands[3])));
+ // We don't know what the new set insn will be in detail. Just make sure
+ // that it still can be recognized and the constraints are satisfied.
+ rtx_insn* i = emit_insn (gen_rtx_SET (operands[2], operands[3]));
+
+ recog_data_d prev_recog_data = recog_data;
+ bool i_invalid = insn_invalid_p (i, false);
+ recog_data = prev_recog_data;
+
+ if (i_invalid)
+ FAIL;
+
+ sh_check_add_incdec_notes (i);
+
emit_insn (gen_tstsi_t (operands[2],
gen_rtx_REG (SImode, (REGNO (operands[1])))));
})
@@ -14755,8 +14766,19 @@
|| REGNO (operands[2]) == REGNO (operands[5]))"
[(const_int 0)]
{
- sh_check_add_incdec_notes (emit_insn (gen_rtx_SET (operands[2],
- operands[3])));
+ // We don't know what the new set insn will be in detail. Just make sure
+ // that it still can be recognized and the constraints are satisfied.
+ rtx_insn* i = emit_insn (gen_rtx_SET (operands[2], operands[3]));
+
+ recog_data_d prev_recog_data = recog_data;
+ bool i_invalid = insn_invalid_p (i, false);
+ recog_data = prev_recog_data;
+
+ if (i_invalid)
+ FAIL;
+
+ sh_check_add_incdec_notes (i);
+
emit_insn (gen_tstsi_t (operands[2],
gen_rtx_REG (SImode, (REGNO (operands[1])))));
})
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [SH][committed] Fix PR 65979 / PR 66611
2015-06-26 2:36 [SH][committed] Fix PR 65979 / PR 66611 Oleg Endo
@ 2015-06-27 1:33 ` Oleg Endo
0 siblings, 0 replies; 2+ messages in thread
From: Oleg Endo @ 2015-06-27 1:33 UTC (permalink / raw)
To: gcc-patches
[-- Attachment #1: Type: text/plain, Size: 407 bytes --]
On 26 Jun 2015, at 08:54, Oleg Endo <oleg.endo@t-online.de> wrote:
> Hi,
>
> This fixes the problem of emitting an insn from within a peephole pattern which can't be recognized due to mismatching operand constraints. Tested by Kaz on sh4-linux. Committed to trunk as r224987. Backport to GCC 5 will follow.
I've committed the attached patch to the GCC 5 branch as r225094.
Cheers,
Oleg
[-- Attachment #2: sh_pr65979_pr66611_gcc5.patch --]
[-- Type: application/octet-stream, Size: 1700 bytes --]
Index: gcc/config/sh/sh.md
===================================================================
--- gcc/config/sh/sh.md (revision 224988)
+++ gcc/config/sh/sh.md (working copy)
@@ -14692,8 +14692,19 @@
if (REGNO (operands[1]) == REGNO (operands[2]))
operands[2] = gen_rtx_REG (SImode, REGNO (operands[0]));
- sh_check_add_incdec_notes (emit_insn (gen_rtx_SET (SImode, operands[2],
- operands[3])));
+ // We don't know what the new set insn will be in detail. Just make sure
+ // that it still can be recognized and the constraints are satisfied.
+ rtx_insn* i = emit_insn (gen_rtx_SET (VOIDmode, operands[2], operands[3]));
+
+ recog_data_d prev_recog_data = recog_data;
+ bool i_invalid = insn_invalid_p (i, false);
+ recog_data = prev_recog_data;
+
+ if (i_invalid)
+ FAIL;
+
+ sh_check_add_incdec_notes (i);
+
emit_insn (gen_tstsi_t (operands[2],
gen_rtx_REG (SImode, (REGNO (operands[1])))));
})
@@ -14720,8 +14731,19 @@
|| REGNO (operands[2]) == REGNO (operands[5]))"
[(const_int 0)]
{
- sh_check_add_incdec_notes (emit_insn (gen_rtx_SET (SImode, operands[2],
- operands[3])));
+ // We don't know what the new set insn will be in detail. Just make sure
+ // that it still can be recognized and the constraints are satisfied.
+ rtx_insn* i = emit_insn (gen_rtx_SET (VOIDmode, operands[2], operands[3]));
+
+ recog_data_d prev_recog_data = recog_data;
+ bool i_invalid = insn_invalid_p (i, false);
+ recog_data = prev_recog_data;
+
+ if (i_invalid)
+ FAIL;
+
+ sh_check_add_incdec_notes (i);
+
emit_insn (gen_tstsi_t (operands[2],
gen_rtx_REG (SImode, (REGNO (operands[1])))));
})
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2015-06-27 0:48 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-06-26 2:36 [SH][committed] Fix PR 65979 / PR 66611 Oleg Endo
2015-06-27 1:33 ` Oleg Endo
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).