[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn) since r11-4623
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 Jakub Jelinek changed: What|Removed |Added Status|NEW |RESOLVED Resolution|--- |FIXED --- Comment #8 from Jakub Jelinek --- Fixed.
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn) since r11-4623
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 --- Comment #7 from CVS Commits --- The master branch has been updated by Segher Boessenkool : https://gcc.gnu.org/g:a0b5843a9b2b3e93bd119d5aef583f53ad2ef0bc commit r11-7695-ga0b5843a9b2b3e93bd119d5aef583f53ad2ef0bc Author: Segher Boessenkool Date: Tue Mar 16 18:29:11 2021 + rs6000: Workaround for PR98092 The bcdinvalid_ RTL instruction uses the "unordered" comparison, which cannot be used if we have -ffinite-math-only. We really need CCMODEs that describe what bits in a CR field are set by other insns than just comparisons, but that is a lot more surgery, and it is stage 4 now. This patch does a simple workaround. 2021-03-16 Segher Boessenkool PR target/98092 * config/rs6000/predicates.md (branch_comparison_operator): Allow ordered and unordered for CCFPmode, if flag_finite_math_only. gcc/testsuite/ PR target/98092 * gcc.target/powerpc/pr98092.c: New.
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn) since r11-4623
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 Segher Boessenkool changed: What|Removed |Added Attachment #50040|0 |1 is obsolete|| --- Comment #6 from Segher Boessenkool --- Created attachment 50401 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50401=edit Patch
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn) since r11-4623
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 Jakub Jelinek changed: What|Removed |Added Summary|[11 Regression] ICE in |[11 Regression] ICE in |extract_insn, at|extract_insn, at |recog.c:2315 (error:|recog.c:2315 (error: |unrecognizable insn)|unrecognizable insn) since ||r11-4623 --- Comment #5 from Jakub Jelinek --- The builtins were added in r11-4623-g05161256d3d2a598966ca1cf676fa0e427570f73 , before that it has been rejected with: pr98092.c: In function ‘h9’: pr98092.c:4:11: warning: implicit declaration of function ‘__builtin_vec_bcdinvalid’; did you mean ‘__builtin_vec_xxeval’? [-Wimplicit-function-declaration] 4 | return (__builtin_vec_bcdinvalid (un)); | ^~~~ | __builtin_vec_xxeval pr98092.c:4:3: error: AltiVec argument passed to unprototyped function 4 | return (__builtin_vec_bcdinvalid (un)); | ^~
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 --- Comment #4 from Jakub Jelinek --- Any progress on this?
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 --- Comment #3 from Segher Boessenkool --- Created attachment 50040 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50040=edit Patch Patch in testing.
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 --- Comment #2 from Carl Love --- Segher: Yup, I saw the buzilla. Will take a look at it. Carl On Fri, 2021-01-22 at 18:49 +, segher at gcc dot gnu.org wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 > > Segher Boessenkool changed: > >What|Removed |Added > --- > - >Assignee|unassigned at gcc dot gnu.org |segher at gcc > dot gnu.org >
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092
Jakub Jelinek changed:
What|Removed |Added
Last reconfirmed||2021-01-22
Ever confirmed|0 |1
Status|UNCONFIRMED |NEW
CC||carll at gcc dot gnu.org,
||jakub at gcc dot gnu.org,
||segher at gcc dot gnu.org
--- Comment #1 from Jakub Jelinek ---
The expander emits an UNORDERED comparison:
(define_expand "bcdinvalid_"
[(parallel [(set (reg:CCFP CR6_REGNO)
(compare:CCFP
(unspec:V2DF [(match_operand:VBCD 1 "register_operand")]
UNSPEC_BCDADD)
(match_dup 2)))
(clobber (match_scratch:VBCD 3))])
(set (match_operand:SI 0 "register_operand")
(unordered:SI (reg:CCFP CR6_REGNO)
(const_int 0)))]
"TARGET_P8_VECTOR"
{
operands[2] = CONST0_RTX (V2DFmode);
})
But set_cc insn will not match with -ffinite-math-only for UNORDERED,
given:
(define_predicate "scc_comparison_operator"
(and (match_operand 0 "branch_comparison_operator")
(match_code "eq,lt,gt,ltu,gtu,unordered")))
(define_predicate "branch_comparison_operator"
(and (match_operand 0 "comparison_operator")
(match_test "GET_MODE_CLASS (GET_MODE (XEXP (op, 0))) == MODE_CC")
(if_then_else (match_test "GET_MODE (XEXP (op, 0)) == CCFPmode
&& !flag_finite_math_only")
(match_code "lt,gt,eq,unordered,unge,unle,ne,ordered")
(match_code "lt,ltu,le,leu,gt,gtu,ge,geu,eq,ne"))
(match_test "validate_condition_mode (GET_CODE (op),
GET_MODE (XEXP (op, 0))),
1")))
Dunno how to get out of this, perhaps for !flag_finite_math_only expand to a
different instruction that will do both the bcdadd. and mfcr and rlwinm and
just clobber the cc reg?
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 Richard Biener changed: What|Removed |Added Priority|P3 |P1
[Bug target/98092] [11 Regression] ICE in extract_insn, at recog.c:2315 (error: unrecognizable insn)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98092 Richard Biener changed: What|Removed |Added Target Milestone|--- |11.0
