https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #17 from Andrew Pinski ---
(In reply to Andrew Pinski from comment #16)
> Basically you update the min/max values and the value/mask by doing the
> shifts. And I think the rest will fall out.
That is actually done. But for some rea
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #18 from Andrew Pinski ---
Oh I think I know how to fix this case.
In scalar_min_max_arsh, Replace:
/* Its not easy to operate on alu32 bounds here because it depends
* on bits being shifted in from upper 32-bits.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #16 from Andrew Pinski ---
(In reply to Andrew Pinski from comment #15)
> (In reply to Andrew Pinski from comment #14)
> > Just a quick update. It looks like undefined code in the verifier producing
> > a verification error.
>
> Sor
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #15 from Andrew Pinski ---
(In reply to Andrew Pinski from comment #14)
> Just a quick update. It looks like undefined code in the verifier producing
> a verification error.
Sorry this is incorrect. basically the problem is we have:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #14 from Andrew Pinski ---
Just a quick update. It looks like undefined code in the verifier producing a
verification error.
scalar32_min_max_arsh does:
u64 umin_val = src_reg->u32_min_value;
/* Upon reaching here,
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
Sam James changed:
What|Removed |Added
Attachment #61078|0 |1
is obsolete|
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #12 from Andrew Pinski ---
(In reply to Andrew Pinski from comment #11)
> (In reply to Andrew Pinski from comment #10)
> > Created attachment 61081 [details]
> > try this untested patch
>
> I had a typo:
> "(register_operand (oper
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #10 from Andrew Pinski ---
Created attachment 61081
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=61081&action=edit
try this untested patch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #11 from Andrew Pinski ---
(In reply to Andrew Pinski from comment #10)
> Created attachment 61081 [details]
> try this untested patch
I had a typo:
"(register_operand (operands[0], mode)
|| reg_or_imm_operand (operands[1], ))"
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #5 from Andrew Pinski ---
GCC:
r1 = r10
r3 += r5
r1 += -16
call113
.LBE7:
r2 = *(u32 *) (r10+-16)
*(u32 *) (r10+-24) = r2
clang:
call 113
.Ltmp27:
.Ltmp28:
.loc
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #9 from Andrew Pinski ---
(In reply to Andrew Pinski from comment #8)
> Note I would not call this wrong code. It might be a verifier not liking the
> produced code but the code is still correct in terms of the underlying ISA.
And i
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
Andrew Pinski changed:
What|Removed |Added
Keywords|wrong-code |
--- Comment #8 from Andrew Pinski ---
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #7 from Andrew Pinski ---
```
void f(long *a, int *b)
{
*b = *a;
}
```
Shows the difference.
propagating insn 7 into insn 8, replacing:
(set (mem:SI (reg/v/f:DI 22 [ bD.1854 ]) [2 *b_5(D)+0 S4 A32])
(subreg:SI (reg:DI 23 [ *a
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #6 from Andrew Pinski ---
<<< Unknown tree: integer_cst >>> (&raw_magic_number, 8, _17);
__t ={v} {CLOBBER(eos)};
raw_magic_number.2_18 = raw_magic_number;
_19 = (unsigned int) raw_magic_number.2_18;
magic_number = _19;
#
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #3 from Sam James ---
Created attachment 61078
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=61078&action=edit
GCC failed output on test-bpf-restrict-fs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #4 from Sam James ---
Created attachment 61080
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=61080&action=edit
Clang successful output on test-bpf-restrict-fs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #2 from Sam James ---
Created attachment 61077
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=61077&action=edit
restrict-fs.bpf.c
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119731
--- Comment #1 from Sam James ---
Created attachment 61076
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=61076&action=edit
restrict-fs-gcc.bpf.i.xz
18 matches
Mail list logo
