This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [testsuite/guality, committed] Prevent optimization of local in vla-1.c


On Sun, Jul 1, 2018 at 9:25 PM Tom de Vries <tdevries@suse.de> wrote:
>
> On 07/01/2018 09:11 PM, Jakub Jelinek wrote:
> > On Sun, Jul 01, 2018 at 06:19:20PM +0200, Tom de Vries wrote:
> >> So, the local vla a is optimized away.
> >>
> >> This patch adds VOLATILE to 'a', which prevents it from being optimized away,
> >> and fixes the non-lto failures.
> >>
> >> Committed as obvious.
> >
> > That isn't obvious, it is just wrong.
> > The intent of the testcase is to test how debugging of optimized code with
> > VLAs works.  With your change we don't test that anymore.  Please revert.
> >
>
> Hi,
>
> Sure, but ... if this is wrong, then for my understanding can you
> explain to me how the fail should be addressed?

Given the array has size i + 1 it's upper bound should be 'i' and 'i'
should be available via DW_OP_[GNU_]entry_value.

I see it is

    <175>   DW_AT_upper_bound : 10 byte block: 75 1 8 20 24 8 20 26 31
1c       (DW_OP_breg5 (rdi): 1; DW_OP_const1u: 32; DW_OP_shl;
DW_OP_const1u: 32; DW_OP_shra; DW_OP_lit1; DW_OP_minus)

and %rdi is 1.  Not sure why gdb fails to print it's length.  Yes, the
storage itself doesn't have a location but the
type specifies the size.

(gdb) ptype a
type = char [variable length]
(gdb) p sizeof(a)
$3 = 0

this looks like a gdb bug to me?

Btw, the location expression looks odd, if I deciper it correctly
we end up with ((%rdi << 32) >> 32) - 1 which computes to 4
but the upper bound should be 5.  The GIMPLE debug stmts compute
the upper bound as (sizetype)((long)(i_1(D) + 1) - 1)

vartrack correctly prints

(debug_insn 7 26 8 2 (var_location:SI D#3 (plus:SI (entry_value:SI
(reg:SI 5 di [ i ]))
        (const_int 1 [0x1])))
"/space/rguenther/src/gcc-slpcost/gcc/testsuite/gcc.dg/guality/vla-1.c":15
-1
     (nil))
(debug_insn 8 7 9 2 (var_location:DI D#2 (sign_extend:DI
(debug_expr:SI D#3)))
"/space/rguenther/src/gcc-slpcost/gcc/testsuite/gcc.dg/guality/vla-1.c":15
-1
     (nil))
(debug_insn 9 8 10 2 (var_location:DI D#1 (plus:DI (debug_expr:DI D#2)
        (const_int -1 [0xffffffffffffffff])))
"/space/rguenther/src/gcc-slpcost/gcc/testsuite/gcc.dg/guality/vla-1.c":15
-1
     (nil))
(debug_insn 10 9 11 2 (var_location:DI D.1914 (debug_expr:DI D#1))
"/space/rguenther/src/gcc-slpcost/gcc/testsuite/gcc.dg/guality/vla-1.c":15
-1
     (nil))

Richard.

> [ FWIW, I considered this obvious, given the ok for "[testsuite] Fix
> guality/pr45882.c for flto"  (
> https://gcc.gnu.org/ml/gcc-patches/2018-06/msg01304.html ) which seemed
> similar to me. ]
>
> Thanks,
> - Tom
>
> >> [testsuite/guality] Prevent optimization of local in vla-1.c
> >>
> >> 2018-07-01  Tom de Vries  <tdevries@suse.de>
> >>
> >>      * gcc.dg/guality/prevent-optimization.h (VOLATILE): Define.
> >>      * gcc.dg/guality/vla-1.c (f1): Mark local vla a as VOLATILE.
> >
> >       Jakub
> >


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]