* [Patch, fortran] PR80442 Handle DATA statement with iteration var in array slice
@ 2017-05-09 20:52 Nicolas Koenig
2017-05-13 11:59 ` Nicolas Koenig
0 siblings, 1 reply; 4+ messages in thread
From: Nicolas Koenig @ 2017-05-09 20:52 UTC (permalink / raw)
To: GCC-Fortran-ML, GCC-Patches-ML
[-- Attachment #1: Type: text/plain, Size: 999 bytes --]
Hello everyone,
since everybody seems to be submitting patches the last few days, I
thought I might as well :)
Attached is a patch that makes the compiler capable of dealing with
implied do variables in
array slices in data statements.
The copying of the expressions is necessary since
gfc_simplify_expr(expr, 1) substitutes every
symbol in expr that is on the iter_stack with its value.
Ok for trunk?
Nicolas
Regression tested for x86_64-pc-linux-gnu.
Changelog:
2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
PR fortran/80442
* array.c (gfc_ref_dimen_size): Simplify stride
expression
* data.c (gfc_advance_section): Simplify start,
end and stride expressions
(gfc_advance_section): Simplify start and end
expressions
(gfc_get_section_index): Simplify start expression
2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
PR fortran/80442
* gfortran.dg/impl_do_var_data.f90: New Test
[-- Attachment #2: p3.diff --]
[-- Type: text/x-patch, Size: 3952 bytes --]
Index: array.c
===================================================================
--- array.c (revision 247809)
+++ array.c (working copy)
@@ -2201,6 +2201,7 @@ gfc_ref_dimen_size (gfc_array_ref *ar, int dimen,
mpz_t upper, lower, stride;
mpz_t diff;
bool t;
+ gfc_expr *stride_expr = NULL;
if (dimen < 0 || ar == NULL || dimen > ar->dimen - 1)
gfc_internal_error ("gfc_ref_dimen_size(): Bad dimension");
@@ -2225,12 +2226,16 @@ gfc_ref_dimen_size (gfc_array_ref *ar, int dimen,
mpz_set_ui (stride, 1);
else
{
- if (ar->stride[dimen]->expr_type != EXPR_CONSTANT)
+ stride_expr = gfc_copy_expr(ar->stride[dimen]);
+ if(!gfc_simplify_expr(stride_expr, 1))
+ gfc_internal_error("Simplification error");
+ if (stride_expr->expr_type != EXPR_CONSTANT)
{
mpz_clear (stride);
return false;
}
- mpz_set (stride, ar->stride[dimen]->value.integer);
+ mpz_set (stride, stride_expr->value.integer);
+ gfc_free_expr(stride_expr);
}
/* Calculate the number of elements via gfc_dep_differce, but only if
Index: data.c
===================================================================
--- data.c (revision 247809)
+++ data.c (working copy)
@@ -539,6 +539,7 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
mpz_t tmp;
bool forwards;
int cmp;
+ gfc_expr *start, *end, *stride;
for (i = 0; i < ar->dimen; i++)
{
@@ -547,12 +548,16 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
if (ar->stride[i])
{
+ stride = gfc_copy_expr(ar->stride[i]);
+ if(!gfc_simplify_expr(stride, 1))
+ gfc_internal_error("Simplification error");
mpz_add (section_index[i], section_index[i],
- ar->stride[i]->value.integer);
- if (mpz_cmp_si (ar->stride[i]->value.integer, 0) >= 0)
- forwards = true;
- else
- forwards = false;
+ stride->value.integer);
+ if (mpz_cmp_si (stride->value.integer, 0) >= 0)
+ forwards = true;
+ else
+ forwards = false;
+ gfc_free_expr(stride);
}
else
{
@@ -561,7 +566,13 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
}
if (ar->end[i])
- cmp = mpz_cmp (section_index[i], ar->end[i]->value.integer);
+ {
+ end = gfc_copy_expr(ar->end[i]);
+ if(!gfc_simplify_expr(end, 1))
+ gfc_internal_error("Simplification error");
+ cmp = mpz_cmp (section_index[i], end->value.integer);
+ gfc_free_expr(end);
+ }
else
cmp = mpz_cmp (section_index[i], ar->as->upper[i]->value.integer);
@@ -569,7 +580,13 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
{
/* Reset index to start, then loop to advance the next index. */
if (ar->start[i])
- mpz_set (section_index[i], ar->start[i]->value.integer);
+ {
+ start = gfc_copy_expr(ar->start[i]);
+ if(!gfc_simplify_expr(start, 1))
+ gfc_internal_error("Simplification error");
+ mpz_set (section_index[i], start->value.integer);
+ gfc_free_expr(start);
+ }
else
mpz_set (section_index[i], ar->as->lower[i]->value.integer);
}
@@ -679,6 +696,7 @@ gfc_get_section_index (gfc_array_ref *ar, mpz_t *s
int i;
mpz_t delta;
mpz_t tmp;
+ gfc_expr *start;
mpz_set_si (*offset, 0);
mpz_init (tmp);
@@ -692,11 +710,15 @@ gfc_get_section_index (gfc_array_ref *ar, mpz_t *s
case DIMEN_RANGE:
if (ar->start[i])
{
- mpz_sub (tmp, ar->start[i]->value.integer,
+ start = gfc_copy_expr(ar->start[i]);
+ if(!gfc_simplify_expr(start, 1))
+ gfc_internal_error("Simplification error");
+ mpz_sub (tmp, start->value.integer,
ar->as->lower[i]->value.integer);
mpz_mul (tmp, tmp, delta);
mpz_add (*offset, tmp, *offset);
- mpz_set (section_index[i], ar->start[i]->value.integer);
+ mpz_set (section_index[i], start->value.integer);
+ gfc_free_expr(start);
}
else
mpz_set (section_index[i], ar->as->lower[i]->value.integer);
[-- Attachment #3: impl_do_var_data.f90 --]
[-- Type: text/x-fortran, Size: 213 bytes --]
! { dg-do run }
! { dg-options "" }
program main
implicit none
integer:: i
integer, dimension(4):: A
data (A(i:i+2:i), i=1,2) /1, 2, 3, 4, 5/
if(any(A .ne. [1,4,3,5])) call abort()
end program
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [Patch, fortran] PR80442 Handle DATA statement with iteration var in array slice
2017-05-09 20:52 [Patch, fortran] PR80442 Handle DATA statement with iteration var in array slice Nicolas Koenig
@ 2017-05-13 11:59 ` Nicolas Koenig
2017-05-13 16:52 ` Jerry DeLisle
0 siblings, 1 reply; 4+ messages in thread
From: Nicolas Koenig @ 2017-05-13 11:59 UTC (permalink / raw)
To: GCC-Fortran-ML, GCC-Patches-ML
[-- Attachment #1: Type: text/plain, Size: 1141 bytes --]
Ping
Also, attached is a better test case.
On 05/09/2017 10:49 PM, Nicolas Koenig wrote:
> Hello everyone,
>
> since everybody seems to be submitting patches the last few days, I
> thought I might as well :)
> Attached is a patch that makes the compiler capable of dealing with
> implied do variables in
> array slices in data statements.
> The copying of the expressions is necessary since
> gfc_simplify_expr(expr, 1) substitutes every
> symbol in expr that is on the iter_stack with its value.
>
> Ok for trunk?
>
> Nicolas
>
> Regression tested for x86_64-pc-linux-gnu.
>
> Changelog:
> 2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
>
> PR fortran/80442
> * array.c (gfc_ref_dimen_size): Simplify stride
> expression
> * data.c (gfc_advance_section): Simplify start,
> end and stride expressions
> (gfc_advance_section): Simplify start and end
> expressions
> (gfc_get_section_index): Simplify start expression
>
> 2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
>
> PR fortran/80442
> * gfortran.dg/impl_do_var_data.f90: New Test
>
>
[-- Attachment #2: impl_do_var_data.f90 --]
[-- Type: text/x-fortran, Size: 305 bytes --]
! { dg-do run }
! PR 80442
! This test case used to produce an bogus error
! about the variables being below the lower
! array bounds
program main
implicit none
integer:: i
integer, dimension(3):: A
data (A(i:i+2:i+1), i=1,2) /1, 2, 3/
if(any(A .ne. [1,3,2])) call abort()
end program
[-- Attachment #3: p3.diff --]
[-- Type: text/x-patch, Size: 3952 bytes --]
Index: array.c
===================================================================
--- array.c (revision 247809)
+++ array.c (working copy)
@@ -2201,6 +2201,7 @@ gfc_ref_dimen_size (gfc_array_ref *ar, int dimen,
mpz_t upper, lower, stride;
mpz_t diff;
bool t;
+ gfc_expr *stride_expr = NULL;
if (dimen < 0 || ar == NULL || dimen > ar->dimen - 1)
gfc_internal_error ("gfc_ref_dimen_size(): Bad dimension");
@@ -2225,12 +2226,16 @@ gfc_ref_dimen_size (gfc_array_ref *ar, int dimen,
mpz_set_ui (stride, 1);
else
{
- if (ar->stride[dimen]->expr_type != EXPR_CONSTANT)
+ stride_expr = gfc_copy_expr(ar->stride[dimen]);
+ if(!gfc_simplify_expr(stride_expr, 1))
+ gfc_internal_error("Simplification error");
+ if (stride_expr->expr_type != EXPR_CONSTANT)
{
mpz_clear (stride);
return false;
}
- mpz_set (stride, ar->stride[dimen]->value.integer);
+ mpz_set (stride, stride_expr->value.integer);
+ gfc_free_expr(stride_expr);
}
/* Calculate the number of elements via gfc_dep_differce, but only if
Index: data.c
===================================================================
--- data.c (revision 247809)
+++ data.c (working copy)
@@ -539,6 +539,7 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
mpz_t tmp;
bool forwards;
int cmp;
+ gfc_expr *start, *end, *stride;
for (i = 0; i < ar->dimen; i++)
{
@@ -547,12 +548,16 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
if (ar->stride[i])
{
+ stride = gfc_copy_expr(ar->stride[i]);
+ if(!gfc_simplify_expr(stride, 1))
+ gfc_internal_error("Simplification error");
mpz_add (section_index[i], section_index[i],
- ar->stride[i]->value.integer);
- if (mpz_cmp_si (ar->stride[i]->value.integer, 0) >= 0)
- forwards = true;
- else
- forwards = false;
+ stride->value.integer);
+ if (mpz_cmp_si (stride->value.integer, 0) >= 0)
+ forwards = true;
+ else
+ forwards = false;
+ gfc_free_expr(stride);
}
else
{
@@ -561,7 +566,13 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
}
if (ar->end[i])
- cmp = mpz_cmp (section_index[i], ar->end[i]->value.integer);
+ {
+ end = gfc_copy_expr(ar->end[i]);
+ if(!gfc_simplify_expr(end, 1))
+ gfc_internal_error("Simplification error");
+ cmp = mpz_cmp (section_index[i], end->value.integer);
+ gfc_free_expr(end);
+ }
else
cmp = mpz_cmp (section_index[i], ar->as->upper[i]->value.integer);
@@ -569,7 +580,13 @@ gfc_advance_section (mpz_t *section_index, gfc_arr
{
/* Reset index to start, then loop to advance the next index. */
if (ar->start[i])
- mpz_set (section_index[i], ar->start[i]->value.integer);
+ {
+ start = gfc_copy_expr(ar->start[i]);
+ if(!gfc_simplify_expr(start, 1))
+ gfc_internal_error("Simplification error");
+ mpz_set (section_index[i], start->value.integer);
+ gfc_free_expr(start);
+ }
else
mpz_set (section_index[i], ar->as->lower[i]->value.integer);
}
@@ -679,6 +696,7 @@ gfc_get_section_index (gfc_array_ref *ar, mpz_t *s
int i;
mpz_t delta;
mpz_t tmp;
+ gfc_expr *start;
mpz_set_si (*offset, 0);
mpz_init (tmp);
@@ -692,11 +710,15 @@ gfc_get_section_index (gfc_array_ref *ar, mpz_t *s
case DIMEN_RANGE:
if (ar->start[i])
{
- mpz_sub (tmp, ar->start[i]->value.integer,
+ start = gfc_copy_expr(ar->start[i]);
+ if(!gfc_simplify_expr(start, 1))
+ gfc_internal_error("Simplification error");
+ mpz_sub (tmp, start->value.integer,
ar->as->lower[i]->value.integer);
mpz_mul (tmp, tmp, delta);
mpz_add (*offset, tmp, *offset);
- mpz_set (section_index[i], ar->start[i]->value.integer);
+ mpz_set (section_index[i], start->value.integer);
+ gfc_free_expr(start);
}
else
mpz_set (section_index[i], ar->as->lower[i]->value.integer);
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [Patch, fortran] PR80442 Handle DATA statement with iteration var in array slice
2017-05-13 11:59 ` Nicolas Koenig
@ 2017-05-13 16:52 ` Jerry DeLisle
2017-05-14 0:45 ` Nicolas Koenig
0 siblings, 1 reply; 4+ messages in thread
From: Jerry DeLisle @ 2017-05-13 16:52 UTC (permalink / raw)
To: Nicolas Koenig, GCC-Fortran-ML, GCC-Patches-ML
On 05/13/2017 04:56 AM, Nicolas Koenig wrote:
> Ping
> Also, attached is a better test case.
>
>
> On 05/09/2017 10:49 PM, Nicolas Koenig wrote:
>> Hello everyone,
>>
>> since everybody seems to be submitting patches the last few days, I thought I
>> might as well :)
>> Attached is a patch that makes the compiler capable of dealing with implied do
>> variables in
>> array slices in data statements.
>> The copying of the expressions is necessary since gfc_simplify_expr(expr, 1)
>> substitutes every
>> symbol in expr that is on the iter_stack with its value.
>>
>> Ok for trunk?
It looks OK and thanks for patch.
Jerry
>>
>> Nicolas
>>
>> Regression tested for x86_64-pc-linux-gnu.
>>
>> Changelog:
>> 2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
>>
>> PR fortran/80442
>> * array.c (gfc_ref_dimen_size): Simplify stride
>> expression
>> * data.c (gfc_advance_section): Simplify start,
>> end and stride expressions
>> (gfc_advance_section): Simplify start and end
>> expressions
>> (gfc_get_section_index): Simplify start expression
>>
>> 2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
>>
>> PR fortran/80442
>> * gfortran.dg/impl_do_var_data.f90: New Test
>>
>>
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [Patch, fortran] PR80442 Handle DATA statement with iteration var in array slice
2017-05-13 16:52 ` Jerry DeLisle
@ 2017-05-14 0:45 ` Nicolas Koenig
0 siblings, 0 replies; 4+ messages in thread
From: Nicolas Koenig @ 2017-05-14 0:45 UTC (permalink / raw)
To: Jerry DeLisle, GCC-Fortran-ML, GCC-Patches-ML
Hello Jerry,
Thanks for the review. Committed as r248012.
Nicolas
On 05/13/2017 06:30 PM, Jerry DeLisle wrote:
> On 05/13/2017 04:56 AM, Nicolas Koenig wrote:
>> Ping
>> Also, attached is a better test case.
>>
>>
>> On 05/09/2017 10:49 PM, Nicolas Koenig wrote:
>>> Hello everyone,
>>>
>>> since everybody seems to be submitting patches the last few days, I thought I
>>> might as well :)
>>> Attached is a patch that makes the compiler capable of dealing with implied do
>>> variables in
>>> array slices in data statements.
>>> The copying of the expressions is necessary since gfc_simplify_expr(expr, 1)
>>> substitutes every
>>> symbol in expr that is on the iter_stack with its value.
>>>
>>> Ok for trunk?
> It looks OK and thanks for patch.
>
> Jerry
>
>
>>> Nicolas
>>>
>>> Regression tested for x86_64-pc-linux-gnu.
>>>
>>> Changelog:
>>> 2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
>>>
>>> PR fortran/80442
>>> * array.c (gfc_ref_dimen_size): Simplify stride
>>> expression
>>> * data.c (gfc_advance_section): Simplify start,
>>> end and stride expressions
>>> (gfc_advance_section): Simplify start and end
>>> expressions
>>> (gfc_get_section_index): Simplify start expression
>>>
>>> 2017-05-09 Nicolas Koenig <koenigni@student.ethz.ch>
>>>
>>> PR fortran/80442
>>> * gfortran.dg/impl_do_var_data.f90: New Test
>>>
>>>
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2017-05-13 23:43 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-05-09 20:52 [Patch, fortran] PR80442 Handle DATA statement with iteration var in array slice Nicolas Koenig
2017-05-13 11:59 ` Nicolas Koenig
2017-05-13 16:52 ` Jerry DeLisle
2017-05-14 0:45 ` Nicolas Koenig
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).