Fortran: runtime bounds-checking in presence of array constructors [PR31059]
Checks
Commit Message
Dear all,
gfortran's array bounds-checking code does a mostly reasonable
job for array sections in expressions and assignments, but
forgot the case that (rank-1) expressions can involve array
constructors, which have a shape ;-)
The attached patch walks over the loops generated by the
scalarizer, checks for the presence of a constructor, and
takes the first shape found as reference. (If several
constructors are present, discrepancies in their shape
seems to be already detected at compile time).
For more details on what will be caught now see testcase.
Regtested on x86_64-pc-linux-gnu. OK for mainline?
Thanks,
Harald
Comments
Le 31/08/2023 à 22:42, Harald Anlauf via Fortran a écrit :
> Dear all,
>
> gfortran's array bounds-checking code does a mostly reasonable
> job for array sections in expressions and assignments, but
> forgot the case that (rank-1) expressions can involve array
> constructors, which have a shape ;-)
>
> The attached patch walks over the loops generated by the
> scalarizer, checks for the presence of a constructor, and
> takes the first shape found as reference. (If several
> constructors are present, discrepancies in their shape
> seems to be already detected at compile time).
>
> For more details on what will be caught now see testcase.
>
> Regtested on x86_64-pc-linux-gnu. OK for mainline?
>
This is OK.
May I suggest to handle functions the same way?
Thanks.
> Thanks,
> Harald
>
Hi Mikael,
On 9/1/23 10:41, Mikael Morin via Gcc-patches wrote:
> Le 31/08/2023 à 22:42, Harald Anlauf via Fortran a écrit :
>> Dear all,
>>
>> gfortran's array bounds-checking code does a mostly reasonable
>> job for array sections in expressions and assignments, but
>> forgot the case that (rank-1) expressions can involve array
>> constructors, which have a shape ;-)
>>
>> The attached patch walks over the loops generated by the
>> scalarizer, checks for the presence of a constructor, and
>> takes the first shape found as reference. (If several
>> constructors are present, discrepancies in their shape
>> seems to be already detected at compile time).
>>
>> For more details on what will be caught now see testcase.
>>
>> Regtested on x86_64-pc-linux-gnu. OK for mainline?
>>
> This is OK.
I've pushed this is the first step.
> May I suggest to handle functions the same way?
I'll have a look at them, but will need to gather a few
suitable testcases first.
Thanks for the review!
Harald
>
> Thanks.
>
>> Thanks,
>> Harald
>>
>
>
Le 01/09/2023 à 22:48, Harald Anlauf a écrit :
> Hi Mikael,
>
> On 9/1/23 10:41, Mikael Morin via Gcc-patches wrote:
>> May I suggest to handle functions the same way?
>
> I'll have a look at them, but will need to gather a few
> suitable testcases first.
>
I have just opened PR111339
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111339) to track the case
of functions separately.
From 944a35909e8eeb79c92e398ae3f27e94708584e6 Mon Sep 17 00:00:00 2001
From: Harald Anlauf <anlauf@gmx.de>
Date: Thu, 31 Aug 2023 22:19:58 +0200
Subject: [PATCH] Fortran: runtime bounds-checking in presence of array
constructors [PR31059]
gcc/fortran/ChangeLog:
PR fortran/31059
* trans-array.cc (gfc_conv_ss_startstride): For array bounds checking,
consider also array constructors in expressions, and use their shape.
gcc/testsuite/ChangeLog:
PR fortran/31059
* gfortran.dg/bounds_check_fail_5.f90: New test.
---
gcc/fortran/trans-array.cc | 23 ++++++++++++++++
.../gfortran.dg/bounds_check_fail_5.f90 | 26 +++++++++++++++++++
2 files changed, 49 insertions(+)
create mode 100644 gcc/testsuite/gfortran.dg/bounds_check_fail_5.f90
@@ -4740,6 +4740,29 @@ done:
for (n = 0; n < loop->dimen; n++)
size[n] = NULL_TREE;
+ /* If there is a constructor involved, derive size[] from its shape. */
+ for (ss = loop->ss; ss != gfc_ss_terminator; ss = ss->loop_chain)
+ {
+ gfc_ss_info *ss_info;
+
+ ss_info = ss->info;
+ info = &ss_info->data.array;
+
+ if (ss_info->type == GFC_SS_CONSTRUCTOR && info->shape)
+ {
+ for (n = 0; n < loop->dimen; n++)
+ {
+ if (size[n] == NULL)
+ {
+ gcc_assert (info->shape[n]);
+ size[n] = gfc_conv_mpz_to_tree (info->shape[n],
+ gfc_index_integer_kind);
+ }
+ }
+ break;
+ }
+ }
+
for (ss = loop->ss; ss != gfc_ss_terminator; ss = ss->loop_chain)
{
stmtblock_t inner;
new file mode 100644
@@ -0,0 +1,26 @@
+! { dg-do run }
+! { dg-additional-options "-fcheck=bounds -g -fdump-tree-original" }
+! { dg-output "At line 13 .*" }
+! { dg-shouldfail "Array bound mismatch for dimension 1 of array 'ivec' (2/3)" }
+!
+! PR fortran/31059 - runtime bounds-checking in presence of array constructors
+
+program p
+ integer :: jvec(3) = [1,2,3]
+ integer, allocatable :: ivec(:), kvec(:), lvec(:), mvec(:), nvec(:)
+ ivec = [1,2] ! (re)allocation
+ kvec = [4,5,6] ! (re)allocation
+ ivec(:) = [4,5,6] ! runtime error (->dump)
+ ! not reached ...
+ print *, jvec + [1,2,3] ! OK & no check generated
+ print *, [4,5,6] + jvec ! OK & no check generated
+ print *, lvec + [1,2,3] ! check generated (->dump)
+ print *, [4,5,6] + mvec ! check generated (->dump)
+ nvec(:) = jvec ! check generated (->dump)
+end
+
+! { dg-final { scan-tree-dump-times "Array bound mismatch " 4 "original" } }
+! { dg-final { scan-tree-dump-times "Array bound mismatch .*ivec" 1 "original" } }
+! { dg-final { scan-tree-dump-times "Array bound mismatch .*lvec" 1 "original" } }
+! { dg-final { scan-tree-dump-times "Array bound mismatch .*mvec" 1 "original" } }
+! { dg-final { scan-tree-dump-times "Array bound mismatch .*nvec" 1 "original" } }
--
2.35.3