PR fortran/91300 - runtime error message with allocate and errmsg=

Message ID trinity-49cbc3a5-27cb-4a53-8096-443a146d5dcb-1653769511628@3c-app-gmx-bs44
State New
Headers
Series PR fortran/91300 - runtime error message with allocate and errmsg= |

Commit Message

Harald Anlauf May 28, 2022, 8:25 p.m. UTC
  Dear Fortranners,

the PR rightfully complained that we did not differentiate errors on
ALLOCATE(...,stat=,errmsg=) for failures from allocation of already
allocated objects or insufficient virtual memory.

The attached patch introduces a new STAT value LIBERROR_NO_MEMORY
that is returned for insufficient virtual memory, and a corresponding
(simple and invariant) ERRMSG: "Insufficient virtual memory".

(In the PR Janne mentions checking for errno, but since the standard
malloc(3) only mentions ENOMEM as possible error value, and the user
may replace malloc by a special library, I believe that won't be easy
to handle in a general way.)

Most compilers I tried (Intel/NAG/Crayftn) behave similarly, except
for Nvidia/Flang, which try to return the size of the allocation in
the error message.

I am not sure that this is worth the effort.  First, ERRMSG is very
compiler-dependent anyway and thus not really portable.  If a user
wants to know what the size of the failed allocation is and really
wants to recover, he/she should find that out himself.  Second, I
think that the more important change is the introduction of a STAT
value that allows the distinction between the different causes of
failure.

The testcase should be able to handle 32 and 64 bit systems.
At least that's what I think.

Regtested on x86_64-pc-linux-gnu.  OK for mainline?  Suggestions?

Thanks,
Harald
  

Comments

Tobias Burnus May 30, 2022, 7:33 a.m. UTC | #1
On 28.05.22 22:25, Harald Anlauf via Fortran wrote:
> the PR rightfully complained that we did not differentiate errors on
> ALLOCATE(...,stat=,errmsg=) for failures from allocation of already
> allocated objects or insufficient virtual memory.
It is even worse: The error message states the wrong reason.
> The attached patch introduces a new STAT value LIBERROR_NO_MEMORY
> that is returned for insufficient virtual memory, and a corresponding
> (simple and invariant) ERRMSG: "Insufficient virtual memory".
I think the message is fine – at least on Linux 'virtual memory' is
what's used and it is clear what's meant, even if I stumble a bit about
the wording. (But do not have a crisp short & comprehensive wording myself.)
> (In the PR Janne mentions checking for errno, but since the standard
> malloc(3) only mentions ENOMEM as possible error value, and the user
> may replace malloc by a special library, I believe that won't be easy
> to handle in a general way.)
I con concur. Especially as POSIX and the Linux manpage only list ENOMEM
as only value.
> Most compilers I tried (Intel/NAG/Crayftn) behave similarly, except
> for Nvidia/Flang, which try to return the size of the allocation in
> the error message.
>
> I am not sure that this is worth the effort.
I think it is not needed – especially as we generate the message in the
FE and not in libgfortran. The advantage for the users is that they know
what value has been requested – but they cannot really do much with the
knowledge, either.
> The testcase should be able to handle 32 and 64 bit systems.
> At least that's what I think.

I hope it is – at least on 64bit, I currently expect it too fail
somewhat reliably, with 32bit I think it won't – but that's also handled.

But you could add a -fdump-tree-original + '! { dg-final {
scan-tree-dump*' to do some checking in addition (e.g. whether both
strings appear in the dump; could be more complex, but that's probably
not needed).

> Regtested on x86_64-pc-linux-gnu.  OK for mainline?  Suggestions?

LGTM – with considering comments on the testcase.


> Fortran: improve runtime error message with ALLOCATE and ERRMSG=

Consider appending [PR91300] in that line – and try to make the
gcc-patches@ and fortran@ lines the same

(Searching for the PR number or case-insignificantly for the string in
the mailing list archive, will fine the message; thus, that's okay.)

> ALLOCATE: generate different STAT,ERRMSG results for failures from
> allocation of already allocated objects or insufficient virtual memory.
>
> gcc/fortran/ChangeLog:
>
>       PR fortran/91300
>       * libgfortran.h: Define new error code LIBERROR_NO_MEMORY.
>       * trans-stmt.cc (gfc_trans_allocate): Generate code for setting
>       ERRMSG depending on result of STAT result of ALLOCATE.
>       * trans.cc (gfc_allocate_using_malloc): Use STAT value of
>       LIBERROR_NO_MEMORY in case of failed malloc.
>
> gcc/testsuite/ChangeLog:
>
>       PR fortran/91300
>       * gfortran.dg/allocate_alloc_opt_15.f90: New test.
> ---
...

> +  stat1   = -1
> +  errmsg1 = ""
> +  allocate (array(1), stat=stat1, errmsg=errmsg1)
> +  if (stat1   /=  0) stop 1
> +  if (errmsg1 /= "") stop 1
Consider to init the errmsg1 and then check that is has not been
touched. (For completeness, I think we already have such tests).
> +  ! Obtain stat,errmsg for attempt to allocate an allocated object
> +  allocate (array(1), stat=stat1, errmsg=errmsg1)
> +  if (stat1   ==  0) stop 2
> +  if (errmsg1 == "") stop 2
Consider to check (either here or as a third test) for the
gfortran-specific error message.
> +  stat2   = -1
> +  errmsg2 = ""
> +  ! Try to allocate very large object
> +  allocate (bigarray(bignumber), stat=stat2, errmsg=errmsg2)
> +  if (stat2 /= 0) then
> +     print *, "stat  =", stat1
> +     print *, "errmsg: ", trim (errmsg1)
> +     print *, "stat  =", stat2
> +     print *, "errmsg: ", trim (errmsg2)
> +     ! Ensure different results for stat, errmsg variables
> +     if (stat2   == stat1                     ) stop 3
> +     if (errmsg2 == "" .or. errmsg2 == errmsg1) stop 4

Likewise for errmsg2

Thanks,

Tobias

-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955
  

Patch

From 19ccd22ee9359bd14b32a95bd9efcaead3593b2f Mon Sep 17 00:00:00 2001
From: Harald Anlauf <anlauf@gmx.de>
Date: Sat, 28 May 2022 22:02:20 +0200
Subject: [PATCH] Fortran: improve runtime error message with ALLOCATE and
 ERRMSG=

ALLOCATE: generate different STAT,ERRMSG results for failures from
allocation of already allocated objects or insufficient virtual memory.

gcc/fortran/ChangeLog:

	PR fortran/91300
	* libgfortran.h: Define new error code LIBERROR_NO_MEMORY.
	* trans-stmt.cc (gfc_trans_allocate): Generate code for setting
	ERRMSG depending on result of STAT result of ALLOCATE.
	* trans.cc (gfc_allocate_using_malloc): Use STAT value of
	LIBERROR_NO_MEMORY in case of failed malloc.

gcc/testsuite/ChangeLog:

	PR fortran/91300
	* gfortran.dg/allocate_alloc_opt_15.f90: New test.
---
 gcc/fortran/libgfortran.h                     |  1 +
 gcc/fortran/trans-stmt.cc                     | 33 +++++++++++++--
 gcc/fortran/trans.cc                          |  4 +-
 .../gfortran.dg/allocate_alloc_opt_15.f90     | 40 +++++++++++++++++++
 4 files changed, 73 insertions(+), 5 deletions(-)
 create mode 100644 gcc/testsuite/gfortran.dg/allocate_alloc_opt_15.f90

diff --git a/gcc/fortran/libgfortran.h b/gcc/fortran/libgfortran.h
index 064795eb469..4328447be04 100644
--- a/gcc/fortran/libgfortran.h
+++ b/gcc/fortran/libgfortran.h
@@ -133,6 +133,7 @@  typedef enum
   LIBERROR_CORRUPT_FILE,
   LIBERROR_INQUIRE_INTERNAL_UNIT, /* Must be different from STAT_STOPPED_IMAGE.  */
   LIBERROR_BAD_WAIT_ID,
+  LIBERROR_NO_MEMORY,
   LIBERROR_LAST			/* Not a real error, the last error # + 1.  */
 }
 libgfortran_error_codes;
diff --git a/gcc/fortran/trans-stmt.cc b/gcc/fortran/trans-stmt.cc
index 79096816c6e..fd6d294147e 100644
--- a/gcc/fortran/trans-stmt.cc
+++ b/gcc/fortran/trans-stmt.cc
@@ -7130,7 +7130,8 @@  gfc_trans_allocate (gfc_code * code)
   if (code->expr1 && code->expr2)
     {
       const char *msg = "Attempt to allocate an allocated object";
-      tree slen, dlen, errmsg_str;
+      const char *oommsg = "Insufficient virtual memory";
+      tree slen, dlen, errmsg_str, oom_str, oom_loc;
       stmtblock_t errmsg_block;

       gfc_init_block (&errmsg_block);
@@ -7151,8 +7152,34 @@  gfc_trans_allocate (gfc_code * code)
 			     gfc_default_character_kind);
       dlen = gfc_finish_block (&errmsg_block);

-      tmp = fold_build2_loc (input_location, NE_EXPR, logical_type_node,
-			     stat, build_int_cst (TREE_TYPE (stat), 0));
+      tmp = fold_build2_loc (input_location, EQ_EXPR, logical_type_node,
+			     stat, build_int_cst (TREE_TYPE (stat),
+						  LIBERROR_ALLOCATION));
+
+      tmp = build3_v (COND_EXPR, tmp,
+		      dlen, build_empty_stmt (input_location));
+
+      gfc_add_expr_to_block (&block, tmp);
+
+      oom_str = gfc_create_var (pchar_type_node, "OOMMSG");
+      oom_loc = gfc_build_localized_cstring_const (oommsg);
+      gfc_add_modify (&errmsg_block, oom_str,
+		      gfc_build_addr_expr (pchar_type_node, oom_loc));
+
+      slen = build_int_cst (gfc_charlen_type_node, strlen (oommsg));
+      dlen = gfc_get_expr_charlen (code->expr2);
+      slen = fold_build2_loc (input_location, MIN_EXPR,
+			      TREE_TYPE (slen), dlen, slen);
+
+      gfc_trans_string_copy (&errmsg_block, dlen, errmsg,
+			     code->expr2->ts.kind,
+			     slen, oom_str,
+			     gfc_default_character_kind);
+      dlen = gfc_finish_block (&errmsg_block);
+
+      tmp = fold_build2_loc (input_location, EQ_EXPR, logical_type_node,
+			     stat, build_int_cst (TREE_TYPE (stat),
+						  LIBERROR_NO_MEMORY));

       tmp = build3_v (COND_EXPR, tmp,
 		      dlen, build_empty_stmt (input_location));
diff --git a/gcc/fortran/trans.cc b/gcc/fortran/trans.cc
index f0a5dfb50fc..912a206f2ed 100644
--- a/gcc/fortran/trans.cc
+++ b/gcc/fortran/trans.cc
@@ -772,7 +772,7 @@  gfc_call_malloc (stmtblock_t * block, tree type, tree size)
       if (newmem == NULL)
       {
         if (stat)
-          *stat = LIBERROR_ALLOCATION;
+	  *stat = LIBERROR_NO_MEMORY;
         else
 	  runtime_error ("Allocation would exceed memory limit");
       }
@@ -807,7 +807,7 @@  gfc_allocate_using_malloc (stmtblock_t * block, tree pointer,
   if (status != NULL_TREE)
     {
       tmp = fold_build2_loc (input_location, MODIFY_EXPR, status_type, status,
-			     build_int_cst (status_type, LIBERROR_ALLOCATION));
+			     build_int_cst (status_type, LIBERROR_NO_MEMORY));
       gfc_add_expr_to_block (&on_error, tmp);
     }
   else
diff --git a/gcc/testsuite/gfortran.dg/allocate_alloc_opt_15.f90 b/gcc/testsuite/gfortran.dg/allocate_alloc_opt_15.f90
new file mode 100644
index 00000000000..04f7bf11970
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/allocate_alloc_opt_15.f90
@@ -0,0 +1,40 @@ 
+! { dg-do run }
+! PR fortran/91300 - runtime error message with allocate and errmsg=
+! Contributed by zed.three
+
+program bigarray_prog
+  use, intrinsic :: iso_c_binding, only: C_INTPTR_T
+  implicit none
+  real, dimension(:), allocatable :: array, bigarray
+  integer :: stat1, stat2
+  character(len=100) :: errmsg1, errmsg2
+  integer(8), parameter :: n1 = huge (1) / 3            ! request more than 2GB
+  integer(8), parameter :: n2 = huge (1_C_INTPTR_T) / 4 ! "safe" for 64bit
+  integer(8), parameter :: bignumber = max (n1, n2)
+
+  stat1   = -1
+  errmsg1 = ""
+  allocate (array(1), stat=stat1, errmsg=errmsg1)
+  if (stat1   /=  0) stop 1
+  if (errmsg1 /= "") stop 1
+
+  ! Obtain stat,errmsg for attempt to allocate an allocated object
+  allocate (array(1), stat=stat1, errmsg=errmsg1)
+  if (stat1   ==  0) stop 2
+  if (errmsg1 == "") stop 2
+
+  stat2   = -1
+  errmsg2 = ""
+  ! Try to allocate very large object
+  allocate (bigarray(bignumber), stat=stat2, errmsg=errmsg2)
+  if (stat2 /= 0) then
+     print *, "stat  =", stat1
+     print *, "errmsg: ", trim (errmsg1)
+     print *, "stat  =", stat2
+     print *, "errmsg: ", trim (errmsg2)
+     ! Ensure different results for stat, errmsg variables
+     if (stat2   == stat1                     ) stop 3
+     if (errmsg2 == "" .or. errmsg2 == errmsg1) stop 4
+  end if
+
+end program bigarray_prog
--
2.35.3