openssl/apps/vms_decc_argv.c
Richard Levitte ce506d27ab testutil: ensure good treatment of argv on non-Unix platforms
From a Unix point of view, some other platform families have certain
quirks.  Windows command prompt doesn't expand globs into actual file
names, so we must do this.  VMS has some oddity with argv pointer size
that can cause crashes if you're not careful (by copying it to a less
surprising pointer size array).

The fixups already exist and are used in the apps/ code.  However, the
testutil code started using the opt routines from apps/ without
including the non-Unix fixups.  This change fixes that.

For VMS' sake, libtestutil gets an app_malloc() shim, to avoid sucking
in all of apps/apps.c.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8381)
2019-03-05 08:53:19 +01:00

68 lines
2.1 KiB
C

/*
* Copyright 2015-2019 The OpenSSL Project Authors. All Rights Reserved.
*
* Licensed under the Apache License 2.0 (the "License"). You may not use
* this file except in compliance with the License. You can obtain a copy
* in the file LICENSE in the source distribution or at
* https://www.openssl.org/source/license.html
*/
#include <stdlib.h>
#include <openssl/crypto.h>
#include "platform.h" /* for copy_argv() */
#include "apps.h" /* for app_malloc() */
char **newargv = NULL;
static void cleanup_argv(void)
{
OPENSSL_free(newargv);
newargv = NULL;
}
char **copy_argv(int *argc, char *argv[])
{
/*-
* The note below is for historical purpose. On VMS now we always
* copy argv "safely."
*
* 2011-03-22 SMS.
* If we have 32-bit pointers everywhere, then we're safe, and
* we bypass this mess, as on non-VMS systems.
* Problem 1: Compaq/HP C before V7.3 always used 32-bit
* pointers for argv[].
* Fix 1: For a 32-bit argv[], when we're using 64-bit pointers
* everywhere else, we always allocate and use a 64-bit
* duplicate of argv[].
* Problem 2: Compaq/HP C V7.3 (Alpha, IA64) before ECO1 failed
* to NULL-terminate a 64-bit argv[]. (As this was written, the
* compiler ECO was available only on IA64.)
* Fix 2: Unless advised not to (VMS_TRUST_ARGV), we test a
* 64-bit argv[argc] for NULL, and, if necessary, use a
* (properly) NULL-terminated (64-bit) duplicate of argv[].
* The same code is used in either case to duplicate argv[].
* Some of these decisions could be handled in preprocessing,
* but the code tends to get even uglier, and the penalty for
* deciding at compile- or run-time is tiny.
*/
int i, count = *argc;
char **p = newargv;
cleanup_argv();
newargv = app_malloc(sizeof(*newargv) * (count + 1), "argv copy");
if (newargv == NULL)
return NULL;
/* Register automatic cleanup on first use */
if (p == NULL)
OPENSSL_atexit(cleanup_argv);
for (i = 0; i < count; i++)
newargv[i] = argv[i];
newargv[i] = NULL;
*argc = i;
return newargv;
}