From: Ahmad Fatoum <a.fatoum@pengutronix.de>
To: barebox@lists.infradead.org
Cc: Ahmad Fatoum <a.fatoum@pengutronix.de>
Subject: [PATCH 1/2] commands: test: support signed comparisons
Date: Wed, 13 Sep 2023 13:59:57 +0200 [thread overview]
Message-ID: <20230913115958.1858470-1-a.fatoum@pengutronix.de> (raw)
So far signed integers in binary comparisons were accepted, but compared
unsigned resulting in [ "-1" -gt 0 ] returning true.
Rework this, so that:
- If one side is signed, the comparison is signed
- If both sides are unsigned, the comparison is unsigned
The latter is not required by POSIX, but it ensures that we don't break
users who so far did comparisons with integers > LONG_MAX without issue.
While at it, we start checking that the arguments in arithmetic
comparisons are indeed integers. This has the result that e.g.
[ "1x" -eq "1x" ] will no longer return success, but that aligns us with
usual shell behavior and protects the users from silent truncation when
a number > LONG_MAX is compared with a negative number.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
---
commands/test.c | 66 ++++++++++++++++++++++++++++++++++++++-----------
1 file changed, 51 insertions(+), 15 deletions(-)
diff --git a/commands/test.c b/commands/test.c
index 1130bf2e5bd6..13005b97deb3 100644
--- a/commands/test.c
+++ b/commands/test.c
@@ -78,11 +78,56 @@ static int string_comp(const char *left_op, const char *right_op, bool bash_test
return strcmp(left_op, right_op);
}
+static int parse_number(const char *str, long *num, bool signedcmp)
+{
+ int ret;
+
+ ret = signedcmp ? kstrtol(str, 0, num) : kstrtoul(str, 0, num);
+ if (ret)
+ printf("test: %s: integer expression expected\n", str);
+
+ return ret;
+}
+
+#define __do_arith_cmp(x, op, y, signedcmp) \
+ ((signedcmp) ? (long)(x) op (long)(y) : (x) op (y))
+
+static int arith_comp(const char *a_str, const char *b_str, int op)
+{
+ ulong a, b;
+ bool signedcmp = a_str[0] == '-' || b_str[0] == '-';
+ int ret;
+
+ ret = parse_number(a_str, &a, signedcmp);
+ if (ret)
+ return ret;
+
+ ret = parse_number(b_str, &b, signedcmp);
+ if (ret)
+ return ret;
+
+ switch (op) {
+ case OPT_ARITH_EQUAL:
+ return __do_arith_cmp(a, ==, b, signedcmp);
+ case OPT_ARITH_NOT_EQUAL:
+ return __do_arith_cmp(a, !=, b, signedcmp);
+ case OPT_ARITH_GREATER_EQUAL:
+ return __do_arith_cmp(a, >=, b, signedcmp);
+ case OPT_ARITH_GREATER_THAN:
+ return __do_arith_cmp(a, >, b, signedcmp);
+ case OPT_ARITH_LESS_EQUAL:
+ return __do_arith_cmp(a, <=, b, signedcmp);
+ case OPT_ARITH_LESS_THAN:
+ return __do_arith_cmp(a, <, b, signedcmp);
+ }
+
+ return -EINVAL;
+}
+
static int do_test(int argc, char *argv[])
{
char **ap;
int left, adv, expr, last_expr, neg, last_cmp, opt, zero;
- ulong a, b;
struct stat statbuf;
bool bash_test = false;
@@ -199,9 +244,8 @@ static int do_test(int argc, char *argv[])
if (left < 3)
break;
- a = simple_strtol(ap[0], NULL, 0);
- b = simple_strtol(ap[2], NULL, 0);
- switch (parse_opt(ap[1])) {
+ opt = parse_opt(ap[1]);
+ switch (opt) {
case OPT_EQUAL:
case OPT_EQUAL_BASH:
expr = string_comp(ap[0], ap[2], bash_test) == 0;
@@ -210,22 +254,14 @@ static int do_test(int argc, char *argv[])
expr = string_comp(ap[0], ap[2], bash_test) != 0;
break;
case OPT_ARITH_EQUAL:
- expr = a == b;
- break;
case OPT_ARITH_NOT_EQUAL:
- expr = a != b;
- break;
case OPT_ARITH_LESS_THAN:
- expr = a < b;
- break;
case OPT_ARITH_LESS_EQUAL:
- expr = a <= b;
- break;
case OPT_ARITH_GREATER_THAN:
- expr = a > b;
- break;
case OPT_ARITH_GREATER_EQUAL:
- expr = a >= b;
+ expr = arith_comp(ap[0], ap[2], opt);
+ if (expr < 0)
+ return 1;
break;
default:
expr = 1;
--
2.39.2
next reply other threads:[~2023-09-13 12:01 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-09-13 11:59 Ahmad Fatoum [this message]
2023-09-13 11:59 ` [PATCH 2/2] test: self: add some test cases for test command Ahmad Fatoum
2023-09-14 8:10 ` [PATCH 1/2] commands: test: support signed comparisons Sascha Hauer
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230913115958.1858470-1-a.fatoum@pengutronix.de \
--to=a.fatoum@pengutronix.de \
--cc=barebox@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox