Skip to contentSkip to conversation

open: fix cleanup condition and use snprintf

PATCH Superseded Jinseok Kim 2 participants 5 messages Feb 12 – Feb 18

v1 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 open: fix cleanup condition and use snprintf

The test uses sprintf() to build temporary file names, which may overflow the fixed-size buffer. Replace it with snprintf() to avoid potential buffer overflows.

The cleanup logic also checked '!first' to decide whether to close file descriptors. Since file descriptor 0 is valid, this condition can incorrectly skip cleanup and leak file descriptors.

---
 testcases/kernel/syscalls/open/open04.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
testcases/kernel/syscalls/open/open04.c +2 -2
index 3dc3486d3..ed5dd27bb 100644
@@ -30,7 +30,7 @@ static void setup(void)
 	fds[0] = first;
 
 	for (i = first + 1; i < fds_limit; i++) {
-		sprintf(fname, FNAME ".%d", i);
+		snprintf(fname, sizeof(fname), FNAME ".%d", i);
 		fd = open(fname, O_RDWR | O_CREAT, 0777);
 		if (fd == -1) {
 			if (errno != EMFILE)
@@ -50,7 +50,7 @@ static void run(void)
 
 static void cleanup(void)
 {
-	if (!first || !fds)
+	if (first < 0 || !fds)
 		return;
 
 	for (i = first; i < fds_limit; i++)

--
2.43.0

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#1 lore raw

PATCH 2/2 open: replace getdtablesize with getrlimit

The test currently uses getdtablesize() to determine the maximum number of file descriptors for the process. This interface is considered legacy and is not specified by POSIX.

Use getrlimit() instead, which provides a well-defined and portable way to obtain the per-process file descriptor limit.

---
 testcases/kernel/syscalls/open/open04.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
testcases/kernel/syscalls/open/open04.c +5 -1
index ed5dd27bb..3d8ea2ae8 100644
@@ -11,6 +11,7 @@
 
 #include <stdio.h>
 #include <stdlib.h>
+#include <sys/resource.h>
 #include "tst_test.h"
 
 #define FNAME "open04"
@@ -23,7 +24,10 @@ static void setup(void)
 {
 	int fd;
 
-	fds_limit = getdtablesize();
+	struct rlimit rlim;
+
+	SAFE_GETRLIMIT(RLIMIT_NOFILE, &rlim);
+	fds_limit = rlim.rlim_cur;
 	first = SAFE_OPEN(FNAME, O_RDWR | O_CREAT, 0777);
 
 	fds = SAFE_MALLOC(sizeof(int) * (fds_limit - first));

--
2.43.0

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#2 lore raw

Re: [LTP] [PATCH 2/2] open: replace getdtablesize with getrlimit

Hi!

Patch is ok, but it might be good to change sprintf() with snprintf() in the setup() as well to ensure safety.

King regards,

Signature · 7 lines
Andrea Cervesato
SUSE QE Automation Engineer Linux
[email protected]


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#3 lore raw

PATCH v2 1/2 open: fix cleanup condition and use snprintf

Hi Cervesato.

Thanks for the review!

Changes since v1:
- Replace remaining sprintf() with snprintf() in setup()

Thanks,
Jinseok.

---
 testcases/kernel/syscalls/open/open04.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
testcases/kernel/syscalls/open/open04.c +3 -3
index 3dc3486d3..5d39c1569 100644
@@ -30,7 +30,7 @@ static void setup(void)
 	fds[0] = first;
 
 	for (i = first + 1; i < fds_limit; i++) {
-		sprintf(fname, FNAME ".%d", i);
+		snprintf(fname, sizeof(fname), FNAME ".%d", i);
 		fd = open(fname, O_RDWR | O_CREAT, 0777);
 		if (fd == -1) {
 			if (errno != EMFILE)
@@ -44,13 +44,13 @@ static void setup(void)
 
 static void run(void)
 {
-	sprintf(fname, FNAME ".%d", fds_limit);
+	snprintf(fname, sizeof(fname), FNAME ".%d", fds_limit);
 	TST_EXP_FAIL2(open(fname, O_RDWR | O_CREAT, 0777), EMFILE);
 }
 
 static void cleanup(void)
 {
-	if (!first || !fds)
+	if (first < 0 || !fds)
 		return;
 
 	for (i = first; i < fds_limit; i++)

--
2.43.0

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#4 lore raw
Quoted 11 lines · Jinseok Kim

Hi Cervesato.

Thanks for the review!

Changes since v1:
- Replace remaining sprintf() with snprintf() in setup()

Thanks,
Jinseok.

Are you using an automatic tool? This seems like a wrong commit message. Please resend the whole patch-set with a proper version and right commit message. I will reject this one.

Kind regards,

Signature · 7 lines
Andrea Cervesato
SUSE QE Automation Engineer Linux
[email protected]


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp