Skip to contentSkip to conversation

syscalls/ustat: Skip tests on Btrfs using TCONF

PATCH Superseded Disha Goel 2 participants 3 messages Mar 18 – Mar 23

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Disha Goel

author
#0 lore raw

PATCH 1/1 syscalls/ustat: Skip tests on Btrfs using TCONF

The ustat(2) system call is known to fail with EINVAL on Btrfs because it uses anonymous device IDs for subvolumes, which the legacy syscall cannot resolve to a physical block device.

Currently, this results in a TFAIL, which causes false negatives in automated CI environments (e.g., SLES). This patch adds dynamic filesystem detection in setup() and uses tst_brk(TCONF, ...) to gracefully skip the test on Btrfs.

Changes:
- Added sys/vfs.h for statfs support.
- Implemented Btrfs magic number detection in setup().
- Used tst_brk(TCONF, ...) to skip the test if Btrfs is detected.

Verified on:
- SLES (Btrfs): Result is now TCONF (Skipped)
- RHEL (XFS): Result remains TPASS

---
 testcases/kernel/syscalls/ustat/ustat01.c | 8 +++++++-
 testcases/kernel/syscalls/ustat/ustat02.c | 8 +++++++-
 2 files changed, 14 insertions(+), 2 deletions(-)
testcases/kernel/syscalls/ustat/ustat01.c +7 -1
index 161006058..07ee2faf7 100644
@@ -13,6 +13,7 @@
 #include <errno.h>
 #include <sys/types.h>
 #include <sys/stat.h>
+#include <sys/vfs.h>
 
 #include "lapi/syscalls.h"
 #include "lapi/ustat.h"
@@ -34,11 +35,16 @@ void run(void)
 static void setup(void)
 {
 	struct stat buf;
+	struct statfs fs_buf;
 
 	/* Find a valid device number */
 	SAFE_STAT("/", &buf);
-
 	dev_num = buf.st_dev;
+
+	statfs("/", &fs_buf);
+	if (fs_buf.f_type == 0x9123683E) {
+		tst_brk(TCONF, "%s", test.tags[0].value);
+	}
 }
 
 static struct tst_test test = {
testcases/kernel/syscalls/ustat/ustat02.c +7 -1
index 84becaa1f..789ef57fc 100644
@@ -14,6 +14,7 @@
 #include <errno.h>
 #include <sys/stat.h>
 #include <sys/types.h>
+#include <sys/vfs.h>
 
 #include "lapi/syscalls.h"
 #include "lapi/ustat.h"
@@ -50,11 +51,16 @@ void run(unsigned int test)
 static void setup(void)
 {
 	struct stat buf;
+	struct statfs fs_buf;
 
 	/* Find a valid device number */
 	SAFE_STAT("/", &buf);
-
 	root_dev = buf.st_dev;
+
+	statfs("/", &fs_buf);
+	if (fs_buf.f_type == 0x9123683E) {
+		tst_brk(TCONF, "%s", test.tags[0].value);
+	}
 }
 
 static struct tst_test test = {
Signature · 5 lines
2.45.1


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

Hi!

Quoted 8 lines

The ustat(2) system call is known to fail with EINVAL on Btrfs because it uses anonymous device IDs for subvolumes, which the legacy syscall cannot resolve to a physical block device.

Currently, this results in a TFAIL, which causes false negatives in automated CI environments (e.g., SLES). This patch adds dynamic filesystem detection in setup() and uses tst_brk(TCONF, ...) to gracefully skip the test on Btrfs.

That's why the test has known-fail tag, but maybe it's time to accept that ustat() is broken on Btrfs and it's not going to get fixed.

However we do have .skip_filesystems in tst_test structure exactly for cases like that, so it should be used instead.

Signature · 5 lines
Cyril Hrubis
[email protected]

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
author
#2 lore raw
Quoted 10 lines · Cyril Hrubis

Hi!

Quoted 8 lines

The ustat(2) system call is known to fail with EINVAL on Btrfs because it uses anonymous device IDs for subvolumes, which the legacy syscall cannot resolve to a physical block device.

Currently, this results in a TFAIL, which causes false negatives in automated CI environments (e.g., SLES). This patch adds dynamic filesystem detection in setup() and uses tst_brk(TCONF, ...) to gracefully skip the test on Btrfs.

Quoted 6 lines

That's why the test has known-fail tag, but maybe it's time to accept that ustat() is broken on Btrfs and it's not going to get fixed.

However we do have .skip_filesystems in tst_test structure exactly for cases like that, so it should be used instead.

Hi Cyril,

Thank you for the review.

I will refactor both ustat01 and ustat02 to use .skip_filesystems and send a v2.

Signature · 6 lines
Regards,
Disha


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