test: fix incorrect hole count reporting on loongarch64

On LoongArch64 with 16KB pages, Linux tmpfs often uses Transparent
Huge Pages (THP) sized at 32MiB. When a test writes data followed
by a 32MiB hole, the kernel may allocate a single huge page folio
to back the entire range in RAM. Because the range is physically
backed by memory, SEEK_HOLE correctly (from the kernel's perspective)
identifies the range as DATA, failing the test's assertion. Increasing
MIN_HOLE to 64MiB ensures the hole exceeds the huge page size on
loongarch64, preserving the sparse nature of the file for the test.

Signed-off-by: Pluto Yang <yangyj.ee@gmail.com>

diff --git a/libarchive/test/test_sparse_basic.c b/libarchive/test/test_sparse_basic.c
index 3914c3e30e977034fbd7238a62a9f9304a455a72..135bcefeb9d13211b69de4db8465b66fdc976794 100644 (file)
--- a/libarchive/test/test_sparse_basic.c
+++ b/libarchive/test/test_sparse_basic.c
@@ -73,10 +73,16 @@ static void create_sparse_file(const char *, const struct sparse *);
 /* A few data points:
  * = ZFS on FreeBSD needs this to be at least 200kB
  * = macOS APFS needs this to be at least 4096x4097 bytes
+ * = Linux tmpfs on 16KB-page architectures (like LoongArch64) uses
+ *   32MiB Transparent Huge Pages (THP). If a hole is exactly the
+ *   size of a THP, the data blocks on either side can end up in
+ *   adjacent physical folios, causing SEEK_HOLE to report the range
+ *   as contiguous data.
  *
- * 32MiB here is bigger than either of the above.
+ * 64MiB here is enough to ensure a hole exists between THP folios on all
+ * common architectures.
  */
-#define MIN_HOLE (32 * 1024UL * 1024UL)
+#define MIN_HOLE (64 * 1024UL * 1024UL)
 
 #if defined(_WIN32) && !defined(__CYGWIN__)
 #include <winioctl.h>