Skip to contentSkip to conversation

Probable ksmbd bug causing Create Response to show bad timestamps

DISCUSSION David Howells 2 participants 2 messages Jul 23, 2025 – Jul 24, 2025

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi Namjae,

I think I've found a bug in ksmbd. Using the attached program to create a file, stat it, do a DIO write to it and stat it again:

~/write /xfstest.test/foo 0 4K
sleep 1
stat /xfstest.test/foo
~/write -d /xfstest.test/foo 0 4K
stat /xfstest.test/foo

and having mounted a ksmbd share with cache=strict and actimeo=1, I see mtime and ctime getting corrupted:

File: /xfstest.test/foo
Size: 4096 Blocks: 8 IO Block: 1048576 regular file
Device: 0,65 Inode: 2033391 Links: 1
Access: (0755/-rwxr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root)
Context: system_u:object_r:cifs_t:s0
Access: 2025-07-23 22:15:30.136051900 +0100
Modify: 2025-07-23 22:15:30.136051900 +0100
Change: 2025-07-23 22:15:30.136051900 +0100
Birth: 2025-07-23 22:15:30.136051900 +0100
File: /xfstest.test/foo
Size: 4096 Blocks: 8 IO Block: 1048576 regular file
Device: 0,65 Inode: 2033391 Links: 1
Access: (0755/-rwxr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root)
Context: system_u:object_r:cifs_t:s0
Access: 2025-07-23 22:15:30.136051900 +0100
Modify: 1970-01-01 01:00:00.000000000 +0100
Change: 1970-01-01 01:00:00.000000000 +0100
Birth: 2025-07-23 22:15:30.136051900 +0100

I've also attached a PCAP file. Frames 16 and 19 show Create Response messages with bad Last Write and Last Change timestamps and I think this is being reflected in the stat output, possibly due to deferred close. Doing another stat a bit later shows the correct timestamps:

Access: 2025-07-23 22:15:32.279094100 +0100
Modify: 2025-07-23 22:15:32.279094100 +0100
Change: 2025-07-23 22:15:32.279094100 +0100
Birth: 2025-07-23 22:15:30.136051900 +0100

I've also added a kernel trace as requested by Steve.

David

#define _GNU_SOURCE
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <getopt.h>
#include <sys/fcntl.h>
#include <sys/mman.h>

#define OSERROR(X, Y) do { if ((long)(X) == -1) { perror(Y); exit(1); } } while(0)

static unsigned long parseval(char *p)

{
	unsigned long val;
	char *end;

	val = strtoul(p, &end, 0);
	switch (*end) {
	case 0: break;
	case 'k': case 'K': val *= 1024; break;
	case 'm': case 'M': val *= 1024 * 1024; break;
	case 'p': case 'P': val *= 4096; break;
	}
	return val;
}

static void format(void)

{
	printf("Format: write [-dD] <file> <off> <len> [<off> <len>]*\n");
	exit(2);
}

int main(int argc, char *argv[])

{
	const char *filename;
	unsigned int flags = O_CREAT|O_WRONLY;
	ssize_t wrote;
	size_t size;
	off_t pos;
	void *buf = NULL;
	int fd, opt;

	while ((opt = getopt(argc, argv, "dD")) && opt != -1) {
		switch (opt) {
		case 'd':
			flags |= O_DIRECT;
			break;
		case 'D':
			flags |= O_DSYNC;
			break;
		}
	}

	argc -= optind;
	argv += optind;
	if (argc < 3)
		format();
	filename = argv[0];
	argc--;
	argv++;

	if (argc & 1)
		format();

	fd = open(filename, flags, 0666);
	OSERROR(fd, filename);

	do {
		pos = parseval(argv[0]);
		size = parseval(argv[1]);
		if (size) {
			buf = mmap(NULL, size, PROT_READ|PROT_WRITE,
				   MAP_PRIVATE|MAP_ANON, -1, 0);
			OSERROR(buf, "mmap");
		} else {
			buf = "";
		}

		wrote = pwrite(fd, buf, size, pos);
		argv += 2;
		argc -= 2;

		OSERROR(wrote, filename);
		if (wrote < size)
			fprintf(stderr, "Wrote %zx not %zx\n", wrote, size);

		if (size)
			OSERROR(munmap(buf, size), "munmap");
	} while (argc >= 2);

	OSERROR(close(fd), "close");
	return 0;
}
maintainer
#1 lore raw
Quoted 3 lines · David Howells

Hi Namjae,

Hi David,

Quoted 3 lines

I think I've found a bug in ksmbd. Using the attached program to create a file, stat it, do a DIO write to it and stat it again:

Okay, I will take a look:)
Thanks for your report!

Quoted 43 lines

~/write /xfstest.test/foo 0 4K
sleep 1
stat /xfstest.test/foo
~/write -d /xfstest.test/foo 0 4K
stat /xfstest.test/foo

and having mounted a ksmbd share with cache=strict and actimeo=1, I see mtime and ctime getting corrupted:

File: /xfstest.test/foo
Size: 4096 Blocks: 8 IO Block: 1048576 regular file
Device: 0,65 Inode: 2033391 Links: 1
Access: (0755/-rwxr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root)
Context: system_u:object_r:cifs_t:s0
Access: 2025-07-23 22:15:30.136051900 +0100
Modify: 2025-07-23 22:15:30.136051900 +0100
Change: 2025-07-23 22:15:30.136051900 +0100
Birth: 2025-07-23 22:15:30.136051900 +0100
File: /xfstest.test/foo
Size: 4096 Blocks: 8 IO Block: 1048576 regular file
Device: 0,65 Inode: 2033391 Links: 1
Access: (0755/-rwxr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root)
Context: system_u:object_r:cifs_t:s0
Access: 2025-07-23 22:15:30.136051900 +0100
Modify: 1970-01-01 01:00:00.000000000 +0100
Change: 1970-01-01 01:00:00.000000000 +0100
Birth: 2025-07-23 22:15:30.136051900 +0100

I've also attached a PCAP file. Frames 16 and 19 show Create Response messages with bad Last Write and Last Change timestamps and I think this is being reflected in the stat output, possibly due to deferred close. Doing another stat a bit later shows the correct timestamps:

Access: 2025-07-23 22:15:32.279094100 +0100
Modify: 2025-07-23 22:15:32.279094100 +0100
Change: 2025-07-23 22:15:32.279094100 +0100
Birth: 2025-07-23 22:15:30.136051900 +0100

I've also added a kernel trace as requested by Steve.

David