2018-03-23 11:26 GMT-07:00 Pavel Shilovsky <piastryyy@xxxxxxxxx>:
> 2018-03-22 20:34 GMT-07:00 Xiong Zhou <xzhou@xxxxxxxxxx>:
>> On Fri, Mar 23, 2018 at 03:08:53AM +0000, Steve French wrote:
>>> Do you mean smb2 vs smb3 dialect? Or do you mean the 10 year old samba
>>> server
>>
>> I mean this:
>> "
>> vers=
>> SMB protocol version. Allowed values are:
>> · 1.0 - The classic CIFS/SMBv1 protocol. This is the default.
>> · 2.0 - The SMBv2.002 protocol. This was initially introduced in Windows Vista Service Pack 1, and Windows Server 2008. Note that the
>> initial release version of Windows Vista spoke a slightly different dialect (2.000) that is not supported.
>> · 2.1 - The SMBv2.1 protocol that was introduced in Microsoft Windows 7 and Windows Server 2008R2.
>> · 3.0 - The SMBv3.0 protocol that was introduced in Microsoft Windows 8 and Windows Server 2012.
>> "
>>
>> Do you mean v2 is not worth the time..
>>
>> Thanks,
>> Xiong
>>
>>>
>>> On Thu, Mar 22, 2018, 22:05 Xiong Zhou <xzhou@xxxxxxxxxx> wrote:
>>>
>>> > Hi,
>>> >
>>> > Process A open(2) file1 in samba 2.0 cifs mountpoint then sleep.
>>> > Process B try to open(2) the same file1. B is blocked in open
>>> > until A stops sleep and exit.
>>> >
>>> > O_NONBLOCK is not working either.
>>> >
>>> > This does not happen on samba 3.0.
>>> >
>>> > Reproduced on 4.16-rc5 Linus tree.
>>> >
>>> > Thanks,
>>> > Xiong
>>> >
>>> > Reproducer:
>>> >
>>> > --------------------- start test.sh ------------------------------
>>> > #! /bin/bash
>>> >
>>> > err_exit() { echo -e $1; exit 1; }
>>> >
>>> > setup_cifs()
>>> > {
>>> > local cifs_tst_export=/export/cifstest
>>> > local cifs_sch_export=/export/cifsscratch
>>> > mkdir -p $cifs_tst_export $cifs_sch_export
>>> >
>>> > \cp -f /etc/samba/smb.conf{,.bak}
>>> > cat > /etc/samba/smb.conf << EOF
>>> > [test]
>>> > path = $cifs_tst_export
>>> > writeable = yes
>>> > [scratch]
>>> > path = $cifs_sch_export
>>> > writeable = yes
>>> > EOF
>>> > which systemctl && systemctl restart smb > /dev/null 2>&1
>>> > service stop smb > /dev/null 2>&1
>>> > restorecon /etc/samba/smb.conf
>>> > testparm -s
>>> > chcon -t samba_share_t $cifs_sch_export
>>> > chcon -t samba_share_t $cifs_tst_export
>>> > echo -e "redhat\nredhat" | smbpasswd -a root
>>> > service start smb > /dev/null 2>&1
>>> > }
>>> >
>>> > preset_cifs()
>>> > {
>>> > local vers=${1:-3.0}
>>> > echo Testing $vers
>>> > TEST_DEV=//localhost/test
>>> > SCRATCH_DEV=//localhost/scratch
>>> > TEST_DIR=/cifsmnt
>>> > SCRATCH_MNT=/cifssch
>>> > umount $TEST_DEV $SCRATCH_DEV $TEST_DIR $SCRATCH_MNT > /dev/null
>>> > 2>&1
>>> > mkdir -p $TEST_DIR $SCRATCH_MNT
>>> > MOUNT_OPTIONS="-o vers=${vers},username=root,password=redhat"
>>> > TEST_FS_MOUNT_OPTS="-o vers=${vers},username=root,password=redhat"
>>> >
>>> > mount.cifs $TEST_DEV $TEST_DIR $MOUNT_OPTIONS || err_exit "cifs
>>> > mount $TEST_DEV failed"
>>> > mount.cifs $SCRATCH_DEV $SCRATCH_MNT $MOUNT_OPTIONS || err_exit
>>> > "cifs mount $SCRATCH_DEV failed"
>>> > }
>>> >
>>> > cleanup_cifs()
>>> > {
>>> > umount $SCRATCH_MNT $TEST_DIR
>>> > \cp -f /etc/samba/smb.conf{.bak,}
>>> > which systemctl && systemctl restart smb
>>> > service stop smb > /dev/null 2>&1
>>> > service start smb > /dev/null 2>&1
>>> > }
>>> >
>>> > do_test()
>>> > {
>>> > touch $SCRATCH_MNT/testfile
>>> > mount | grep $SCRATCH_MNT
>>> > ./open_wait $SCRATCH_MNT/testfile &
>>> > ./open $SCRATCH_MNT/testfile
>>> > echo ""
>>> > wait
>>> > }
>>> >
>>> > rpm -qv cifs-utils || yum intsall -y cifs-utils
>>> > rpm -qv samba || yum intsall -y samba samba-client
>>> > setup_cifs
>>> > cc open_wait.c -o open_wait
>>> > cc open.c -o open
>>> > preset_cifs 3.0
>>> > do_test
>>> > preset_cifs 2.0
>>> > do_test
>>> > preset_cifs 2.1
>>> > do_test
>>> > preset_cifs 1.0
>>> > do_test
>>> > cleanup_cifs
>>> > rm -f open_wait open a.out
>>> > exit
>>> > --------------------- end test.sh ------------------------------
>>> > --------------------- start open.c ------------------------------
>>> > #ifndef _GNU_SOURCE
>>> > #define _GNU_SOURCE
>>> > #endif
>>> > #include <unistd.h>
>>> > #include <fcntl.h>
>>> > #include <errno.h>
>>> > #include <stdio.h>
>>> > #include <unistd.h>
>>> > #include <string.h>
>>> > #include <stdlib.h>
>>> > #include <sched.h>
>>> > #include <sys/stat.h>
>>> > #include <sys/types.h>
>>> > #include <sys/wait.h>
>>> > #include <sys/ipc.h>
>>> > #include <sys/sem.h>
>>> >
>>> > static int fd;
>>> >
>>> > static void err_exit(char *op, int errn)
>>> > {
>>> > fprintf(stderr, "%s: %s\n", op, strerror(errn));
>>> > if (fd > 0)
>>> > close(fd);
>>> > exit(errn);
>>> > }
>>> >
>>> > int main(int argc, char **argv)
>>> > {
>>> > struct timeval tv1, tv2;
>>> > long double t1, t2;
>>> >
>>> > gettimeofday(&tv1, NULL);
>>> > //fd = open(argv[1], O_RDWR|O_NONBLOCK);
>>> > fd = open(argv[1], O_RDWR);
>>> > gettimeofday(&tv2, NULL);
>>> >
>>> > t1 = (long double)(tv1.tv_sec) + (long
>>> > double)(tv1.tv_usec/1000000.0);
>>> > t2 = (long double)(tv2.tv_sec) + (long
>>> > double)(tv2.tv_usec/1000000.0);
>>> > printf("open %s: %llf seconds\n", argv[1], t2 - t1);
>>> > err_exit("open", errno);
>>> > }
>>> > --------------------- end open.c ------------------------------
>>> > --------------------- start open_wait.c ------------------------------
>>> > #ifndef _GNU_SOURCE
>>> > #define _GNU_SOURCE
>>> > #endif
>>> > #include <errno.h>
>>> > #include <stdio.h>
>>> > #include <stdlib.h>
>>> > #include <unistd.h>
>>> > #include <sys/types.h>
>>> > #include <sys/stat.h>
>>> > #include <fcntl.h>
>>> >
>>> > static int fd;
>>> >
>>> > static void err_exit(char *op, int errn)
>>> > {
>>> > fprintf(stderr, "%s: %s\n", op, strerror(errn));
>>> > if (fd > 0)
>>> > close(fd);
>>> > exit(errn);
>>> > }
>>> >
>>> > int main(int argc, char **argv)
>>> > {
>>> > fd = open(argv[1], O_RDWR);
>>> > if (fd == -1)
>>> > err_exit("open", errno);
>>> >
>>> > sleep(2);
>>> > close(fd);
>>> > return 0;
>>> > }
>>> > --------------------- end open_wait.c ------------------------------
>
> I suspect there might be a problem with handling SMB 2.0 oplock break
> by the client. Adding Ronnie and Long to take a look since they
> changed the request handling code recently.
>
Xiong,
In the meantime could you provide a kernel debug output during the
execution of your test program as described in
https://wiki.samba.org/index.php/LinuxCIFS_troubleshooting ?
--
Best regards,
Pavel Shilovsky
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
