Manoj Joseph wrote:
> Manoj Joseph wrote:
>> Manoj Joseph wrote:
>>> Hi,
>>>
>>> In brief, what I am trying to do is to use libzpool to access a zpool
>>> - like ztest does.
>>
>> [snip]
>>
>>> No, AFAIK, the pool is not damaged. But yes, it looks like the device
>>> can't be written to by the userland zfs.
>>
>> Well, I might have figured out something.
>>
>> Turssing the process shows this:
>>
>> /1: open64("/dev/rdsk/c2t0d0s0", O_RDWR|O_LARGEFILE) = 3
>> /108: pwrite64(3, " X0101\0140104\n $\0\r ".., 638, 4198400) Err#22
>> EINVAL
>> /108: pwrite64(3, "FC BFC BFC BFC BFC BFC B".., 386, 4199038) Err#22
>> EINVAL
>> [more failures...]
>>
>> The writes are not aligned to a block boundary. And, apparantly,
>> unlike files, this does not work for devices.
>>
>> Question: were ztest and libzpool not meant to be run on real devices?
>> Or could there be an issue in how I setup up things?
>
> The failing write has this call stack:
>
> pwrite64:return
> libc.so.1`_pwrite64+0x15
> libzpool.so.1`vn_rdwr+0x5b
> libzpool.so.1`vdev_file_io_start+0x17e
> libzpool.so.1`vdev_io_start+0x18
> libzpool.so.1`zio_vdev_io_start+0x33d
> [snip]
>
> usr/src/uts/common/fs/zfs/vdev_file.c has this:
>
> /*
> * From userland we access disks just like files.
> */
> #ifndef _KERNEL
>
> vdev_ops_t vdev_disk_ops = {
> vdev_file_open,
> vdev_file_close,
> vdev_default_asize,
> vdev_file_io_start,
> vdev_file_io_done,
> NULL,
> VDEV_TYPE_DISK, /* name of this vdev type */
> B_TRUE /* leaf vdev */
> };
>
> Guess vdev_file_io_start() does not work very well for devices.
Unlike what I had assumed earlier, zio_t that is passed to
vdev_file_io_start() has aligned offset and size.
The libzpool library, when writing data to the devices below a zpool,
splits the write into two. This is done for the sake of testing. The
comment in the routine, vn_rdwr() says this:
/*
* To simulate partial disk writes, we split writes into two
* system calls so that the process can be killed in between.
*/
This has the effect of creating misaligned writes to raw devices which
fail with errno=EINVAL.
Patching that solves the problem for me. :)
End of this thread! ;)
Cheers
Manoj
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
