Re: [Qemu-devel] [Nbd] [PATCH] nbd: fix trim/discard commands with a length bigger than NBD_MAX_BUFFER_SIZE

From: Eric Blake <eblake@redhat.com>
To: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Alex Bligh <alex@alex.org.uk>,
	"qemu-devel@nongnu.org" <qemu-devel@nongnu.org>,
	"qemu-trivial@nongnu.org" <qemu-trivial@nongnu.org>,
	Paolo Bonzini <pbonzini@redhat.com>,
	"nbd-general@lists.sourceforge.net"
	<nbd-general@lists.sourceforge.net>,
	"qemu-stable@nongnu.org" <qemu-stable@nongnu.org>,
	qemu block <qemu-block@nongnu.org>
Subject: Re: [Qemu-devel] [Nbd] [PATCH] nbd: fix trim/discard commands with a length bigger than NBD_MAX_BUFFER_SIZE
Date: Tue, 10 May 2016 14:24:56 -0600	[thread overview]
Message-ID: <57324398.7060106@redhat.com> (raw)
In-Reply-To: <20160510163309.GF28315@chrystal.uk.oracle.com>

[-- Attachment #1: Type: text/plain, Size: 2005 bytes --]

On 05/10/2016 10:33 AM, Quentin Casasnovas wrote:

> Looks like there's an easier way:
> 
>  $ qemu-img create -f qcow2 foo.qcow2 10G
>  $ qemu-nbd --discard=on -c /dev/nbd0 foo.qcow2
>  $ mkfs.ext4 /dev/nbd0
>  mke2fs 1.42.13 (17-May-2015)
>  Discarding device blocks: failed - Input/output error

strace on mkfs.ext4 shows:
...
open("/dev/nbd1", O_RDWR|O_EXCL)        = 3
stat("/dev/nbd1", {st_mode=S_IFBLK|0660, st_rdev=makedev(43, 1), ...}) = 0
ioctl(3, BLKDISCARDZEROES, [0])         = 0
ioctl(3, BLKROGET, [0])                 = 0
uname({sysname="Linux", nodename="red", ...}) = 0
ioctl(3, BLKDISCARD, [0, 1000000])      = 0
write(1, "Discarding device blocks: ", 26) = 26
write(1, "   4096/2621440", 15   4096/2621440)         = 15
write(1, "\10\10\10\10\10\10\10\10\10\10\10\1) = 15
ioctl(3, BLKDISCARD, [1000000, 80000000]) = -1 EIO (Input/output error)
write(1, "               ", 15               )         = 15
write(1, "\10\10\10\10\10\10\10\10\10\10\10\1) = 15
write(1, "failed - ", 9failed - )                = 9

so it does indeed look like the smaller request [0, 0x1000000] succeeded
while the larger [0x1000000, 0x80000000] failed with EIO.  But I
instrumented my qemu-nbd server to output all of the incoming requests
it was receiving from the kernel, and I never saw a mention of
NBD_CMD_TRIM being received.  Maybe it's dependent on what version of
the kernel and/or NBD kernel module you are running?

I also tried fallocate(1) on /dev/nbd1, as in:
 # strace fallocate -o 0 -l 64k -p /dev/nbd1
but it looks like the kernel doesn't yet wire up fallocate(2) to forward
to the appropriate device ioctls and/or NBD_CMD_TRIM, where strace says:

open("/dev/nbd1", O_RDWR)               = 3
fallocate(3, FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE, 0, 65536) = -1
ENODEV (No such device)

The ENODEV is a rather unusual choice of error.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 604 bytes --]