qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Michael S. Tsirkin" <mst@redhat.com>
To: Denis Plotnikov <dplotnikov@virtuozzo.com>
Cc: fam@euphon.net, kwolf@redhat.com, stefanha@redhat.com,
	qemu-block@nongnu.org, den@virtuozzo.com, qemu-devel@nongnu.org,
	mreitz@redhat.com, ehabkost@redhat.com
Subject: Re: [PATCH v1 3/4] virtio: increase virtuqueue sizes in new machine types
Date: Tue, 5 Nov 2019 15:52:53 -0500	[thread overview]
Message-ID: <20191105155224-mutt-send-email-mst@kernel.org> (raw)
In-Reply-To: <20191105161105.19016-4-dplotnikov@virtuozzo.com>

On Tue, Nov 05, 2019 at 07:11:04PM +0300, Denis Plotnikov wrote:
> Linux guests submit IO requests no longer than PAGE_SIZE * max_seg
> field reported by SCSI controler. Thus typical sequential read with
> 1 MB size results in the following pattern of the IO from the guest:
>   8,16   1    15754     2.766095122  2071  D   R 2095104 + 1008 [dd]
>   8,16   1    15755     2.766108785  2071  D   R 2096112 + 1008 [dd]
>   8,16   1    15756     2.766113486  2071  D   R 2097120 + 32 [dd]
>   8,16   1    15757     2.767668961     0  C   R 2095104 + 1008 [0]
>   8,16   1    15758     2.768534315     0  C   R 2096112 + 1008 [0]
>   8,16   1    15759     2.768539782     0  C   R 2097120 + 32 [0]
> The IO was generated by
>   dd if=/dev/sda of=/dev/null bs=1024 iflag=direct
> 
> This effectively means that on rotational disks we will observe 3 IOPS
> for each 2 MBs processed. This definitely negatively affects both
> guest and host IO performance.
> 
> The cure is relatively simple - we should report lengthy scatter-gather
> ability of the SCSI controller. Fortunately the situation here is very
> good. VirtIO transport layer can accomodate 1024 items in one request
> while we are using only 128. This situation is present since almost
> very beginning. 2 items are dedicated for request metadata thus we
> should publish VIRTQUEUE_MAX_SIZE - 2 as max_seg.
> 
> The following pattern is observed after the patch:
>   8,16   1     9921     2.662721340  2063  D   R 2095104 + 1024 [dd]
>   8,16   1     9922     2.662737585  2063  D   R 2096128 + 1024 [dd]
>   8,16   1     9923     2.665188167     0  C   R 2095104 + 1024 [0]
>   8,16   1     9924     2.665198777     0  C   R 2096128 + 1024 [0]
> which is much better.
> 
> To fix this particular case, the patch adds new machine types with
> extended virtqueue sizes to 256 which also increases max_seg to 254
> implicitly.
> 
> Suggested-by: Denis V. Lunev <den@openvz.org>
> Signed-off-by: Denis Plotnikov <dplotnikov@virtuozzo.com>
> ---



the way we normally do this is change the defaults to 256.
what is wrong with doing that?


>  hw/core/machine.c   | 14 ++++++++++++++
>  hw/i386/pc_piix.c   | 16 +++++++++++++---
>  hw/i386/pc_q35.c    | 14 ++++++++++++--
>  include/hw/boards.h |  6 ++++++
>  4 files changed, 45 insertions(+), 5 deletions(-)
> 
> diff --git a/hw/core/machine.c b/hw/core/machine.c
> index 55b08f1466..28013a0e3f 100644
> --- a/hw/core/machine.c
> +++ b/hw/core/machine.c
> @@ -24,6 +24,13 @@
>  #include "hw/pci/pci.h"
>  #include "hw/mem/nvdimm.h"
>  
> +GlobalProperty hw_compat_4_0_1[] = {
> +    { "virtio-blk-device", "queue-size", "128" },
> +    { "virtio-scsi-device", "virtqueue_size", "128" },
> +    { "vhost-scsi-device", "virtqueue_size", "128" },
> +};
> +const size_t hw_compat_4_0_1_len = G_N_ELEMENTS(hw_compat_4_0_1);
> +
>  GlobalProperty hw_compat_4_0[] = {
>      { "virtio-balloon-device", "qemu-4-0-config-size", "true" },
>  };
> @@ -157,6 +164,13 @@ GlobalProperty hw_compat_2_1[] = {
>  };
>  const size_t hw_compat_2_1_len = G_N_ELEMENTS(hw_compat_2_1);
>  
> +GlobalProperty hw_compat[] = {
> +    { "virtio-blk-device", "queue-size", "256" },
> +    { "virtio-scsi-device", "virtqueue_size", "256" },
> +    { "vhost-scsi-device", "virtqueue_size", "256" },
> +};
> +const size_t hw_compat_len = G_N_ELEMENTS(hw_compat);
> +
>  static char *machine_get_accel(Object *obj, Error **errp)
>  {
>      MachineState *ms = MACHINE(obj);
> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
> index 8ad8e885c6..2260a61b1b 100644
> --- a/hw/i386/pc_piix.c
> +++ b/hw/i386/pc_piix.c
> @@ -426,15 +426,27 @@ static void pc_i440fx_machine_options(MachineClass *m)
>      m->default_machine_opts = "firmware=bios-256k.bin";
>      m->default_display = "std";
>      machine_class_allow_dynamic_sysbus_dev(m, TYPE_RAMFB_DEVICE);
> +    compat_props_add(m->compat_props, hw_compat, hw_compat_len);
>  }
>  
> -static void pc_i440fx_4_0_machine_options(MachineClass *m)
> +static void pc_i440fx_4_0_2_machine_options(MachineClass *m)
>  {
>      pc_i440fx_machine_options(m);
>      m->alias = "pc";
>      m->is_default = 1;
>  }
>  
> +DEFINE_I440FX_MACHINE(v4_0_2, "pc-i440fx-4.0.2", NULL,
> +                      pc_i440fx_4_0_2_machine_options);
> +
> +static void pc_i440fx_4_0_machine_options(MachineClass *m)
> +{
> +    pc_i440fx_4_0_2_machine_options(m);
> +    m->alias = NULL;
> +    m->is_default = 0;
> +    compat_props_add(m->compat_props, hw_compat_4_0_1, hw_compat_4_0_1_len);
> +}
> +
>  DEFINE_I440FX_MACHINE(v4_0, "pc-i440fx-4.0", NULL,
>                        pc_i440fx_4_0_machine_options);
>  
> @@ -443,9 +455,7 @@ static void pc_i440fx_3_1_machine_options(MachineClass *m)
>      PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
>  
>      pc_i440fx_4_0_machine_options(m);
> -    m->is_default = 0;
>      m->smbus_no_migration_support = true;
> -    m->alias = NULL;
>      pcmc->pvh_enabled = false;
>      compat_props_add(m->compat_props, hw_compat_3_1, hw_compat_3_1_len);
>      compat_props_add(m->compat_props, pc_compat_3_1, pc_compat_3_1_len);
> diff --git a/hw/i386/pc_q35.c b/hw/i386/pc_q35.c
> index 45cc29d1ad..50ccd9ebcf 100644
> --- a/hw/i386/pc_q35.c
> +++ b/hw/i386/pc_q35.c
> @@ -363,14 +363,25 @@ static void pc_q35_machine_options(MachineClass *m)
>      machine_class_allow_dynamic_sysbus_dev(m, TYPE_INTEL_IOMMU_DEVICE);
>      machine_class_allow_dynamic_sysbus_dev(m, TYPE_RAMFB_DEVICE);
>      m->max_cpus = 288;
> +    compat_props_add(m->compat_props, hw_compat, hw_compat_len);
>  }
>  
> -static void pc_q35_4_0_1_machine_options(MachineClass *m)
> +static void pc_q35_4_0_2_machine_options(MachineClass *m)
>  {
>      pc_q35_machine_options(m);
>      m->alias = "q35";
>  }
>  
> +DEFINE_Q35_MACHINE(v4_0_2, "pc-q35-4.0.2", NULL,
> +                   pc_q35_4_0_2_machine_options);
> +
> +static void pc_q35_4_0_1_machine_options(MachineClass *m)
> +{
> +    pc_q35_4_0_2_machine_options(m);
> +    m->alias = NULL;
> +    compat_props_add(m->compat_props, hw_compat_4_0_1, hw_compat_4_0_1_len);
> +}
> +
>  DEFINE_Q35_MACHINE(v4_0_1, "pc-q35-4.0.1", NULL,
>                     pc_q35_4_0_1_machine_options);
>  
> @@ -378,7 +389,6 @@ static void pc_q35_4_0_machine_options(MachineClass *m)
>  {
>      pc_q35_4_0_1_machine_options(m);
>      m->default_kernel_irqchip_split = true;
> -    m->alias = NULL;
>      compat_props_add(m->compat_props, hw_compat_4_0, hw_compat_4_0_len);
>      compat_props_add(m->compat_props, pc_compat_4_0, pc_compat_4_0_len);
>  }
> diff --git a/include/hw/boards.h b/include/hw/boards.h
> index fe1885cbff..cf10632dac 100644
> --- a/include/hw/boards.h
> +++ b/include/hw/boards.h
> @@ -293,6 +293,9 @@ struct MachineState {
>      } \
>      type_init(machine_initfn##_register_types)
>  
> +extern GlobalProperty hw_compat_4_0_1[];
> +extern const size_t hw_compat_4_0_1_len;
> +
>  extern GlobalProperty hw_compat_4_0[];
>  extern const size_t hw_compat_4_0_len;
>  
> @@ -338,4 +341,7 @@ extern const size_t hw_compat_2_2_len;
>  extern GlobalProperty hw_compat_2_1[];
>  extern const size_t hw_compat_2_1_len;
>  
> +extern GlobalProperty hw_compat[];
> +extern const size_t hw_compat_len;
> +
>  #endif
> -- 
> 2.17.0


  reply	other threads:[~2019-11-05 21:06 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-05 16:11 [PATCH v1 0/4] virtio: fix IO request length in virtio SCSI/block Denis Plotnikov
2019-11-05 16:11 ` [PATCH v1 1/4] virtio: protect non-modern devices from too big virtqueue size setting Denis Plotnikov
2019-11-05 20:56   ` Michael S. Tsirkin
2019-11-06  7:46     ` Denis Plotnikov
2019-11-06  9:01       ` Stefan Hajnoczi
2019-11-06  9:19       ` Stefan Hajnoczi
2019-11-06 11:33       ` Michael S. Tsirkin
2019-11-06  9:18     ` Stefan Hajnoczi
2019-11-06 11:51       ` Michael S. Tsirkin
2019-11-05 16:11 ` [PATCH v1 2/4] virtio: make seg_max virtqueue size dependent Denis Plotnikov
2019-11-05 20:51   ` Michael S. Tsirkin
2019-11-06 10:07     ` Denis Lunev
2019-11-06 11:54       ` Michael S. Tsirkin
2019-11-08  7:43         ` Denis Plotnikov
2019-11-08  9:52           ` Michael S. Tsirkin
2019-11-05 16:11 ` [PATCH v1 3/4] virtio: increase virtuqueue sizes in new machine types Denis Plotnikov
2019-11-05 20:52   ` Michael S. Tsirkin [this message]
2019-11-05 16:11 ` [PATCH v1 4/4] iotests: add test for virtio-scsi and virtio-blk machine type settings Denis Plotnikov
2019-11-06  9:24   ` Stefan Hajnoczi
2019-11-06 10:04     ` Max Reitz
2019-11-06 19:26       ` Eduardo Habkost
2019-11-07 16:30         ` Cleber Rosa
2019-11-08  7:08           ` Denis Plotnikov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191105155224-mutt-send-email-mst@kernel.org \
    --to=mst@redhat.com \
    --cc=den@virtuozzo.com \
    --cc=dplotnikov@virtuozzo.com \
    --cc=ehabkost@redhat.com \
    --cc=fam@euphon.net \
    --cc=kwolf@redhat.com \
    --cc=mreitz@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).