[Qemu-devel] [PATCH 0/3] vmw_pvscsi: Fine-tune device capabilities

[Qemu-devel] [PATCH 0/3] vmw_pvscsi: Fine-tune device capabilities

[Shmulik Ladkani]

Various fixes to what the vmw_pvscsi device reports in its PCI
configuration space, to better align with VMware virtual hardware
as exposed by ESXi/Workstation.

Shmulik Ladkani (3):
  vmw_pvscsi: Set device subsystem and revision
  vmw_pvscsi: The pvscsi device is a PCIE endpoint
  vmw_pvscsi: Change offsets of PCI capabilities

 hw/scsi/vmw_pvscsi.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

-- 
1.9.1

[Qemu-devel] [PATCH 1/3] vmw_pvscsi: Set device subsystem and revision

[Shmulik Ladkani]

To be VMware PVSCSI SCSI Controller, rev 02.
As reported by the VMware virtual hardware.

Signed-off-by: Shmulik Ladkani <shmulik.ladkani@ravellosystems.com>
---
 hw/scsi/vmw_pvscsi.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/hw/scsi/vmw_pvscsi.c b/hw/scsi/vmw_pvscsi.c
index 9c71f31..df4d383 100644
--- a/hw/scsi/vmw_pvscsi.c
+++ b/hw/scsi/vmw_pvscsi.c
@@ -1069,10 +1069,6 @@ pvscsi_init(PCIDevice *pci_dev)
 
     trace_pvscsi_state("init");
 
-    /* PCI subsystem ID */
-    pci_dev->config[PCI_SUBSYSTEM_ID] = 0x00;
-    pci_dev->config[PCI_SUBSYSTEM_ID + 1] = 0x10;
-
     /* PCI latency timer = 255 */
     pci_dev->config[PCI_LATENCY_TIMER] = 0xff;
 
@@ -1190,7 +1186,9 @@ static void pvscsi_class_init(ObjectClass *klass, void *data)
     k->vendor_id = PCI_VENDOR_ID_VMWARE;
     k->device_id = PCI_DEVICE_ID_VMWARE_PVSCSI;
     k->class_id = PCI_CLASS_STORAGE_SCSI;
-    k->subsystem_id = 0x1000;
+    k->subsystem_vendor_id = PCI_VENDOR_ID_VMWARE;
+    k->subsystem_id = PCI_DEVICE_ID_VMWARE_PVSCSI;
+    k->revision = 0x2;
     dc->reset = pvscsi_reset;
     dc->vmsd = &vmstate_pvscsi;
     dc->props = pvscsi_properties;
-- 
1.9.1

Re: [Qemu-devel] [PATCH 1/3] vmw_pvscsi: Set device subsystem and revision

[Dmitry Fleytman]

Acked-by: Dmitry Fleytman <dmitry@daynix.com>

> On 19 Nov 2015, at 16:21 PM, Shmulik Ladkani <shmulik.ladkani@ravellosystems.com> wrote:
> 
> To be VMware PVSCSI SCSI Controller, rev 02.
> As reported by the VMware virtual hardware.
> 
> Signed-off-by: Shmulik Ladkani <shmulik.ladkani@ravellosystems.com>
> ---
> hw/scsi/vmw_pvscsi.c | 8 +++-----
> 1 file changed, 3 insertions(+), 5 deletions(-)
> 
> diff --git a/hw/scsi/vmw_pvscsi.c b/hw/scsi/vmw_pvscsi.c
> index 9c71f31..df4d383 100644
> --- a/hw/scsi/vmw_pvscsi.c
> +++ b/hw/scsi/vmw_pvscsi.c
> @@ -1069,10 +1069,6 @@ pvscsi_init(PCIDevice *pci_dev)
> 
>     trace_pvscsi_state("init");
> 
> -    /* PCI subsystem ID */
> -    pci_dev->config[PCI_SUBSYSTEM_ID] = 0x00;
> -    pci_dev->config[PCI_SUBSYSTEM_ID + 1] = 0x10;
> -
>     /* PCI latency timer = 255 */
>     pci_dev->config[PCI_LATENCY_TIMER] = 0xff;
> 
> @@ -1190,7 +1186,9 @@ static void pvscsi_class_init(ObjectClass *klass, void *data)
>     k->vendor_id = PCI_VENDOR_ID_VMWARE;
>     k->device_id = PCI_DEVICE_ID_VMWARE_PVSCSI;
>     k->class_id = PCI_CLASS_STORAGE_SCSI;
> -    k->subsystem_id = 0x1000;
> +    k->subsystem_vendor_id = PCI_VENDOR_ID_VMWARE;
> +    k->subsystem_id = PCI_DEVICE_ID_VMWARE_PVSCSI;
> +    k->revision = 0x2;
>     dc->reset = pvscsi_reset;
>     dc->vmsd = &vmstate_pvscsi;
>     dc->props = pvscsi_properties;
> -- 
> 1.9.1
> 

[Qemu-devel] [PATCH 2/3] vmw_pvscsi: The pvscsi device is a PCIE endpoint

[Shmulik Ladkani]

Report the 'express endpoint' capability if on a PCIE bus.

Signed-off-by: Shmulik Ladkani <shmulik.ladkani@ravellosystems.com>
---
 hw/scsi/vmw_pvscsi.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/hw/scsi/vmw_pvscsi.c b/hw/scsi/vmw_pvscsi.c
index df4d383..8854fd9 100644
--- a/hw/scsi/vmw_pvscsi.c
+++ b/hw/scsi/vmw_pvscsi.c
@@ -1081,6 +1081,10 @@ pvscsi_init(PCIDevice *pci_dev)
 
     pvscsi_init_msi(s);
 
+    if (pci_bus_is_express(pci_dev->bus)) {
+        pcie_endpoint_cap_init(pci_dev, 0);
+    }
+
     s->completion_worker = qemu_bh_new(pvscsi_process_completion_queue, s);
     if (!s->completion_worker) {
         pvscsi_cleanup_msi(s);
@@ -1189,6 +1193,7 @@ static void pvscsi_class_init(ObjectClass *klass, void *data)
     k->subsystem_vendor_id = PCI_VENDOR_ID_VMWARE;
     k->subsystem_id = PCI_DEVICE_ID_VMWARE_PVSCSI;
     k->revision = 0x2;
+    k->is_express = 1;
     dc->reset = pvscsi_reset;
     dc->vmsd = &vmstate_pvscsi;
     dc->props = pvscsi_properties;
-- 
1.9.1

Re: [Qemu-devel] [PATCH 2/3] vmw_pvscsi: The pvscsi device is a PCIE endpoint

[Dmitry Fleytman]

Acked-by: Dmitry Fleytman <dmitry@daynix.com>

> On 19 Nov 2015, at 16:21 PM, Shmulik Ladkani <shmulik.ladkani@ravellosystems.com> wrote:
> 
> Report the 'express endpoint' capability if on a PCIE bus.
> 
> Signed-off-by: Shmulik Ladkani <shmulik.ladkani@ravellosystems.com>
> ---
> hw/scsi/vmw_pvscsi.c | 5 +++++
> 1 file changed, 5 insertions(+)
> 
> diff --git a/hw/scsi/vmw_pvscsi.c b/hw/scsi/vmw_pvscsi.c
> index df4d383..8854fd9 100644
> --- a/hw/scsi/vmw_pvscsi.c
> +++ b/hw/scsi/vmw_pvscsi.c
> @@ -1081,6 +1081,10 @@ pvscsi_init(PCIDevice *pci_dev)
> 
>     pvscsi_init_msi(s);
> 
> +    if (pci_bus_is_express(pci_dev->bus)) {
> +        pcie_endpoint_cap_init(pci_dev, 0);
> +    }
> +
>     s->completion_worker = qemu_bh_new(pvscsi_process_completion_queue, s);
>     if (!s->completion_worker) {
>         pvscsi_cleanup_msi(s);
> @@ -1189,6 +1193,7 @@ static void pvscsi_class_init(ObjectClass *klass, void *data)
>     k->subsystem_vendor_id = PCI_VENDOR_ID_VMWARE;
>     k->subsystem_id = PCI_DEVICE_ID_VMWARE_PVSCSI;
>     k->revision = 0x2;
> +    k->is_express = 1;
>     dc->reset = pvscsi_reset;
>     dc->vmsd = &vmstate_pvscsi;
>     dc->props = pvscsi_properties;
> -- 
> 1.9.1
> 

[Qemu-devel] [PATCH 3/3] vmw_pvscsi: Change offsets of PCI capabilities

[Shmulik Ladkani]

Place device reported PCI capabilities at the same offsets as placed by
the VMware virtual hardware:
  Express Endpoint at [40], MSI at [7c]

Signed-off-by: Shmulik Ladkani <shmulik.ladkani@ravellosystems.com>
---
 hw/scsi/vmw_pvscsi.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/hw/scsi/vmw_pvscsi.c b/hw/scsi/vmw_pvscsi.c
index 8854fd9..5c8cae1 100644
--- a/hw/scsi/vmw_pvscsi.c
+++ b/hw/scsi/vmw_pvscsi.c
@@ -32,7 +32,8 @@
 #include "trace.h"
 
 
-#define PVSCSI_MSI_OFFSET        (0x50)
+#define PVSCSI_EXP_EP_OFFSET     (0x40)
+#define PVSCSI_MSI_OFFSET        (0x7c)
 #define PVSCSI_USE_64BIT         (true)
 #define PVSCSI_PER_VECTOR_MASK   (false)
 
@@ -1082,7 +1083,7 @@ pvscsi_init(PCIDevice *pci_dev)
     pvscsi_init_msi(s);
 
     if (pci_bus_is_express(pci_dev->bus)) {
-        pcie_endpoint_cap_init(pci_dev, 0);
+        pcie_endpoint_cap_init(pci_dev, PVSCSI_EXP_EP_OFFSET);
     }
 
     s->completion_worker = qemu_bh_new(pvscsi_process_completion_queue, s);
-- 
1.9.1

Re: [Qemu-devel] [PATCH 3/3] vmw_pvscsi: Change offsets of PCI capabilities

[Dmitry Fleytman]

Acked-by: Dmitry Fleytman <dmitry@daynix.com>

> On 19 Nov 2015, at 16:21 PM, Shmulik Ladkani <shmulik.ladkani@ravellosystems.com> wrote:
> 
> Place device reported PCI capabilities at the same offsets as placed by
> the VMware virtual hardware:
>  Express Endpoint at [40], MSI at [7c]
> 
> Signed-off-by: Shmulik Ladkani <shmulik.ladkani@ravellosystems.com>
> ---
> hw/scsi/vmw_pvscsi.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/hw/scsi/vmw_pvscsi.c b/hw/scsi/vmw_pvscsi.c
> index 8854fd9..5c8cae1 100644
> --- a/hw/scsi/vmw_pvscsi.c
> +++ b/hw/scsi/vmw_pvscsi.c
> @@ -32,7 +32,8 @@
> #include "trace.h"
> 
> 
> -#define PVSCSI_MSI_OFFSET        (0x50)
> +#define PVSCSI_EXP_EP_OFFSET     (0x40)
> +#define PVSCSI_MSI_OFFSET        (0x7c)
> #define PVSCSI_USE_64BIT         (true)
> #define PVSCSI_PER_VECTOR_MASK   (false)
> 
> @@ -1082,7 +1083,7 @@ pvscsi_init(PCIDevice *pci_dev)
>     pvscsi_init_msi(s);
> 
>     if (pci_bus_is_express(pci_dev->bus)) {
> -        pcie_endpoint_cap_init(pci_dev, 0);
> +        pcie_endpoint_cap_init(pci_dev, PVSCSI_EXP_EP_OFFSET);
>     }
> 
>     s->completion_worker = qemu_bh_new(pvscsi_process_completion_queue, s);
> -- 
> 1.9.1
> 

Re: [Qemu-devel] [PATCH 0/3] vmw_pvscsi: Fine-tune device capabilities

[Paolo Bonzini]

On 19/11/2015 15:21, Shmulik Ladkani wrote:
> Various fixes to what the vmw_pvscsi device reports in its PCI
> configuration space, to better align with VMware virtual hardware
> as exposed by ESXi/Workstation.
> 
> Shmulik Ladkani (3):
>   vmw_pvscsi: Set device subsystem and revision
>   vmw_pvscsi: The pvscsi device is a PCIE endpoint
>   vmw_pvscsi: Change offsets of PCI capabilities
> 
>  hw/scsi/vmw_pvscsi.c | 16 ++++++++++------
>  1 file changed, 10 insertions(+), 6 deletions(-)
> 

Looks good, thanks!  I'll queue these patches for 2.5.

Paolo

Re: [Qemu-devel] [PATCH 0/3] vmw_pvscsi: Fine-tune device capabilities

[Shmulik Ladkani]

Hi,

On Thu, 19 Nov 2015 16:25:47 +0100 Paolo Bonzini <pbonzini@redhat.com> wrote:
> On 19/11/2015 15:21, Shmulik Ladkani wrote:
> > Various fixes to what the vmw_pvscsi device reports in its PCI
> > configuration space, to better align with VMware virtual hardware
> > as exposed by ESXi/Workstation.
> > 
> > Shmulik Ladkani (3):
> >   vmw_pvscsi: Set device subsystem and revision
> >   vmw_pvscsi: The pvscsi device is a PCIE endpoint
> >   vmw_pvscsi: Change offsets of PCI capabilities
> 
> Looks good, thanks!  I'll queue these patches for 2.5.

Thanks Paolo,

Since these are not yet pulled, I'm submitting a v2 of the series.

The above changes are guest visible, thus I've added compatability
properties for these changes - to allow migrating from old versions,
as suggested by Jason Wang for similar changes done to the vmxnet3
device (see [1]).

Regards,
Shmulik

[1]
https://lists.nongnu.org/archive/html/qemu-devel/2015-11/msg05515.html