Re: [Qemu-devel] [RFC PATCH 2/4] Xen: add a dummy vIOMMU to create/destroy vIOMMU in Xen

[Chao Gao]

On Thu, Mar 30, 2017 at 05:24:52PM +0100, Anthony PERARD wrote:
>Hi,
>
>On Fri, Mar 17, 2017 at 07:29:15PM +0800, Lan Tianyu wrote:
>> From: Chao Gao 
>> 
>> Since adding a dynamic sysbus device is forbidden, so choose TYPE_DEVICE
>> as parent class.
>> 
>> Signed-off-by: Chao Gao 
>> Signed-off-by: Lan Tianyu 
>> ---
>>  hw/xen/Makefile.objs |   1 +
>>  hw/xen/xen_viommu.c  | 116 
>> +++
>>  2 files changed, 117 insertions(+)
>>  create mode 100644 hw/xen/xen_viommu.c
>> 
>> +static void xen_viommu_realize(DeviceState *dev, Error **errp)
>> +{
>> +int rc;
>> +uint64_t cap;
>> +char *dom;
>> +char viommu_path[1024];
>> +XenVIOMMUState *s = XEN_VIOMMU_DEVICE(dev);
>> +
>> +s->id = -1;
>> +
>> +/* Read vIOMMU attributes from Xenstore. */
>> +dom = xs_get_domain_path(xenstore, xen_domid);
>> +snprintf(viommu_path, sizeof(viommu_path), "%s/viommu", dom);
>> +rc = xenstore_read_uint64(viommu_path, "base_addr", >base_addr);  
>
>Could these informations (base_addr and cap) be read from the command
>line instead of via xenstore?
>Any reason for these informations to be on xenstore?

Actually, we passed both via command line at first. We just concerned
whether it was ok to pass a address through command line since
we find no device does the similar thing.

>
>> +if (rc) {
>> +error_report("Can't get base address of vIOMMU");
>
>I think error_setg should be used instead of error_report.
>
>> +exit(1);
>
>Also, exit should be remove, and return instead. error_setg would be
>enough to signal that the device can not work.
>
>> +}
>> +
>> +rc = xenstore_read_uint64(viommu_path, "cap", >cap);
>> +if (rc) {
>> +error_report("Can't get capabilities of vIOMMU");
>> +exit(1);
>> +}
>> +
>> +rc = xc_viommu_query_cap(xen_xc, xen_domid, );
>
>Since xc_viommu_* seems to be new, you should use
>xendevicemodel_viommu_* instead, also you will need to define a stub for
>them to be able to compile QEMU against older version of Xen.

Will follow your suggestions above.

Thanks
Chao

>
>
>The patch looks good otherwise.
>
>Thanks,
>
>-- 
>Anthony PERARD

Re: [Qemu-devel] [RFC PATCH 2/4] Xen: add a dummy vIOMMU to create/destroy vIOMMU in Xen

[Anthony PERARD]

Hi,

On Fri, Mar 17, 2017 at 07:29:15PM +0800, Lan Tianyu wrote:
> From: Chao Gao 
> 
> Since adding a dynamic sysbus device is forbidden, so choose TYPE_DEVICE
> as parent class.
> 
> Signed-off-by: Chao Gao 
> Signed-off-by: Lan Tianyu 
> ---
>  hw/xen/Makefile.objs |   1 +
>  hw/xen/xen_viommu.c  | 116 
> +++
>  2 files changed, 117 insertions(+)
>  create mode 100644 hw/xen/xen_viommu.c
> 
> diff --git a/hw/xen/Makefile.objs b/hw/xen/Makefile.objs
> index d367094..e37d808 100644
> --- a/hw/xen/Makefile.objs
> +++ b/hw/xen/Makefile.objs
> @@ -3,3 +3,4 @@ common-obj-$(CONFIG_XEN_BACKEND) += xen_backend.o 
> xen_devconfig.o
>  
>  obj-$(CONFIG_XEN_PCI_PASSTHROUGH) += xen-host-pci-device.o
>  obj-$(CONFIG_XEN_PCI_PASSTHROUGH) += xen_pt.o xen_pt_config_init.o 
> xen_pt_graphics.o xen_pt_msi.o
> +obj-$(CONFIG_XEN) += xen_viommu.o
> diff --git a/hw/xen/xen_viommu.c b/hw/xen/xen_viommu.c
> new file mode 100644
> index 000..9bf9158
> --- /dev/null
> +++ b/hw/xen/xen_viommu.c
> @@ -0,0 +1,116 @@
> +/*
> + * Xen virtual IOMMU (virtual VT-D)
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> +
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> +
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, see .
> + */
> +
> +#include "qemu/osdep.h"
> +#include "qemu/error-report.h"
> +
> +#include "hw/qdev-core.h"
> +#include "hw/sysbus.h"
> +#include "hw/xen/xen.h"
> +#include "hw/xen/xen_backend.h"
> +
> +#define TYPE_XEN_VIOMMU_DEVICE "xen_viommu"
> +#define  XEN_VIOMMU_DEVICE(obj) \
> +OBJECT_CHECK(XenVIOMMUState, (obj), TYPE_XEN_VIOMMU_DEVICE)
> +
> +typedef struct XenVIOMMUState XenVIOMMUState;
> +
> +struct XenVIOMMUState {
> +DeviceState dev;
> +uint32_t id;
> +uint64_t cap;
> +uint64_t base_addr;
> +};
> +
> +static void xen_viommu_realize(DeviceState *dev, Error **errp)
> +{
> +int rc;
> +uint64_t cap;
> +char *dom;
> +char viommu_path[1024];
> +XenVIOMMUState *s = XEN_VIOMMU_DEVICE(dev);
> +
> +s->id = -1;
> +
> +/* Read vIOMMU attributes from Xenstore. */
> +dom = xs_get_domain_path(xenstore, xen_domid);
> +snprintf(viommu_path, sizeof(viommu_path), "%s/viommu", dom);
> +rc = xenstore_read_uint64(viommu_path, "base_addr", >base_addr);  

Could these informations (base_addr and cap) be read from the command
line instead of via xenstore?
Any reason for these informations to be on xenstore?

> +if (rc) {
> +error_report("Can't get base address of vIOMMU");

I think error_setg should be used instead of error_report.

> +exit(1);

Also, exit should be remove, and return instead. error_setg would be
enough to signal that the device can not work.

> +}
> +
> +rc = xenstore_read_uint64(viommu_path, "cap", >cap);
> +if (rc) {
> +error_report("Can't get capabilities of vIOMMU");
> +exit(1);
> +}
> +
> +rc = xc_viommu_query_cap(xen_xc, xen_domid, );

Since xc_viommu_* seems to be new, you should use
xendevicemodel_viommu_* instead, also you will need to define a stub for
them to be able to compile QEMU against older version of Xen.


The patch looks good otherwise.

Thanks,

-- 
Anthony PERARD