| Message ID | 20230830103754.36461-8-zhenzhong.duan@intel.com |
|---|---|
| State | New |
| Headers | show |
| Series | vfio: Adopt iommufd | expand |
On 8/30/23 12:37, Zhenzhong Duan wrote: > So that it doesn't need to be moved into container.c as done > in following patch. This is a bit weird to refer to container.c which is not yet created. I would suggested just reuse the commit title as a commit msg + this will become easier to handle multiple IOMMU BEs Eric > > Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> > --- > hw/vfio/common.c | 17 ++++++++++++++++- > 1 file changed, 16 insertions(+), 1 deletion(-) > > diff --git a/hw/vfio/common.c b/hw/vfio/common.c > index 51c6e7598e..fda5fc87b9 100644 > --- a/hw/vfio/common.c > +++ b/hw/vfio/common.c > @@ -219,7 +219,22 @@ void vfio_unblock_multiple_devices_migration(void) > > bool vfio_viommu_preset(VFIODevice *vbasedev) > { > - return vbasedev->group->container->space->as != &address_space_memory; > + VFIOAddressSpace *space; > + VFIOContainer *container; > + VFIODevice *tmp_dev; > + > + QLIST_FOREACH(space, &vfio_address_spaces, list) { > + QLIST_FOREACH(container, &space->containers, next) { > + tmp_dev = NULL; > + while ((tmp_dev = vfio_container_dev_iter_next(container, > + tmp_dev))) { > + if (vbasedev == tmp_dev) { > + return space->as != &address_space_memory; > + } > + } > + } > + } > + g_assert_not_reached(); > } > > static void vfio_set_migration_error(int err)
On Wed, 30 Aug 2023 18:37:39 +0800 Zhenzhong Duan <zhenzhong.duan@intel.com> wrote: > So that it doesn't need to be moved into container.c as done > in following patch. > > Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> > --- > hw/vfio/common.c | 17 ++++++++++++++++- > 1 file changed, 16 insertions(+), 1 deletion(-) > > diff --git a/hw/vfio/common.c b/hw/vfio/common.c > index 51c6e7598e..fda5fc87b9 100644 > --- a/hw/vfio/common.c > +++ b/hw/vfio/common.c > @@ -219,7 +219,22 @@ void vfio_unblock_multiple_devices_migration(void) > > bool vfio_viommu_preset(VFIODevice *vbasedev) > { > - return vbasedev->group->container->space->as != &address_space_memory; > + VFIOAddressSpace *space; > + VFIOContainer *container; > + VFIODevice *tmp_dev; > + > + QLIST_FOREACH(space, &vfio_address_spaces, list) { > + QLIST_FOREACH(container, &space->containers, next) { > + tmp_dev = NULL; > + while ((tmp_dev = vfio_container_dev_iter_next(container, > + tmp_dev))) { > + if (vbasedev == tmp_dev) { > + return space->as != &address_space_memory; > + } > + } > + } > + } > + g_assert_not_reached(); Should the VFIODevice just have a pointer to the VFIOAddressSpace? Thanks, Alex > } > > static void vfio_set_migration_error(int err)
>-----Original Message----- >From: Eric Auger <eric.auger@redhat.com> >Sent: Wednesday, September 20, 2023 9:01 PM >Subject: Re: [PATCH v1 07/22] vfio/common: Refactor vfio_viommu_preset() to >be group agnostic > > > >On 8/30/23 12:37, Zhenzhong Duan wrote: >> So that it doesn't need to be moved into container.c as done >> in following patch. >This is a bit weird to refer to container.c which is not yet created. I >would suggested just reuse the commit title as a commit msg + this will >become easier to handle multiple IOMMU BEs Will fix, thanks Eric. BRs. Zhenzhong
>-----Original Message----- >From: Alex Williamson <alex.williamson@redhat.com> >Sent: Thursday, September 21, 2023 6:51 AM >Subject: Re: [PATCH v1 07/22] vfio/common: Refactor vfio_viommu_preset() to >be group agnostic > >On Wed, 30 Aug 2023 18:37:39 +0800 >Zhenzhong Duan <zhenzhong.duan@intel.com> wrote: > >> So that it doesn't need to be moved into container.c as done >> in following patch. >> >> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> >> --- >> hw/vfio/common.c | 17 ++++++++++++++++- >> 1 file changed, 16 insertions(+), 1 deletion(-) >> >> diff --git a/hw/vfio/common.c b/hw/vfio/common.c >> index 51c6e7598e..fda5fc87b9 100644 >> --- a/hw/vfio/common.c >> +++ b/hw/vfio/common.c >> @@ -219,7 +219,22 @@ void vfio_unblock_multiple_devices_migration(void) >> >> bool vfio_viommu_preset(VFIODevice *vbasedev) >> { >> - return vbasedev->group->container->space->as != &address_space_memory; >> + VFIOAddressSpace *space; >> + VFIOContainer *container; >> + VFIODevice *tmp_dev; >> + >> + QLIST_FOREACH(space, &vfio_address_spaces, list) { >> + QLIST_FOREACH(container, &space->containers, next) { >> + tmp_dev = NULL; >> + while ((tmp_dev = vfio_container_dev_iter_next(container, >> + tmp_dev))) { >> + if (vbasedev == tmp_dev) { >> + return space->as != &address_space_memory; >> + } >> + } >> + } >> + } >> + g_assert_not_reached(); > >Should the VFIODevice just have a pointer to the VFIOAddressSpace? After "[PATCH v1 13/22] vfio: Add base container", VFIODevice has pointer to base container which has pointer further to VFIOAddressSpace. Is that meet your expectation? Thanks Zhenzhong
diff --git a/hw/vfio/common.c b/hw/vfio/common.c index 51c6e7598e..fda5fc87b9 100644 --- a/hw/vfio/common.c +++ b/hw/vfio/common.c @@ -219,7 +219,22 @@ void vfio_unblock_multiple_devices_migration(void) bool vfio_viommu_preset(VFIODevice *vbasedev) { - return vbasedev->group->container->space->as != &address_space_memory; + VFIOAddressSpace *space; + VFIOContainer *container; + VFIODevice *tmp_dev; + + QLIST_FOREACH(space, &vfio_address_spaces, list) { + QLIST_FOREACH(container, &space->containers, next) { + tmp_dev = NULL; + while ((tmp_dev = vfio_container_dev_iter_next(container, + tmp_dev))) { + if (vbasedev == tmp_dev) { + return space->as != &address_space_memory; + } + } + } + } + g_assert_not_reached(); } static void vfio_set_migration_error(int err)
So that it doesn't need to be moved into container.c as done in following patch. Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> --- hw/vfio/common.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-)