Re: [PATCH V2 2/4] vdpasim: customize allocation size

From: Jason Wang
Date: Thu Dec 22 2022 - 03:50:03 EST


On Thu, Dec 22, 2022 at 4:22 PM Eugenio Perez Martin
<eperezma@xxxxxxxxxx> wrote:
>
> On Thu, Dec 22, 2022 at 6:01 AM Jason Wang <jasowang@xxxxxxxxxx> wrote:
> >
> > Allow individual simulator to customize the allocation size.
> >
> > Signed-off-by: Jason Wang <jasowang@xxxxxxxxxx>
> > ---
> > drivers/vdpa/vdpa_sim/vdpa_sim.c | 8 ++++++--
> > drivers/vdpa/vdpa_sim/vdpa_sim.h | 1 +
> > drivers/vdpa/vdpa_sim/vdpa_sim_blk.c | 1 +
> > drivers/vdpa/vdpa_sim/vdpa_sim_net.c | 1 +
> > 4 files changed, 9 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.c b/drivers/vdpa/vdpa_sim/vdpa_sim.c
> > index 757afef86ba0..55aaa023a6e2 100644
> > --- a/drivers/vdpa/vdpa_sim/vdpa_sim.c
> > +++ b/drivers/vdpa/vdpa_sim/vdpa_sim.c
> > @@ -253,7 +253,10 @@ struct vdpasim *vdpasim_create(struct vdpasim_dev_attr *dev_attr,
> > struct vdpa_device *vdpa;
> > struct vdpasim *vdpasim;
> > struct device *dev;
> > - int i, ret = -ENOMEM;
> > + int i, ret = -EINVAL;
> > +
> > + if (!dev_attr->alloc_size)
> > + goto err_alloc;
>
> It's weird that we need an error goto here and the next chunk of code
> may return ERR_PTR(-EINVAL) directly. It's better to return directly
> here in my opinion.

Right, let me fix it.

>
> >
> > if (config->mask & BIT_ULL(VDPA_ATTR_DEV_FEATURES)) {
> > if (config->device_features &
> > @@ -268,9 +271,10 @@ struct vdpasim *vdpasim_create(struct vdpasim_dev_attr *dev_attr,
> > else
> > ops = &vdpasim_config_ops;
> >
> > + ret = -ENOMEM;
>
> Nitpicking again,
>
> To set ret = -ENOMEM here is weird to me, since the next failure on
> __vdpa_alloc_device will override it. I'd set it right before
> dma_set_mask_and_coherent, closer to its actual usage.

Yes.

>
> Actually, I'd propagate dma_set_mask_and_coherent error and set ret =
> -ENOMEM right before *alloc functions, but it wasn't done that way
> before this series, so not a reason to NAK to me.

Yes and I think this needs a separate fix.

Thanks

>
> Thanks!
>
> > vdpa = __vdpa_alloc_device(NULL, ops,
> > dev_attr->ngroups, dev_attr->nas,
> > - sizeof(struct vdpasim),
> > + dev_attr->alloc_size,
> > dev_attr->name, false);
> > if (IS_ERR(vdpa)) {
> > ret = PTR_ERR(vdpa);
> > diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.h b/drivers/vdpa/vdpa_sim/vdpa_sim.h
> > index 0e78737dcc16..51c070a543f1 100644
> > --- a/drivers/vdpa/vdpa_sim/vdpa_sim.h
> > +++ b/drivers/vdpa/vdpa_sim/vdpa_sim.h
> > @@ -37,6 +37,7 @@ struct vdpasim_dev_attr {
> > struct vdpa_mgmt_dev *mgmt_dev;
> > const char *name;
> > u64 supported_features;
> > + size_t alloc_size;
> > size_t config_size;
> > size_t buffer_size;
> > int nvqs;
> > diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c b/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c
> > index c6db1a1baf76..4f7c35f59aa5 100644
> > --- a/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c
> > +++ b/drivers/vdpa/vdpa_sim/vdpa_sim_blk.c
> > @@ -378,6 +378,7 @@ static int vdpasim_blk_dev_add(struct vdpa_mgmt_dev *mdev, const char *name,
> > dev_attr.nvqs = VDPASIM_BLK_VQ_NUM;
> > dev_attr.ngroups = VDPASIM_BLK_GROUP_NUM;
> > dev_attr.nas = VDPASIM_BLK_AS_NUM;
> > + dev_attr.alloc_size = sizeof(struct vdpasim);
> > dev_attr.config_size = sizeof(struct virtio_blk_config);
> > dev_attr.get_config = vdpasim_blk_get_config;
> > dev_attr.work_fn = vdpasim_blk_work;
> > diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim_net.c b/drivers/vdpa/vdpa_sim/vdpa_sim_net.c
> > index c3cb225ea469..20cd5cdff919 100644
> > --- a/drivers/vdpa/vdpa_sim/vdpa_sim_net.c
> > +++ b/drivers/vdpa/vdpa_sim/vdpa_sim_net.c
> > @@ -249,6 +249,7 @@ static int vdpasim_net_dev_add(struct vdpa_mgmt_dev *mdev, const char *name,
> > dev_attr.nvqs = VDPASIM_NET_VQ_NUM;
> > dev_attr.ngroups = VDPASIM_NET_GROUP_NUM;
> > dev_attr.nas = VDPASIM_NET_AS_NUM;
> > + dev_attr.alloc_size = sizeof(struct vdpasim);
> > dev_attr.config_size = sizeof(struct virtio_net_config);
> > dev_attr.get_config = vdpasim_net_get_config;
> > dev_attr.work_fn = vdpasim_net_work;
> > --
> > 2.25.1
> >
>