Re: [PATCH drm-misc-next v6 1/6] drm/gpuvm: add common dma-resv per struct drm_gpuvm

From: Thomas Hellström
Date: Fri Oct 13 2023 - 09:00:15 EST


On Fri, 2023-10-13 at 13:51 +0200, Danilo Krummrich wrote:
> On 10/13/23 13:38, Thomas Hellström wrote:
> > On Mon, 2023-10-09 at 01:32 +0200, Danilo Krummrich wrote:
> > > Provide a common dma-resv for GEM objects not being used outside
> > > of
> > > this
> > > GPU-VM. This is used in a subsequent patch to generalize dma-
> > > resv,
> > > external and evicted object handling and GEM validation.
> > >
> > > Signed-off-by: Danilo Krummrich <dakr@xxxxxxxxxx>
> > > ---
> > >   drivers/gpu/drm/drm_gpuvm.c            | 56
> > > +++++++++++++++++++++++++-
> > >   drivers/gpu/drm/nouveau/nouveau_uvmm.c | 13 +++++-
> > >   include/drm/drm_gpuvm.h                | 35 +++++++++++++++-
> > >   3 files changed, 99 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/drm_gpuvm.c
> > > b/drivers/gpu/drm/drm_gpuvm.c
> > > index 02ecb45a2544..ebda9d594165 100644
> > > --- a/drivers/gpu/drm/drm_gpuvm.c
> > > +++ b/drivers/gpu/drm/drm_gpuvm.c
> > > @@ -61,6 +61,15 @@
> > >    * contained within struct drm_gpuva already. Hence, for
> > > inserting
> > > &drm_gpuva
> > >    * entries from within dma-fence signalling critical sections
> > > it is
> > > enough to
> > >    * pre-allocate the &drm_gpuva structures.
> > > + *
> > > + * &drm_gem_objects which are private to a single VM can share a
> > > common
> > > + * &dma_resv in order to improve locking efficiency (e.g. with
> > > &drm_exec).
> > > + * For this purpose drivers must pass a &drm_gem_object to
> > > drm_gpuvm_init(), in
> > > + * the following called 'root object', which serves as the
> > > container
> >
> > Nit: Perhaps resv object altough it might typically be the root
> > page-
> > table object, that doesn't have any meaning to drm_gpuvm, which
> > uses it
> > solely as a container for the resv?
>
> With "root" I didn't want to refer to the object representing the
> root
> page-table object, but being *the* object every other (internal)
> object
> needs to keep a reference to.

OK, yes but the reason they need a reference is because of the shared
resv, so IMO resv_object is a good fit. (I later noticed there's even
the function name drm_gpuvm_resv_obj()). And it will probably get
confused with the driver's "root" page table object, but up to you.

> Maybe I should be more explicit here and say
> that drivers need to make sure every internal object requires a
> reference
> to take a reference to this root object.
>
> >
> > > of the
> > > + * GPUVM's shared &dma_resv. This root object can be a driver
> > > specific
> > > + * &drm_gem_object, such as the &drm_gem_object containing the
> > > root
> > > page table,
> > > + * but it can also be a 'dummy' object, which can be allocated
> > > with
> > > + * drm_gpuvm_root_object_alloc().
> > >    */
> > >  
> > >   /**
> > > @@ -652,9 +661,47 @@ drm_gpuvm_range_valid(struct drm_gpuvm
> > > *gpuvm,
> > >                 !drm_gpuvm_in_kernel_node(gpuvm, addr, range);
> > >   }
> > >  
> > > +static void
> > > +drm_gpuvm_gem_object_free(struct drm_gem_object *obj)
> > > +{
> > > +       drm_gem_object_release(obj);
> > > +       kfree(obj);
> > > +}
> > > +
> > > +static const struct drm_gem_object_funcs drm_gpuvm_object_funcs
> > > = {
> > > +       .free = drm_gpuvm_gem_object_free,
> > > +};
> > > +
> > > +/**
> > > + * drm_gpuvm_root_object_alloc() - allocate a dummy
> > > &drm_gem_object
> > > + * @drm: the drivers &drm_device
> > > + *
> > > + * Allocates a dummy &drm_gem_object which can be passed to
> > > drm_gpuvm_init() in
> > > + * order to serve as root GEM object providing the &drm_resv
> > > shared
> > > across
> > > + * &drm_gem_objects local to a single GPUVM.
> > > + *
> > > + * Returns: the &drm_gem_object on success, NULL on failure
> > > + */
> > > +struct drm_gem_object *
> > > +drm_gpuvm_root_object_alloc(struct drm_device *drm)
> > > +{
> > > +       struct drm_gem_object *obj;
> > > +
> > > +       obj = kzalloc(sizeof(*obj), GFP_KERNEL);
> > > +       if (!obj)
> > > +               return NULL;
> > > +
> > > +       obj->funcs = &drm_gpuvm_object_funcs;
> > > +       drm_gem_private_object_init(drm, obj, 0);
> > > +
> > > +       return obj;
> > > +}
> > > +EXPORT_SYMBOL_GPL(drm_gpuvm_root_object_alloc);
> > > +
> > >   /**
> > >    * drm_gpuvm_init() - initialize a &drm_gpuvm
> > >    * @gpuvm: pointer to the &drm_gpuvm to initialize
> > > + * @r_obj: the root &drm_gem_object providing the GPUVM's common
> > > &dma_resv
> > >    * @name: the name of the GPU VA space
> > >    * @start_offset: the start offset of the GPU VA space
> > >    * @range: the size of the GPU VA space
> > > @@ -668,7 +715,7 @@ drm_gpuvm_range_valid(struct drm_gpuvm
> > > *gpuvm,
> > >    * &name is expected to be managed by the surrounding driver
> > > structures.
> > >    */
> > >   void
> > > -drm_gpuvm_init(struct drm_gpuvm *gpuvm,
> > > +drm_gpuvm_init(struct drm_gpuvm *gpuvm, struct drm_gem_object
> > > *r_obj,
> > >                 const char *name,
> > >                 u64 start_offset, u64 range,
> > >                 u64 reserve_offset, u64 reserve_range,
> > > @@ -683,6 +730,9 @@ drm_gpuvm_init(struct drm_gpuvm *gpuvm,
> > >  
> > >          gpuvm->name = name ? name : "unknown";
> > >          gpuvm->ops = ops;
> > > +       gpuvm->r_obj = r_obj;
> > > +
> > > +       drm_gem_object_get(r_obj);
> > >  
> > >          memset(&gpuvm->kernel_alloc_node, 0, sizeof(struct
> > > drm_gpuva));
> > >  
> > > @@ -713,7 +763,9 @@ drm_gpuvm_destroy(struct drm_gpuvm *gpuvm)
> > >                  __drm_gpuva_remove(&gpuvm->kernel_alloc_node);
> > >  
> > >          WARN(!RB_EMPTY_ROOT(&gpuvm->rb.tree.rb_root),
> > > -            "GPUVA tree is not empty, potentially leaking
> > > memory.");
> > > +            "GPUVA tree is not empty, potentially leaking
> > > memory.\n");
> >
> > Should we cache the drm device in struct drm_gpuvm and use
> > drm_warn()
> > here instead of WARN?
>
> I'd guess the additional backtrace of WARN() isn't overly useful in
> this
> case. However, it might be a bit more obvious in dmesg due to its
> verboseness. Not a strong opinion on that, though.

My bad. I meant drm_WARN(). In a multi-gpu environment it's nice to
have the extra device info.

/Thomas


>
> >
> > > +
> > > +       drm_gem_object_put(gpuvm->r_obj);
> > >   }
> > >   EXPORT_SYMBOL_GPL(drm_gpuvm_destroy);
> > >  
> > > diff --git a/drivers/gpu/drm/nouveau/nouveau_uvmm.c
> > > b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
> > > index 5cf892c50f43..4dea847ef989 100644
> > > --- a/drivers/gpu/drm/nouveau/nouveau_uvmm.c
> > > +++ b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
> > > @@ -1808,8 +1808,9 @@ int
> > >   nouveau_uvmm_init(struct nouveau_uvmm *uvmm, struct nouveau_cli
> > > *cli,
> > >                    u64 kernel_managed_addr, u64
> > > kernel_managed_size)
> > >   {
> > > -       int ret;
> > > +       struct drm_gem_object *r_obj;
> > >          u64 kernel_managed_end = kernel_managed_addr +
> > > kernel_managed_size;
> > > +       int ret;
> > >  
> > >          mutex_init(&uvmm->mutex);
> > >          dma_resv_init(&uvmm->resv);
> > > @@ -1833,14 +1834,22 @@ nouveau_uvmm_init(struct nouveau_uvmm
> > > *uvmm,
> > > struct nouveau_cli *cli,
> > >                  goto out_unlock;
> > >          }
> > >  
> > > +       r_obj = drm_gpuvm_root_object_alloc(cli->drm->dev);
> > > +       if (!r_obj) {
> > > +               ret = -ENOMEM;
> > > +               goto out_unlock;
> > > +       }
> > > +
> > >          uvmm->kernel_managed_addr = kernel_managed_addr;
> > >          uvmm->kernel_managed_size = kernel_managed_size;
> > >  
> > > -       drm_gpuvm_init(&uvmm->base, cli->name,
> > > +       drm_gpuvm_init(&uvmm->base, r_obj, cli->name,
> > >                         NOUVEAU_VA_SPACE_START,
> > >                         NOUVEAU_VA_SPACE_END,
> > >                         kernel_managed_addr, kernel_managed_size,
> > >                         NULL);
> > > +       /* GPUVM takes care from here on. */
> > > +       drm_gem_object_put(r_obj);
> > >  
> > >          ret = nvif_vmm_ctor(&cli->mmu, "uvmm",
> > >                              cli->vmm.vmm.object.oclass, RAW,
> > > diff --git a/include/drm/drm_gpuvm.h b/include/drm/drm_gpuvm.h
> > > index c7ed6bf441d4..0aec14d8b259 100644
> > > --- a/include/drm/drm_gpuvm.h
> > > +++ b/include/drm/drm_gpuvm.h
> > > @@ -238,9 +238,15 @@ struct drm_gpuvm {
> > >           * @ops: &drm_gpuvm_ops providing the split/merge steps
> > > to
> > > drivers
> > >           */
> > >          const struct drm_gpuvm_ops *ops;
> > > +
> > > +       /**
> > > +        * @r_obj: Root GEM object; representing the GPUVM's
> > > common
> > > &dma_resv.
> > > +        */
> > > +       struct drm_gem_object *r_obj;
> > >   };
> > >  
> > > -void drm_gpuvm_init(struct drm_gpuvm *gpuvm, const char *name,
> > > +void drm_gpuvm_init(struct drm_gpuvm *gpuvm, struct
> > > drm_gem_object
> > > *r_obj,
> > > +                   const char *name,
> > >                      u64 start_offset, u64 range,
> > >                      u64 reserve_offset, u64 reserve_range,
> > >                      const struct drm_gpuvm_ops *ops);
> > > @@ -248,6 +254,33 @@ void drm_gpuvm_destroy(struct drm_gpuvm
> > > *gpuvm);
> > >  
> > >   bool drm_gpuvm_interval_empty(struct drm_gpuvm *gpuvm, u64
> > > addr, u64
> > > range);
> > >  
> > > +struct drm_gem_object *
> > > +drm_gpuvm_root_object_alloc(struct drm_device *drm);
> > > +
> > > +/**
> > > + * drm_gpuvm_resv() - returns the &drm_gpuvm's &dma_resv
> > > + * @gpuvm__: the &drm_gpuvm
> > > + *
> > > + * Returns: a pointer to the &drm_gpuvm's shared &dma_resv
> > > + */
> > > +#define drm_gpuvm_resv(gpuvm__) ((gpuvm__)->r_obj->resv)
> > > +
> > > +/**
> > > + * drm_gpuvm_resv_obj() - returns the &drm_gem_object holding
> > > the
> > > &drm_gpuvm's
> > > + * &dma_resv
> > > + * @gpuvm__: the &drm_gpuvm
> > > + *
> > > + * Returns: a pointer to the &drm_gem_object holding the
> > > &drm_gpuvm's shared
> > > + * &dma_resv
> > > + */
> > > +#define drm_gpuvm_resv_obj(gpuvm__) ((gpuvm__)->r_obj)
> > > +
> > > +#define drm_gpuvm_resv_held(gpuvm__) \
> > > +       dma_resv_held(drm_gpuvm_resv(gpuvm__))
> > > +
> > > +#define drm_gpuvm_resv_assert_held(gpuvm__) \
> > > +       dma_resv_assert_held(drm_gpuvm_resv(gpuvm__))
> > > +
> > >   static inline struct drm_gpuva *
> > >   __drm_gpuva_next(struct drm_gpuva *va)
> > >   {
> >
> > Reviewed-by: Thomas Hellström <thomas.hellstrom@xxxxxxxxxxxxxxx>
> >
> >
>