Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220330204804.660819-1-robdclark@gmail.com> <20220330204804.660819-8-robdclark@gmail.com>
 <83979c7b-8a8a-5006-6af3-f3ca8b0d8ced@collabora.com>
In-Reply-To: <83979c7b-8a8a-5006-6af3-f3ca8b0d8ced@collabora.com>
From:   Rob Clark <robdclark@gmail.com>
Date:   Thu, 31 Mar 2022 11:58:36 -0700
Message-ID: <CAF6AEGtEczCSzwMNcr2EJJ=OcncABC2ZM2dVpAYoJM+5TBTKXQ@mail.gmail.com>
Subject: Re: [PATCH v2 07/10] drm/msm/gem: Rework vma lookup and pin
To:     Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc:     dri-devel <dri-devel@lists.freedesktop.org>,
        freedreno <freedreno@lists.freedesktop.org>,
        linux-arm-msm <linux-arm-msm@vger.kernel.org>,
        Dmitry Baryshkov <dmitry.baryshkov@linaro.org>,
        Rob Clark <robdclark@chromium.org>,
        Sean Paul <sean@poorly.run>,
        Abhinav Kumar <quic_abhinavk@quicinc.com>,
        David Airlie <airlied@linux.ie>,
        Daniel Vetter <daniel@ffwll.ch>,
        open list <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Mar 31, 2022 at 11:27 AM Dmitry Osipenko
<dmitry.osipenko@collabora.com> wrote:
>
> On 3/30/22 23:47, Rob Clark wrote:
> > From: Rob Clark <robdclark@chromium.org>
> >
> > Combines duplicate vma lookup in the get_and_pin path.
> >
> > Signed-off-by: Rob Clark <robdclark@chromium.org>
> > ---
> >  drivers/gpu/drm/msm/msm_gem.c | 50 ++++++++++++++++++-----------------
> >  1 file changed, 26 insertions(+), 24 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/msm/msm_gem.c b/drivers/gpu/drm/msm/msm_gem.c
> > index deafae6feaa8..218744a490a4 100644
> > --- a/drivers/gpu/drm/msm/msm_gem.c
> > +++ b/drivers/gpu/drm/msm/msm_gem.c
> > @@ -376,39 +376,40 @@ put_iova_vmas(struct drm_gem_object *obj)
> >       }
> >  }
> >
> > -static int get_iova_locked(struct drm_gem_object *obj,
> > -             struct msm_gem_address_space *aspace, uint64_t *iova,
> > +static struct msm_gem_vma *get_vma_locked(struct drm_gem_object *obj,
> > +             struct msm_gem_address_space *aspace,
> >               u64 range_start, u64 range_end)
> >  {
> >       struct msm_gem_vma *vma;
> > -     int ret = 0;
> >
> >       GEM_WARN_ON(!msm_gem_is_locked(obj));
> >
> >       vma = lookup_vma(obj, aspace);
> >
> >       if (!vma) {
> > +             int ret;
> > +
> >               vma = add_vma(obj, aspace);
> >               if (IS_ERR(vma))
> > -                     return PTR_ERR(vma);
> > +                     return vma;
> >
> >               ret = msm_gem_init_vma(aspace, vma, obj->size,
> >                       range_start, range_end);
> >               if (ret) {
> You're allocation range_start -> range_end
>
>
> >                       del_vma(vma);
> > -                     return ret;
> > +                     return ERR_PTR(ret);
> >               }
> > +     } else {
> > +             GEM_WARN_ON(vma->iova < range_start);
> > +             GEM_WARN_ON((vma->iova + obj->size) > range_end);
>
> and then comparing range_start -> range_start + obj->size, hence you're
> assuming that range_end always equals to obj->size during the allocation.
>
> I'm not sure what is the idea here.. this looks inconsistent. I think
> you wanted to write:
>
>                 GEM_WARN_ON(vma->iova < range_start);
>                 GEM_WARN_ON(vma->iova + (vma->node.size << PAGE_SHIFT) > range_end);
>
> But is it really useful to check whether the new range is inside of the
> old range? Shouldn't it be always a error to change the IOVA range
> without reallocating vma?

There are a few cases (for allocations for GMU) where the range is
larger than the bo.. see a6xx_gmu_memory_alloc()

BR,
-R

>
> I'd expect to see:
>
>                 GEM_WARN_ON(vma->iova != range_start);
>                 GEM_WARN_ON(vma->iova + (vma->node.size << PAGE_SHIFT) != range_end);
>
> and then error out if range mismatches.