From: "T.J. Mercier" <tjmercier@google.com>
To: "Christian König" <christian.koenig@amd.com>
Cc: devicetree@vger.kernel.org, Conor Dooley <conor+dt@kernel.org>,
jianjiao.zeng@mediatek.com,
Benjamin Gaignard <benjamin.gaignard@collabora.com>,
kuohong.wang@mediatek.com, linux-kernel@vger.kernel.org,
dri-devel@lists.freedesktop.org, linaro-mm-sig@lists.linaro.org,
Rob Herring <robh+dt@kernel.org>,
John Stultz <jstultz@google.com>,
linux-arm-kernel@lists.infradead.org,
Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
Matthias Brugger <matthias.bgg@gmail.com>,
linux-mediatek@lists.infradead.org, linux-media@vger.kernel.org,
Sumit Semwal <sumit.semwal@linaro.org>,
Yong Wu <yong.wu@mediatek.com>,
AngeloGioacchino Del Regno
<angelogioacchino.delregno@collabora.com>
Subject: Re: [PATCH 2/9] dma-heap: Add proper kref handling on dma-buf heaps
Date: Fri, 22 Sep 2023 11:19:46 -0700 [thread overview]
Message-ID: <CABdmKX1-5ujRRVQHNVMkoC4Ls0mruxE8GCXNavRLyywD7SobOw@mail.gmail.com> (raw)
In-Reply-To: <71c435a6-ba43-5d26-e658-f55bddbb8b98@amd.com>
On Mon, Sep 11, 2023 at 2:49 AM Christian König
<christian.koenig@amd.com> wrote:
>
> Am 11.09.23 um 04:30 schrieb Yong Wu:
> > From: John Stultz <jstultz@google.com>
> >
> > Add proper refcounting on the dma_heap structure.
> > While existing heaps are built-in, we may eventually
> > have heaps loaded from modules, and we'll need to be
> > able to properly handle the references to the heaps
> >
> > Also moves minor tracking into the heap structure so
> > we can properly free things.
>
> This is completely unnecessary, see below.
>
> >
> > Signed-off-by: John Stultz <jstultz@google.com>
> > Signed-off-by: T.J. Mercier <tjmercier@google.com>
> > Signed-off-by: Yong Wu <yong.wu@mediatek.com>
> > [Yong: Just add comment for "minor" and "refcount"]
> > ---
> > drivers/dma-buf/dma-heap.c | 38 ++++++++++++++++++++++++++++++++++----
> > include/linux/dma-heap.h | 6 ++++++
> > 2 files changed, 40 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/dma-buf/dma-heap.c b/drivers/dma-buf/dma-heap.c
> > index 51030f6c9d6e..dcc0e38c61fa 100644
> > --- a/drivers/dma-buf/dma-heap.c
> > +++ b/drivers/dma-buf/dma-heap.c
> > @@ -11,6 +11,7 @@
> > #include <linux/dma-buf.h>
> > #include <linux/dma-heap.h>
> > #include <linux/err.h>
> > +#include <linux/kref.h>
> > #include <linux/list.h>
> > #include <linux/nospec.h>
> > #include <linux/syscalls.h>
> > @@ -30,6 +31,8 @@
> > * @heap_devt: heap device node
> > * @list: list head connecting to list of heaps
> > * @heap_cdev: heap char device
> > + * @minor: heap device node minor number
> > + * @refcount: reference counter for this heap device
> > *
> > * Represents a heap of memory from which buffers can be made.
> > */
> > @@ -40,6 +43,8 @@ struct dma_heap {
> > dev_t heap_devt;
> > struct list_head list;
> > struct cdev heap_cdev;
> > + int minor;
> > + struct kref refcount;
> > };
> >
> > static LIST_HEAD(heap_list);
> > @@ -205,7 +210,6 @@ struct dma_heap *dma_heap_add(const struct dma_heap_export_info *exp_info)
> > {
> > struct dma_heap *heap, *h, *err_ret;
> > struct device *dev_ret;
> > - unsigned int minor;
> > int ret;
> >
> > if (!exp_info->name || !strcmp(exp_info->name, "")) {
> > @@ -222,12 +226,13 @@ struct dma_heap *dma_heap_add(const struct dma_heap_export_info *exp_info)
> > if (!heap)
> > return ERR_PTR(-ENOMEM);
> >
> > + kref_init(&heap->refcount);
> > heap->name = exp_info->name;
> > heap->ops = exp_info->ops;
> > heap->priv = exp_info->priv;
> >
> > /* Find unused minor number */
> > - ret = xa_alloc(&dma_heap_minors, &minor, heap,
> > + ret = xa_alloc(&dma_heap_minors, &heap->minor, heap,
> > XA_LIMIT(0, NUM_HEAP_MINORS - 1), GFP_KERNEL);
> > if (ret < 0) {
> > pr_err("dma_heap: Unable to get minor number for heap\n");
> > @@ -236,7 +241,7 @@ struct dma_heap *dma_heap_add(const struct dma_heap_export_info *exp_info)
> > }
> >
> > /* Create device */
> > - heap->heap_devt = MKDEV(MAJOR(dma_heap_devt), minor);
> > + heap->heap_devt = MKDEV(MAJOR(dma_heap_devt), heap->minor);
> >
> > cdev_init(&heap->heap_cdev, &dma_heap_fops);
> > ret = cdev_add(&heap->heap_cdev, heap->heap_devt, 1);
> > @@ -280,12 +285,37 @@ struct dma_heap *dma_heap_add(const struct dma_heap_export_info *exp_info)
> > err2:
> > cdev_del(&heap->heap_cdev);
> > err1:
> > - xa_erase(&dma_heap_minors, minor);
> > + xa_erase(&dma_heap_minors, heap->minor);
> > err0:
> > kfree(heap);
> > return err_ret;
> > }
> >
> > +static void dma_heap_release(struct kref *ref)
> > +{
> > + struct dma_heap *heap = container_of(ref, struct dma_heap, refcount);
> > +
> > + /* Note, we already holding the heap_list_lock here */
> > + list_del(&heap->list);
> > +
> > + device_destroy(dma_heap_class, heap->heap_devt);
> > + cdev_del(&heap->heap_cdev);
> > + xa_erase(&dma_heap_minors, heap->minor);
>
> You can just use MINOR(heap->heap_devt) here instead.
>
Got it, thanks.
> > +
> > + kfree(heap);
> > +}
> > +
> > +void dma_heap_put(struct dma_heap *h)
> > +{
> > + /*
> > + * Take the heap_list_lock now to avoid racing with code
> > + * scanning the list and then taking a kref.
> > + */
>
> This is usually considered a bad idea since it makes the kref approach
> superfluous.
>
> There are multiple possibilities how handle this, the most common one is
> to use kref_get_unless_zero() in your list traversal code and ignore the
> entry when that fails.
>
> Alternatively you could use kref_put_mutex() instead. This gives you the
> same functionality as this here, but as far as I know it's normally only
> used in a couple of special cases.
>
Ok, I'll move this mutex acquisition to dma_heap_release so that it
guards just the list_del, and change dma_heap_find to use
kref_get_unless_zero. Thanks.
> > + mutex_lock(&heap_list_lock);
> > + kref_put(&h->refcount, dma_heap_release);
> > + mutex_unlock(&heap_list_lock);
> > +}
> > +
> > static char *dma_heap_devnode(const struct device *dev, umode_t *mode)
> > {
> > return kasprintf(GFP_KERNEL, "dma_heap/%s", dev_name(dev));
> > diff --git a/include/linux/dma-heap.h b/include/linux/dma-heap.h
> > index c7c29b724ad6..f3c678892c5c 100644
> > --- a/include/linux/dma-heap.h
> > +++ b/include/linux/dma-heap.h
> > @@ -64,4 +64,10 @@ const char *dma_heap_get_name(struct dma_heap *heap);
> > */
> > struct dma_heap *dma_heap_add(const struct dma_heap_export_info *exp_info);
> >
> > +/**
> > + * dma_heap_put - drops a reference to a dmabuf heap, potentially freeing it
> > + * @heap: the heap whose reference count to decrement
> > + */
>
> Please don't add kerneldoc to the definition, add it to the
> implementation of the function.
>
Will fix.
> Regards,
> Christian.
>
> > +void dma_heap_put(struct dma_heap *heap);
> > +
> > #endif /* _DMA_HEAPS_H */
>
next prev parent reply other threads:[~2023-09-22 18:20 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-09-11 2:30 [PATCH 0/9] dma-buf: heaps: Add MediaTek secure heap Yong Wu
2023-09-11 2:30 ` [PATCH 1/9] dma-buf: heaps: Deduplicate docs and adopt common format Yong Wu
2023-09-11 9:36 ` Christian König
2023-09-11 23:51 ` T.J. Mercier
2023-09-11 2:30 ` [PATCH 2/9] dma-heap: Add proper kref handling on dma-buf heaps Yong Wu
2023-09-11 9:48 ` Christian König
2023-09-22 18:19 ` T.J. Mercier [this message]
2023-09-11 2:30 ` [PATCH 3/9] dma-heap: Provide accessors so that in-kernel drivers can allocate dmabufs from specific heaps Yong Wu
2023-09-11 10:13 ` Christian König
2023-09-11 18:29 ` John Stultz
2023-09-12 7:06 ` Christian König
2023-09-12 8:52 ` Yong Wu (吴勇)
2023-09-12 14:46 ` Christian König
2023-09-12 14:58 ` Nicolas Dufresne
2023-09-13 8:30 ` Christian König
2023-09-12 14:50 ` Nicolas Dufresne
2023-09-11 16:12 ` Nicolas Dufresne
2023-09-12 8:47 ` Yong Wu (吴勇)
2023-09-12 15:05 ` Nicolas Dufresne
2023-09-18 10:46 ` Yong Wu (吴勇)
2023-09-11 2:30 ` [PATCH 4/9] dma-buf: heaps: Initialise MediaTek secure heap Yong Wu
2023-09-11 8:05 ` kernel test robot
2023-09-27 14:42 ` Joakim Bech
2023-09-28 8:03 ` Yong Wu (吴勇)
2023-10-19 4:45 ` Vijayanand Jitta
2023-10-20 9:59 ` Yong Wu (吴勇)
2023-10-26 4:48 ` Vijayanand Jitta
2023-10-27 7:47 ` Yong Wu (吴勇)
2023-10-30 8:06 ` Vijayanand Jitta
2023-09-11 2:30 ` [PATCH 5/9] dma-buf: heaps: mtk_sec_heap: Initialise tee session Yong Wu
2023-09-11 9:29 ` AngeloGioacchino Del Regno
2023-09-11 10:15 ` Christian König
2023-09-12 6:17 ` Yong Wu (吴勇)
2023-09-12 9:32 ` AngeloGioacchino Del Regno
2023-09-25 12:49 ` Yong Wu (吴勇)
2023-09-27 13:46 ` Joakim Bech
2023-09-27 15:17 ` Benjamin Gaignard
2023-09-27 18:56 ` Jeffrey Kardatzke
2023-09-28 8:30 ` Benjamin Gaignard
2023-09-28 17:48 ` Jeffrey Kardatzke
2023-09-29 6:54 ` Benjamin Gaignard
2023-10-13 19:09 ` Jeffrey Kardatzke
2023-10-13 19:10 ` Jeffrey Kardatzke
2023-09-27 18:54 ` Jeffrey Kardatzke
2023-09-13 13:35 ` kernel test robot
2023-09-11 2:30 ` [PATCH 6/9] dma-buf: heaps: mtk_sec_heap: Add tee service call for buffer allocating/freeing Yong Wu
2023-09-14 10:18 ` kernel test robot
2023-09-27 14:37 ` Joakim Bech
2023-09-28 5:24 ` Yong Wu (吴勇)
2023-10-19 4:45 ` Vijayanand Jitta
2023-10-20 10:01 ` Yong Wu (吴勇)
2023-09-11 2:30 ` [PATCH 7/9] dma-buf: heaps: mtk_sec_heap: Add dma_ops Yong Wu
2023-09-11 2:30 ` [PATCH 8/9] dt-bindings: reserved-memory: MediaTek: Add reserved memory for SVP Yong Wu
2023-09-11 15:44 ` Rob Herring
2023-09-12 6:16 ` Yong Wu (吴勇)
2023-09-12 8:28 ` Krzysztof Kozlowski
2023-09-12 10:13 ` Robin Murphy
2023-09-12 15:53 ` Rob Herring
2023-09-12 16:05 ` Robin Murphy
2023-09-18 10:47 ` Yong Wu (吴勇)
2023-09-19 22:15 ` Jeffrey Kardatzke
2023-10-12 6:54 ` Yong Wu (吴勇)
2023-10-12 7:07 ` Krzysztof Kozlowski
2023-10-12 11:15 ` Yong Wu (吴勇)
2023-10-19 4:46 ` Vijayanand Jitta
2023-10-20 9:50 ` Yong Wu (吴勇)
2023-11-01 5:50 ` Jaskaran Singh
2023-11-06 5:56 ` Yong Wu (吴勇)
2023-11-20 8:20 ` Jaskaran Singh
2023-09-11 2:30 ` [PATCH 9/9] dma_buf: heaps: mtk_sec_heap: Add a new CMA heap Yong Wu
2023-09-11 9:33 ` AngeloGioacchino Del Regno
2023-10-19 4:44 ` [PATCH 0/9] dma-buf: heaps: Add MediaTek secure heap Vijayanand Jitta
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CABdmKX1-5ujRRVQHNVMkoC4Ls0mruxE8GCXNavRLyywD7SobOw@mail.gmail.com \
--to=tjmercier@google.com \
--cc=angelogioacchino.delregno@collabora.com \
--cc=benjamin.gaignard@collabora.com \
--cc=christian.koenig@amd.com \
--cc=conor+dt@kernel.org \
--cc=devicetree@vger.kernel.org \
--cc=dri-devel@lists.freedesktop.org \
--cc=jianjiao.zeng@mediatek.com \
--cc=jstultz@google.com \
--cc=krzysztof.kozlowski+dt@linaro.org \
--cc=kuohong.wang@mediatek.com \
--cc=linaro-mm-sig@lists.linaro.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=linux-mediatek@lists.infradead.org \
--cc=matthias.bgg@gmail.com \
--cc=robh+dt@kernel.org \
--cc=sumit.semwal@linaro.org \
--cc=yong.wu@mediatek.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).