RE: [PATCH] media: staging/intel-ipu3: Implement lock for stream on/off operations
Hi Sakari, [snip] > > > > > fail_stop_pipeline: > > > > > @@ -543,6 +546,8 @@ static void ipu3_vb2_stop_streaming(struct > > > > vb2_queue *vq) > > > > > dev_err(>pci_dev->dev, > > > > > "failed to stop subdev streaming\n"); > > > > > > > > > > + mutex_lock(>streaming_lock); > > > > > + > > > > > /* Was this the first node with streaming disabled? */ > > > > > if (imgu->streaming && ipu3_all_nodes_streaming(imgu, node)) { > > > > > /* Yes, really stop streaming now */ @@ -552,6 +557,7 @@ > > > > static > > > > > void ipu3_vb2_stop_streaming(struct vb2_queue *vq) > > > > > imgu->streaming = false; > > > > > } > > > > > > > > > > + mutex_unlock(>streaming_lock); > > > > > ipu3_return_all_buffers(imgu, node, VB2_BUF_STATE_ERROR); > > > > I'd also call ipu3_return_all_buffers() before releasing the lock: in > > principle the user may have queued new buffers on the devices before > > the driver marks the buffers as faulty. > > Ack (Somehow I missed this comment.) [snip] ___ devel mailing list de...@linuxdriverproject.org http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
RE: [PATCH] media: staging/intel-ipu3: Implement lock for stream on/off operations
Hi Sakari, > Subject: Re: [PATCH] media: staging/intel-ipu3: Implement lock for stream > on/off operations > > Hi Raj, > > On Wed, Jan 30, 2019 at 05:17:15PM +0000, Mani, Rajmohan wrote: > > Hi Sakari, > > > > > -Original Message- > > > From: Sakari Ailus [mailto:sakari.ai...@linux.intel.com] > > > Sent: Wednesday, January 30, 2019 12:59 AM > > > To: Mani, Rajmohan > > > Cc: Mauro Carvalho Chehab ; Greg Kroah-Hartman > > > ; linux-me...@vger.kernel.org; > > > de...@driverdev.osuosl.org; linux-ker...@vger.kernel.org; Laurent > > > Pinchart ; Jacopo Mondi > > > ; Qiu, Tian Shu ; Cao, > > > Bingbu ; z...@paasikivi.fi.intel.com; Zhi, Yong > > > ; hverk...@xs4all.nl; tf...@chromium.org > > > Subject: Re: [PATCH] media: staging/intel-ipu3: Implement lock for > > > stream on/off operations > > > > > > Hi Rajmohan, > > > > > > On Tue, Jan 29, 2019 at 02:27:36PM -0800, Rajmohan Mani wrote: > > > > Currently concurrent stream off operations on ImgU nodes are not > > > > synchronized, leading to use-after-free bugs (as reported by KASAN). > > > > > > > > [ 250.090724] BUG: KASAN: use-after-free in > > > > ipu3_dmamap_free+0xc5/0x116 [ipu3_imgu] [ 250.090726] Read of > > > > size 8 at addr 888127b29bc0 by task yavta/18836 [ 250.090731] > > > > Hardware > > > > name: HP Soraka/Soraka, BIOS Google_Soraka.10431.17.0 03/22/2018 [ > > > 250.090732] Call Trace: > > > > [ 250.090735] dump_stack+0x6a/0xb1 [ 250.090739] > > > > print_address_description+0x8e/0x279 > > > > [ 250.090743] ? ipu3_dmamap_free+0xc5/0x116 [ipu3_imgu] [ > > > > 250.090746] kasan_report+0x260/0x28a [ 250.090750] > > > > ipu3_dmamap_free+0xc5/0x116 [ipu3_imgu] [ 250.090754] > > > > ipu3_css_pool_cleanup+0x24/0x37 [ipu3_imgu] [ 250.090759] > > > > ipu3_css_pipeline_cleanup+0x61/0xb9 [ipu3_imgu] [ 250.090763] > > > > ipu3_css_stop_streaming+0x1f2/0x321 [ipu3_imgu] [ 250.090768] > > > > imgu_s_stream+0x94/0x443 [ipu3_imgu] [ 250.090772] ? > > > > ipu3_vb2_buf_queue+0x280/0x280 [ipu3_imgu] [ 250.090775] ? > > > > vb2_dma_sg_unmap_dmabuf+0x16/0x6f [videobuf2_dma_sg] [ > > > > 250.090778] > > > ? > > > > vb2_buffer_in_use+0x36/0x58 [videobuf2_common] [ 250.090782] > > > > ipu3_vb2_stop_streaming+0xf9/0x135 [ipu3_imgu] > > > > > > > > Implemented a lock to synchronize imgu stream on / off operations > > > > and the modification of streaming flag (in struct imgu_device), to > > > > prevent these issues. > > > > > > > > Reported-by: Laurent Pinchart > > > > Suggested-by: Laurent Pinchart > > > > > > > > Signed-off-by: Rajmohan Mani > > > > --- > > > > drivers/staging/media/ipu3/ipu3-v4l2.c | 6 ++ > > > > drivers/staging/media/ipu3/ipu3.c | 3 +++ > > > > drivers/staging/media/ipu3/ipu3.h | 4 > > > > 3 files changed, 13 insertions(+) > > > > > > > > diff --git a/drivers/staging/media/ipu3/ipu3-v4l2.c > > > > b/drivers/staging/media/ipu3/ipu3-v4l2.c > > > > index c7936032beb9..cf7e917cd0c8 100644 > > > > --- a/drivers/staging/media/ipu3/ipu3-v4l2.c > > > > +++ b/drivers/staging/media/ipu3/ipu3-v4l2.c > > > > @@ -507,12 +507,15 @@ static int ipu3_vb2_start_streaming(struct > > > vb2_queue *vq, unsigned int count) > > > > goto fail_stop_pipeline; > > > > } > > > > > > > > + mutex_lock(>streaming_lock); > > > > + > > > > > > You appear to be using imgu_device.lock (while searching buffers to > > > queue to the device) as well as imgu_video_device.lock (qbuf, dqbuf) > > > to serialise access to imgu_video_device.buffers list. > > > > Ack > > > > > The two locks may be acquired at the same time but each by different > > > processes. That needs to be addressed, but probably not in this > > > patch. > > > > > > > The node specific locks will be used by different processes and all of > > these processes will be competing commonly (and successfully) for the > imgu_device lock. > > I will look into this more. > > > > > I wonder if it'd be more simple to use imgu->lock here instead of > > > adding a new one. > > > > > > > Extending img
RE: [PATCH] media: staging/intel-ipu3: Implement lock for stream on/off operations
Hi Sakari, > -Original Message- > From: Sakari Ailus [mailto:sakari.ai...@linux.intel.com] > Sent: Wednesday, January 30, 2019 12:59 AM > To: Mani, Rajmohan > Cc: Mauro Carvalho Chehab ; Greg Kroah-Hartman > ; linux-me...@vger.kernel.org; > de...@driverdev.osuosl.org; linux-ker...@vger.kernel.org; Laurent Pinchart > ; Jacopo Mondi ; > Qiu, Tian Shu ; Cao, Bingbu > ; z...@paasikivi.fi.intel.com; Zhi, Yong > ; hverk...@xs4all.nl; tf...@chromium.org > Subject: Re: [PATCH] media: staging/intel-ipu3: Implement lock for stream > on/off operations > > Hi Rajmohan, > > On Tue, Jan 29, 2019 at 02:27:36PM -0800, Rajmohan Mani wrote: > > Currently concurrent stream off operations on ImgU nodes are not > > synchronized, leading to use-after-free bugs (as reported by KASAN). > > > > [ 250.090724] BUG: KASAN: use-after-free in > > ipu3_dmamap_free+0xc5/0x116 [ipu3_imgu] [ 250.090726] Read of size 8 > > at addr 888127b29bc0 by task yavta/18836 [ 250.090731] Hardware > > name: HP Soraka/Soraka, BIOS Google_Soraka.10431.17.0 03/22/2018 [ > 250.090732] Call Trace: > > [ 250.090735] dump_stack+0x6a/0xb1 > > [ 250.090739] print_address_description+0x8e/0x279 > > [ 250.090743] ? ipu3_dmamap_free+0xc5/0x116 [ipu3_imgu] [ > > 250.090746] kasan_report+0x260/0x28a [ 250.090750] > > ipu3_dmamap_free+0xc5/0x116 [ipu3_imgu] [ 250.090754] > > ipu3_css_pool_cleanup+0x24/0x37 [ipu3_imgu] [ 250.090759] > > ipu3_css_pipeline_cleanup+0x61/0xb9 [ipu3_imgu] [ 250.090763] > > ipu3_css_stop_streaming+0x1f2/0x321 [ipu3_imgu] [ 250.090768] > > imgu_s_stream+0x94/0x443 [ipu3_imgu] [ 250.090772] ? > > ipu3_vb2_buf_queue+0x280/0x280 [ipu3_imgu] [ 250.090775] ? > > vb2_dma_sg_unmap_dmabuf+0x16/0x6f [videobuf2_dma_sg] [ 250.090778] > ? > > vb2_buffer_in_use+0x36/0x58 [videobuf2_common] [ 250.090782] > > ipu3_vb2_stop_streaming+0xf9/0x135 [ipu3_imgu] > > > > Implemented a lock to synchronize imgu stream on / off operations and > > the modification of streaming flag (in struct imgu_device), to prevent > > these issues. > > > > Reported-by: Laurent Pinchart > > Suggested-by: Laurent Pinchart > > > > Signed-off-by: Rajmohan Mani > > --- > > drivers/staging/media/ipu3/ipu3-v4l2.c | 6 ++ > > drivers/staging/media/ipu3/ipu3.c | 3 +++ > > drivers/staging/media/ipu3/ipu3.h | 4 > > 3 files changed, 13 insertions(+) > > > > diff --git a/drivers/staging/media/ipu3/ipu3-v4l2.c > > b/drivers/staging/media/ipu3/ipu3-v4l2.c > > index c7936032beb9..cf7e917cd0c8 100644 > > --- a/drivers/staging/media/ipu3/ipu3-v4l2.c > > +++ b/drivers/staging/media/ipu3/ipu3-v4l2.c > > @@ -507,12 +507,15 @@ static int ipu3_vb2_start_streaming(struct > vb2_queue *vq, unsigned int count) > > goto fail_stop_pipeline; > > } > > > > + mutex_lock(>streaming_lock); > > + > > You appear to be using imgu_device.lock (while searching buffers to queue to > the device) as well as imgu_video_device.lock (qbuf, dqbuf) to serialise > access > to imgu_video_device.buffers list. Ack > The two locks may be acquired at the same > time but each by different processes. That needs to be addressed, but > probably not in this patch. > The node specific locks will be used by different processes and all of these processes will be competing commonly (and successfully) for the imgu_device lock. I will look into this more. > I wonder if it'd be more simple to use imgu->lock here instead of adding a new > one. > Extending imgu->lock here, does not work in this case, as imgu_queue_buffers() will be stuck acquiring imgu->lock, which was already acquired by imgu_s_stream() through ipu3_vb2_start_streaming(). > > /* Start streaming of the whole pipeline now */ > > dev_dbg(dev, "IMGU streaming is ready to start"); > > r = imgu_s_stream(imgu, true); > > if (!r) > > imgu->streaming = true; > > > > + mutex_unlock(>streaming_lock); > > return 0; > > > > fail_stop_pipeline: > > @@ -543,6 +546,8 @@ static void ipu3_vb2_stop_streaming(struct > vb2_queue *vq) > > dev_err(>pci_dev->dev, > > "failed to stop subdev streaming\n"); > > > > + mutex_lock(>streaming_lock); > > + > > /* Was this the first node with streaming disabled? */ > > if (imgu->streaming && ipu3_all_nodes_streaming(imgu, node)) { > > /* Yes, really stop streaming now */ @@ -552,6 +557,7 @@ > static > > void ipu3_vb