The DMA device can't be registered if it doesn't have any channels registered at all. Moreover, it leads to memory leak and is reported by kmemleak as (on 3.10 kernel, and same shall happen on mainline):
unreferenced object 0xffffffc09e597240 (size 64): comm "swapper/0", pid 1, jiffies 4294877736 (age 7060.280s) hex dump (first 32 bytes): 00 00 00 00 c0 ff ff ff 30 00 00 ff 00 00 00 ff ........0....... 00 00 00 ff 00 00 00 ff 00 00 00 ff 00 00 00 ff ................ backtrace: [<ffffffc0003079ec>] create_object+0x148/0x2a0 [<ffffffc000cc150c>] kmemleak_alloc+0x80/0xbc [<ffffffc000303a7c>] kmem_cache_alloc_trace+0x120/0x1ac [<ffffffc00054771c>] dma_async_device_register+0x160/0x46c [<ffffffc000548958>] foo_probe+0x1a0/0x264 [<ffffffc0005d6658>] platform_drv_probe+0x14/0x20 [<ffffffc0005d50cc>] driver_probe_device+0x160/0x374 [<ffffffc0005d538c>] __driver_attach+0x60/0x90 [<ffffffc0005d3e78>] bus_for_each_dev+0x7c/0xb0 [<ffffffc0005d4a0c>] driver_attach+0x1c/0x28 [<ffffffc0005d459c>] bus_add_driver+0x124/0x248 [<ffffffc0005d59cc>] driver_register+0x90/0x110 [<ffffffc0005d6bf4>] platform_driver_register+0x58/0x64 [<ffffffc00142a70c>] foo_driver_init+0x10/0x1c [<ffffffc000200878>] do_one_initcall+0xac/0x148 [<ffffffc00140096c>] kernel_init_freeable+0x1a0/0x258
Return -ENODEV from dma_async_device_register() on such a case.
Signed-off-by: Viresh Kumar viresh.kumar@linaro.org --- Hi Vinod,
Sorry if the fundamentals behind this patch are completely incorrect, i.e. We *can't* register a dma device with 0 channels. Its been long that I have worked on dma stuff :)
drivers/dma/dmaengine.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c index 8c9f45fd55fc..6b535262ac5d 100644 --- a/drivers/dma/dmaengine.c +++ b/drivers/dma/dmaengine.c @@ -997,6 +997,13 @@ int dma_async_device_register(struct dma_device *device) } chan->client_count = 0; } + + if (!chancnt) { + dev_err(device->dev, "%s: device has no channels!\n", __func__); + rc = -ENODEV; + goto err_out; + } + device->chancnt = chancnt;
mutex_lock(&dma_list_mutex);
On Wed, Jul 27, 2016 at 02:32:58PM -0700, Viresh Kumar wrote:
The DMA device can't be registered if it doesn't have any channels registered at all. Moreover, it leads to memory leak and is reported by kmemleak as (on 3.10 kernel, and same shall happen on mainline):
unreferenced object 0xffffffc09e597240 (size 64): comm "swapper/0", pid 1, jiffies 4294877736 (age 7060.280s) hex dump (first 32 bytes): 00 00 00 00 c0 ff ff ff 30 00 00 ff 00 00 00 ff ........0....... 00 00 00 ff 00 00 00 ff 00 00 00 ff 00 00 00 ff ................ backtrace: [<ffffffc0003079ec>] create_object+0x148/0x2a0 [<ffffffc000cc150c>] kmemleak_alloc+0x80/0xbc [<ffffffc000303a7c>] kmem_cache_alloc_trace+0x120/0x1ac [<ffffffc00054771c>] dma_async_device_register+0x160/0x46c [<ffffffc000548958>] foo_probe+0x1a0/0x264 [<ffffffc0005d6658>] platform_drv_probe+0x14/0x20 [<ffffffc0005d50cc>] driver_probe_device+0x160/0x374 [<ffffffc0005d538c>] __driver_attach+0x60/0x90 [<ffffffc0005d3e78>] bus_for_each_dev+0x7c/0xb0 [<ffffffc0005d4a0c>] driver_attach+0x1c/0x28 [<ffffffc0005d459c>] bus_add_driver+0x124/0x248 [<ffffffc0005d59cc>] driver_register+0x90/0x110 [<ffffffc0005d6bf4>] platform_driver_register+0x58/0x64 [<ffffffc00142a70c>] foo_driver_init+0x10/0x1c [<ffffffc000200878>] do_one_initcall+0xac/0x148 [<ffffffc00140096c>] kernel_init_freeable+0x1a0/0x258
Return -ENODEV from dma_async_device_register() on such a case.
Signed-off-by: Viresh Kumar viresh.kumar@linaro.org
Hi Vinod,
Sorry if the fundamentals behind this patch are completely incorrect, i.e. We *can't* register a dma device with 0 channels. Its been long that I have worked on dma stuff :)
This sounds okay, but why would anyone register a device without having a channel?
drivers/dma/dmaengine.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c index 8c9f45fd55fc..6b535262ac5d 100644 --- a/drivers/dma/dmaengine.c +++ b/drivers/dma/dmaengine.c @@ -997,6 +997,13 @@ int dma_async_device_register(struct dma_device *device) } chan->client_count = 0; }
- if (!chancnt) {
dev_err(device->dev, "%s: device has no channels!\n", __func__);
rc = -ENODEV;
goto err_out;
- }
- device->chancnt = chancnt;
mutex_lock(&dma_list_mutex); -- 2.7.4
On 28-07-16, 09:18, Vinod Koul wrote:
On Wed, Jul 27, 2016 at 02:32:58PM -0700, Viresh Kumar wrote:
The DMA device can't be registered if it doesn't have any channels registered at all. Moreover, it leads to memory leak and is reported by kmemleak as (on 3.10 kernel, and same shall happen on mainline):
unreferenced object 0xffffffc09e597240 (size 64): comm "swapper/0", pid 1, jiffies 4294877736 (age 7060.280s) hex dump (first 32 bytes): 00 00 00 00 c0 ff ff ff 30 00 00 ff 00 00 00 ff ........0....... 00 00 00 ff 00 00 00 ff 00 00 00 ff 00 00 00 ff ................ backtrace: [<ffffffc0003079ec>] create_object+0x148/0x2a0 [<ffffffc000cc150c>] kmemleak_alloc+0x80/0xbc [<ffffffc000303a7c>] kmem_cache_alloc_trace+0x120/0x1ac [<ffffffc00054771c>] dma_async_device_register+0x160/0x46c [<ffffffc000548958>] foo_probe+0x1a0/0x264 [<ffffffc0005d6658>] platform_drv_probe+0x14/0x20 [<ffffffc0005d50cc>] driver_probe_device+0x160/0x374 [<ffffffc0005d538c>] __driver_attach+0x60/0x90 [<ffffffc0005d3e78>] bus_for_each_dev+0x7c/0xb0 [<ffffffc0005d4a0c>] driver_attach+0x1c/0x28 [<ffffffc0005d459c>] bus_add_driver+0x124/0x248 [<ffffffc0005d59cc>] driver_register+0x90/0x110 [<ffffffc0005d6bf4>] platform_driver_register+0x58/0x64 [<ffffffc00142a70c>] foo_driver_init+0x10/0x1c [<ffffffc000200878>] do_one_initcall+0xac/0x148 [<ffffffc00140096c>] kernel_init_freeable+0x1a0/0x258
Return -ENODEV from dma_async_device_register() on such a case.
/Signed-off-by: Viresh Kumar viresh.kumar@linaro.org
Hi Vinod,
Sorry if the fundamentals behind this patch are completely incorrect, i.e. We *can't* register a dma device with 0 channels. Its been long that I have worked on dma stuff :)
This sounds okay,
Great..
but why would anyone register a device without having a channel?
A buggy driver? The core should be handling this no matter what, right?
The out of tree driver, 3.10 based driver I have looks to be buggy to me. It is doing something like this (please ignore naming/formatting):
xlate() { chan = alloc(); list_add(chan.device_node, &dma_dev.channels) }
probe() { ... dma_async_device_register() of_dma_controller_register(xlate); ... }
But with this it looks like the 'chan' structure wouldn't get initialized properly as that is done from within dma_async_device_register(). Right? Also, I am not sure if DMA is getting used at all right now on my platform. Just noticed the memleak, which I tried to fix.
On Wed, Jul 27, 2016 at 02:32:58PM -0700, Viresh Kumar wrote:
The DMA device can't be registered if it doesn't have any channels registered at all. Moreover, it leads to memory leak and is reported by kmemleak as (on 3.10 kernel, and same shall happen on mainline):
unreferenced object 0xffffffc09e597240 (size 64): comm "swapper/0", pid 1, jiffies 4294877736 (age 7060.280s) hex dump (first 32 bytes): 00 00 00 00 c0 ff ff ff 30 00 00 ff 00 00 00 ff ........0....... 00 00 00 ff 00 00 00 ff 00 00 00 ff 00 00 00 ff ................ backtrace: [<ffffffc0003079ec>] create_object+0x148/0x2a0 [<ffffffc000cc150c>] kmemleak_alloc+0x80/0xbc [<ffffffc000303a7c>] kmem_cache_alloc_trace+0x120/0x1ac [<ffffffc00054771c>] dma_async_device_register+0x160/0x46c [<ffffffc000548958>] foo_probe+0x1a0/0x264 [<ffffffc0005d6658>] platform_drv_probe+0x14/0x20 [<ffffffc0005d50cc>] driver_probe_device+0x160/0x374 [<ffffffc0005d538c>] __driver_attach+0x60/0x90 [<ffffffc0005d3e78>] bus_for_each_dev+0x7c/0xb0 [<ffffffc0005d4a0c>] driver_attach+0x1c/0x28 [<ffffffc0005d459c>] bus_add_driver+0x124/0x248 [<ffffffc0005d59cc>] driver_register+0x90/0x110 [<ffffffc0005d6bf4>] platform_driver_register+0x58/0x64 [<ffffffc00142a70c>] foo_driver_init+0x10/0x1c [<ffffffc000200878>] do_one_initcall+0xac/0x148 [<ffffffc00140096c>] kernel_init_freeable+0x1a0/0x258
Return -ENODEV from dma_async_device_register() on such a case.
Signed-off-by: Viresh Kumar viresh.kumar@linaro.org
Hi Vinod,
Sorry if the fundamentals behind this patch are completely incorrect, i.e. We *can't* register a dma device with 0 channels. Its been long that I have worked on dma stuff :)
drivers/dma/dmaengine.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c index 8c9f45fd55fc..6b535262ac5d 100644 --- a/drivers/dma/dmaengine.c +++ b/drivers/dma/dmaengine.c @@ -997,6 +997,13 @@ int dma_async_device_register(struct dma_device *device) } chan->client_count = 0; }
- if (!chancnt) {
dev_err(device->dev, "%s: device has no channels!\n", __func__);
rc = -ENODEV;
goto err_out;
- }
- device->chancnt = chancnt;
mutex_lock(&dma_list_mutex);
To follow up on this, as you found out, this patch breaks working hardware, so please don't apply it anywhere. It needs more debugging to figure out exactly what is going on with some very strange hardware that seems to be relying on some strange code...
thanks,
greg k-h
On 03-08-16, 21:34, Greg Kroah-Hartman wrote:
On Wed, Jul 27, 2016 at 02:32:58PM -0700, Viresh Kumar wrote:
The DMA device can't be registered if it doesn't have any channels registered at all. Moreover, it leads to memory leak and is reported by kmemleak as (on 3.10 kernel, and same shall happen on mainline):
unreferenced object 0xffffffc09e597240 (size 64): comm "swapper/0", pid 1, jiffies 4294877736 (age 7060.280s) hex dump (first 32 bytes): 00 00 00 00 c0 ff ff ff 30 00 00 ff 00 00 00 ff ........0....... 00 00 00 ff 00 00 00 ff 00 00 00 ff 00 00 00 ff ................ backtrace: [<ffffffc0003079ec>] create_object+0x148/0x2a0 [<ffffffc000cc150c>] kmemleak_alloc+0x80/0xbc [<ffffffc000303a7c>] kmem_cache_alloc_trace+0x120/0x1ac [<ffffffc00054771c>] dma_async_device_register+0x160/0x46c [<ffffffc000548958>] foo_probe+0x1a0/0x264 [<ffffffc0005d6658>] platform_drv_probe+0x14/0x20 [<ffffffc0005d50cc>] driver_probe_device+0x160/0x374 [<ffffffc0005d538c>] __driver_attach+0x60/0x90 [<ffffffc0005d3e78>] bus_for_each_dev+0x7c/0xb0 [<ffffffc0005d4a0c>] driver_attach+0x1c/0x28 [<ffffffc0005d459c>] bus_add_driver+0x124/0x248 [<ffffffc0005d59cc>] driver_register+0x90/0x110 [<ffffffc0005d6bf4>] platform_driver_register+0x58/0x64 [<ffffffc00142a70c>] foo_driver_init+0x10/0x1c [<ffffffc000200878>] do_one_initcall+0xac/0x148 [<ffffffc00140096c>] kernel_init_freeable+0x1a0/0x258
Return -ENODEV from dma_async_device_register() on such a case.
Signed-off-by: Viresh Kumar viresh.kumar@linaro.org
Hi Vinod,
Sorry if the fundamentals behind this patch are completely incorrect, i.e. We *can't* register a dma device with 0 channels. Its been long that I have worked on dma stuff :)
drivers/dma/dmaengine.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c index 8c9f45fd55fc..6b535262ac5d 100644 --- a/drivers/dma/dmaengine.c +++ b/drivers/dma/dmaengine.c @@ -997,6 +997,13 @@ int dma_async_device_register(struct dma_device *device) } chan->client_count = 0; }
- if (!chancnt) {
dev_err(device->dev, "%s: device has no channels!\n", __func__);
rc = -ENODEV;
goto err_out;
- }
- device->chancnt = chancnt;
mutex_lock(&dma_list_mutex);
To follow up on this, as you found out, this patch breaks working hardware, so please don't apply it anywhere. It needs more debugging to figure out exactly what is going on with some very strange hardware that seems to be relying on some strange code...
Thanks for your email Greg, and I know what the issue is.
So, the dmaengine core actually allows registering a device without any channels and the dma driver can add channels later (directly to the dma-device channel list), for example, from the of-xlate callback once someone requests for a channel.
Though its not the right way of doing things, but it still works. The only thing which breaks in that case is sysfs interface to dmaengine as the channels aren't registered as devices and so they aren't visible in userspace, as that code is *only* executed while the dma-device is registered.
I think its dmaengine core's decision to keep supporting such drivers or not and so I would like Vinod to comment on that..
But if we do want to support such drivers, then the core must be updated to fix the memleak I reported here, as idr_ref is allocated and not used at all (also dma-id allocated for it).
Or provide another API to properly and fully add channels at a later point of time.
On Fri, Aug 05, 2016 at 09:30:50AM -0700, Viresh Kumar wrote:
On 03-08-16, 21:34, Greg Kroah-Hartman wrote:
To follow up on this, as you found out, this patch breaks working hardware, so please don't apply it anywhere. It needs more debugging to figure out exactly what is going on with some very strange hardware that seems to be relying on some strange code...
Thanks for your email Greg, and I know what the issue is.
So, the dmaengine core actually allows registering a device without any channels and the dma driver can add channels later (directly to the dma-device channel list), for example, from the of-xlate callback once someone requests for a channel.
I do not think that is good idea
Though its not the right way of doing things, but it still works. The only thing which breaks in that case is sysfs interface to dmaengine as the channels aren't registered as devices and so they aren't visible in userspace, as that code is *only* executed while the dma-device is registered.
I think its dmaengine core's decision to keep supporting such drivers or not and so I would like Vinod to comment on that..
My view would be to ensure that controllers have channels before they register. If they dont, then please register later.
But if we do want to support such drivers, then the core must be updated to fix the memleak I reported here, as idr_ref is allocated and not used at all (also dma-id allocated for it).
Or provide another API to properly and fully add channels at a later point of time.
And looking at drivers, I am not seeing anyone who needs such a support, so am not inclined to support such a feature atm.
On 08-08-16, 13:47, Vinod Koul wrote:
On Fri, Aug 05, 2016 at 09:30:50AM -0700, Viresh Kumar wrote:
On 03-08-16, 21:34, Greg Kroah-Hartman wrote:
To follow up on this, as you found out, this patch breaks working hardware, so please don't apply it anywhere. It needs more debugging to figure out exactly what is going on with some very strange hardware that seems to be relying on some strange code...
Thanks for your email Greg, and I know what the issue is.
So, the dmaengine core actually allows registering a device without any channels and the dma driver can add channels later (directly to the dma-device channel list), for example, from the of-xlate callback once someone requests for a channel.
I do not think that is good idea
Though its not the right way of doing things, but it still works. The only thing which breaks in that case is sysfs interface to dmaengine as the channels aren't registered as devices and so they aren't visible in userspace, as that code is *only* executed while the dma-device is registered.
I think its dmaengine core's decision to keep supporting such drivers or not and so I would like Vinod to comment on that..
My view would be to ensure that controllers have channels before they register. If they dont, then please register later.
But if we do want to support such drivers, then the core must be updated to fix the memleak I reported here, as idr_ref is allocated and not used at all (also dma-id allocated for it).
Or provide another API to properly and fully add channels at a later point of time.
And looking at drivers, I am not seeing anyone who needs such a support, so am not inclined to support such a feature atm.
Right and so I believe that applying this patch is the right thing to do..
On Mon, Aug 08, 2016 at 08:24:49AM -0700, Viresh Kumar wrote:
On 08-08-16, 13:47, Vinod Koul wrote:
On Fri, Aug 05, 2016 at 09:30:50AM -0700, Viresh Kumar wrote:
On 03-08-16, 21:34, Greg Kroah-Hartman wrote:
To follow up on this, as you found out, this patch breaks working hardware, so please don't apply it anywhere. It needs more debugging to figure out exactly what is going on with some very strange hardware that seems to be relying on some strange code...
Thanks for your email Greg, and I know what the issue is.
So, the dmaengine core actually allows registering a device without any channels and the dma driver can add channels later (directly to the dma-device channel list), for example, from the of-xlate callback once someone requests for a channel.
I do not think that is good idea
Though its not the right way of doing things, but it still works. The only thing which breaks in that case is sysfs interface to dmaengine as the channels aren't registered as devices and so they aren't visible in userspace, as that code is *only* executed while the dma-device is registered.
I think its dmaengine core's decision to keep supporting such drivers or not and so I would like Vinod to comment on that..
My view would be to ensure that controllers have channels before they register. If they dont, then please register later.
But if we do want to support such drivers, then the core must be updated to fix the memleak I reported here, as idr_ref is allocated and not used at all (also dma-id allocated for it).
Or provide another API to properly and fully add channels at a later point of time.
And looking at drivers, I am not seeing anyone who needs such a support, so am not inclined to support such a feature atm.
Right and so I believe that applying this patch is the right thing to do..
So only the drivers that are out-of-tree are broken with this patch?
I like it :)
greg k-h
On Wed, Jul 27, 2016 at 02:32:58PM -0700, Viresh Kumar wrote:
The DMA device can't be registered if it doesn't have any channels registered at all. Moreover, it leads to memory leak and is reported by kmemleak as (on 3.10 kernel, and same shall happen on mainline):
unreferenced object 0xffffffc09e597240 (size 64): comm "swapper/0", pid 1, jiffies 4294877736 (age 7060.280s) hex dump (first 32 bytes): 00 00 00 00 c0 ff ff ff 30 00 00 ff 00 00 00 ff ........0....... 00 00 00 ff 00 00 00 ff 00 00 00 ff 00 00 00 ff ................ backtrace: [<ffffffc0003079ec>] create_object+0x148/0x2a0 [<ffffffc000cc150c>] kmemleak_alloc+0x80/0xbc [<ffffffc000303a7c>] kmem_cache_alloc_trace+0x120/0x1ac [<ffffffc00054771c>] dma_async_device_register+0x160/0x46c [<ffffffc000548958>] foo_probe+0x1a0/0x264 [<ffffffc0005d6658>] platform_drv_probe+0x14/0x20 [<ffffffc0005d50cc>] driver_probe_device+0x160/0x374 [<ffffffc0005d538c>] __driver_attach+0x60/0x90 [<ffffffc0005d3e78>] bus_for_each_dev+0x7c/0xb0 [<ffffffc0005d4a0c>] driver_attach+0x1c/0x28 [<ffffffc0005d459c>] bus_add_driver+0x124/0x248 [<ffffffc0005d59cc>] driver_register+0x90/0x110 [<ffffffc0005d6bf4>] platform_driver_register+0x58/0x64 [<ffffffc00142a70c>] foo_driver_init+0x10/0x1c [<ffffffc000200878>] do_one_initcall+0xac/0x148 [<ffffffc00140096c>] kernel_init_freeable+0x1a0/0x258
Return -ENODEV from dma_async_device_register() on such a case.
Applied, thanks
linaro-kernel@lists.linaro.org