Re: [PATCH v3.18-rc3] drm: msm: Allow exported dma-bufs to be mapped

From: Daniel Thompson
Date: Tue Nov 11 2014 - 09:29:05 EST


On 10/11/14 17:36, Rob Clark wrote:
> On Mon, Nov 10, 2014 at 12:16 PM, Daniel Thompson
> <daniel.thompson@xxxxxxxxxx> wrote:
>> diff --git a/drivers/gpu/drm/msm/msm_gem_prime.c b/drivers/gpu/drm/msm/msm_gem_prime.c
>> index ad772fe36115..4e4fa5828d5d 100644
>> --- a/drivers/gpu/drm/msm/msm_gem_prime.c
>> +++ b/drivers/gpu/drm/msm/msm_gem_prime.c
>> @@ -20,6 +20,14 @@
>>
>> #include <linux/dma-buf.h>
>>
>> +struct dma_buf *msm_gem_prime_export(struct drm_device *dev,
>> + struct drm_gem_object *obj, int flags)
>> +{
>> + /* we want to be able to write in mmapped buffer */
>> + flags |= O_RDWR;
>> + return drm_gem_prime_export(dev, obj, flags);
>> +}
>> +
>
> seems like this probably should be done more centrally.. and in fact,
> might be better to have something like this in
> drm_prime_handle_to_fd_ioctl:
>
> /* check flags are valid */
> - if (args->flags & ~DRM_CLOEXEC)
> + if (args->flags & ~(DRM_CLOEXEC | O_RDWR))
> return -EINVAL;
>
> so exporter can specify whether to allow mmap or not.

That makes sense I'll try this.

Do we need to wrap O_RDWR in the same way we wrap O_CLOEXEC? (I don't
really understand why DRM_CLOEXEC exists; even the patch description
from when the symbol was introduced names it O_CLOEXEC).

Also the "flags |= O_RDWR" approach is copied from the sti driver. I'll
share a patch to remove it but that will definitely needs Benjamin's ack
because it will stop some userspaces working correctly (however I
suspect that Benjamin may be the only person currently with such a
userspace and that he can be persuaded not to call it a regression).


Daniel.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/