Re: mmap fails for DRM_IOCTL_MODE_MAP_DUMB for Radeon, Nouveau, and VMWGFX

[Rian Quinn <rianquinn@xxxxxxxxx>]

Verified that this fixed the issue for all of the video cards. 

On Mon, Nov 10, 2014 at 5:38 PM, Rian Quinn <rianquinn@xxxxxxxxx> wrote:

Your a life savor man. That did the trick. I cannot test on the Radeon and Nouveau until I get back into the office tomorrow, but it worked great on the VMWGFX driver. 

I'll also email David to let him know as his examples (which a lot of people reference) also need to that #define. 

Thanks again,

- Rian

On Mon, Nov 10, 2014 at 5:33 PM, Dave Airlie <airlied@xxxxxxxxx> wrote:

On 11 November 2014 07:33, Rian Quinn <rianquinn@xxxxxxxxx> wrote:
> I would be surprised if the size argument was the issue. That comes right
> out out of the create IOCTL unmodified. Here is the full source that I am
> using.
>
>     // Clear the arg buffers.
>     memset(&create_arg, 0, sizeof(create_arg));
>     memset(&map_arg, 0, sizeof(map_arg));
>
>     //
> ------------------------------------------------------------------------
>     // Allocate
>
>     // Fill in the arguments for the IOCTL. Note that we only support 32bpp
>     // dumb buffers so that is hard coded here.
>     create_arg.width = width;
>     create_arg.height = height;
>     create_arg.bpp = 32;
>
>     // Allocate the dumb buffer
>     if (drmIoctl(gpu->fd(), DRM_IOCTL_MODE_CREATE_DUMB, &create_arg))
>     {
>         svWarning(0) << "Failed: DRM_IOCTL_MODE_CREATE_DUMB - " <<
> strerror(errno);
>         return;
>     }
>
>     // Store the dumb buffer properties.
>     mSize = create_arg.size;
>     mStride = create_arg.pitch;
>     mHandle = create_arg.handle;
>
>     //
> ------------------------------------------------------------------------
>     // Map
>
>     // Fill in the arguments for the IOCTL.
>     map_arg.handle = mHandle;
>
>     // Allocate the dumb buffer
>     if (drmIoctl(gpu->fd(), DRM_IOCTL_MODE_MAP_DUMB, &map_arg))
>     {
>         svWarning(0) << "Failed: DRM_IOCTL_MODE_MAP_DUMB - " <<
> strerror(errno);
>         return;
>     }
>
>     // Store the mapped memory
>     if ((mVaddr = (svuint32 *)mmap(0, mSize, PROT_READ | PROT_WRITE,
> MAP_SHARED, gpu->fd(), map_arg.offset)) == MAP_FAILED)
>     {
>         // Need to clear out the address so that we don't try to use it.
>         mVaddr = NULL;
>
>         // Error out.
>         svWarning(0) << "Failed: Dumb Buffer mmap - " << strerror(errno);
>         return;
>     }
>

mmap is probably truncating the offset,

look into,
#define _FILE_OFFSET_BITS 64

AC_SYS_LARGEFILE

Dave.

_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/dri-devel