[Development] Running Qt 5.2.1 without udev

[Thiago Macieira]

Em seg 09 jun 2014, às 12:06:10, Jacob Kroon escreveu:
> Hi,
> 
> I'm testing Qt 5.2.1 on a Wandboard using Yocto. If I build Qt5 without
> udev support,
> I get 3 or 4 of these kernel warnings when I start my application:
> 
>     idr_remove called for id=xxx which is not allocated.
>     CPU: 0 PID: 538 Comm: xxx Not tainted xxx
>     [<80013c7c>] (unwind_backtrace+0x0/0xec) from [<80011634>]
> (show_stack+0x10/0x14)
>     [<80011634>] (show_stack+0x10/0x14) from [<80279198>]
> (idr_remove.part.9+0x290/0x29c)
>     [<80279198>] (idr_remove.part.9+0x290/0x29c) from [<802fdaec>]
> (drm_ctxbitmap_free+0x24/0x30)
>     [<802fdaec>] (drm_ctxbitmap_free+0x24/0x30) from [<802fe01c>]
> (drm_rmctx+0x64/0x110)
>     [<802fe01c>] (drm_rmctx+0x64/0x110) from [<802fe714>]
> (drm_ioctl+0x3d4/0x50c)
>     [<802fe714>] (drm_ioctl+0x3d4/0x50c) from [<800d8f70>]
> (do_vfs_ioctl+0x3f4/0x5b4)
>     [<800d8f70>] (do_vfs_ioctl+0x3f4/0x5b4) from [<800d9168>]
> (SyS_ioctl+0x38/0x60)
>     [<800d9168>] (SyS_ioctl+0x38/0x60) from [<8000e180>]
> (ret_fast_syscall+0x0/0x30)
> 
> Could anyone shed some light on the matter ?

Looks like a kernel bug. Maybe you should talk to the libdrm guys. It would be 
a good idea to run the application with strace so you can get the actual 
ioctls done.

-- 
Thiago Macieira - thiago.macieira (AT) intel.com
  Software Architect - Intel Open Source Technology Center