Bug#502362: kernel 2.6.26 x86_64: some ioctls not mapped from 32bit userland

[Moritz Muehlenhoff <jmm@inutil.org>]

On Thu, Oct 16, 2008 at 12:58:05AM +0200, Mario 'BitKoenig' Holbe wrote:
> Package: linux-image-2.6.26-1-amd64
> Version: 2.6.26-8
> 
> Hello,
> 
> while testing 64bit kernel under 32bit userland I found some ioctls that
> definitely work on a 32bit kernel and are likely just not correctly
> mapped:
> 
> ioctl32(hdparm:7607): Unknown cmd fd(3) cmd(0000031c){t:03;sz:0} arg(00000000) on /dev/hda
> 	HDIO_DRIVE_RESET - hdparm -w /dev/hda
> 	That's not funny, I cannot reset IDE devices anymore.
> 
> ioctl32(conserver:4500): Unknown cmd fd(5) cmd(0000530b){t:'S';sz:0} arg(0806b17e) on /dev/ttyS5
> 	This one is probably safe - I just have no idea, why (and where)
> 	conserver does CDROM ioctls on a tty - maybe some probing
> 	thingy.
> 
> 
> compat_ioctl32: VIDIOC_QUERYCAPioctl32(v4lctl:6890): Unknown cmd fd(5) cmd(80585600){t:'V';sz:88} arg(097847b0) on /dev/video0
> compat_ioctl32: VIDIOC_S_EXT_CTRLSioctl32(vlc:8081): Unknown cmd fd(7) cmd(c0185648){t:'V';sz:24} arg(f37fe940) on /dev/video0
> compat_ioctl32: VIDIOC_ENUM_FRAMEINTERVALSioctl32(vlc:8081): Unknown cmd fd(7) cmd(c034564b){t:'V';sz:52} arg(f37fee90) on /dev/video0
> 	Those are strange, they seem to be mapped correctly and
> 	transmitted to video4linux compat_ioctl32 but the V4L1->V4L2
> 	re-mapping done there seems to be broken...

Does this still occur with recent kernels?

Cheers,
        Moritz