[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#642198: ITP: r8168 -- dkms source for the r8168 network driver


On Tuesday 20 September 2011, Andreas Beckmann wrote:
> Package: wnpp
> Severity: wishlist
> Owner: Andreas Beckmann <debian@abeckmann.de>
> * Package name    : r8168
>   Version         : 8.025.00
>   Upstream Author : Realtek NIC software team <nicfae@realtek.com>
> * URL             : http://www.realtek.com/downloads/downloadsView.aspx?Langid=1&PNid=13&PFid=5&Level=5&Conn=4&DownTypeID=3&GetDown=false#2
> * License         : GPL2+
>   Programming Lang: C
>   Description     : dkms source for the r8168 network driver
>  r8168 is the Linux device driver released for RealTek RTL8168B/8111B,
>  RTL8168C/8111C, RTL8168CP/8111CP, RTL8168D/8111D, and RTL8168DP/8111DP, and
>  RTK8168E/8111E Gigabit Ethernet controllers with PCI-Express interface.
>  .
>  This driver should only used for devices not yet supported by the in-kernel
>  driver r8169.

Personally speaking I'd assume this package would create much more 
problems than it would solve, due to the PCI ID overlap with r8169.ko 
shipped by the kernel packages themselves. This driver claims 
10ec:8168, which is also taken by the in-kernel r8169.ko module. 

Upstream, all r8168 devices are (or will be) supported by the r8169.ko
module (r8168.ko doesn't exist in the kernel) and RealTek recently 
started to participate in developing r8169.ko mainline. Which means the
overlap will only increase from kernel version to kernel version in 
unstable, even if you'd restrict this package to squeeze (and it's not 
likely that a completely new packages would be accepted for upcoming 
point releases) the package would conflict with the kernel team's 
efforts to backport support for newer devices[1], as those also 
backport support for newer devices in stable.

Please also consider that systems using both onboard PCIe r8168 and 
additional PCI r8169 cards are not an uncommon situation, which is not
possible using this proposed driver.

>  This package provides the dkms source code for the r8168 kernel modules.
>  Kernel source or headers are required to compile these modules.
> I just came along this in #debian last week, helping someone to get this
> module built and installed (upstream build system does not work properly
> on kernel 3.x) because the in-kernel module r8169 did not support his
> NIC.

Is this driver really needed with kernel 3.0, what about 3.1~rc in 
experimental? If it's still missing, looking at net-next[2] it might be
easier to backport a small patch adding support for the affected device
than dealing with conflicting modules both claiming the same device IDs.

from 2.6.39 to 3.0, r8169.ko gained support for:
- RTL8168E/RTL8111E

and learned about new chipset variants for:
- RTL8105
- RTL8168DP

from 3.0 to 3.1~rc, r8169.ko gained support for:
- RTL8111E-VL

and learned about new cards like:
- D-Link DGE-530T rev C1 (DLG10028C)

For even longer, r8169.ko supports (looking at 3.1~rc):
- 3 different RTL8168b/8111b variants
- 4 RTL8168c/8111c variants
- 3 RTL8168cp/8111cp variants
- 2 RTL8168d/8111 variants
- 3 RTL8168dp/8111dp variants
- 2 RTL8168e/8111e variants 

…which doesn't leave much of r8168.ko uncovered.

> I did this package just as a first experiment with dkms, the module
> builds fine with squeeze and sid kernels. I don't have the hardware to
> actually test it. So if someone who actually needs it wants to take over
> maintainership, he is welcome.

	Stefan Lippers-Hollmann

[1]	http://lists.debian.org/debian-kernel/2011/09/msg00540.html
[2]	git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next.git
	git://github.com/davem330/net-next.git (temporarily)

Reply to: