Re: git: re(4): Use MSI if device supports it.

[Sepherosa Ziehau <sepherosa@xxxxxxxxx>]

On Fri, Oct 5, 2012 at 5:51 PM, Francois Tigeot
<ftigeot@crater.dragonflybsd.org> wrote:
>
> commit 4dce912a82d188446de8e4ac7bef52877acbca51
> Author: Tim Bisson <bissont@mac.com>
> Date:   Sun Mar 18 19:39:37 2012 -0700
>
>     re(4): Use MSI if device supports it.

Well, as far as I know, OpenBSD tried to enable MSI on their re(4),
which causes many problems.  FreeBSD probably has the magic eeprom
adjustment which is necessary to enable MSI on all re(4) (I will have
to say adjust eeprom could be dangerous, as I have damaged one of my
re(4) back to the time I worked on it).  So I think we probably should
disable MSI by default or only enable it by default for PCI-E devices
(even for that I am not quite sure about).  After all, no one will
expect any performance out of the re(4) (the worst gigE chips I have
seen), just get it work should be enough.

Best Regards,
sephe

>
>     RT8168E supports MSI.
>
> Summary of changes:
>  sys/dev/netif/re/if_re.c    |   17 +++++++++++++----
>  sys/dev/netif/re/if_revar.h |    2 ++
>  2 files changed, 15 insertions(+), 4 deletions(-)
>
> http://gitweb.dragonflybsd.org/dragonfly.git/commitdiff/4dce912a82d188446de8e4ac7bef52877acbca51
>
>
> --
> DragonFly BSD source repository



-- 
Tomorrow Will Never Die