Discussion:
newer broadcom chip support
(too old to reply)
Manuel Bouyer
2006-12-16 17:46:16 UTC
Permalink
Hi,
I have a new server which has a new broadcom chip (5708), not supported by
NetBSD. What is the best approach; porting the bce(4) driver from FreeBSD or
trying to integrate support to our bge(4) ? Has someone already looked at
this ?
--
Manuel Bouyer <***@antioche.eu.org>
NetBSD: 26 ans d'experience feront toujours la difference
--

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de
Quentin Garnier
2006-12-17 09:48:17 UTC
Permalink
Post by Manuel Bouyer
Hi,
I have a new server which has a new broadcom chip (5708), not supported by
NetBSD. What is the best approach; porting the bce(4) driver from FreeBSD or
trying to integrate support to our bge(4) ? Has someone already looked at
this ?
I have a port of OpenBSD's bnx(4) (which is a port of FreeBSD's bce(4)).

It needs more testing (it worked fine on i386, but it had an unexplained
issue on amd64; I don't have the hardware myself so debugging was hard
and my tester kind of disappeared), and a final bit of polishing.

I'll make sure it compiles in -current and send it to you if you want.
--
Quentin Garnier - ***@cubidou.net - ***@NetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.
Manuel Bouyer
2006-12-17 10:37:55 UTC
Permalink
Post by Quentin Garnier
Post by Manuel Bouyer
Hi,
I have a new server which has a new broadcom chip (5708), not supported by
NetBSD. What is the best approach; porting the bce(4) driver from FreeBSD or
trying to integrate support to our bge(4) ? Has someone already looked at
this ?
I have a port of OpenBSD's bnx(4) (which is a port of FreeBSD's bce(4)).
It needs more testing (it worked fine on i386, but it had an unexplained
issue on amd64; I don't have the hardware myself so debugging was hard
and my tester kind of disappeared), and a final bit of polishing.
I'll make sure it compiles in -current and send it to you if you want.
YES ! :)
--
Manuel Bouyer <***@antioche.eu.org>
NetBSD: 26 ans d'experience feront toujours la difference
--

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de
Thor Lancelot Simon
2006-12-17 18:15:48 UTC
Permalink
Post by Quentin Garnier
Post by Manuel Bouyer
Hi,
I have a new server which has a new broadcom chip (5708), not supported by
NetBSD. What is the best approach; porting the bce(4) driver from FreeBSD or
trying to integrate support to our bge(4) ? Has someone already looked at
this ?
I have a port of OpenBSD's bnx(4) (which is a port of FreeBSD's bce(4)).
I know the OpenBSD manual page says that bnx/bce can do full TCP offload;
I'm not sure I believe that (or that we'd want it if it actually could).

But do you know if it can do header splitting? The newer Intel cards can,
and using it gives a significant performance boost under Linux by allowing
zero-copy receive to userspace.

Thor

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de
Quentin Garnier
2006-12-17 18:52:49 UTC
Permalink
Post by Thor Lancelot Simon
Post by Quentin Garnier
Post by Manuel Bouyer
Hi,
I have a new server which has a new broadcom chip (5708), not supported by
NetBSD. What is the best approach; porting the bce(4) driver from FreeBSD or
trying to integrate support to our bge(4) ? Has someone already looked at
this ?
I have a port of OpenBSD's bnx(4) (which is a port of FreeBSD's bce(4)).
I know the OpenBSD manual page says that bnx/bce can do full TCP offload;
I'm not sure I believe that (or that we'd want it if it actually could).
I asked the author about TSO, he replied that he didn't think FreeBSD
supported it, so it's not in the driver anyway.
Post by Thor Lancelot Simon
But do you know if it can do header splitting? The newer Intel cards can,
and using it gives a significant performance boost under Linux by allowing
zero-copy receive to userspace.
The author of the driver was very nice and replied very quickly to my
questions; you'd better ask him directly.
--
Quentin Garnier - ***@cubidou.net - ***@NetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.
Loading...