Date: Thu, 7 Sep 2006 10:24:01 -0700 From: "David Christensen" <davidch@broadcom.com> To: "Sam Leffler" <sam@errno.com> Cc: Oleg Bulyzhin <oleg@freebsd.org>, freebsd-current@freebsd.org, LI Xin <delphij@delphij.net>, "Simon L. Nielsen" <simon@freebsd.org> Subject: RE: bge(4) on BCM 5752 A02 panic due to media autoselect Message-ID: <09BFF2FA5EAB4A45B6655E151BBDD90301F128C8@NT-IRVA-0750.brcm.ad.broadcom.com> In-Reply-To: <450038BD.6010401@errno.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> > Can you test the attached patch? I've simply removed the test > > altogether. > > The Linux driver doesn't perform this test and the fact=20 > that the test is > > performed after bge_reset() is called simply means we're=20 > checking the > > same > > information twice (i.e. did the NIC firmware load). Since=20 > the firmware > > handshake was successful we know the firmware loaded so the test is > > redundant. >=20 > This appears to fix problems I was seeing on a 5703 part (sorry David, > told you 5702 before): >=20 > bge0: <Broadcom BCM5703 A2, ASIC rev. 0x1002> mem=20 > 0xfaff0000-0xfaffffff > irq 11 at device 0.0 on pci2 >=20 There is actually a defect in the bge driver that will identify a 5702 as a 5703. This is caused by the fact that both the 5702 and the 5703 use the same ASIC IDs in hardware and the bge driver doesn't do any further checking to separate the two. My testing confirmed that the bge driver incorrectly=20 identified a 5702 as a 5703. > This is a Dell 600m. Prior to the patch I'd get the RX diagnostic > failed msg. dhclient would fetch a lease but I couldn't mount an nfs > volume (w/ UDP). With your patch things appear to work as before. That's good to know. If at least one of the other users who reported a=20 problem with my last patch can confirm that the issue is also fixed by=20 this change then I'll go ahead and commit it. Dave
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?09BFF2FA5EAB4A45B6655E151BBDD90301F128C8>