Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 20 Apr 2011 10:18:13 -0700
From:      YongHyeon PYUN <pyunyh@gmail.com>
To:        Paul Thornton <prt@prt.org>
Cc:        freebsd-net@freebsd.org, Marius Strobl <marius@alchemy.franken.de>
Subject:   Re: Broadcom BCM57765 support?
Message-ID:  <20110420171813.GA8566@michelle.cdnetworks.com>
In-Reply-To: <4DAEF9F8.5070305@prt.org>
References:  <4DADD913.5010208@prt.org> <20110419220907.GD1637@michelle.cdnetworks.com> <4DAEBB67.7070303@prt.org> <20110420131613.GA84319@alchemy.franken.de> <4DAEDE27.8080205@prt.org> <20110420134459.GL38455@alchemy.franken.de> <4DAEF9F8.5070305@prt.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Apr 20, 2011 at 04:21:28PM +0100, Paul Thornton wrote:
> Hi,
> 
> On 20/04/2011 14:44, Marius Strobl wrote:
> > Hrm, looks like neither NetBSD nor OpenBSD support this variant so
> > far, however Linux also doesn't seem to have special handling for it.
> > Could you please give the attached patch in addition to the existing
> > one a try?
> 
> I've added your patch to brgphy as well - but at a user level I'm still
> seeing the same thing:
> - Ethernet frames are transmitted by the box, switch sees them with no
> errors.
> - Frames received don't seem to make it to/beyond the driver.
> - After I hit CTRL-C on the non-working ping, I get a "bge0: watchdog
> timeout -- resetting" message.
> 
> If I tcpdump the interface, I see nothing, but the dev.bge.0.stats
> counters are incrementing when I attempt to ping.
> 
> The switch is showing counts on both its tx and rx packets counters with
> no errors.
> 
> Is there somewhere I need to add more debug in the driver to understand
> what it is doing?
> 

Ok, that would indicate there is an interrupt delivery issue if all
others changes(including Marius' patch) are correct. Because there
is no publicly available data sheet for BCM57765 yet I'm not sure
what is real difference between BCM5717 and BCM57765. However I
guess they require similar hardware configuration. Both BCM5717 and
BCM57765 will use tagged status feature of controller if MSI is
available. And NVIDIA bridge controller is known to have MSI issues
for a long time in FreeBSD. Please check whether you received any
interrupts for bge(4) with vmstat(8). If you see no interrupt from
the output, either try disabling MSI or apply r219737 and r219740
and see whether that makes any difference.

> The current dmesg looks like:
> 
> bge0: <Broadcom unknown BCM57765, ASIC rev. 0x57785000> mem
> 0x93100000-0x9310ffff,0x93110000-0x9311ffff irq 18 at device 0.0 on pci4
> bge0: Reserved 0x10000 bytes for rid 0x10 type 3 at 0x93100000
> bge0: attempting to allocate 1 MSI vectors (8 supported)
> msi: routing MSI IRQ 256 to local APIC 0 vector 56
> bge0: using IRQ 256 for MSI
> bge0: CHIP ID 0x57785000; ASIC REV 0x57785; CHIP REV 0x577850; PCI-E
> bge0: Disabling fastboot
> bge0: Disabling fastboot
> miibus0: <MII bus> on bge0
> brgphy0: <BCM57785 10/100/1000baseTX PHY> PHY 1 on miibus0
> brgphy0: OUI 0x00d897, model 0x0024, rev. 0
> brgphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseT,
> 1000baseT-master, 1000baseT-FDX, 1000baseT-FDX-master, auto, auto-flow
> bge0: bpf attached
> bge0: Ethernet address: c4:2c:03:08:0b:9d
> bge0: [MPSAFE]
> bge0: [FILTER]
> 
> Paul.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110420171813.GA8566>