Re: Call for bfe(4) testers.

On Wed, 30.07.2008 at 20:34:49 +0900, Pyun YongHyeon wrote:
> 
> Hi,
> 
> I have been using the following patched bfe(4) and it seems it
> works as expected. I've fixed long standing bus_dma(9) bugs in it
> and it should work on systems with > 1GB memory. In theory bfe(4)
> should also work on big-endian architectures. However the 1GB DMA
> address space of the hardware is much lower than that of DVMA base
> address of sparc64 so the possibility of using bfe(4) on sparc64 is
> zero. :-(
> 
> As added bonus of patched driver, bfe(4) now supports hardware
> statistics counter. You can use sysctl(8) to see what counters are
> maintained in the hardware. For example, the following command will
> show detailed statistics for Tx/Rx frames.
> #sysctl dev.bfe.0.stats=1
> 
> If you have one of bfe(4) hardwares please give it try and let me
> know how it goes. The latest bfe(4) can be found at the following
> URL.
> http://people.freebsd.org/~yongari/bfe/if_bfe.c
> http://people.freebsd.org/~yongari/bfe/if_bfereg.h

Hi Pyun,

I recompiled a fresh RELENG_7 kernel with the files above, and it led to
a panic, shortly after the system was up. I can't provide you with a
dump for now, but here's the handwritten backtrace:

last dmesg: no toe capability of 0xc421d800
trace:
        device_is_attached
        cf_set_method
        cpufreq_curr_sysctl
        sysctl_root
        userland_sysctl

If you need more info, I'll crash the machine again, I'm also happy to
test further patches.

Cheers,
Ulrich Spoerlein
-- 
It is better to remain silent and be thought a fool,
than to speak, and remove all doubt.