Re: My boss want to kill debian, please help !
On Fri, 28 Jan 2005, George Alexandru Dragoi wrote:
> and everything got back to normal (guess how charming the tech manager
> has begun to talk about debian :( .
> this is the lspci line about that e100:
Still, to make things clear. The *hardware* issue with e100 that I know off
causes the e100 PHY (not the machine. not the PCI bus. not even the entire
e100 board) to lock up. Nothing gets sent, nothing is received. The driver
resets the PHY after about 30s I think. It tells you that it thinks the
hardware bug is present, and displays something to the effect of "foo
workaround activated" on startup.
Also, there's something else I should point out: the debian *stable* kernel
is too old (I won't say "crap", but it is certainly useless for what I need
it). Get the Debian kernel in testing or sid, after verifying that it is
currently sane (as of this writing, 2.6.10-4 and 2.4.27-8, both are sane).
Backporting policy is well and good, but it doesn't work for the kernel if
you are using non-ancient, non-stale hardware.
The e100/e1000 drivers have had a *ton* of fixes in the latest revisions
(mostly due to new features such as NAPI, but...)
My experiences with NICs tell me that you get about 1% chance of receiving a
bad one. I got a 3COM that would cause PCI busmaster errors during high
loads after a while, and a e100 that would hang the entire machine. But
they were defective parts, and would have hung FreeBSD just the same...
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot