tree 204e049cee7f7e2f167bd1580c91c78eaddfe816
parent f5e7ebb31ce7415b68c10e30405527ca8025e410
author Barret Rhoden <brho@cs.berkeley.edu> 1425496346 -0500
committer Barret Rhoden <brho@cs.berkeley.edu> 1425501270 -0500

BNX2X: usability fixups

For whatever reason, we can run with the max number of queues on ether0,
but not on ether1.  From poking around, this happened once we started
attaching after resetting both NIC functions (ether0 and ether1).  I had
tried this at one point, but it didn't work.  For whateve reason, it
does now.

But ether1 will still die with the same old MC assert - unless both NICs
are limited to 1 queue.  Then they both work fine.  Note that if the
NICs have different values for NQ (say, 15 for ether0 and 1 for ether1),
*ether0* will get the MC assert.

Other changes include making sure init only happens when we actually
have a bnx2x NIC (by contrast, pnp always happens), and cleaning up the
print debugging.

Unfortunately, we don't have a way to send parameters to attach yet, so
you'll have to manually toggle the debug level if you want to debug
while attaching.