Garrett D'Amore wrote:
> The err_summary bit will be set along with the frame2long bit for 
> full-size VLAN frames.  You need to ignore those bits, and instead check 
> against the various other error bits.  (E.g. runt_frame, dribbling, 
> mii_err, overflow, and crc.)
> 
> The concern here is less about updating stats and more about making sure 
> you don't drop the vlan packet on the floor.

Okay, I've done a bit more research into this. Right now, it appears 
that the only references which may need fixing wrt err_summary are at 
lines: 1823, 1856, and 1859 in dnet.c (patched revision).

Would I be correct in assuming the only uses of err_summary we are 
particularly interested in for vlan tagging is on the rx_desc_type? I 
did not see a problem with the current uses of err_summary on tx_desc_type.

Cheers,

Steve

-- 
Yet magic and hierarchy
arise from the same source,
and this source has a null pointer.

Reference the NULL within NULL,
it is the gateway to all wizardry.
_______________________________________________
driver-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/driver-discuss

Reply via email to