4 - Use extended TX block?
10 - Type tracking has been implemented in an unusual way in order to make it
11 more suitable for multiple users. Each client that tracks a particular
12 packet type is given the impression that it is the only one doing so.
13 When a client begins tracking a packet type that is already being
14 tracked, the current packet statistics are recorded internally. These
15 initial statistics are then subtracted from the figures returned from
16 every S2_GETTYPESTATS call the client makes.
17 - The driver is designed to work best with protocols that provide the S2R3
18 DMA hooks. Other protocols may suffer poor transmission performance, as
19 the driver's single transmit buffer must be used.
20 - first_cb points to the next CB to finish (even if it hasn't started yet).
21 last_cb points to the most recently sent CB. Unused CBs have their
22 control field set to zero (this could potentially be confused with a NOP
24 - We can't use the S2_CopyToBuff32 function for non-raw packets because
25 the non-raw portion of the receive buffer is not aligned to 32 bits.
28 TCP/IP STACK BUGS TO CONSIDER:
30 - Miami(DX) doesn't like requests being aborted upon S2_OFFLINE.
31 - Genesis calls S2_ONLINE before S2_CONFIGINTERFACE?
32 - When a device goes back online without Genesis's intervention, Genesis
33 displays the message "eth0 is online again" or similar, but keeps the
34 interface's indicator red and doesn't disable the "Online" button or
35 enable the "Offline" button.
38 OPENPCI/POWERPCI BUGS:
40 - 0xffff doesn't work as a product wildcard in pci_find_device().
41 - IRQ 12 not reported on my machine.
42 - powerpci.library sometimes reports an IO base address that is 0x10000
43 less than it should be.