sergiotarxz/nuttx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
2d56197792
nuttx/net/Kconfig
Michael Jung 2e27698d6c SLIP: Switch to poll based design 
This is a refactored version of the SLIP network driver.  Updates
include:

1. The original design started two kernel threads per SLIP device.
   The refactored version uses file_poll to essentially be driven
   by the UART RX and TX interrupts and pushes work to the low
   priority work queue.

2. The SLIP byte un-/stuffing is more efficient now, using memcpy
   instead of handling each byte individually.

3. The switch of the old SLIP driver to IOBs caused buffer overwrites
   if packets were sent that would not fit into a single IOB.  This is
   fixed now.

Signed-off-by: Michael Jung <michael.jung@secore.ly>
2023-05-26 19:57:57 -03:00

398 lines
11 KiB
Plaintext
Raw Blame History

#
# For a description of the syntax of this configuration file,
# see the file kconfig-language.txt in the NuttX tools repository.
#
config ARCH_HAVE_NET
bool
default n
config ARCH_HAVE_PHY
bool
default n
config ARCH_PHY_INTERRUPT
bool
default n
config ARCH_PHY_POLLED
bool
default n
config ARCH_HAVE_NETDEV_STATISTICS
bool
default n
config NET_WRITE_BUFFERS
bool
default n
select MM_IOB
config NET_READAHEAD
bool
default n
select MM_IOB
config NET_MCASTGROUP
bool
default n
config NET
bool "Networking support"
default n
select ARCH_HAVE_NET
---help---
Enable or disable all network features
if NET
config NET_PROMISCUOUS
bool "Promiscuous mode"
default n
---help---
Force the Ethernet driver to operate in promiscuous mode (if supported
by the Ethernet driver).
menu "Driver buffer configuration"
config NET_ETH_PKTSIZE
int "Ethernet packet buffer size"
default 1294 if NET_IPv6
default 590 if !NET_IPv6
range 1294 1518 if NET_IPv6
range 590 1518 if !NET_IPv6
depends on NET_ETHERNET
---help---
Packet buffer size. This size includes the TCP/UDP payload plus the
size of TCP/UDP header, the IP header, and the Ethernet header.
This value is related to the MTU (Maximum Transmission Unit), except
that it includes the size of the link layer header; the payload is
the MSS (Maximum Segment Size).
IPv4 hosts are required to be able to handle an MSS of at least
536 octets, resulting in a minimum buffer size of 536+20+20+14 =
590.
IPv6 hosts are required to be able to handle an MSS of 1220 octets,
resulting in a minimum buffer size of 1220+20+40+14 = 1294
To get an MTU of 1500, for example, you would need packet buffer of
size 1514.
config NET_SLIP_PKTSIZE
int "SLIP packet buffer size"
default 296
depends on NET_SLIP
range 296 1518
---help---
Provides the size of the SLIP packet buffers. This size includes
the TCP/UDP payload plus the size of TCP/UDP header and the IP header.
This value is related to the MTU (Maximum Transmission Unit), except
that it includes the size of the link layer header; the payload is
the MSS (Maximum Segment Size). SLIP has no link layer header so for
SLIP the MTU is the same as the PKTSIZE.
SLIP is required to support at least 256+20+20 = 296. Values other than
296 are not recommended.
The Linux slip module hard-codes its MTU size to 296 (40 bytes for
the IP+TCP headers plus 256 bytes of data). So you might as well
set CONFIG_NET_SLIP_PKTSIZE to 296 as well.
There may be an issue with this setting, however. I see that Linux
uses a MTU of 296 and window of 256, but actually only sends 168
bytes of data: 40 + 128. I believe that is to allow for the 2x
worst cast packet expansion. Ideally we would like to advertise the
256 MSS, but restrict transfers to 128 bytes (possibly by modifying
the MSS value in the TCP connection structure).
config NET_GUARDSIZE
int "Driver I/O guard size"
default 2
---help---
Network drivers often receive packets with garbage at the end and
are longer than the size of packet in the TCP header. The following
"fudge" factor increases the size of the I/O buffering by a small
amount to allocate slightly oversize packets. After receipt, the
packet size will be chopped down to the size indicated in the TCP
header.
config NET_LL_GUARDSIZE
int "Data Link Layer(L2) Guard size of Network buffer(IOB)"
default 16 if NET_CAN && NET_TIMESTAMP
default 14 if NET_ETHERNET
default 0
---help---
This is reserved l2 buffer header size of network buffer to isolate
the L2/L3 (MAC/IP) data on Network layer, which will be beneficial
to L3 network layer protocol transparent transmission and forwarding
config NET_RECV_BUFSIZE
int "Net Receive buffer size"
default 0
---help---
This is the default value for receive buffer size.
config NET_SEND_BUFSIZE
int "Net Send buffer size"
depends on NET_TCP_WRITE_BUFFERS || NET_UDP_WRITE_BUFFERS
default 0
---help---
This is the default value for send buffer size.
endmenu # Driver buffer configuration
menu "Link layer support"
config NET_ETHERNET
bool "Ethernet support"
default y
---help---
If NET_SLIP is not selected, then Ethernet will be used (there is
no need to define anything special in the configuration file to use
Ethernet -- it is the default).
config NET_CELLULAR
bool "Cellular Link support"
default n
---help---
Add support for the cellular network device. Unlike Ethernet, cellular
network transmit pure IP packets.
config NET_LOOPBACK
bool "Local loopback"
select ARCH_HAVE_NETDEV_STATISTICS
default n
---help---
Add support for the local network loopback device, lo.
config NET_LOOPBACK_PKTSIZE
int "Loopback packet buffer size"
default 0
depends on NET_LOOPBACK
range 0 65535
---help---
The loopback driver packet buffer should be quite large. The larger
the loopback packet buffer, the better will be TCP performance of
the loopback transfers. The Linux loopback device historically used
packet buffers of size 16Kb, but that was increased in recent Linux
versions to 64Kb. Those sizes may be excessive for resource
constrained MCUs, however.
The network enforces a lower limit that is the maximum packet size
of all enabled link layer protocols. The default value of
CONFIG_NET_LOOPBACK_PKTSIZE is zero, meaning that this maximum
packet size will be used by loopback driver.
menuconfig NET_MBIM
bool "MBIM modem support"
depends on USBHOST_CDCMBIM
default n
menuconfig NET_SLIP
bool "SLIP support"
select ARCH_HAVE_NETDEV_STATISTICS
depends on NET_IPv4
default n
---help---
Enables building of the SLIP driver. SLIP requires
at least one IP protocol selected.
SLIP supports point-to-point IP communications over a serial port.
The default link layer for network layer is Ethernet. If NET_SLIP
is defined in the NuttX configuration file, then SLIP will be
supported. The basic differences between the SLIP and Ethernet
configurations is that when SLIP is selected:
* The link level header (that comes before the IP header) is omitted.
* All MAC address processing is suppressed.
* ARP is disabled.
If NET_SLIP is not selected, then Ethernet will be used (there is
no need to define anything special in the configuration file to use
Ethernet -- it is the default).
if NET_SLIP
config NET_SLIP_NINTERFACES
int "Number of SLIP interfaces"
default 1
---help---
Selects the number of physical SLIP
interfaces to support.
Default: 1
endif # NET_SLIP
menuconfig NET_TUN
bool "TUN Virtual Network Device support"
default n
depends on ALLOW_BSD_COMPONENTS
select ARCH_HAVE_NETDEV_STATISTICS
if NET_TUN
config TUN_NINTERFACES
int "Number of TUN interfaces"
default 1
range 1 8
---help---
Selects the number of TUN
interfaces to support.
Default: 1
config NET_TUN_PKTSIZE
int "TUN packet buffer size"
default 296
range 296 1518
---help---
Provides the size of the TUN packet buffers. This size includes
the TCP/UDP payload plus the size of TCP/UDP header and the IP header.
This value is related to the MTU (Maximum Transmission Unit), except
that it includes the size of the link layer header; the payload is
the MSS (Maximum Segment Size). TUN has no link layer header so for
TUN the MTU is the same as the PKTSIZE.
endif # NET_TUN
config NETDEV_LATEINIT
bool "Late driver initialization"
default n
---help---
Normally, networking initialization occur in the later phase of the
boot process in the function up_initialize() when it calls the
driver initialization function, up_netinitialize(). This
initialization occurs after a sufficient amount of the OS has been
initialized so that driver registration can be performed, but
before the completion of OS initialization and before the first
application is started.
In a few situations, however, you may want to suppress this early
network driver initialization. As examples:
- If you are using SLIP or PPPD, then there will be no network
driver to be initialized,
- Certain multi-network configurations where a simple call to
up_netinitialize() may be insufficient, and
- Situations where there are other board-level hardware
dependencies so that the hardware is not in an appropriate
state for up_netinitialize() to be called.
Examples of this latter situation include such things as network
drivers that required some setup via an I2C I/O expander, or network
drivers that depend on USB, SPI, I2C, PCI, serial, or other
interfaces that may not be ready when up_netinitialize() is normally
called.
endmenu # Link layer support
source "net/netdev/Kconfig"
menu "Internet Protocol Selection"
config NET_IPv4
bool "IPv4"
default y
---help---
Build in support for IPv4.
config NET_IPv6
bool "IPv6"
default n
---help---
Build in support for IPv6.
source "net/neighbor/Kconfig"
menuconfig NET_6LOWPAN
bool "6LoWPAN support"
default n
select NETDEV_IOCTL
select NET_HAVE_STAR
depends on NET_IPv6
---help---
Enable support for Low power Wireless Personal Area Networking (6LoWPAN)
for IEEE 802.15.4 or other packet radios.
source "net/sixlowpan/Kconfig"
source "net/ipforward/Kconfig"
source "net/nat/Kconfig"
source "net/netfilter/Kconfig"
source "net/ipfrag/Kconfig"
endmenu # Internet Protocol Selection
source "net/socket/Kconfig"
source "net/inet/Kconfig"
source "net/pkt/Kconfig"
source "net/local/Kconfig"
source "net/rpmsg/Kconfig"
source "net/can/Kconfig"
source "net/netlink/Kconfig"
source "net/tcp/Kconfig"
source "net/udp/Kconfig"
source "net/bluetooth/Kconfig"
source "net/ieee802154/Kconfig"
source "net/icmp/Kconfig"
source "net/icmpv6/Kconfig"
source "net/mld/Kconfig"
source "net/igmp/Kconfig"
source "net/arp/Kconfig"
source "net/procfs/Kconfig"
source "net/usrsock/Kconfig"
source "net/utils/Kconfig"
config NET_STATISTICS
bool "Collect network statistics"
default n
---help---
Network layer statistics on or off
config NET_HAVE_STAR
bool
default n
---help---
Automatically enabled if at least one selected L2 protocol supports
a STAR topology. In order to support the star topology, the L2
protocol must support relaying all packets to a well-known hub node.
menu "Network Topologies"
config NET_STAR
bool "Enable star topology"
default n
depends on NET_HAVE_STAR && NET_IPv6
---help---
Enable support for a star network topology.
NOTE: Currently only supported by 6LoWPAN.
NOTE: L2 forwarding only supported for IPv6.
choice
prompt "Node role"
depends on NET_STAR
default NET_STARPOINT
---help---
Specifies the role of this not in the star configuration.
config NET_STARPOINT
bool "Point node in star"
---help---
The node is a "point" in the star configuration and must send all
packets to the star hub node.
config NET_STARHUB
bool "Hub node of star"
select NET_IPFORWARD
---help---
This is the "hub" node in the star configurations. It will receive
packets packets from all "point" nodes and perform L2 forwarding of
the packets that are not destined for this node.
endchoice # Node role
endmenu # Network Topologies
source "net/route/Kconfig"
endif # NET
Reference in New Issue View Git Blame Copy Permalink