Before this commit, weak symbols were not overridden by non-weak symbols
found in archive libraries when linking with recent versions of
binutils. As stated in the System V ABI, "the link editor does not
extract archive members to resolve undefined weak symbols".
This commit changes all Makefiles to use partial linking (ld -r) instead
of creating library archives, which forces all symbols to participate in
linking, allowing non-weak symbols to override weak symbols as intended.
This approach is also used by Linux, from which the gmake function
cmd_link_o_target (defined in config.mk and used in all Makefiles) is
inspired.
The name of each former library archive is preserved except for
extensions which change from ".a" to ".o". This commit updates
references accordingly where needed, in particular in some linker
scripts.
This commit reveals board configurations that exclude some features but
include source files that depend these disabled features in the build,
resulting in undefined symbols. Known such cases include:
- disabling CMD_NET but not CMD_NFS;
- enabling CONFIG_OF_LIBFDT but not CONFIG_QE.
Signed-off-by: Sebastien Carlier <sebastien.carlier@gmail.com>
This has always been confusing, and the idea of these functions returning the
number of interfaces initialized was half-baked and ultimately pointless.
Instead, act more like regular functions and return < 0 on failure, >= 0 on
success.
This change shouldn't break anything.
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Can't get IP address with dhcp due to the dhcp server not
allow the empty param list request under some network env
This patch is based on Gray Remlin's initial patch.
Signed-off-by: Jason Liu <r64343@freescale.com>
Signed-off-by: Gray Remlin <g_remlin@rocketmail.com>
Most people don't use the 'rarpboot' command, so only enable it when
CONFIG_CMD_RARP is defined.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Commit 6e37b1a3a25004d3df5867de49fff6b3fc9c4f04 modifies several net calls
to take a (const char *) parameter instead of (char *), but in some cases
the modified functions call other functions taking (char *). The end result
is warnings about discarding the const qualifier.
This patch fixes these other function signatures.
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Some commands operate on eth device names (like 'mii'), but those cannot
be passed on the command line as one argument. So detect devices like
these and warn about them so someone will fix it.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Move it inside the #ifdef CONFIG_NET_MULTI to avoid
eth.c:64: warning: 'eth_mac_skip' defined but not used
messages from a number of old, non-CONFIG_NET_MULTI boards.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
TFTP'ing a file of size 1747851 bytes with CONFIG_IP_DEFRAG and
CONFIG_TFTP_BLOCKSIZE set to 4096 fails with a timeout, because
the last fragment is not taken into account. This patch fixes
IP fragments having less than 8 bytes of payload.
Signed-off-by: Stephane Fillod <stephane.fillod@grassvalley.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Add a new function to the eth_device struct for programming a network
controller's hardware address.
After all network devices have been initialized and the proper MAC address
for each has been determined, make a device driver call to program the
address into the device. Only device instances with valid unicast addresses
will be programmed.
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Acked-by: Detlev Zundel <dzu@denx.de>
Tested-by: Prafulla Wadaskar <prafulla@marvell.com>
Tested-by: Heiko Schocher <hs@denx.de>
Tested-by: Thomas Chou <thomas@wytron.com.tw>
For some reason, (which I can't find any documentation on), if U-Boot
gives a port number higher than 17500 to a Microsoft DNS server, the
server will reply to port 17500, and U-Boot will ignore things (since
that isn't the port it asked the DNS server to reply to).
This fixes that by ensuring the random port number is less than 17500.
Signed-off-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
So far, TFTP negotiated a fixed retransmission timeout of 5 seconds.
In some cases (busy networks, slow TFTP servers) this caused very
slow transfers. A new environment variable "tftptimeout" allows to
set this timeout. Lowering this value may make downloads succeed
faster in networks with high packet loss rates or with unreliable
TFTP servers.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Cc: Ben Warren <biggerbadderben@gmail.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
There are boards out there that do not have network support in
U-Boot (CONFIG_CMD_NET not set), but they do so in Linux. This
makes it desirable to be able to port network configuration (like
the IP address) to the Linux kernel.
We should not make the passing of the IP configuration to Linux
dependent on U-Boot features / settings.
For this, make getenv_IPaddr() global. This fixes build error
u-boot/lib_xxx/board.c:360: undefined reference to `getenv_IPaddr'
on various architectures.
Signed-off-by: Dirk Behme <dirk.behme@googlemail.com>
Acked-by: Ben Warren <biggerbadderben@gmail.com>
- NfsTimeout() does not correctly update the NFS timeout value which
results in NfsTimeout() only being called once in certain situations.
This can result in the 'nfs' command hanging indefinetly. For
example, the command:
nfs 192.168.0.1:/home/user/file
will not exit until ctrl-c is pressed if 192.168.0.1 does not have an
NFS server running.
This issue is resolved by reinitializting the NFS timeout value inside
NfsTimeout() when a timeout occurs.
- Make the 'nfs' command print the 'T' character when a timeout occurs.
Previously there was no indication that timeouts were occuring.
- Mimic the 'tftpboot' command and when a download fails print "Retry
count exceeded; starting again", and restart the download taking the
'netretry' environment variable into account.
Signed-off-by: Evan Samanas <esamanas@xes-inc.com>
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Tested on TQM8xxL.
Tested by: Wolfgang Denk <wd@denx.de>
Tested on MPC8527DS.
Tested by: Ed Swarthout <Ed.Swarthout@freescale.com>
An off by one error may cause nfs readlink lookup fail if
nfs_path_buff has non-zero data from a previous use.
Loading: *** ERROR: File lookup fail
Signed-off-by: Ed Swarthout <Ed.Swarthout@freescale.com>
'netretry = once' does the same as 'netretry = yes', because it is not stored
when it was tried once.
Signed-off-by: Remy Bohmer <linux@bohmer.net>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Fix warning: bootp.c:695: warning: dereferencing type-punned pointer
will break strict-aliasing rules
Signed-off-by: Wolfgang Denk <wd@denx.de>
Cc: Ben Warren <biggerbadderben@gmail.com>
Optionally add RFC 2349 "Transfer Size Option", so we can minimize the
time spent sending data over the UART (now print a single line during a
tftp transfer).
- If turned on (CONFIG_TFTP_TSIZE), U-Boot asks for the size of the file.
- if receives the file size, a single line (50 chars) are printed.
one hash mark == 2% of the file downloaded.
- if it doesn't receive the file size (the server doesn't support RFC
2349, prints standard hash marks (one mark for each UDP frame).
Signed-off-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
To take advantage of defragmented packets, the config file
can define CONFIG_NFS_READ_SIZE to override the 1kB default.
No support is there for an environment variable by now.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Increasing the block size is useful if CONFIG_IP_DEFRAG is
used. Howerver, the last fragments in a burst may overflow the
receiving ethernet, so the default is left at 1468, with thre new
CONFIG_TFTP_BLOCKSIZE for config files. Further, "tftpblocksize"
can be set in the environment.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
The defragmenting code is enabled by CONFIG_IP_DEFRAG; the code is
useful for TFTP and NFS transfers. The user can specify the maximum
defragmented payload as CONFIG_NET_MAXDEFRAG (default 16k).
Since NFS has a bigger per-packet overhead than TFTP, the static
reassembly buffer can hold CONFIG_NET_MAXDEFRAG + the NFS overhead.
The packet buffer is used as an array of "hole" structures, acting as
a double-linked list. Each new fragment can split a hole in two,
reduce a hole or fill a hole. No support is there for a fragment
overlapping two diffrent holes (i.e., thre new fragment is across an
already-received fragment).
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
tftp.c:294: warning: pointer targets in passing argument 1 of 'strlen'
differ in signedness
This was only visible for the utx8245 board which seems to have DEBUG
enabled.
Signed-off-by: Wolfgang Denk <wd@denx.de>
In the previous enetaddr refactoring, the assumption with commit 56b555a644
was that the eth layer would handle the env -> device enetaddr syncing.
This was not the case as eth_initialize() is called only once and the sync
occurs there. So make sure the eth_init() function does the env -> device
sync with every network init.
Reported-by: Andrzej Wolski <awolski@poczta.fm>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Minor ./net cleanups - no functional changes
- change #ifdef DEBUG printf(); #endif to just debug()
- changed __FUNCTION__ to __func__
- got rid of extra whitespace between function and opening brace
- removed unnecessary braces on if statements
gcc dead code elimination should make this functionally/size equivalent
when DEBUG is not defined. (confirmed on Blackfin, with gcc 4.3.3).
Signed-off-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Linux's netconsole works much better when you can pass it the MAC address of
the server. (otherwise it just uses broadcast, which everyone else on my
network complains about :)
This sets the env var "serveraddr" (to match ethaddr), so that you can pass
it to linux with whatever bootargs you want to....
addnetconsole=set bootargs $(bootargs) netconsole=@$(ipaddr)/eth0,@$(serverip)/$(serveraddr)
Signed-of-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Fixed the DHCP/BOOTP/RARP regression introduced in u-boot-2009.06
by initializing our IP addr to 0 in order to accept any IP addr
assigned to us by the DHCP/BOOTP/RARP server.
Ack-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Michael Zaidman <michael.zaidman@gmail.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
On 04 Oct 2008 Pieter posted a dns implementation for U-Boot.
http://www.mail-archive.com/u-boot-users@lists.sourceforge.net/msg10216.html
>
> DNS can be enabled by setting CFG_CMD_DNS. After performing a query,
> the serverip environment var is updated.
>
> Probably there are some cosmetic issues with the patch. Unfortunatly I
> do not have the time to correct these. So if anybody else likes DNS
> support in U-Boot and has the time, feel free to patch it in the main tree.
Here it is again - slightly modified & smaller:
- update to 2009-06 (Pieter's patch was for U-Boot 1.2.0)
- README.dns is added
- syntax is changed (now takes a third option, the env var to store
the result in)
- add a random port() function in net.c
- sort Makefile in ./net/Makefile
- dns just returns unless a env var is given
- run through checkpatch, and clean up style issues
- remove packet from stack
- cleaned up some comments
- failure returns much faster (if server responds, don't wait for
timeout)
- use built in functions (memcpy) rather than byte copy.
Signed-off-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Pieter Voorthuijsen <pieter.voorthuijsen@prodrive.nl>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
The net code is mostly consistent in using 'Packet' rather than 'Pkt', so
rename the minor detractor to follow suite.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
The problem is that BOOTP_SIZE uses ETHER_HDR_SIZE which is 14 bytes.
If sending a VLAN tagged frame (when env variable vlan is set) this
should be VLAN_ETHER_HDR_SIZE=18 which is what NetSetEther returns.
Signed-off-by: Norbert van Bolhuis <nvbolhuis@aimvalley.nl>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Added CONFIG_NET_MULTI to all Davinci boards
Removed all calls to Davinci network driver from board code
Added cpu_eth_init() to cpu/arm926ejs/cpu.c
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Fix some issues introduced from commit:
2f70c49e5b
suggested by Mike Frysinger.
- added some comment for the env_id variable in common_cmd_nvedit.c
- moved some variables in fn scope instead of file scope
- NetInitLoop now static void
Signed-off-by: Heiko Schocher <hs@denx.de>
Acked-by: Ben Warren <biggerbadderben@gmail.com>
__attribute__ follows gcc's documented syntax and is generally more
common than __attribute. This change is only asthetic and should not
affect functionality.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
The patch fixes the bug of partial initialization of global network
parameters.
Upon u-boot's start up the first ping command causes a failure of the
consequent TFTP command. It happens in the recently added mechanism of
the NetLoop initialization where initialization of global network
parameters is separated in the NetInitLoop routine which is called per
env_id change. Thus, ping request will initialize the network parameters
necessary for ping operation only, afterwards the env_changed_id will be
set to the env_id that will prevent all following initialization requests
from other protocols.
The problem is that the initialized by ping subset of network parameters
is not sufficient for other protocols and particularly for TFTP which
requires the NetServerIp also.
Signed-off-by: Michael Zaidman <michael.zaidman@gmail.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Declare new utility functions for converting between the environment
variables (eth*addr) and the binary MAC address representation. This way
we can unify all the random places that already do this kind of thing.
The functions in question:
eth_parse_enetaddr - "..." -> {...}
eth_getenv_enetaddr - env -> {...}
eth_setenv_enetaddr - {...} -> env
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
CC: Ben Warren <biggerbadderben@gmail.com>
Now that our printf functions support the %pI4 modifier like the kernel,
let's drop the inflexible print_IPaddr() function and covert over to the
%pI4 modifier.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
CC: Ben Warren <biggerbadderben@gmail.com>
NetLoop polls every cycle with getenv some environment variables.
This is horribly slow, especially when the environment is big.
This patch reads only the environment variables in NetLoop,
when they were changed.
Also moved the init part of the NetLoop function in a seperate
function.
Signed-off-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
This patch removes volatile from:
volatile IP_t *ip = (IP_t *)xip;
Due to a bug, avr32-gcc will assume that ip is aligned on a word boundary when
using volatile, which causes an exception since xip isn't aligned on a word
boundary.
Signed-off-by: Gunnar Rangoy <gunnar@rangoy.com>
Signed-off-by: Paul Driveklepp <pauldriveklepp@gmail.com>
Signed-off-by: Olav Morken <olavmrk@gmail.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Mike Frysinger