Decouple network stack from transceiver #644

miri64 · 2014-02-07T00:05:37Z

Follow-Up to #460

Depends on #460 and #557.

miri64 · 2014-02-07T00:05:58Z

Not tested functionality yet, but it builds on native.

OlegHahm · 2014-02-07T11:30:32Z

Needs rebase already.

miri64 · 2014-02-07T14:51:57Z

Yes, because of #648 I guess ;-)

miri64 · 2014-02-07T14:58:07Z

Done + Rebased all depending PRs (#460 and #557).

OlegHahm · 2014-02-07T17:51:06Z

examples/rpl_udp/rpl.c

@@ -100,10 +103,10 @@ void rpl_udp_init(char *str)
    ipv6_addr_init(&std_addr, 0xABCD, 0xEF12, 0, 0, 0x1034, 0x00FF, 0xFE00, id);
    ipv6_addr_init_prefix(&prefix, &std_addr, 64);
    plist_add(&prefix, 64, NDP_OPT_PI_VLIFETIME_INFINITE, 0, 1, ICMPV6_NDP_OPT_PI_FLAG_AUTONOM);


Needs to be replaced by ndp_add_prefix_info() (?).

Yes, sorry tested the last change only on the 6LoWPAN plugtests

miri64 · 2014-02-07T18:24:29Z

For those wondering why I removed the ip command from the applications: ifconfig [<interface>] from net_if shell commands does this + more info

OlegHahm · 2014-02-07T18:28:15Z

needs rebase again.

mehlis · 2014-02-12T21:30:47Z

probably the example/rpl_udp needs changes too:

$ valgrind ./bin/native/rpl_udp.elf tap0
==5912== Memcheck, a memory error detector
==5912== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.
==5912== Using Valgrind-3.8.1 and LibVEX; rerun with -h for copyright info
==5912== Command: ./bin/native/rpl_udp.elf tap0
==5912== 
RIOT native uart0 initialized.
RIOT native interrupts/signals initialized.
LED_GREEN_OFF
LED_RED_ON
RIOT native board initialized.
RIOT native hardware initialization complete.

kernel_init(): This is RIOT! (Version: 2013.08-1210-gf168)
Scheduler...[OK]
kernel_init(): jumping into first task...
Auto init hwtimer module.
Auto init vtimer module.
Auto init uart0 module.
UART0 thread started.
uart0_init() [OK]
Auto init net_if module.
Set radio address on interface 0 to 1
Change this value at compile time with macro CONF_RADIO_ADDR
Set PAN ID on interface 0 to 0xabcd
Change this value at compile time with macro CONF_PAN_ID
Interface 0 initialized
Auto init 6LoWPAN module.
Auto init transport layer [destiny] module.
Initializing transport layer packages. Size of socket_type: 177
RPL router v1.1
> set 1
set 1
Set node ID to 1
> init r
init r
INFO: Initialize as root on address 1
lowpan.c, 1699: sixlowpan_lowpan_init(): add link local address to interface 0: fe80::abcd:ff:fe00:1
lowpan.c, 1712: sixlowpan_lowpan_init(): add solicited nodes multicast address to of link layer address interface 0: ff02::1:0:ff00:1
==5912== Conditional jump or move depends on uninitialised value(s)
==5912==    at 0x400B7AF: bcmp (in /usr/lib/valgrind/vgpreload_memcheck-x86-linux.so)
==5912==    by 0x80565E1: ipv6_net_if_addr_match (ip.c:508)
==5912==    by 0x8056284: ipv6_net_if_add_addr (ip.c:445)
==5912==    by 0x805ADF5: sixlowpan_lowpan_init_interface (lowpan.c:1715)
==5912==    by 0x8062F5B: rpl_init (rpl.c:211)
==5912==    by 0x8050E35: rpl_udp_init (rpl.c:74)
==5912==    by 0x8051B8C: handle_input_line (shell.c:104)
==5912==    by 0x80519C2: shell_run (shell.c:162)
==5912==    by 0x8050305: main (main.c:54)
==5912== 
==5912== Conditional jump or move depends on uninitialised value(s)
==5912==    at 0x80565E7: ipv6_net_if_addr_match (ip.c:508)
==5912==    by 0x8056284: ipv6_net_if_add_addr (ip.c:445)
==5912==    by 0x805ADF5: sixlowpan_lowpan_init_interface (lowpan.c:1715)
==5912==    by 0x8062F5B: rpl_init (rpl.c:211)
==5912==    by 0x8050E35: rpl_udp_init (rpl.c:74)
==5912==    by 0x8051B8C: handle_input_line (shell.c:104)
==5912==    by 0x80519C2: shell_run (shell.c:162)
==5912==    by 0x8050305: main (main.c:54)
==5912== 
lowpan.c, 1725: sixlowpan_lowpan_init(): add all nodes multicast address to interface 0: ff02::1
lowpan.c, 1738: sixlowpan_lowpan_init(): add loopback address to interface 0: ::1
6LoWPAN and RPL initialized.
Channel set to 10
Destiny initialized
> ==5912== Invalid read of size 2
==5912==    at 0x80552A0: ipv6_send_packet (ip.c:70)
==5912==    by 0x80636E6: rpl_send (rpl.c:949)
==5912==    by 0x80634C3: send_DIO (rpl.c:328)
==5912==    by 0x80625B8: trickle_timer_over (trickle.c:134)
==5912==    by 0x4A121CEA: makecontext (in /usr/lib/libc-2.17.so)
==5912==    by 0x8082E03: ???
==5912==  Address 0x4 is not stack'd, malloc'd or (recently) free'd
==5912== 
==5912== 
==5912== Process terminating with default action of signal 11 (SIGSEGV)
==5912==  Access not within mapped region at address 0x4
==5912==    at 0x80552A0: ipv6_send_packet (ip.c:70)
==5912==    by 0x80636E6: rpl_send (rpl.c:949)
==5912==    by 0x80634C3: send_DIO (rpl.c:328)
==5912==    by 0x80625B8: trickle_timer_over (trickle.c:134)
==5912==    by 0x4A121CEA: makecontext (in /usr/lib/libc-2.17.so)
==5912==    by 0x8082E03: ???
==5912==  If you believe this happened as a result of a stack
==5912==  overflow in your program's main thread (unlikely but
==5912==  possible), you can try to increase the size of the
==5912==  main thread stack using the --main-stacksize= flag.
==5912==  The main thread stack size used in this run was 8388608.
==5912== 
==5912== HEAP SUMMARY:
==5912==     in use at exit: 0 bytes in 0 blocks
==5912==   total heap usage: 27 allocs, 27 frees, 2,737 bytes allocated
==5912== 
==5912== All heap blocks were freed -- no leaks are possible
==5912== 
==5912== For counts of detected and suppressed errors, rerun with: -v
==5912== Use --track-origins=yes to see where uninitialised values come from
==5912== ERROR SUMMARY: 3 errors from 3 contexts (suppressed: 0 from 0)

miri64 · 2014-02-13T06:19:32Z

@mehlis nope, thats the network stack (and I think some false positives, because some of this "uninitialised variables" are definitely initialized at some point).

miri64 · 2014-02-13T06:20:40Z

BTW I prefer https://github.com/authmillenon/projects/tree/6LoWPAN_PLUGTEST/6lpplugtest for testing ;-)

miri64 · 2014-02-18T13:09:17Z

LoWPAN layer should be working now. Tested (but not valgrinded) with des-testbed/RIOT-projects#97. All other apps will follow.
Only the tests with short addresses will work on native, since native does not support sending and receiving of packets with long addresses (yet), though at least the sending packages are shown in wireshark (keep in mind that fragmented, uncompressed packages are either generated wrong by us or parsed wrong by the 6LoWPAN dissector; I actually guess the last one is right, since otherwise distinguishing uncompressed, fragmented packages and compressed, fragmented packages will be really, really hard except I miss something).
ND as it was should work and tests will follow this night.

miri64 · 2014-02-23T20:23:33Z

No longer WIP but might need some follow-up PRs for neighbor discovery.

miri64 · 2014-02-23T20:24:10Z

(neighbor discovery never worked right before anyway and it was not target of this PR)

mehlis · 2014-02-24T17:28:17Z

@authmillenon PR needs rebase

OlegHahm · 2014-02-28T23:51:52Z

sys/net/network_layer/sixlowpan/icmp.c

-        return SIXLOWERROR_SUCCESS;
+    if (prefix_len > 128) {
+        prefix_len = 128;
+    }


prefix_len & 0x80 - but probably the compiler will optimize this anyway.

But this could benefit from documentation.

Why do you thing that your proposed code would be more efficient? I doubt it very much.

Without any optimization it's a branch less, but I doubt any gained efficiency, too. Mostly I find my proposal neater.

OlegHahm · 2014-03-01T00:00:05Z

Review completed.

miri64 · 2014-03-01T00:13:16Z

Critiques applied and/or discussed.

OlegHahm · 2014-03-01T14:32:00Z

ACK - from my perspective we are ready to merge this.

Decouple network stack from transceiver

OlegHahm · 2014-03-01T14:55:41Z

Hooray!

mehlis · 2014-03-01T17:07:38Z

whooohoooo, let's focus on #799 which is the next big one

miri64 mentioned this pull request Feb 7, 2014

Make applications work with #644 des-testbed/RIOT-projects#95

Open

miri64 mentioned this pull request Feb 7, 2014

Seperate IPv6 and 6LoWPAN #357

Closed

OlegHahm reviewed Feb 7, 2014
View reviewed changes

This was referenced Feb 9, 2014

Add wireshark dissector for native packets #665

Merged

Make ICMP echo requests routable #670

Closed

miri64 mentioned this pull request Feb 23, 2014

Plugtest hotfixes miri64/RIOT#4

Closed

OlegHahm reviewed Feb 28, 2014
View reviewed changes

Martin Lenders and others added 13 commits March 1, 2014 14:44

Fix coding conventions for sixlowpan and rpl module

bc4befb

Fix ieee802154_frame module

90e946e

Put 6LoWPAN border router in its own module

4e39cbd

Decouple ICMPv6 from 6LoWPAN

78b5f2a

Add capability to set and get variable default hop limit

470d874

Use net_if in 6LoWPAN MAC layer

108989b

Use net_if in 6LoWPAN layer

c9e8adc

Byte order and rpl problems

e2e6259

Let IPv6 and ICMP use new 6LoWPAN layer (+fixes)

00acab9

Add sixlowpan to auto_init

6c9578c

Fix apps for previous changes

3a8c177

nbr_cache_auto_rem does not need an extra thread

2c76f99

Eliminate side effects in icmpv6_csum and make it public

bdced31

miri64 added a commit that referenced this pull request Mar 1, 2014

Merge pull request #644 from authmillenon/decouple_network_stack

79a16df

Decouple network stack from transceiver

miri64 merged commit 79a16df into RIOT-OS:master Mar 1, 2014

miri64 deleted the decouple_network_stack branch March 1, 2014 14:46

LudwigKnuepfer mentioned this pull request Mar 1, 2014

test_pnet is broken #829

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Decouple network stack from transceiver #644

Decouple network stack from transceiver #644

miri64 commented Feb 7, 2014

miri64 commented Feb 7, 2014

OlegHahm commented Feb 7, 2014

miri64 commented Feb 7, 2014

miri64 commented Feb 7, 2014

OlegHahm Feb 7, 2014

miri64 Feb 7, 2014

miri64 commented Feb 7, 2014

OlegHahm commented Feb 7, 2014

mehlis commented Feb 12, 2014

miri64 commented Feb 13, 2014

miri64 commented Feb 13, 2014

miri64 commented Feb 18, 2014

miri64 commented Feb 23, 2014

miri64 commented Feb 23, 2014

mehlis commented Feb 24, 2014

OlegHahm Feb 28, 2014

Kijewski Feb 28, 2014

OlegHahm Mar 1, 2014

OlegHahm commented Mar 1, 2014

miri64 commented Mar 1, 2014

OlegHahm commented Mar 1, 2014

OlegHahm commented Mar 1, 2014

mehlis commented Mar 1, 2014

Decouple network stack from transceiver #644

Decouple network stack from transceiver #644

Conversation

miri64 commented Feb 7, 2014

miri64 commented Feb 7, 2014

OlegHahm commented Feb 7, 2014

miri64 commented Feb 7, 2014

miri64 commented Feb 7, 2014

OlegHahm Feb 7, 2014

Choose a reason for hiding this comment

miri64 Feb 7, 2014

Choose a reason for hiding this comment

miri64 commented Feb 7, 2014

OlegHahm commented Feb 7, 2014

mehlis commented Feb 12, 2014

miri64 commented Feb 13, 2014

miri64 commented Feb 13, 2014

miri64 commented Feb 18, 2014

miri64 commented Feb 23, 2014

miri64 commented Feb 23, 2014

mehlis commented Feb 24, 2014

OlegHahm Feb 28, 2014

Choose a reason for hiding this comment

Kijewski Feb 28, 2014

Choose a reason for hiding this comment

OlegHahm Mar 1, 2014

Choose a reason for hiding this comment

OlegHahm commented Mar 1, 2014

miri64 commented Mar 1, 2014

OlegHahm commented Mar 1, 2014

OlegHahm commented Mar 1, 2014

mehlis commented Mar 1, 2014