From 663f5dfbdde689812059e64bb737d303c328a35f Mon Sep 17 00:00:00 2001 From: Gregory Nutt Date: Tue, 8 Aug 2017 18:14:40 -0600 Subject: [PATCH 1/3] Update readme. Fix stray newline in shenzhou Make.defs --- configs/b-l475e-iot01a/README.txt | 26 ++++++++++---------------- configs/shenzhou/scripts/Make.defs | 3 +-- 2 files changed, 11 insertions(+), 18 deletions(-) diff --git a/configs/b-l475e-iot01a/README.txt b/configs/b-l475e-iot01a/README.txt index a1fe6c50278..3432944a7ee 100644 --- a/configs/b-l475e-iot01a/README.txt +++ b/configs/b-l475e-iot01a/README.txt @@ -583,26 +583,19 @@ Configuration sub-directories be beefed up to handle this routinely without asserting and without leaving the Spirit in a bad state. - One remaining issue with the above is that when we fail to go to the TX - state, there is a lot of warning debug output. ANY debug output while - the Spirit is heavily loaded WILL cause failures and packet loss! - Perhaps using a RAMLOG would remedy this. - The TCP test beats the radio very hard and it is actually heartening that there are no failures that lead to data loss in this environment. I would say it is functional but fragile in this usage, but probably robust in a less busy environment. - 2017-08-08: The warning debug messages noted yesterday are no longer - present. No clue why. - - Added broadcast packet transfers using the hub-based broadcast UDP - client. This appears to be a problem the HC06 compression and/or - decompression. The decompression logic comes up with the - destination address of ff02::ff00:00fe:3500 (which derives from the - receiving node address of 37) instead of the all-nodes multicast - address of ff02::0001. It is then out of sync with the IPHC headers - and is unable to uncompress the rest of the packet correctly. + 2017-08-08: Added broadcast packet transfers using the hub-based + broadcast UDP client. This appears to be a problem the HC06 + compression and/or decompression. The decompression logic comes up + with the destination address of ff02::ff00:00fe:3500 (which derives + from the receiving node address of 37) instead of the all-nodes + multicast address of ff02::0001. It is then out of sync with the + IPHC headers and is unable to uncompress the rest of the packet + correctly. Trying again with HC1 compression, I see other isses. The first frame is received correctly, but the following frames have an incorrect @@ -614,4 +607,5 @@ Configuration sub-directories we are sending to the multicast or broadcast address, should we not also disable ACKs, retries, and RX timeouts? What will happen if multiple radios ACK? At a minimum it could keep the driver - unnecessarily busy. + unnecessarily busy. There is some prototype code to do just this + in the driver, but does not seem to work. diff --git a/configs/shenzhou/scripts/Make.defs b/configs/shenzhou/scripts/Make.defs index 8135e642f5c..9ab3b8d8d84 100644 --- a/configs/shenzhou/scripts/Make.defs +++ b/configs/shenzhou/scripts/Make.defs @@ -99,8 +99,7 @@ CPPFLAGS = $(ARCHINCLUDES) $(ARCHDEFINES) $(EXTRADEFINES) AFLAGS = $(CFLAGS) -D__ASSEMBLY__ NXFLATLDFLAGS1 = -r -d -warn-common -NXFLATLDFLAGS2 = $(NXFLATLDFLAGS1) -T$(TOPDIR)/binfmt/libnxflat/gnu-nxflat-gotoff.ld --no-check-sections +NXFLATLDFLAGS2 = $(NXFLATLDFLAGS1) -T$(TOPDIR)/binfmt/libnxflat/gnu-nxflat-gotoff.ld -no-check-sections #NXFLATLDFLAGS2 = $(NXFLATLDFLAGS1) -T$(TOPDIR)/binfmt/libnxflat/gnu-nxflat-pcrel.ld -no-check-sections LDNXFLATFLAGS = -e main -s 2048 From 7c142be7ec419e6ea74dd2eda1a711b59869266f Mon Sep 17 00:00:00 2001 From: Gregory Nutt Date: Tue, 8 Aug 2017 18:37:42 -0600 Subject: [PATCH 2/3] Fix small build problems found in build testing. --- net/ipforward/ipfwd_dropstats.c | 1 + net/ipforward/ipfwd_poll.c | 1 + 2 files changed, 2 insertions(+) diff --git a/net/ipforward/ipfwd_dropstats.c b/net/ipforward/ipfwd_dropstats.c index c6ebc68626b..b79d6435876 100644 --- a/net/ipforward/ipfwd_dropstats.c +++ b/net/ipforward/ipfwd_dropstats.c @@ -41,6 +41,7 @@ #include +#include #include #include diff --git a/net/ipforward/ipfwd_poll.c b/net/ipforward/ipfwd_poll.c index 00a367a478e..138440c942c 100644 --- a/net/ipforward/ipfwd_poll.c +++ b/net/ipforward/ipfwd_poll.c @@ -42,6 +42,7 @@ #include #include +#include #include "devif/devif.h" #include "ipforward/ipforward.h" From b5f5b07b18f967916d927e3537782c22517521fd Mon Sep 17 00:00:00 2001 From: Gregory Nutt Date: Tue, 8 Aug 2017 18:43:05 -0600 Subject: [PATCH 3/3] Another compiler error found in build testing. --- net/netdev/netdev_lladdrsize.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/netdev/netdev_lladdrsize.c b/net/netdev/netdev_lladdrsize.c index 27b7a055b19..e5470837c38 100644 --- a/net/netdev/netdev_lladdrsize.c +++ b/net/netdev/netdev_lladdrsize.c @@ -108,7 +108,7 @@ static inline int netdev_pktradio_addrlen(FAR struct net_driver_s *dev) * ****************************************************************************/ -int netdev_type_lladdrsize(FAR struct net_driver_s *dev) +int netdev_dev_lladdrsize(FAR struct net_driver_s *dev) { DEBUGASSERT(dev != NULL);