aboutsummaryrefslogtreecommitdiffstats
path: root/host/lib/transport
Commit message (Collapse)AuthorAgeFilesLines
* Merge branch 'rfx' of git@ettus.sourcerepo.com:ettus/uhd into ioJosh Blum2010-04-122-555/+16
|\ | | | | | | | | Conflicts: host/lib/usrp/usrp2/io_impl.cpp
| * added python+cheetah build requirement, generating vrt.cppJosh Blum2010-04-092-550/+2
| |
| * converted timespec to use nanoseconds for fractional partJosh Blum2010-04-082-39/+43
| |
* | store the mtu and hdr len stuff only in the usrp2 implJosh Blum2010-04-121-8/+4
| |
* | Added data type conversion routines to transport api.Josh Blum2010-04-122-4/+151
| |
* | Created zero copy interface/framework, made use of it in usrp2 udp transport ↵Josh Blum2010-04-121-34/+71
|/ | | | stuff.
* tweak the ifaddrs address discoveryJosh Blum2010-03-261-11/+10
|
* get interface addresses on windowsJosh Blum2010-03-261-272/+39
|
* made python app to generate vrt jump tables, seems to run faster...Josh Blum2010-03-262-56/+700
|
* Merge branch 'addrs' of git@ettus.sourcerepo.com:ettus/uhd into timingJosh Blum2010-03-263-0/+345
|\ | | | | | | | | Conflicts: host/lib/usrp/usrp2/usrp2_impl.hpp
| * added interface address discoveryJosh Blum2010-03-263-0/+345
| |
* | Overhaullllllled the way we do streaming. There is an odd bug whereJosh Blum2010-03-251-1/+1
|/ | | | | | | | | a zero length command (now, no chain) used to stop the streaming. Now it seems to do the reverse... must investigate. Made all clock configuration into enums. The strings were painful and there cant be that many variations that enums cant cover them. The enums will make more sense to developers than mystery strings.
* Added example app to receive timed samples.Josh Blum2010-03-231-2/+2
| | | | | Added useful calls to simple device. Fixed vrt frac time usage (wrong word).
* Moved typedefs from props.hpp into new file types.hpp.Josh Blum2010-03-211-0/+0
| | | | | | | Created structs to replace range tuples, and clock config struct. Merged clock config props into one property using config struct. Added templated dict construction to use the assign::map_list_of. Added gcc flag to set visibility to hidden and use the api macro.
* got uhd almost compiling in windowze. figured out special flags. also had to ↵Josh Blum2010-03-172-14/+17
| | | | use boost stdint because its missing in visual c++, added a bunch of numeric casts to reduce warnings
* Added recovery app to use raw socket to burn known ip addr.Josh Blum2010-03-161-1/+1
| | | | Fixed up fw so the device can reply after new addrs are set.
* Moved timeouts into the udp transports.Josh Blum2010-03-042-4/+64
| | | | | | Simplified the fast path checking in the fw, but it turns out this was not the issue. Fixed some bad bit operations with the 16sc words (dont forget sign extension). Added some more documentation to the headers....
* Split metadata into rx and tx specific metadata.Josh Blum2010-03-031-8/+8
| | | | | | The rx metadata has fragment flags and the tx metatdata has burst flags. Made the io impl for usrp2 rx routine fill in the rx metatdata fragment flag. Added device documentation for send and recv in regards to fragmentation.
* Making use of vrt lib in the usrp2 io_impl.Josh Blum2010-03-031-3/+7
| | | | Added a packet size param to the vrt pack and unpack.
* Added a vrt library to pack and unpack from metadata.Josh Blum2010-03-021-0/+104
| | | | Added a vrt test app that packs and unpacks the data.
* Expanded the UDP api:Josh Blum2010-03-023-98/+250
| | | | | | | | We can make simple udp transports for discovery and control. We can support a udp zero copy transport (currently just asio). Reworked the io_impl for usrp2 to work with the zero copy api. So far, all of this untested other than compiling. A cut-down vrt library is in the works to simplify the io impl.
* Added IF data io handing within the usrp2 impl.Josh Blum2010-02-231-14/+6
| | | | | It packs and unpacks vrt headers/metadata. NOT YET TESTED IN ANY WAY...
* Work on the io interface for a device (and some implementation work in usrp2).Josh Blum2010-02-221-8/+12
| | | | | Modified the udp transport to reflect some of these changes. Got the fw compiling again, and it will not set data to true for small payloads (configuration ones).
* Moved the udp implementation guts into the cpp fileJosh Blum2010-02-221-5/+40
|
* moved host code into host directory for clean (unambiguous) top levelJosh Blum2010-02-171-0/+67