Commit message (Collapse) | Author | Age | Files | Lines | ||
---|---|---|---|---|---|---|
... | ||||||
| * | uhd: always link winsock2 on windows, disable pthread SCHED_RR for cygwin | Josh Blum | 2011-04-06 | 1 | -1/+3 | |
| | | ||||||
| * | uhd: tweaks for cygwin/mingw, always link winsock2, findusb1, __USE_W32_SOCKETS | Josh Blum | 2011-04-05 | 1 | -0/+4 | |
| | | ||||||
* | | usb: mark libusb callbacks with LIBUSB_CALL to ensure correct calling convention | Josh Blum | 2011-04-05 | 1 | -2/+12 | |
|/ | ||||||
* | usb: newer libusb1 does not need to link with setupapi.lib | Josh Blum | 2011-04-04 | 1 | -4/+0 | |
| | ||||||
* | uhd: specify msvc for implementations known only to work on msvc | Josh Blum | 2011-04-04 | 1 | -2/+2 | |
| | ||||||
* | uhd: implemented boost barriers on all code that creates threads | Josh Blum | 2011-03-31 | 1 | -2/+6 | |
| | | | | | | | | | The barrier ensures that the thread must spawn before the caller exits. Some of the code already used a mutex to accomplish this, however cygwin chokes when a mutex is locked twice by the same thread. Mutex implementations were replaced with the barrier implementation. Also the barrier implementation is far cleaner. | |||||
* | usb: changes to allow for static linking of libusb on windows | Josh Blum | 2011-03-28 | 1 | -0/+4 | |
| | ||||||
* | uhd: update copyright headers with automated script | Josh Blum | 2011-03-23 | 1 | -1/+1 | |
| | ||||||
* | usb: fix callback cast in libusb zero copy under msvc | Josh Blum | 2011-03-21 | 1 | -1/+1 | |
| | ||||||
* | Merge branch 'usb-cancel' into next | Josh Blum | 2011-03-17 | 1 | -6/+14 | |
|\ | | | | | | | | | Conflicts: host/lib/transport/libusb1_zero_copy.cpp | |||||
| * | USB zero copy impl: proper cleanup for canceled transfers -- wait for cancel ↵ | Nick Foster | 2011-03-16 | 1 | -6/+14 | |
| | | | | | | | | before freeing | |||||
* | | usrp1: safe destruction for usrp1 device | Josh Blum | 2011-03-10 | 1 | -6/+7 | |
| | | ||||||
* | | Merge branch 'master' into next | Josh Blum | 2011-03-03 | 1 | -0/+2 | |
|\| | | | | | | | | | | | Conflicts: host/lib/usrp/usrp2/mboard_impl.cpp host/lib/usrp/usrp2/usrp2_impl.cpp | |||||
| * | udp: return the managed recv buffer to the buffer queue on timeout | Josh Blum | 2011-03-01 | 1 | -0/+2 | |
| | | ||||||
* | | uhd: replaced instanced of std::exception with the uhd exceptions | Josh Blum | 2011-02-24 | 3 | -10/+10 | |
| | | ||||||
* | | uhd: renamed the assert header to assert has | Josh Blum | 2011-02-24 | 3 | -3/+3 | |
| | | | | | | | | | | | | | | | | only the assert has implementation is in this header uhd assert throw moved to the exception header updated code base includes to match | |||||
* | | uhd: moved exception to top level include | Josh Blum | 2011-02-24 | 1 | -1/+1 | |
| | | ||||||
* | | Merge branch 'usrp2_fw_clock_cleanup' into usrp2_dual_dsp | Josh Blum | 2011-02-22 | 5 | -167/+130 | |
|\| | ||||||
| * | uhd: replace header checks in cmake files with more robust compile checks ↵ | Josh Blum | 2011-02-21 | 2 | -15/+26 | |
| | | | | | | | | | | | | for features implemented different ifdefs in the cpp files | |||||
| * | udp: try non-blocking recv first for performance | Josh Blum | 2011-02-20 | 3 | -14/+18 | |
| | | ||||||
| * | udp: update docs for transport, create common header for wait implementation | Josh Blum | 2011-02-20 | 4 | -143/+91 | |
| | | | | | | | | | | | | | | Reimplemented simple udp transport with one impl class. Moved wait for ready/select implementation into common header. Important note on select, timeval should have usecs < 1 second or it may error on some platforms. Fixed in this implementation. | |||||
* | | usrp2: code working again in a completed form but did not test dual dsp | Josh Blum | 2011-02-20 | 1 | -2/+2 | |
|/ | ||||||
* | uhd: use source properties to set flags and defs not globally, but only for ↵ | Josh Blum | 2011-02-17 | 1 | -2/+7 | |
| | | | | the source | |||||
* | uhd: tweaks for windows msvc compiler warnings and errors | Josh Blum | 2011-02-16 | 1 | -2/+2 | |
| | ||||||
* | usrp-e100: replaced safe managed buffers in usrp-e100 mmap with custom ones | Josh Blum | 2011-02-15 | 2 | -115/+0 | |
| | | | | | | | | the buffers are reusable and the overhead is reduced (nothing to malloc) also removed the zero_copy.cpp (not needed anymore) need to test on device... | |||||
* | uhd: reusable buffers for libusb zero copy implementation | Josh Blum | 2011-02-15 | 2 | -347/+186 | |
| | | | | tweaks on udp implementation to simplify | |||||
* | uhd: misc speedups w/ look up tables | Josh Blum | 2011-02-10 | 1 | -1/+2 | |
| | | | | | | | use a look up table for io type size (in the case its used in the fast-path) move the static const pred table in vrt unpacker to the global level, for some reason this was incurring a malloc (perhaps because there were 2 tables). | |||||
* | udp: udp_zero_copy_asio comments and tweaks | Josh Blum | 2011-02-10 | 1 | -10/+2 | |
| | ||||||
* | uhd: simplified converter calls in vrt pkt handler with ref vector changes | Josh Blum | 2011-02-10 | 1 | -9/+4 | |
| | ||||||
* | uhd: use ref vector class for the conversion routines I/O | Josh Blum | 2011-02-10 | 1 | -2/+2 | |
| | ||||||
* | uhd: tweaks to bounded buffer | Josh Blum | 2011-02-10 | 2 | -7/+7 | |
| | | | | | | | Added push with haste. Call with haste first in the wait methods to avoid time compare/wait when not needed. Added new calls to the libusb and udp zero copy impls tests pass | |||||
* | udp: comments and minor code tweaks for udp zero copy impl | Josh Blum | 2011-02-10 | 1 | -18/+41 | |
| | ||||||
* | uhd: tweaks to vrt pkt handler and usrp2 fc monitor | Josh Blum | 2011-02-09 | 1 | -3/+5 | |
| | | | | | | pkt handler, only resize once per call to send/recv (not per fragment) fc monitor, code tweaks, check ready before touching time wait stuff (faster) | |||||
* | uhd: tweaks for windows warnings and errors | Josh Blum | 2011-02-09 | 1 | -3/+3 | |
| | ||||||
* | uhd: replaced std::vector<type> for buffer arguments in send/recv | Josh Blum | 2011-02-09 | 1 | -4/+4 | |
| | | | | | | | | | Created new type ref_vector for representing a vector of pointers. Can be created from std::vector or a pointer. Removes the convenience constrcutors for send/recv, its not needed. Removes malloc/free overhead when using send/recv with pointer. | |||||
* | udp: removed asio implementation, created custom managed buffer classes to ↵ | Josh Blum | 2011-02-09 | 1 | -155/+115 | |
| | | | | re-use | |||||
* | uhd: replace asio buffer in make safe w/ memory and length, makes things simpler | Josh Blum | 2011-02-07 | 3 | -26/+33 | |
| | ||||||
* | uhd: removed instances of shared_from_this() in transports | Josh Blum | 2011-02-05 | 2 | -4/+3 | |
| | | | | | | Its known to be slow and using sptrs of large objects. This should be a slight performance improvement for libusb and usrp_e100_mmap implementation. | |||||
* | uhd: change bounded_buffer implementation and code using it | Josh Blum | 2011-02-05 | 3 | -26/+23 | |
| | | | | | | | The bounded buffer now uses the detail idiom to hide implementation to inline better. The whole sptr/make idiom was removed from bounded buffer to just construct directly. The code using bounded buffer was changed for the new api: replaces access operators and calls to the factory function. | |||||
* | uhd: work to remove dynamic allocations of std::vector in vrt_packet_handler ↵ | Josh Blum | 2011-02-05 | 1 | -22/+25 | |
| | | | | calls | |||||
* | udp: simplfy zero copy asio overhead with less shared_from_this, and timed ↵ | Josh Blum | 2011-02-05 | 1 | -19/+25 | |
| | | | | waits when not needed | |||||
* | uhd: various performance tweaks | Josh Blum | 2011-02-04 | 2 | -24/+41 | |
| | ||||||
* | Merge branch 'udp_xport_work' | Josh Blum | 2011-01-26 | 1 | -45/+9 | |
|\ | ||||||
| * | udp_zero_copy_asio: removed the #ifdefed num frames and min buff size stuff | Josh Blum | 2011-01-26 | 1 | -45/+9 | |
| | | | | | | | | | | | | basically, everything can share a common number of frames, and there is no min buff size, only resize if a size was specified, this simplifies the warnings and resize code | |||||
* | | uhd: if_addrs check that iter->ifa_addr is NULL before continuing | Josh Blum | 2011-01-26 | 1 | -0/+1 | |
|/ | | | | | Happened when building UHD within the OpenWRT environment where ifa_addr can be null it seems! | |||||
* | uhd: fixed maxosx bug, was resizing the transport buffer | Josh Blum | 2011-01-26 | 1 | -4/+0 | |
| | | | | | | but it cant be resized on macos and I messed up the code that blocked that behavior reimplemented block in usrp2_impl, seems more correct here | |||||
* | udp: revert depadding hack for recv_frame_size | Josh Blum | 2011-01-21 | 1 | -1/+1 | |
| | ||||||
* | usb: restored disable_interruption on get_lut_with_wait (gets called externally) | Josh Blum | 2011-01-21 | 1 | -0/+1 | |
| | ||||||
* | udp: shrink default recv_frame_size by 4 bytes until FPGA handles 2 byte pad ↵ | Josh Blum | 2011-01-21 | 1 | -1/+1 | |
| | | | | correctly (1516) | |||||
* | usb: use thread interruption in usb zero copy on deconstruction | Josh Blum | 2011-01-17 | 1 | -8/+9 | |
| |