Attention is currently required from: Felix Singer, Nico Huber, Thomas Heijligen, Alexander Goncharov.
4 comments:
Patchset:
Hi Nico, thanks for your feedback. I'm still hesitating about limiting the amount of memcpy in dirtyjtag_djtag1_spi_send_command. It would make the code run faster for sure, but it would also make it harder to read.
File dirtyjtag_spi.c:
Patch Set #5, Line 18: * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
Please remove this paragraph. The address became outdated in the past so […]
Done
Patch Set #5, Line 121: size_t num_xfer = (len + max_xfer_size - 1 ) / max_xfer_size; // ceil(len/max_xfer_size)
1. Lift the `.max_data_*` limits so this function gets called for bigger
chunks. There should be less overhead as we wouldn't have to send the
TMS/STOP sequence as often. I assume this could speed up long reads.
Some of the memcpy() also seems unnecessary, but the USB transfers
probably limit speed so much that it doesn't matter.
I went with that option.
Btw. is it necessary to drain input buffers with the receive
call even if we don't need the data?
Unfortunately, yes. We'll probably fix this in DJTAG2 (see https://github.com/jeanthom/DirtyJTAG/issues/77#issuecomment-758623004 for more info), but I still want to keep backwards compatibility with DJTAG1.
Patch Set #5, Line 275: except TRST
What about SRST and TMS? I might read the code wrong, but it looks like […]
Yes SRST and TMS are also high during init. I added a little table right after the copyright to better explain how the JTAG interface should be connected to an SPI flash.
To view, visit change 67878. To unsubscribe, or for help writing mail filters, visit settings.