Philipp Wagner

Better check the permissions *after* the opening failed to give a more helpful error message, but avoiding corner cases where our checking fails but the device can still be opened.

small correction here (agree otherwise): we use a 16 bit firmware for 16 bit FIFO width, so no bandwidth is wasted (e.g. on the USB bus). We're just limited by...

cbuf already has single reader/writer semantics, as expected from a FIFO (see http://www.glip.io/group__development-cbuf.html). What kind of thread-safety are you looking for?

What's the specific use case for that? And how common is it to justify putting the locking overhead into the library itself, as opposed to letting the user of the...

From all cases I can think of and have implemented, the application that uses GLIP is implementing a higher-level protocol on top of the FIFO semantics provided by GLIP. Almost...

approach looks good to me. Let's have a detailled review once the ZTEX firmware is ready and we know that the abstraction actually covers all cases we want to have...

@wallento that's done, isn't it?

currently auto-detect seems to be broken, connecting with speed=12000000 works for our standard OpTiMSoC design, without the speed option I get this: ``` $ opensocdebugd uart device=/dev/ttyUSB1 Open SoC Debug...

@koenenwmn is this something you could have a look at?

Thanks for your contribution! Can you add a test to show the problem you're fixing, and to ensure that it actually stays fixed? And please add a version constraint on...