fpga-open-speech-tools / component_library Goto Github PK

When writing to the volume register, values without a decimal point do not get parsed correctly. For example, writing -20 to the volume register results in

[  914.313055] Maximum attenuation exceeded.
[  914.313062] Setting attenuation to -100 dB.

being printed to the kernel log.

In light of this, perhaps all of our drivers should use shared (and tested) string parsing/conversion functions.

In newer kernel versions, i2c_new_device() needs to be replaced with i2c_new_client_device().

I tested this on 5.10, though it appears to affect at least 5.8+.

During testing of the programmable FIR filter, we've found a couple bugs in the AD1939 component

• maxChannel was set to 0 in the tcl script, which caused any incoming channel greater than 0 to not be valid; this was causing the valid signal to never go high for the right channel.
• left and right DAC channels get swapped in the VHDL code

We can group our Platform Designer components into a group in the IP Catalog by putting something like set_module_property GROUP "FPGA Open Speech Tools" in the *_hw.tcl files.

The file is really just fixed point to string and vice versa conversion funcitons. There is probably a name that better describes what the file is actually doing.

Changing the file name will necessitate changing the device driver autogen to reflect the new file name.

Linking to either a product page that contains links to documentation or the documentation itself will make it easier to find the documentation on parts referenced in the repo.