riscv-non-isa / tg-nexus-trace

RISC-V Nexus Trace TG documentation and reference code
https://jira.riscv.org/browse/RVG-96
Creative Commons Attribution 4.0 International
44 stars 32 forks source link

Section 3.4 - PIB Idle cycles #40

Closed ved-rivos closed 8 months ago

ved-rivos commented 12 months ago

The section 3.4 indicates that idle sequences may be transmitted between packets and the idle sequences may be a sequence of n where n may be 0 or any integer. One key aspect that is not specified is that for the probe to be able to synchronize with the bit stream, the encoder must provide a sequence of 8 idle bits to be able to disambiguate a packet boundary. These 8 idle bits may not occur naturally in the trace stream and may need to be explicitly inserted periodically. Guidance is needed on the need for these synchronizing sequences and the periodicity at which they should be generated if they dont occur naturally.

RobertChyla commented 11 months ago

E-Trace just added 'Encapsulation Protocol'. For N-Trace idle is defined as 0xFF, which CANNOT appear inside of normal N-Trace message (0xFF may appear as last byte of message, but it is OK). I will look at this topic again.

mipsrobert commented 8 months ago

This was addresses in related chapter in N-Trace PDF (as this is specific to different protocols).

mipsrobert commented 8 months ago

I am closing all Connectors PDF related issues with same comments as all issues were handled via comments/discussions in SINGLE Google Docs. Relevant links are as follows:

Notes to Connectors PDF: https://docs.google.com/document/d/1iNbB7-nTiQQ4vQBzLqxtBDQxT51QeUQMd9MFOe0jsBs

Connectors PDF rc20: https://github.com/riscv-non-isa/tg-nexus-trace/commit/221f6b1cef94c5f503554e0cb0deb4046dd49686 Connectors for ARC review: https://github.com/riscv-non-isa/tg-nexus-trace/commit/1de77dcff1a75197232f72b8fe6863892da0cc74