RISE RP005 QEMU weekly report 2024-11-27

Milestone 2

Make requested revisions to Paolo's upstream patches:

REOPENED.
this work is now being handled by Craig Blackmore
the first patch had been awaiting merge. We are investigating whether an exception during the load/store could lead to an invalid value in the vstart CSR (see this mailing list post).
for the second patch, we are exploring a possible solution to ensure that interrupts are not taken in the middle of an elements during memcpy. (see this mailing list post). We anticipate this will be a new patch, since the current patch is incompatible with upstream.

SiFive benchmarks:

COMPLETE

Milestone 3

Generate TCG Ops for vector load/store

IN PROGRESS.
We have a preliminary solution, but it is not yet working as intended. We have asked Daniel Barboza to review the code for suggestions.

Improve first-fault handling for vector load/store helper functions.

IN PROGRESS.

Improve strided load/store helper functions.

IN PROGRESS.

Statistics

SiFive benchmarks

There are no new statistics to report since our last report.

Actions

2024-11-13:

Craig Blackmore (transferred from Paolo Savini) to investigate whether patch #1 may cause exceptions to be taken at the wrong byte address, thereby causing a failure on resumption.
- ONGOING
- under investigation as described above.
Craig Blackmore (transferred from Paolo Savini) to investigate whether patch #2 can use larger memcpy than Max and therefore may offer further improvements.
- ONGOING
- potential solution identified - see above.
Jeremy Bennett to redo the SiFive benchmark scripts without any plugins enabled and to report just total execution time when comparing.
- COMPLETE

Other

Next meeting 4 December 2024.