Avoid READ command in stream mode #5408
Conversation
Some mouse modules (in my case, X19 trackball) seems not responding to PS2_MOUSE_READ command when operating in stream mode. Instead, new data are uploaded by push from the device to the host. Because of that, when mouse cursor is not moving, read commands succeed but no data arrive, so the following routine must wait for 4 x 25 = 100 ms. This slows down all other proocedures in the chip. This change is to use pushed data, and stop pulling when no data are available.
drashna
requested review from
ezuk,
fredizzimo,
jackhumbert,
mechmerlin,
skullydazed and
yanfali
|
My only concern with this change is testing it with hardware which does not exhibit this behavior. Can we find another user of ps2 mouse? |
|
Better yet, making this configurable, so you can enable if if needed, without breaking the implementation for others. |
|
@tomykaira This change looks good at its core but I agree with drashna that we should make this configurable with a #define. Is that something you're working on? |
|
There are a number of merge conflicts here. Could you update the PR to fix these? |
|
Sorry for necrobumping, but has there been any follow-up on this? I'm experiencing the same issue as this PR attempts to solve. Matrix scans become ~10 times slower with both interrupt & USART mode, which renders my rotary encoder pretty useless. I'm willing to do the testing if necessary. |
Description
Some mouse modules (in my case, X19 trackball) seems not responding to
PS2_MOUSE_READ command when operating in stream mode. Instead, new data
are uploaded by push from the device to the host. Because of that, when
mouse cursor is not moving, read commands succeed but no data arrive, so
the following routine must wait for 4 x 25 = 100 ms. This slows down
all other proocedures in the chip.
This change is to use pushed data, and stop pulling when no data are
available.
Types of Changes
Issues Fixed or Closed by This PR
Checklist