pset_read'ing a slot which has uninitalized data crashes grid

[dndrks]

steps to repro

• execute pset_read(n) where n is a slot with no data yet saved to it

expected
no data is returned

experienced
grid freezes and is unresponsive to press + diii commands until hot-plugged

[tehn]

fixed in 250129

[dndrks]

hihi! still experiencing pset_read crashes with 250201.

test script:

print("hi!")
metro.new(function(x)
  if x == 1 then
    print("going to read a blank pset!")
  elseif x == 2 then
    print('reading PSET "10"')
    pset_read(10)
  elseif x == 3 then
    print("did we get here?")
  end
end, 600, 3)

of note, the 2nd pass's print command doesn't execute ahead of trying the pset_read(10)...
lmk if there's anything else i can provide!

[tehn]

confirmed.

if you try to flash_read() on an uninitialized flash slot it will potentially try to return a huge long string with unpredictable results. if run from pset_read it will attempt to execute this and break.

i added a firmware fix to clear the flash area when uploading the firmware for the first time, which will be in the next release. however this won't fix devices that have an earlier beta installed.

to fix this in beta, run:

for i=1,256 do flash_clear(i) end

or better yet, i'm considering adding some metadata to the pset system so it is aware of which flash slots have been written, hence which are safe to load.

before that happens, you can roll your own. (i did this in another script, just a table that gets stored to slot 1 with metadata of which psets are legit)

i'm going to leave this open until it's truly fixed, not just partial