Roman Zeyde
Roman Zeyde
Many thanks!
@dpc would it please be possible to share the commits being used in https://twitter.com/dpc_pw/status/1857090601042198921?
When running with `jemalloc` (#1115), there is a ~10% memory usage improvement (right-side plot vs. left-side):  Tested with Electrum client loading the `1BitcoinEaterAddressDontSendf59kuE` address (with >3.8k txs).
`LOG_STACK_USAGE=1` shows that the `SdProtect` command takes half of the available stack - so probably something else causes the stack overflow. ``` [15:36:14.183] 56331000 trezor.wire.message_handler DEBUG estimated stack usage=6280 (unused...
Another reproduction - RSOD happened after clicking `Retry` ``` ============================================================================ test session starts ============================================================================= platform linux -- Python 3.10.10, pytest-8.3.2, pluggy-1.5.0 -- /home/rzeyde/src/trezor-firmware/.venv/bin/python cachedir: .pytest_cache hypothesis profile 'default' -> database=DirectoryBasedExampleDatabase('/home/rzeyde/src/trezor-firmware/.hypothesis/examples')...
It would be interesting also to test the latest upstream FatFs 0.16 release (https://elm-chan.org/fsw/ff/). Currently we use FatFs 0.15 (8f7cbf7b9d7971ab17ab5b9f0f709ce04805c527).
Indeed, Electrum currently is not using this feature: https://github.com/spesmilo/electrum/issues/4900
 https://github.com/trezor/trezor-firmware/actions/runs/15564481949/job/43825018802?pr=5177
IIUC `PYOPT=1` emulators should be built and tested during the nightly CI run.