search

stakwork / sphinx-key

Lightning hardware signer on ESP32
28 stars 1 forks source link

ESP32 crash #108

Closed Evanfeenstra closed 9 months ago

Evanfeenstra commented 10 months ago 
I (14250) sphinx_signer::persist: NODE LIST LEN 1
I (14270) vls_protocol_signer::handler: Restore node 02702120bb0b21446b0cae862322657890435641dc32b6a411f3d73633644db5f7
I (15340) lightning_signer::node:   Restore channel 0379bb7397a354dc8bfb07d3d8a0d24d88d84c0eae2c6a68fd15149aaafbbaca950100000000000000 outpoint None
I (15620) lightning_signer::node:   Restore channel 0379bb7397a354dc8bfb07d3d8a0d24d88d84c0eae2c6a68fd15149aaafbbaca950200000000000000 outpoint None
I (15920) lightning_signer::node:   Restore channel 0379bb7397a354dc8bfb07d3d8a0d24d88d84c0eae2c6a68fd15149aaafbbaca950300000000000000 outpoint None
I (16200) lightning_signer::node:   Restore channel 0379bb7397a354dc8bfb07d3d8a0d24d88d84c0eae2c6a68fd15149aaafbbaca950400000000000000 outpoint None
I (16480) lightning_signer::node:   Restore channel 0379bb7397a354dc8bfb07d3d8a0d24d88d84c0eae2c6a68fd15149aaafbbaca950500000000000000 outpoint Some(OutPoint { txid: 71ddfbbb86d152c6677020806148236213ef1677377bd91ff50f2abf75e6719f, vout: 0 })
I (16770) lightning_signer::node: Restore node 02702120bb0b21446b0cae862322657890435641dc32b6a411f3d73633644db5f7 on bitcoin
I (16950) sphinx_signer::persist: => update_tracker
I (16990) sphinx_signer::persist: => update_tracker complete
I (17060) vls_protocol_signer::handler: vls_protocol_signer::handler::RootHandlerBuilder::do_build:: NodeState::summary 0270: 0 invoices, 0 issued_invoices, 0 payments, excess_amount 0
memory allocation of 11508 bytes failed

abort() was called at PC 0x420a7ac7 on core 0
irriden commented 9 months ago

This was after we initiated many fundchannel_start on CLN, but never completed the opening of the channel via fundchannel_complete.

This creates a bunch of superfluous channels on the ESP32, and we don't have space for all of them.

Should be alleviated by recent memory work with VLS team - but for now is not a priority, especially with splicing coming online that should allow us to use a single channel for the vast majority of cases.

irriden commented 9 months ago

Definitely should ask VLS team what happens if we do fundchannel_start, but then do fundchannel_cancel instead of fundchannel_complete.