Merge pull request #3707 from TomHarkness/doc/pm5-flipper-usage-update-2

doc: update PM5-FLIPPER-USAGE.md with recent CEP/FAP fixes
This commit is contained in:
Iceman authored and GitHub committed 2026-10-16 11:31:00 +07:00
commit e519c1af94
1 file changed
+10
@@ -42,6 +42,9 @@ Build as usual, CEP is available by default. (`SKIP_CEP=1` to remove CEP support
- `hf`/`lf` reads in general, same as over USB, once the device's one-time
FPGA bitstream step is done (see `Getting_Started_With_PM5.md`) — CEP
itself doesn't block RF.
- The link survives idle periods cleanly — the FAP's idle dummy-poll bytes
(`0xFFFF` length header, no data pending) no longer trigger a false SPI
resync.
## What doesn't (yet)
@@ -49,6 +52,10 @@ Build as usual, CEP is available by default. (`SKIP_CEP=1` to remove CEP support
replies `PM3_EFAILED` with stale cached UID data, even though the same
reader succeeds moments later over plain USB against the same tag.
Firmware-side, root cause open, out of scope for the CEP transport itself.
- `CMD_MAIN_CHIP_UNIQUEID` intermittently times out on its first attempt
(Hardware Status page, occasionally slow/incomplete on first load),
always recovers on the FAP's retry. PM5 handler and FAP cache both
confirmed clean — likely a transport/timing-layer race, root cause open.
---
@@ -70,3 +77,6 @@ Build as usual, CEP is available by default. (`SKIP_CEP=1` to remove CEP support
- CEP uses its own reply-routing flag (`g_reply_via_cep`).
- SPI bit-order/clock-phase settings are confirmed correct against real
Flipper traffic (logic analyzer): MSB-first, mode 0, CS active-low.
- FAP-side: a short fast-poll window after sending a command, and blocking
deep-sleep for the life of the RX thread (fixes unreliable reads on
battery), both merged in the FAP's own repo.