ESP32 boards hanging with 8.0.0-beta.2

[dhalbert]

(Split off from #7060)

Some but not all ESP32 boards are hanging when loaded with 8.0.0-beta.2.

Feather HUZZAH32 (WROOM) hangs. Feather ESP32 V2 and QT Py ESP32 (Pico module) do not. @microdev1 notes the former is 4MB, the latter two are 8MB (also latter two have PSRAM).

ets Jun  8 2016 00:22:57

rst:0x1 (POWERON_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
flash read err, 1000
ets_main.c 371 
ets Jun  8 2016 00:22:57

rst:0x10 (RTCWDT_RTC_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:2
load:0x3fff0030,len:1184
load:0x40078000,len:12756
ho 0 tail 12 room 4
load:0x40080400,len:3024
entry 0x400805dc
Serial console setup
[hangs here]

I believe the next step to do a bisect in CircuitPython, and then possibly in ESP-IDF to track down where the problem started for these ESP32 boards. I tried rolling back the toolchain to patch3 from patch5, and doing nothing else, and did not fix the hanging, whereas it did for C3. So I think this is a different problem. Also we could compare the generated build-*/esp-idf/sdkconfig files for working and non-working versions and see if there are interesting differences.

[dhalbert]

@microdev1 I did a bisect on this, testing on a Feather HUZZAH32 (ESP32 WROOM module) and the critical commit is different than #7060.

The first commit that causes the problem is 7d8ff20, the first commit in #7000, in your storage extension PR. I don't know what the issue is, but I also tested that same commit on Feather ESP32 V2, and it works on that board. So there's something different about the build on those two boards that causes one to fail and one not.

I double-checked by recompiling the commit just before 7d8ff20, and that works.