Yes, the problem still is there, at least in the version dl'ed in March, not sure what version-no.

I recently varied the delay value
delayUs(150);

It worked down to 40us when the same core is used.
I have not tried to go lower to find out the limits.

It would be great so see this finally fixed.

For TCP, there is a non-standard IDF extension which allows you to call call setsockopt(TCP_SNDBUF) with an int-sized parameter which is the send buffer size as a multiple of the TCP_MSS value (rather than in bytes.) ie the send buffer size in bytes is the parameter value multiplied by TCP_MSS ...

Yes, I had also asked here and got a reply
https://github.com/espressif/esp-idf/issues/4581

I just downloaded the current IDF.
Compiling my project now results in errors:
uart_set_line_inverse(n, UART_INVERSE_TXD);

UART_INVERSE_TXD is not defined anymore in the new IDF. The doucmentation is also unclear:
Choose the wires that need to be inverted. Using the ORred mask of uart_signal_inv_t ...

My ESP32 module was connected to a WLAN-repeater (Fritz-Repeater).
Now I moved it back near to the router (Fritzbox) but it still connects to the far away repeater, even that the router signal is much stronger. The result: slow & difficult traffic.
Why doesn't ESP32 select the stronger signal?
An ...

I'm sorry I have to correct myself. Checking the schematics again I see that IO17 is used as clock output, so not IO0. I mixed it up with the ESP32-CAM that generates 20 MHz at IO0 and that is no problem even with the 30cm programming cable.
I guess I had problems with IO0 and therefore used IO17 ...

With ESP32-CAM working as STA I have observed several times, that it and the AP report that connection still exists (no SYSTEM_EVENT_STA_DISCONNECTED occured in ESP32-CAM), but no more IP messages can be exchanged. No ping / UDP / HTTP. The ESP32-CAM sits in the basement so signal strength is weak ...

I have several ESP32-CAM modules. One module doesn't connect to the AP when using correct credentials. Only after executing WPS it connects. That means I have to start WPS after each reset.
The same firmware works well with other modules.
The credentials are stored in the firmware, to be on the ...

GPIO0 works fine here as clock-out, even with a 30cm long programming cable that I initially feared would cause problems.
Have you an extra load connected to it that impairs the waveform?

The problem I finally solved.
I used a second board and that worked fine. It seemed the first board had a LAN8720 solder problem, I could not find out in detail the reason, and gave up as it was a one time problem. So after all no s/w problem at all.