Problem with sharing serial port after upgrade from 5.1.26 to 6.0.X

[t.ruffier@…]

Bonjour,

Suite à la montée de version 5.1.26 vers la 6.x la communication via les ports séries ne fonctionnent pas sur les VM montées en OS2.

Suite au ticket 18668 nous rencontrons toujours des difficultés. La personne ayant crée le ticket a résolu sa problématique.

Pouvons nous avoir la correction de ce souci svp ?

[aeichner]

Please describe your problem in English and attacha VBox.log of the affected VM.

[t.ruffier@…]

Good morning,

Following the upgrade from version 5.1.26 to 6.x, communication via serial ports does not work on VMs mounted in OS2.

Following ticket 18668 we are still encountering difficulties. The person who created the ticket at the time had resolved their problem.

Can we have the description of the correction relating to ticket 18668 please?

[t.ruffier@…]

Hello we ugrapde Virtualbox to latest realease (V7) The problem is the same. The translation of physical port RS232 (pc GETAC S410) to virtualbox dont work already.

Is this version supported with version 7 ?

[sylcad]

Are you still encountering this issue?

We are currently using and testing Virtual Box 5, 6 and 7 with GETAC S410 too and Dell Latitude 5430 Rugged. They both have a native serial port. Guest type : Windows (XP, 7, 10) Host type : Windows 10

For now, I arrive at the same conclusion as you. VirtualBox 5 works completely with the serial communication, but bring some security problems as it's not supported anymore.

In my research, I also rebuild completely using the #18668 tickets and as you said, it would be great to know what was the solution. Two solutions where provided, but only one is still present in the source Uartcore.cpp:335.

    /*
     * Set interrupt value accordingly. As this is an ISA device most guests
     * configure the IRQ as edge triggered instead of level triggered.
     * So this needs to be done everytime, even if the internal interrupt state
     * doesn't change in order to avoid the guest losing interrupts (reading one byte at
     * a time from the FIFO for instance which doesn't change the interrupt source).
     */
    if (uRegIirNew == UART_REG_IIR_IP_NO_INT)
        pThisCC->pfnUartIrqReq(pDevIns, pThis, pThis->iLUN, 0);
    else
        pThisCC->pfnUartIrqReq(pDevIns, pThis, pThis->iLUN, 1);

What was wrong with the second proposition? And if no one knows, could it be possible to get more informations about the change done between VB5 and 6? (register set calculation, architecture...)

Thanks in advance