From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stas Sergeev Subject: Re: Can I use DOSEMU for testing device driver code? Date: Sun, 15 Feb 2004 14:15:42 +0300 Sender: linux-msdos-owner@vger.kernel.org Message-ID: <402F54DE.9080600@aknet.ru> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: List-Id: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: linux-msdos@vger.kernel.org Hello. Ryan Underwood wrote: >> The problem in (B) seems to be that all of my transmitted characters >> disappear into a sort of 'virtual' UART to be transmitted later, whilst >> the RTS signal seems to come out of the port in 'real' time. > I don't think that is happening. i.e. in,out instructions are directly > executed, there is no emulation involved. No, the "problem B" was about an emulated access, which is indeed going via an emulated UART. Its latency was considerably reduced in dosemu-1.2.0 (not in any 1.1), but still of course can't match the direct access. The fact that RTS gets out of sync with the TX looks like a bug, most likely the emulated FIFO is being switched on when it shouldn't. But sigh, looks like that bug will have to wait for the next time:) Btw, from the docs I have, it is unclear whether there are some ways to disable the FIFO, other than by writing to FCR directly. But I strongly suspect that at least changing the DLAB state also should disables it, which may explain this problem. If someone have an info on that - would be nice to have a look.