From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1Ny66Y-0004lb-KR for qemu-devel@nongnu.org; Sat, 03 Apr 2010 12:21:50 -0400 Received: from [140.186.70.92] (port=36144 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1Ny66X-0004ke-1y for qemu-devel@nongnu.org; Sat, 03 Apr 2010 12:21:49 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.69) (envelope-from ) id 1Ny66V-0005js-VS for qemu-devel@nongnu.org; Sat, 03 Apr 2010 12:21:48 -0400 Received: from chello212186124096.11.vie.surfer.at ([212.186.124.96]:51947 helo=wiesinger.com) by eggs.gnu.org with esmtp (Exim 4.69) (envelope-from ) id 1Ny66V-0005di-KQ for qemu-devel@nongnu.org; Sat, 03 Apr 2010 12:21:47 -0400 Date: Sat, 3 Apr 2010 18:20:03 +0200 (CEST) From: Gerhard Wiesinger Subject: Re: [Qemu-devel] Absolute Mouse in VNC and debug ports patch In-Reply-To: <201004031626.52399.paul@codesourcery.com> Message-ID: References: <201004031626.52399.paul@codesourcery.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paul Brook Cc: qemu-devel@nongnu.org On Sat, 3 Apr 2010, Paul Brook wrote: >> This is my first contribution to QEMU: I've written a virtual piece of >> hardware (e.g. ports) that the mouse pointer can be used as an absolute >> pointing device even with non USB devices, e.g. with DOS and doesn't need >> any high memory usage USB driver. This is necessary for e.g. VNC remote >> access, this was my primary motivation. > > Why not just use the existing vmmouse? > First I didn't know that QEMU has already implemented vmmouse. I'm very new to QEMU. After looking at the code vmmouse requires 386 instructions in the DOS driver to access the 32 bit port access (eax, ebx, etc.) which might be problematic in real mode. and even on 286 (don't know whether this is relevant or not). Further protocol is a little bit more complex to implement in Assembler in the DOS TSR and therefore performance would be lower (DOS polls the mouse, no standard event or interrupt based system specified). >> To use it you need a mouse driver (e.g. cutemouse, >> http://cutemouse.sourceforge.net/) and an QEMU mouse TSR. >> Further the mouse TSR emulates also a HP-HIL 45911 tablet digitizer. This >> device has been chosen because an Interrupt interface exists at INT 6Fh and >> therefore it is relativly easy to emulate (all other devices I've evaluated >> use the serial port and are therefore much more complex to emulate) > > However emulating an actual tablet connected to the serial port means you > don't need any custom guest software. But I thing it is only relevant for very old software like DOS. Every modern OS has USB on board and therefore the USB tablet emulation is IHMO sufficient and works well. The guest software could also be implemented as option rom (~2k), then no TSR would be necessary. As the typical use case is for DOS I don't see any relevant problems here because resident TSR size is very small (~2k) and common practice in DOS. BTW: is it possible to load more than one option ROM at a time, e.g. SCSI and another? > >> Further I've written a simple logging debug ports functionality to debug the >> driver > > This looks a lot like the existing isa-debugcon. > Is this in the development tree only? I had only a look at the stable release and didn't find anything. Ciao, Gerhard -- http://www.wiesinger.com/