chem194@canterbury.ac.nz (John Davis, chem dept, UoC NZ) (07/20/90)
In an earlier article there was discussion of problems using himem.sys 2.60 with win3. This should rightly be a followup article, but our news system just went brain dead and decided followup were no go .... >I have found three bugs in himem.sys 2.60 that I will outline in this >article. They are (1) himem.sys 2.60 doesn't correctly support the >resize function which will cause it to incorrectly assign handles, (2) >it doesn't correctly preserve the 32-bit AX register which can cause problems >on 386 machines, and (3) the interrupt 15 handler incorrectly assumes >the state of interrupts which can cause conflicts between himem.sys >and the keyboard controller chip. I think that himem.sys does an >iret instead of a far ret. It either incorrectly assumes the >interrupts are always off or on. I think that it assumes that they >are off. We too have been having trouble with himem.sys on one AT here. The machine ( a hyundai AT clone ) has been losing the keyboard interrupt as you stated. We've tried switching BIOSes to see if the problem lies with the BIOS in use (initially an Award bios, we've now also tried the Phoenix BIOS off a machine which did run himem 2.60 ok ), and as the problems persisted it is clear that the problem is entirely the blame of himem ( and not dependent on the specific BIOS ). Interestingly, we set the machine to go into win3 on bootup, largely to see if any other interrupts ( disk, mouse, serial ) were being trashed. Turned out that only the keyboard interrupt was dead, and what's more on exiting Windows ( back to dos ) we found the keyboard was then re-enabled. To say that this behaviour has us mystified is an understatement, while I can understand himem hanging the keyboard int, windows somehow 'fixing' it is strange! Anyway, the net result is that as long as you boot into windows direct you can work around the the problem. I'd be interested to hear what other people have been finding with himem, and even more so what response there's been ( if any ) from microsoft as regards fixing the bug .... ----------------------------------------------------------- | o John Davis - CHEM194@canterbury.ac.nz o | | o (Depart)mental Programmer,Chemistry Department o | | o University of Canterbury, Christchurch, New Zealand o | | o o |