|Re: [hatari-devel] (Missing) TT SCSI emulation problems|
[ Thread Index |
| More lists.tuxfamily.org/hatari-devel Archives
- To: hatari-devel@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [hatari-devel] (Missing) TT SCSI emulation problems
- From: Uwe Seimet <Uwe.Seimet@xxxxxxxxx>
- Date: Sun, 6 Sep 2015 12:23:00 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; t=1441534981; l=1578; s=domk; d=seimet.de; h=In-Reply-To:Content-Disposition:Content-Type:MIME-Version: References:Subject:To:From:Date; bh=itE5DsvROf9Y78tvySQm/KrdW2JSm7Q+NTuWfQO00CU=; b=W6e4qoQwcPdmqWT5jMRrHN2302RYnvEtswG3dO79DfcGa7KYiUw8hL9mEWcAPffIop5 24Eirenhhe0vfIJpfPPa/CiCJEhntvI1r405h8MM5MUi9ZaItL5jAnb1ll73TSd8+EgIh 4BJm7Y9wL+xipxn0l3IfHkK/APJuCCzbtek=
Just tested with ioMemTabTT: When I disable the bus error for only 14
instead of 16 bytes and at the same time modify HDDRIVER to access $ff878e
instead of $ff8780 in order to test for the presence of the SCSI chip
it appears to work, i.e. HDDRIVER assumes that there is no TT-compatible
SCSI and does not hang anymore.
It does not help with CBHD, though, but there will most likely not be any
fix for CBHD anyway, given that it is not maintained anymore.
Would you agree that we should test this change with the next development
version after the release of Hatari 1.9.0?
For TOS 3.06 not to crash when booting at least for 10 bytes starting with
$ff8780 there must not be a bus error, by the way.
> Am Sun, 6 Sep 2015 11:42:22 +0200
> schrieb Uwe Seimet <Uwe.Seimet@xxxxxxxxx>:
> > Hi Thomas,
> > Indeed, TOS 3.0x will probably fail because there is no reason for TOS
> > to expect that the SCSI hardware is not available.
> > It may be possible to resolve this by only raising a bus error for one
> > of the SCSI chip registers, so that TOS does not crash but at least
> > HDDRIVER (any maybe CBHD) is able to detect that there is no SCSI
> > chip. How should I modify the Hatari sources to test this, i.e. how
> > can I raise a bus error when reading from a particular memory address?
> Search for "SCSI" in ioMemTabFalcon. There is an entry for 16 bytes at
> address 0xff8780. If you comment that out, the default behaviour - bus
> errors - will be used for that memory region instead.