home *** CD-ROM | disk | FTP | other *** search
- Path: god.bel.alcatel.be!nlev00!barnhoorn
- From: barnhoorn@nlev00 ()
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: Does MUNGWALL have errors ?
- Date: 23 Feb 1996 07:55:03 GMT
- Organization: Alcatel Bell
- Distribution: world
- Message-ID: <4gjrso$i68@btmpjg.god.bel.alcatel.be>
- References: <1996Feb19.135354@mathematik.tu-muenchen.de> <4ga4ff$im7@btmpjg.god.bel.alcatel.be> <1996Feb21.141618@mathematik.tu-muenchen.de>
- Reply-To: barnhoorn@nlev00 ()
- NNTP-Posting-Host: 138.203.178.61
- X-Newsreader: mxrn 6.18-10
-
-
- In article <1996Feb21.141618@mathematik.tu-muenchen.de>, ganterth@mathematik.tu-muenchen.de (Thomas Ganter) writes:
- |> In article <4ga4ff$im7@btmpjg.god.bel.alcatel.be>, barnhoorn@nlev00 () writes:
- |> |> In article <1996Feb19.135354@mathematik.tu-muenchen.de>, ganterth@mathematik.tu-muenchen.de (Thomas Ganter) writes:
- |> |> |>
- |> |>
- |> |> Mungwall is doing the job it should do: initializing unused and used
- |> |> memory to 0xDEADBEEF, 0xC0DEDBAD, etcetera. It is obvious that every
- |> |> program that produces an Enforcer-hit when Mungwall is running,
- |> |> is badly coded,
- |>
- |> So enforcer hits in mungwall segments (says SegTracker) indicate that
- |> there perhaps might be a badly coded routine inside mungwall?
- |>
- This is a whole new point of view to the problem. You did not mention
- that Enforcer found hits in Mungwall!!
-
- |> |> [...obvious explanation of badly coded programmes ommitted...]
- |>
- It wasn't that obvious to me, when I found out, five years ago.
-
- |> |> I think something like this is your problem. The only solution is
- |> |> to stop using Mungwall, or rewriting your software.
- |>
- |> Neither can I rewrite commercial software like the MFC-drivers, nor
- |> can I pinpoint any fault in the NDK's serial device example code.
- |> And still, the errors occur while the PC is inside a MungWall segment.
- |> Perhaps there is a MungWall problem?
- |>
- Could still be a problem of the device-driver, that it is freeing
- memory that does not actually exist. By some lucky chance FreeMem() does
- not produce an enforcer-hit, but Mungwall, who is going to fill the
- memory for you, will cause the enforcer-hit.
-
- Jaco.
-