home *** CD-ROM | disk | FTP | other *** search
- From: Kay Roemer <roemer@informatik.uni-frankfurt.de>
- Posted-Date: Fri, 20 May 94 8:37:38 MESZ
- Received-Date: Fri, 20 May 94 08:37:38 +0200
- Message-Id: <9405200637.AA16319@hera.rbi.informatik.uni-frankfurt.de>
- Subject: MintNet 0.41
- To: mint@atari.archive.umich.edu
- Date: Fri, 20 May 94 8:37:38 MESZ
- Mailer: Elm [revision: 70.85]
-
- Probably to your surprise there is again a new version of MintNet - 0.41
-
- This is because there was a very nasty bug in the timeout handling in
- Mint, caused by the addroottimeout() patches.
-
- In the original patches I modified canceltimeout() to be able to remove
- root timeouts too.
-
- But since Mint does eventually cancel timeouts in Fselect() and nap() and
- probably elsewhere using canceltimeout() even if they have already expired
- (ie the timeout occured), this could result in freeing another timeout which
- has by conincidence the same address in memory.
-
- This is especially annoing with the new Minixfs PL 10, which uses
- addroottimeout() to write out buffered data, ie when the above bug occures
- no updating is done any longer....
-
- Therefore there is a new call cancelroottimeout() which does only cancel
- root timeouts and the old canceltimeout() works again. The new MintNet
- uses this new call and contains the new patches for Mint.
-
- YOU SHOULD REALLY GET 0.41 AND APPLY THE NEW PACTHES!
-
- My next mail contains the new patches, they are relative to clean Mint 1.10,
- sorry...
-
- Cheers,
- Kay.
-