X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-workers-bounces using -f Message-ID: <3CE37BCA.F5DE2858@yahoo.com> Date: Thu, 16 May 2002 05:28:42 -0400 From: CBFalconer Organization: Ched Research X-Mailer: Mozilla 4.75 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 To: djgpp-workers AT delorie DOT com Subject: Re: Malloc/free DJGPP code References: <10204222312 DOT AA16592 AT clio DOT rice DOT edu> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Reply-To: djgpp-workers AT delorie DOT com Charles Sandmann wrote: > > > I considered it, but did not test for it exclusively. > > That's good then. I know how to test for it in a real application. > > > I have never seen unexpected values under W98 AFTER initialization. > > Typically only seen when multitasking. DOS apps share the same > address space, so if DPMI #1 allocates some blocks, DPMI #2 starts > running, DPMI #1 exits, then DPMI #2 can get blocks below it's > initial address space. If the address space is fragmented you can > also see the same issues. > > > If anything were actually happening it would be trivial to add > > just such a test to the fakesbrk generator in tnmalloc. As it is, > > I shall avoid disturbing anything unless I decide a real bug > > exists. It is hard to test/evaluate a moving target. > > Not so easy - need to show some negative addresses and some "fill" > blocks which are > 0x80000000 in size. In this case it's easier > just to test with the real thing. > > > Notice that there is a serious realloc problem with the existing > > code, as shown by the evilalgo program. > > In V2.03 yes. Does this problem still exist in the CVS/2.04 current > release? It had changes made to it's realloc() algorithm, so it > might not have this problem. I don't know if anyone has tested either > way. What's in 2.03 isn't usually the most recent version. ... from another thread .... > > Once more, available at: > > > > > > I haven't forgotten, but I got kicked really hard this weekend when > the SCSI drive that had all the refresh++ stuff (in addition to > the malloc tests and 8Gb of stuff that's not replaceable) died. > Two months out of warranty, of course. Why does evaluation depend solely on you? Although both free and realloc have at least one serious problem each in the current distribution, nobody else seems to give the proverbial rat's a ss. The other thread about emacs seems to indicate that problems arise from releasing sbrk segments back to the system. As a slight inducement the organization in nmalloc can tell when such a segment has been completely freed. One of the possible problems inherent in actually doing so is also obvious, at least to me at the moment. I am not about to do anything about it. (Resent - would you believe:) Your message was rejected by delorie.com for the following reason: Hey, offensive language ("a ss") is not allowed. -- Chuck F (cbfalconer AT yahoo DOT com) (cbfalconer AT worldnet DOT att DOT net) Available for consulting/temporary embedded and systems. USE worldnet address!