Xref: news2.mv.net comp.lang.asm.x86:14407 comp.os.msdos.djgpp:1294
From: ulfg AT groundhog DOT phy DOT nist DOT gov (Ulf Griesmann)
Newsgroups: comp.os.msdos.djgpp,comp.lang.asm.x86
Subject: reading DMA residue count
Date: 17 Feb 1996 17:01:10 GMT
Organization: NIST
Lines: 25
Message-ID: <4g51km$mg7@dove.nist.gov>
NNTP-Posting-Host: groundhog.phy.nist.gov
Summary: Is it O.K. to read DMA residue count while DMA in progress?
Keywords: DMA
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

Hello All,

I am posting this to both the djgpp and the asm.x86 groups in the hope 
that the message will reach more DMA experts that way.

We have a home made data acquisition board based on a soundboard chip that
we program to transfer data via DMA on channel 5 to a Pentium machine at a
rate of 40kHz. Simultaneously, additional data are being transferred on
DMA channel 6 at a rate of 1kHz. Whenever a transfer on the fast channel
has finished we try to figure out how many data were transferred on the
slow channel in the meantime by reading the DMA residue count. This works
o.k. several times but suddenly veers off and returns nonsense as the
buffer for DMA channel 6 fills up. Has anyone out there any ideas why it
might be a Bad Thing to read the residue count while DMA is in progress?
Could this be the reason why it does not work? 

Thank you,
   Ulf

-- 
  Ulf Griesmann                       \\\   Tel.: (301) 975 3212 (x73220)
  National Institute of                \\\   Fax.: (301) 990 1350
  Standards & Technology                \\\   e-mail:
  Atomic Physics Division, Bld. 221/A167 \\\   ulfg AT groundhog DOT phy DOT nist DOT gov
  Gaithersburg MD 20899, U.S.A.           \\\   ulfg AT aeldata DOT phy DOT nist DOT gov