[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Data corruption.
- To: Mailing List Recipients <pci-sig-request@znyx.com>
- Subject: Data corruption.
- From: Rafi Boneh <rafib@gilat.com>
- Date: Tue, 24 Jun 1997 16:17:05 -0500
- Organization: Gilat Satellite Networks Ltd.
- Resent-Date: Tue, 24 Jun 1997 16:17:05 -0500
- Resent-From: pci-sig-request@znyx.com
- Resent-Message-Id: <"GFZON.0.2I1.NA9ip"@dart>
- Resent-Sender: pci-sig-request@znyx.com
- Sender: Rafi Boneh <rafib@gilat.com>
Hi.
I am facing a very strange phenomena regarding PCI transfers.
I am doing memory burst writes from a master PCI agent, directly to a
PC’s
memory. It appears that once in a while (I can’t point out the
regularity,
but it looks like it depends on the data itself) bit 26 in only one
transfer
of the burst, passes to the PC’s memory as "0", although it was written
"1".
A PCI bus analyzer shows that the bit is written correctly, means "1".
Even
a digital scope agrees with the PCI bus analyzer, that the bit is written
as
"1" on the PCI bus.
Moreover, the PAR line on the PCI is generated correct for this transfer,
and there is no PERR back from the PC, which means that the PC’s PCI
bridge
dose not detect any error. But still this bit is corrupted when viewed in
the PC’s memory.
Could it be a problem between the PC’s PCI bridge and the main memory?
Did someone faced a similar problem?
Thanks for every information.
Rafi Boneh.
Ë L 9