Re: "movb" for spin-unlock

From: Gérard Roudier (
Date: Fri Apr 28 2000 - 13:14:05 EST

On Fri, 28 Apr 2000, Jamie Lokier wrote:

> > Indeed this is complex. This happens to be at least as complex and
> > sometime more when having from a driver to deal with a PCI device through
> > all the variations of bridges that have been invented, and for such
> > exercise we donnot have a usable LOCK protocol.
> As Alan pointed out, to flush your PCI writes, you have to do a PCI
> read.

Hmmm... Is `Alan' some secret name for the PCI specifications. ;-)

This is true on paper, unfortunately this is not so simple, given all
existing bridge implementations and CPU/cache architecture optimizations
to make a PCI device driver work reliably using memory mapped IOs.

PCI specs 2.1 have detailed PCI ordering rules and only guys that were
actually aware of PCI ordering rules have been able to provide and
maintain reliable drivers that use memory mapped IOs.

AFAICT, what I know about PCI device drivers under Linux and FreeBSD let
me think that the list of such PCI devices drivers (using reliably MMIO)
is very short. Obviously, le drivers I maintain use memory mapped IOs.


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

This archive was generated by hypermail 2b29 : Sun Apr 30 2000 - 21:00:15 EST