[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: supporting large memory on alphas today



>>> Greg Lindahl said:
> 
> > Or for example ones which push lumps of
> > machine code to the SCSI controller for them to execute?
> 
> I *think* these are OK since they will never see a buffer to use with
> an address bigger than 4 GB.

IMHO, no, as it's not typically a "buffer" setup by sd.c that they are told
about, it's usually some memory kmalloc'd during driver setup/probe time,
and as we all know, memory is normally allocated from the top (ie highest
addresses) down.

This will break the NCR driver, for example; I know, because when I tried
an 875-based card in a DP264 with 2Gb of physical *before* I changed the
windows to support more than 1Gb, the card couldn't fetch the scripts...

And this will prolly break most of the Ethernet drivers, de4x5.c and tulip.c
included, as that's how they set up their ring buffers and such.

--Jay++

-------------------------------------------------------------------------------
  American Non Sequitur Society: we don't make sense, but we do like pizza...

Jay A Estabrook                            Alpha Motherboards - LINUX Project
Compaq Computer Corporation                (508) 841-3241 or (DTN) 237-3241
334 South Street, Shrewsbury, MA 01545     Jay.Estabrook@digital.com
-------------------------------------------------------------------------------



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index] []