Gilad Ben-Yossef wrote:
No. What makes a block device a block device is that the actions done on it (rad/write etc) go through the buffer cache.
I'm willing to be that the memory area in question is used to communicate with some card/DSP/chip and the last thing Ilan want is to cache access to those pages in the buffer cache.
Ok. How does the kernel handle firewire and shared SCSI devices then?
The same way.
Data on those can change without the current machine doing anything about it.
Which is only a problem if you try to open the *same* file from two different systems and expect to find the same content without doing something to assure it is so.
Shared media != cache coherent access.
There are some filesystems (CXFS, for example), where this is actually done. I guess that the caches are managed through the network (CXFS stores the actual data on shared storage devices, and coordinates the access to them through the network).
At the file system level there is no problem to cordinate a cache sync before every read.
Gilad
-- Gilad Ben-Yossef <[EMAIL PROTECTED]> Codefidence. A name you can trust(tm) Web: http://codefidence.com | SIP: [EMAIL PROTECTED] Tel: +972.9.8650475 ext. 201 | Fax: +972.9.8850643
"I am Jack's Overwritten Stack Pointer"
-- Hackers Club, the movie================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
