Thanks for the answer. 

No for a follow-up question: when I read a block from disk the page it gets 
read into will get its mod bit set (in this architecture the mod/ref bits are 
associated with the physical page). This would mean fsflush() would detect this 
page and attempt to flush it back to disk. Therefore, there should/must be some 
mechanism that saves the page state before the read and restores it after so 
that no false positives would happen. What I'm seeing is that at boot time if I 
slow the system down enough (by putting in various traces) the fsflush thread 
can get up before the root device is remounted r/w and attempt to putpage these 
false positives out to disk. Naturally we panic if it happens at this stage of 
play. Even when the device is mounted r/w the false positives would cause 
unnecessary I/O.
 
 
This message posted from opensolaris.org
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code

Reply via email to