| |||||||||||||
"doesn't support DPO or FUA""doesn't support DPO or FUA"Posted Aug 19, 2010 1:41 UTC (Thu) by cesarb (subscriber, #6266)Parent article: The end of block barriers
I often see things like the following in the kernel log for SATA disks:
sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
What happens to these when a filesystem requests a force unit access?
(Log in to post comments)
"doesn't support DPO or FUA" Posted Aug 19, 2010 20:58 UTC (Thu) by willy (subscriber, #9762) [Link]
If the drive doesn't support FUA, the block layer sends down the FUA-tagged command, then sends down another flush. So you get:
write, write, write, flush, important write, flush, write, write, write
instead of:
write, write, write, flush, important write FUA, write, write, write
This is documented in Documentation/block/barrier.txt if you're curious in learning more.
"doesn't support DPO or FUA" Posted Aug 20, 2010 11:14 UTC (Fri) by Hknr (guest, #67789) [Link]
From Documentation/block/barrier.txt:
»All low level drivers have to are implementing its prepare_flush_fn»...
Does nobody ever (proof-)read these Documents?
"doesn't support DPO or FUA" Posted Aug 20, 2010 22:37 UTC (Fri) by cesarb (subscriber, #6266) [Link]
> Does nobody ever (proof-)read these Documents?
Some people do. Every once in a while I see a batch of spelling fix patches on the git history of the kernel. After all, anyone can send a spelling fix patch.
"doesn't support DPO or FUA" Posted Aug 29, 2010 22:24 UTC (Sun) by Duncan (guest, #6647) [Link]
>> "drivers have are implementing its prepare_flush_fn"...
Gaah! Spammers in the kernel docs!
|
|||||||||||||