[H-GEN] Clock Time Loss When Copying Files

Martin Pool mbp at linuxcare.com.au
Tue Jun 6 22:43:25 EDT 2000


[ Humbug *General* list - semi-serious discussions about Humbug and ]
[ Unix-related topics.  Please observe the list's charter.          ]

On Wed, 7 Jun 2000, Bruce Campbell wrote:

> Not quite correct.  There are two things occuring in the above.  One,
> reading/writing data from/to an IDE HDD (mucho interrupts).  Two,
> transferring data out to the network card.  As you mentioned that it is a
> BNC (ie, Coax) network, a valid assumption would be older network cards
> (again, mucho interrupts required)

Yes, and also: it's not just the frequency of interrupts that matters, but
also how long they take to process.  It should be quite harmless to mask
out a clock interrupt briefly, but if it's blocked for too long it will be
lost.

Older hardware will either require the CPU to do more work in handling the
interrupt, or will have an old and simple driver which does too much work
while interrupts are blocked.  

(A better approach is for the driver to remember the interrupt and crucial
information and then RTI.  In Linux this is handled by BH's
(bottom-halves) or more recently by tasklets.  (I think.)  The Lions book
has far too much information about this.)

-- 
Martin Pool, Linuxcare, Inc.
+61 2 6262 8990
mbp at linuxcare.com, http://www.linuxcare.com/
Linuxcare. Support for the revolution.



--
* This is list (humbug) general handled by majordomo at lists.humbug.org.au .
* Postings to this list are only accepted from subscribed addresses of
* lists 'general' or 'general-post'.



More information about the General mailing list