* [9fans] cfs
@ 2021-01-19 22:50 Steve Simon
2021-01-20 16:52 ` [9fans] cfs Anthony Martin
0 siblings, 1 reply; 6+ messages in thread
From: Steve Simon @ 2021-01-19 22:50 UTC (permalink / raw)
To: 9fans
Hi all,
I am playing with cfs(4). It works nicely dialing to my fileserver but
fails if I try to use the existing /srv/boot channel.
failed with /srv/boot:
sceolan% 5.out -d -n -F /srv/boot -f $cfs -S /n/cache
<-client: Tversion tag 65535 msize 8216 version '9P2000'
->server: Tversion tag 65535 msize 8216 version '9P2000'
5.out: write failed, inappropriate use of fd
works using dial:
sceolan% 5.out -d -F /srv/boot -f $cfs -S /n/cache
<-client: Tversion tag 65535 msize 8216 version '9P2000'
->server: Tversion tag 65535 msize 8216 version '9P2000'
<-server: Rversion tag 65535 msize 8216 version '9P2000'
->client: Rversion tag 65535 msize 8216 version '9P2000'
<-client: Tauth tag 13 afid 402 uname steve aname
->server: Tauth tag 13 afid 402 uname steve aname
<-server: Rauth tag 13 qid (0000000000000192 0 A)
->client: Rauth tag 13 qid (0000000000000192 0 A)
<-client: Tread tag 13 fid 402 offset 0 count 2048
Note: using -n or not makes no differnce to the failed version,
though I would expect to need -n with /srv/boot.
The error occurs on the first write to /srv/boot.
I can mount /srv/boot with mount(1) without problems.
What have I missed?
-Steve
------------------------------------------
9fans: 9fans
Permalink: https://9fans.topicbox.com/groups/9fans/T1ff0c2bf191f9858-M8e031b5e1276934aa63964f2
Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
^ permalink raw reply [flat|nested] 6+ messages in thread* [9fans] cfs
@ 2009-05-13 13:01 Steve Simon
0 siblings, 0 replies; 6+ messages in thread
From: Steve Simon @ 2009-05-13 13:01 UTC (permalink / raw)
To: 9fans
I think I have been silly,
I confused the block cache for cfs.
Sorry for the noise.
-Steve
^ permalink raw reply [flat|nested] 6+ messages in thread
* [9fans] cfs
@ 2009-05-13 11:20 Steve Simon
2009-05-13 12:34 ` erik quanstrom
0 siblings, 1 reply; 6+ messages in thread
From: Steve Simon @ 2009-05-13 11:20 UTC (permalink / raw)
To: 9fans
Why is cfs in the kernel and not a userlevel program?
Is it historic, perhaps this was done before /boot appeared so it was not
easy to add a userlevel program into the boot process.
I guess there might be performance issues also, though these would
be swamped by the performance improvment of plugging in CFS in the
first place (IMHO).
Just curious
-Steve
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [9fans] cfs
2009-05-13 11:20 Steve Simon
@ 2009-05-13 12:34 ` erik quanstrom
0 siblings, 0 replies; 6+ messages in thread
From: erik quanstrom @ 2009-05-13 12:34 UTC (permalink / raw)
To: 9fans
> Why is cfs in the kernel and not a userlevel program?
i don't know. but by putting cfs in the kernel, you can use the kernel's
built-in concurrency. cfs predates the thread library.
> I guess there might be performance issues also, though these would
> be swamped by the performance improvment of plugging in CFS in the
> first place (IMHO).
why guess? we can get a good approximation. the difference
is a round trip to user space which has bandwidth and latency
costs. i used dd and ramfs to simulate.
userspace kernel
b/w latency b/w latency
i7 2666 460 MB/s 1680µs 3900 MB/s 118µs
Xeon5000 1859 156 MB/s 1205 868 265
AMD64 2604 be 142 MB/s 2511µs 2600 332
AMD64 2004 135 MB/s 2510 2600 324
if you're limited by 1gbe, the bandwidth would be good
but the latency is spoiled. if you're limited by 10gbe,
the vodka's weak, too.
- erik
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2021-01-20 17:25 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-01-19 22:50 [9fans] cfs Steve Simon
2021-01-20 16:52 ` [9fans] cfs Anthony Martin
2021-01-20 17:25 ` Richard Miller
-- strict thread matches above, loose matches on Subject: below --
2009-05-13 13:01 [9fans] cfs Steve Simon
2009-05-13 11:20 Steve Simon
2009-05-13 12:34 ` erik quanstrom
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).