2003-10-21 16:08:16

by Sundareswar Jayakumar

[permalink] [raw]
Subject: query

Hi,
I am working on adistributed filesytem on linux.I am using
nfs source code as a learning resource.I require a clarification.


In many entry points the Big Kernel Lock lock/unlock_kernel is
used instead of locking just the private data structures of the
filesystem.why is it so?.kindly clarify w.r.t to a smp m/c.

Thanks
sundar


-------------------------------------------------------
This SF.net email is sponsored by OSDN developer relations
Here's your chance to show off your extensive product knowledge
We want to know what you know. Tell us and you have a chance to win $100
http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs


2003-10-22 16:48:31

by Trond Myklebust

[permalink] [raw]
Subject: Re: query

>>>>> " " == Sundareswar Jayakumar <[email protected]> writes:

> Hi,
> I am working on adistributed filesytem on linux.I am using
> nfs source code as a learning resource.I require a
> clarification.


> In many entry points the Big Kernel Lock lock/unlock_kernel is
> used instead of locking just the private data structures of the
> filesystem.why is it so?.kindly clarify w.r.t to a smp m/c.

When working on performance, you pursue the largest sources of latency
first...
BKL contention has not yet turned up on the radar screen. There has
therefore not yet been enough incentive to replace it as a mechanism
for serialization.

Cheers,
Trond


-------------------------------------------------------
This SF.net email is sponsored by OSDN developer relations
Here's your chance to show off your extensive product knowledge
We want to know what you know. Tell us and you have a chance to win $100
http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

2003-10-22 23:52:09

by NeilBrown

[permalink] [raw]
Subject: Re: query

On Tuesday October 21, [email protected] wrote:
> Hi,
> I am working on adistributed filesytem on linux.I am using
> nfs source code as a learning resource.I require a clarification.
>
>
> In many entry points the Big Kernel Lock lock/unlock_kernel is
> used instead of locking just the private data structures of the
> filesystem.why is it so?.kindly clarify w.r.t to a smp m/c.

Because tghe BKL was the easiest way to make it work under SMP, and
the process of refining the locks hasn't made it into 2.4.

The 2.6.0 kernel does not use the BKL for nfsd at all, and I have
patches that remove the BKL for 2.4 but marcelo didn't want them when
I offered them some months ago, and I haven't offered again since.

NeilBrown

>
> Thanks
> sundar
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by OSDN developer relations
> Here's your chance to show off your extensive product knowledge
> We want to know what you know. Tell us and you have a chance to win $100
> http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54
> _______________________________________________
> NFS maillist - [email protected]
> https://lists.sourceforge.net/lists/listinfo/nfs


-------------------------------------------------------
This SF.net email is sponsored by OSDN developer relations
Here's your chance to show off your extensive product knowledge
We want to know what you know. Tell us and you have a chance to win $100
http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

2003-10-25 08:21:30

by Lever, Charles

[permalink] [raw]
Subject: RE: query

hi-

what trond says is true, and i'd like to supplement that
idea by saying that the types of workloads where the BKL
would be a bottleneck are rare these days, compared to
the problems more common workloads experience, so it has
been a lower priority.

i think as database on NFS becomes more popular, for example,
this issue will become more important.

> -----Original Message-----
> From: Trond Myklebust [mailto:[email protected]]
> Sent: Wednesday, October 22, 2003 9:46 AM
> To: Sundareswar Jayakumar
> Cc: [email protected]
> Subject: Re: [NFS] query
>=20
>=20
> >>>>> " " =3D=3D Sundareswar Jayakumar <[email protected]> writes:
>=20
> > Hi,
> > I am working on adistributed filesytem on=20
> linux.I am using
> > nfs source code as a learning resource.I require a
> > clarification.
>=20
>=20
> > In many entry points the Big Kernel Lock lock/unlock_kernel is
> > used instead of locking just the private data structures of the
> > filesystem.why is it so?.kindly clarify w.r.t to a smp m/c.
>=20
> When working on performance, you pursue the largest sources of latency
> first...
> BKL contention has not yet turned up on the radar screen. There has
> therefore not yet been enough incentive to replace it as a mechanism
> for serialization.
>=20
> Cheers,
> Trond
>=20
>=20
> -------------------------------------------------------
> This SF.net email is sponsored by OSDN developer relations
> Here's your chance to show off your extensive product knowledge
> We want to know what you know. Tell us and you have a chance=20
> to win $100
> http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54
> _______________________________________________
> NFS maillist - [email protected]
> https://lists.sourceforge.net/lists/listinfo/nfs
>=20


-------------------------------------------------------
This SF.net email is sponsored by: The SF.net Donation Program.
Do you like what SourceForge.net is doing for the Open
Source Community? Make a contribution, and help us add new
features and functionality. Click here: http://sourceforge.net/donate/
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs