List Info

Thread: nfs optimization and veriexec




nfs optimization and veriexec
country flaguser name
Japan
2007-10-28 17:59:03
hi,

i'd like to make a change like the following, which optimize
a RPC away
in some cases.
however, i think it breaks veriexec. (specifically, breaks
assumptions
held by veriexec_openchk.)
can any veriexec people take a look and suggest how it
should be solved?

thanks.

YAMAMOTO Takashi

  
Re: nfs optimization and veriexec
country flaguser name
United States
2007-10-28 23:57:35
>i'd like to make a change like the following, which
optimize a RPC away
>in some cases.
>however, i think it breaks veriexec. (specifically,
breaks assumptions
>held by veriexec_openchk.)
>can any veriexec people take a look and suggest how it
should be solved?

I'm still using an ancient version of veriexec with several
mods,
but I think the current version has a state variable you can
check.
This would allow you to skip the optimization if veriexec is
enabled.
You could perhaps even test if the kernel has any
finger-prints loaded for 
the mounted fs - if not, then veriexec won't matter - the fs
is presumably 
being used for data only. 

--sjg
 

Re: nfs optimization and veriexec
country flaguser name
Japan
2007-10-29 05:49:59
> hi,
> 
> i'd like to make a change like the following, which
optimize a RPC away
> in some cases.
> however, i think it breaks veriexec. (specifically,
breaks assumptions
> held by veriexec_openchk.)
> can any veriexec people take a look and suggest how it
should be solved?
> 
> thanks.
> 
> YAMAMOTO Takashi

as some people asked me to explain more...

currently, create basically works as the following,

	1. VOP_LOOKUP: issue a LOOKUP RPC to check if the file
already exists.
	2. veriexec_openchk.
	3. VOP_CREATE: issue a CREATE RPC to create a file.

the patch changes it to delegate more to nfs server.
ie.
	1. VOP_LOOKUP: just return.
	2. veriexec_openchk ???
	3. VOP_CREATE: issue a CREATE RPC to do "create a file
unless
	   it already exists."

YAMAMOTO Takashi

Re: nfs optimization and veriexec
country flaguser name
Australia
2007-10-29 06:00:09
On Mon, Oct 29, 2007 at 07:49:59PM +0900, YAMAMOTO Takashi
wrote:
> 
> as some people asked me to explain more...
> 

I think the approach that Simon suggested is probably
workable.  If
veriexec is active on the mount then there should be a
fileassoc table
for the mount point, you can look up the handle in that
fileassoc, if
it is there then don't short circuit otherwise it is safe to
do so.

If you like I can fix that for you.

-- 
Brett Lymn
"Warning:
The information contained in this email and any attached
files is
confidential to BAE Systems Australia. If you are not the
intended
recipient, any use, disclosure or copying of this email or
any
attachments is expressly prohibited.  If you have received
this email
in error, please notify us immediately. VIRUS: Every care
has been
taken to ensure this email and its attachments are virus
free,
however, any loss or damage incurred in using this email is
not the
sender's responsibility.  It is your responsibility to
ensure virus
checks are completed before installing any data sent in this
email to
your computer."


[1-4]

about | contact  Other archives ( Real Estate discussion Medical topics )