[nfs-discuss] Code review for 6758280

2009-02-11 Thread Marcel Telka
On Tue, Feb 10, 2009 at 03:26:02PM -0600, james wahlig wrote:
> Looks alright to me.

Thank you.

-- 
Marcel Telka
Solaris RPE



[nfs-discuss] Code review for 6758280

2009-02-11 Thread Marcel Telka
On Tue, Feb 10, 2009 at 11:56:50AM -0600, james wahlig wrote:
> I started looking at this a while ago, but got distracted...  There is  
> really only one important change in nfs_client.c, right?  The rest are  

And the exactly same in nfs4_client.c file.

> just cstyle things?

Yes.

-- 
Marcel Telka
Solaris RPE



[nfs-discuss] Code review for 6758280

2009-02-10 Thread james wahlig
Looks alright to me.

jim

Marcel Telka wrote:

>Hi all,
>
>Nobody interested? :-(
>Is there anything important missing what prevents you to do a review?
>
>Thanks.
>
>On Fri, Feb 06, 2009 at 07:46:49PM +0100, Marcel Telka wrote:
>  
>
>>Hi,
>>
>>I prepared fix for:
>>
>>6758280 nfs_async_manager doesn't schedule nfs_async_start even if async reqs
>>are enqueued
>>
>>webrev: http://cr.opensolaris.org/~aragorn/6758280/
>>root cause: http://cr.opensolaris.org/~aragorn/6758280.root_cause
>>
>>Short root cause summary:
>>
>>There is a race condition between nfs_async_manager() and a function what is
>>enqueuing a new asynch request if such function is called early after the
>>nfs_async_manager() thread is created. An example of the racing function is
>>nfs_async_putapage(). The same problem is with nfs4_async_manager().
>>
>>
>
>  
>




[nfs-discuss] Code review for 6758280

2009-02-10 Thread james wahlig
I started looking at this a while ago, but got distracted...  There is 
really only one important change in nfs_client.c, right?  The rest are 
just cstyle things?

I should be done before the end of the day.

jim

Marcel Telka wrote:

>Hi all,
>
>Nobody interested? :-(
>Is there anything important missing what prevents you to do a review?
>
>Thanks.
>
>On Fri, Feb 06, 2009 at 07:46:49PM +0100, Marcel Telka wrote:
>  
>
>>Hi,
>>
>>I prepared fix for:
>>
>>6758280 nfs_async_manager doesn't schedule nfs_async_start even if async reqs
>>are enqueued
>>
>>webrev: http://cr.opensolaris.org/~aragorn/6758280/
>>root cause: http://cr.opensolaris.org/~aragorn/6758280.root_cause
>>
>>Short root cause summary:
>>
>>There is a race condition between nfs_async_manager() and a function what is
>>enqueuing a new asynch request if such function is called early after the
>>nfs_async_manager() thread is created. An example of the racing function is
>>nfs_async_putapage(). The same problem is with nfs4_async_manager().
>>
>>
>
>  
>




[nfs-discuss] Code review for 6758280

2009-02-10 Thread Marcel Telka
Hi all,

Nobody interested? :-(
Is there anything important missing what prevents you to do a review?

Thanks.

On Fri, Feb 06, 2009 at 07:46:49PM +0100, Marcel Telka wrote:
> Hi,
> 
> I prepared fix for:
> 
> 6758280 nfs_async_manager doesn't schedule nfs_async_start even if async reqs
> are enqueued
> 
> webrev: http://cr.opensolaris.org/~aragorn/6758280/
> root cause: http://cr.opensolaris.org/~aragorn/6758280.root_cause
> 
> Short root cause summary:
> 
> There is a race condition between nfs_async_manager() and a function what is
> enqueuing a new asynch request if such function is called early after the
> nfs_async_manager() thread is created. An example of the racing function is
> nfs_async_putapage(). The same problem is with nfs4_async_manager().

-- 
Marcel Telka
Solaris RPE



[nfs-discuss] Code review for 6758280

2009-02-06 Thread Marcel Telka
Hi,

I prepared fix for:

6758280 nfs_async_manager doesn't schedule nfs_async_start even if async reqs
are enqueued

webrev: http://cr.opensolaris.org/~aragorn/6758280/
root cause: http://cr.opensolaris.org/~aragorn/6758280.root_cause

Short root cause summary:

There is a race condition between nfs_async_manager() and a function what is
enqueuing a new asynch request if such function is called early after the
nfs_async_manager() thread is created. An example of the racing function is
nfs_async_putapage(). The same problem is with nfs4_async_manager().


Thanks.

-- 
Marcel Telka
Solaris RPE