Home » Mailing list archives » linux-kernel » 2008 » January » 8

Re: [PATCH 1/6] SUNRPC: spin svc_rqst initialization to its own function

Previous thread: [PATCH -mm] x86 boot : export boot_params via debugfs for debugging by Huang, Ying on Tuesday, January 8, 2008 - 1:40 am. (2 messages)

Next thread: Re: [PATCH 5/6] NLM: Convert lockd to use kthreads by Neil Brown on Tuesday, January 8, 2008 - 2:16 am. (1 message)
[view original message]
From: Neil Brown <neilb@...>
To: Jeff Layton <jlayton@...>
Cc: <akpm@...>, <linux-nfs@...>, <linux-kernel@...>
Subject: Re: [PATCH 1/6] SUNRPC: spin svc_rqst initialization to its own function
Date: Tuesday, January 8, 2008 - 1:53 am

I realise that the bug existed before your change, but calling

svc_exit_thread at this point is not good.

The 'goto out_thread' is *before* "pool->sp_nrthreads++", but

svc_exit_thread does "pool->sp_nrthreads--;".  Not good.


As you are playing is this code, do you feel like fixing that error

path??


Otherwise, patch looks good.

Acked-By: NeilBrown <neilb@suse.de>


Thanks,

NeilBrown

--
[view original message]
From: Jeff Layton <jlayton@...>
To: Neil Brown <neilb@...>
Cc: <akpm@...>, <linux-nfs@...>, <linux-kernel@...>
Subject: Re: [PATCH 1/6] SUNRPC: spin svc_rqst initialization to its own function
Date: Tuesday, January 8, 2008 - 8:11 am

On Tue, 8 Jan 2008 16:53:09 +1100


Good catch. That bug will be a NULL pointer dereference since

rqstp->rq_pool isn't even set yet. There's a similar bug with

serv->sv_nrthreads too.


Still looking over your other comments, but I'll plan to post an

updated patchset that includes a fix for that.


--

Jeff Layton <jlayton@redhat.com>

--
Previous thread: [PATCH -mm] x86 boot : export boot_params via debugfs for debugging by Huang, Ying on Tuesday, January 8, 2008 - 1:40 am. (2 messages)

Next thread: Re: [PATCH 5/6] NLM: Convert lockd to use kthreads by Neil Brown on Tuesday, January 8, 2008 - 2:16 am. (1 message)