Re: [Qemu-devel] [RFC][PATCH 11/12] qcow2: Convert qcow2 to use coroutines for async I/O

From: Avi Kivity <avi@redhat.com>
To: Anthony Liguori <anthony@codemonkey.ws>
Cc: Kevin Wolf <kwolf@redhat.com>,
	Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>,
	qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [RFC][PATCH 11/12] qcow2: Convert qcow2 to use	coroutines for async I/O
Date: Thu, 27 Jan 2011 11:25:22 +0200	[thread overview]
Message-ID: <4D413A02.5060700@redhat.com> (raw)
In-Reply-To: <4D4055FB.6090104@codemonkey.ws>

On 01/26/2011 07:12 PM, Anthony Liguori wrote:
>> What do you actually save?  The longjmp() to the coroutine code, 
>> linking in to the mutex wait queue, and another switch back to the 
>> main coroutine?  Given that we don't expect to block often, it seems 
>> hardly a cost worth optimizing.
>
>
> It's a matter of correctness, not optimization.
>
> Consider the following example:
>
> coroutine {
>    l2 = find_l2(offset);
>    // allocates but does not increment max cluster offset
>    l2[l2_offset(offset)] = alloc_cluster();
>
>    co_mutex_lock(&lock);
>    write_l2(l2);
>    co_mutex_unlock(&lock);
>
>    l1[l1_offset(offset)] = l2;
>
>    co_mutex_lock(&lock);
>    write_l1(l1);
>    co_mutex_unlock(&lock);
>
>    commit_cluster(l2[l2_offset(offset)]);
> }
>
> This code is incorrect.

Yes it's incorrect, but it isn't what I'm proposing.

-- 
error compiling committee.c: too many arguments to function