[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [Patch] support of lock profiling in Xen


  • To: Keir Fraser <keir.fraser@xxxxxxxxxxxxx>
  • From: Juergen Gross <juergen.gross@xxxxxxxxxxxxxx>
  • Date: Thu, 08 Oct 2009 11:35:47 +0200
  • Cc: "xen-devel@xxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Thu, 08 Oct 2009 02:36:11 -0700
  • Domainkey-signature: s=s1536a; d=ts.fujitsu.com; c=nofws; q=dns; h=X-SBRSScore:X-IronPort-AV:Received:X-IronPort-AV: Received:Received:Message-ID:Date:From:Organization: User-Agent:MIME-Version:To:CC:Subject:References: In-Reply-To:X-Enigmail-Version:Content-Type: Content-Transfer-Encoding; b=b2CUQ3usSlSXCbbjFFGwXh8dcOO5fZni+jKB1GpEkGM3WYIGXTDCUnsf bCsE0+GNNZVF+/uXTjEDEZoNxbIv9MP8ZOBEcK7f5EDC5eimSGKktxww1 E7TW8vE5YsF5jlqyKQMY+BSNbhIDYTB8Ock1TAGWHjupztSNDJ8YBHR78 dazdZNM28FBd+IVPz/oSyizAlw1AaoVoHOE0dqQG9jnkQ187kGEDU/M9D vlZ5JUmDzKtIlIqmhtIgPdiI4JMHt;
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>

Keir Fraser wrote:
> On 08/10/2009 10:04, "Juergen Gross" <juergen.gross@xxxxxxxxxxxxxx> wrote:
> 
>>>> It would even be possible to define lock arrays if summing up all data for
>>>> e.g. domain specific locks is not desired.
>>> Allowing spinlock_init() to specify not only a name but also a
>>> 'category-specific index' might be okay. For example, domid, in the case of
>>> a domain-specific lock. Hiding the allocation of profiling info structs in
>>> the spinlock subsystem somehow... (e.g., radix tree to make a simple
>>> growable array).
>> I'll try another patch...
> 
> Well, see my email just now. I could perhaps live with perfc-style
> declarations after all... Because this might make sense if locks can be
> broadly categorised as global or as per-domain, and it could avoid you
> having to make your implementation much more complex?
> 
> After all, now I think about it, an 'arbitrary' index specified to
> spinlock_init() is just making our lives more complicated if it will
> basically *always* be a domid. And knowing at a higher level which locks are
> actually per-domain could help your pretty printing?

I thought in this direction, too.
I would, however, try not to limit it to per domain locks, even if the first
implementation would support only global locks and per domain locks.

The perfc style declarations should only be necessary for the dynamically
initialized locks, and those could live near the domain structure definition
(or any other structure where they are needed).

I think the statically defined locks can be handled via an own data section
requiring no centrally defined index.


Juergen

-- 
Juergen Gross                 Principal Developer Operating Systems
TSP ES&S SWE OS6                       Telephone: +49 (0) 89 636 47950
Fujitsu Technolgy Solutions               e-mail: juergen.gross@xxxxxxxxxxxxxx
Otto-Hahn-Ring 6                        Internet: ts.fujitsu.com
D-81739 Muenchen                 Company details: ts.fujitsu.com/imprint.html

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.