Deadlock Scenario (?)
gil.romo at qlogic.com
Thu Jun 24 14:16:06 PDT 2004
* From the T10 Reflector (t10 at t10.org), posted by:
* Gil Romo <gil.romo at qlogic.com>
When a pathway is blocked by the ECR pathway and not the destination port,
should the expander resources be released using pathway recovery?
The ECM currently does not recognize this condition and does not present this
information to the requesting phy.
This scenario is similar to requests crossing on a link, but this link is wide
and crossing requests are handled only by the phys on the link.
> X-Authentication-Warning: t10.org: lohmeyer set sender to owner-t10 at t10.org
> X-MimeOLE: Produced By Microsoft Exchange V6.5.6944.0
> Content-class: urn:content-classes:message
> MIME-Version: 1.0
> Subject: RE: Deadlock Scenario (?)
> Date: Thu, 24 Jun 2004 09:18:45 -0500
> Thread-Topic: Deadlock Scenario (?)
> Thread-Index: AcRZi0YSnRHyXaRfQfGl1YJeP+gEAwAaT8Mg
> From: "Elliott, Robert (Server Storage)" <elliott at hp.com>
> To: <t10 at t10.org>
> X-OriginalArrivalTime: 24 Jun 2004 14:20:24.0420 (UTC)
> Content-Transfer-Encoding: 8bit
> X-MIME-Autoconverted: from quoted-printable to 8bit by t10.org id i5OE8oA23159
> X-Message-Number: 5017
> X-PMX-Version: 126.96.36.199886, Antispam-Core: 188.8.131.52542, Antispam-Data:
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Elliott, Robert (Server Storage)" <elliott at hp.com>
> I think we should change the message used when "there are insufficient
> routing resources" from Arbitrating (Waiting on Connection) to
> Arbitrating (Waiting on Partial).
> In your example, the requests are not really waiting on a connection,
> they are waiting on partial connections. On the left, phys W and X are
> returning Phy Status (Partial Pathway), not Phy Status (Connection), at
> this time. If there really were a narrow port on which the Phy Y
> request was waiting, that would cause Arbitrating (Waiting on Partial).
> Rob Elliott, elliott at hp.com
> Hewlett-Packard Industry Standard Server Storage Advanced Technology
> > -----Original Message-----
> > From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf
> > Of Fred Dutton
> > Sent: Wednesday, June 23, 2004 7:46 PM
> > To: t10 at t10.org
> > Subject: Deadlock Scenario (?)
> > * From the T10 Reflector (t10 at t10.org), posted by:
> > * Fred Dutton <fred.dutton at qlogic.com>
> > *
> > Assuming an edge expander supports a limited number of
> > connections (by limited I mean it only supports a number of
> > concurrent connections less than what is physically
> > possible). For example if an expander includes eight external
> > phys, then the maximum number of concurrent connections --end
> > device to end device-- is four but the SAS
> > specification only requires a minimum of one connection (see
> > 4.6.4). So the expander could, in theory, support a number of
> > connections less than four.
> > Suppose two edge expanders supporting N connections each are
> > connected to each other forming a wide port with M physical
> > links, where M > N.
> > Is it possible to get into a deadlock situation because there
> > are more links in the wide port than connections supported by
> > each expander?
> > For discussion purposes, suppose two edge expanders that
> > support only one connection each, are connected to each other
> > through a wide X2 port.
> > How should the following deadlock be resolved? (You may have
> > use mono space fonts to get this message to display correctly)
> > ...
> > Thanks,
> > Fred Dutton
> > ----------------------------------------------------
> > Fred Dutton * QLogic * 949.389.6340
> > *
> > * For T10 Reflector information, send a message with
> > * 'info t10' (no quotes) in the message body to majordomo at t10.org
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org
Circuits & Integration
QLogic Corporation, Aliso Viejo, California
E-mail: gil.romo at qlogic.com
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org
More information about the T10