<feed xmlns='http://www.w3.org/2005/Atom'>
<title>litmus-rt.git/drivers/s390, branch v2.6.31-rc1</title>
<subtitle>The LITMUS^RT kernel.</subtitle>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/'/>
<entry>
<title>[S390] dasd: fix refcounting in dasd_change_state</title>
<updated>2009-06-22T10:08:23+00:00</updated>
<author>
<name>Sebastian Ott</name>
<email>sebott@linux.vnet.ibm.com</email>
</author>
<published>2009-06-22T10:08:21+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=181d95229b0931ee2ce6aad7348079cbc10e8d05'/>
<id>181d95229b0931ee2ce6aad7348079cbc10e8d05</id>
<content type='text'>
To set a dasd online dasd_change_state is called twice. The first
cycle will schedule initial analysis of the device, set the rc to
-EAGAIN and will not touch the device state any more.
The initial analysis will in turn call dasd_change_state to increase
the state to the final DASD_STATE_ONLINE.

If the dasd_change_state on the second thread outruns the other one
both finish with the state set to DASD_STATE_ONLINE and the device
refcount will be decreased by 2.

Fix this by leaving dasd_change_state on rc == -EAGAIN so that the
refcount will always be decreased by 1.

Signed-off-by: Sebastian Ott &lt;sebott@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
To set a dasd online dasd_change_state is called twice. The first
cycle will schedule initial analysis of the device, set the rc to
-EAGAIN and will not touch the device state any more.
The initial analysis will in turn call dasd_change_state to increase
the state to the final DASD_STATE_ONLINE.

If the dasd_change_state on the second thread outruns the other one
both finish with the state set to DASD_STATE_ONLINE and the device
refcount will be decreased by 2.

Fix this by leaving dasd_change_state on rc == -EAGAIN so that the
refcount will always be decreased by 1.

Signed-off-by: Sebastian Ott &lt;sebott@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] driver_data access</title>
<updated>2009-06-22T10:08:23+00:00</updated>
<author>
<name>Martin Schwidefsky</name>
<email>schwidefsky@de.ibm.com</email>
</author>
<published>2009-06-22T10:08:19+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=4f0076f77fb64889d4e5e425b63333e5764b446d'/>
<id>4f0076f77fb64889d4e5e425b63333e5764b446d</id>
<content type='text'>
Replace the remaining direct accesses to the driver_data pointer
with calls to the dev_get_drvdata() and dev_set_drvdata() functions.

Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Replace the remaining direct accesses to the driver_data pointer
with calls to the dev_get_drvdata() and dev_set_drvdata() functions.

Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] dasd_pm: fix stop flag handling</title>
<updated>2009-06-22T10:08:22+00:00</updated>
<author>
<name>Stefan Haberland</name>
<email>stefan.haberland@de.ibm.com</email>
</author>
<published>2009-06-22T10:08:17+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=e6125fba81e362d9b314d10893af1d9dc5658f33'/>
<id>e6125fba81e362d9b314d10893af1d9dc5658f33</id>
<content type='text'>
The stop flags are handled in the generic restore function so the
stop flag is removed also for FBA and DIAG devices.

Signed-off-by: Stefan Haberland &lt;stefan.haberland@de.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
The stop flags are handled in the generic restore function so the
stop flag is removed also for FBA and DIAG devices.

Signed-off-by: Stefan Haberland &lt;stefan.haberland@de.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] ap/zcrypt: Suspend/Resume ap bus and zcrypt</title>
<updated>2009-06-22T10:08:21+00:00</updated>
<author>
<name>Felix Beck</name>
<email>felix.beck@de.ibm.com</email>
</author>
<published>2009-06-22T10:08:16+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=772f54720ab82a6e88f0a8a84d76e7af15ca1f0c'/>
<id>772f54720ab82a6e88f0a8a84d76e7af15ca1f0c</id>
<content type='text'>
Add Suspend/Resume support to ap bus and zcrypt. All enhancements are
done in the ap bus. No changes in the crypto card specific part are
necessary.

Signed-off-by: Felix Beck &lt;felix.beck@de.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Add Suspend/Resume support to ap bus and zcrypt. All enhancements are
done in the ap bus. No changes in the crypto card specific part are
necessary.

Signed-off-by: Felix Beck &lt;felix.beck@de.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] qdio: Sanitize do_QDIO sanity checks</title>
<updated>2009-06-22T10:08:21+00:00</updated>
<author>
<name>Jan Glauber</name>
<email>jang@linux.vnet.ibm.com</email>
</author>
<published>2009-06-22T10:08:15+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=6618241b47cd131503610d8df68dd6f4948e5c1a'/>
<id>6618241b47cd131503610d8df68dd6f4948e5c1a</id>
<content type='text'>
Remove unneeded sanity checks from do_QDIO since this is the hot path.
Change the type of bufnr and count to unsigned int so the check for the
maximum value works.

Reported-by: Roel Kluin &lt;roel.kluin@gmail.com&gt;
Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Remove unneeded sanity checks from do_QDIO since this is the hot path.
Change the type of bufnr and count to unsigned int so the check for the
maximum value works.

Reported-by: Roel Kluin &lt;roel.kluin@gmail.com&gt;
Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] qdio: leave inbound SBALs primed</title>
<updated>2009-06-22T10:08:21+00:00</updated>
<author>
<name>Jan Glauber</name>
<email>jang@linux.vnet.ibm.com</email>
</author>
<published>2009-06-22T10:08:14+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=f0a0b15e0f3aff0a25f21f58bef8e40e80b16dc6'/>
<id>f0a0b15e0f3aff0a25f21f58bef8e40e80b16dc6</id>
<content type='text'>
It is not required to change the state of primed SBALs. Leaving them
primed saves a SQBS instruction under z/VM.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
It is not required to change the state of primed SBALs. Leaving them
primed saves a SQBS instruction under z/VM.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] qdio: merge AI tasklet into interrupt handler</title>
<updated>2009-06-22T10:08:20+00:00</updated>
<author>
<name>Jan Glauber</name>
<email>jang@linux.vnet.ibm.com</email>
</author>
<published>2009-06-22T10:08:13+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=cf9a031c2cc881e9873ab9ccf5e1f59f5b5167aa'/>
<id>cf9a031c2cc881e9873ab9ccf5e1f59f5b5167aa</id>
<content type='text'>
Since the adapter interrupt tasklet only schedules the queue tasklets
and contains no code that requires serialization in can be merged
with the adapter interrupt handler. That possibly safes some CPU
cycles.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Since the adapter interrupt tasklet only schedules the queue tasklets
and contains no code that requires serialization in can be merged
with the adapter interrupt handler. That possibly safes some CPU
cycles.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] qdio: extract all primed SBALs at once</title>
<updated>2009-06-22T10:08:20+00:00</updated>
<author>
<name>Jan Glauber</name>
<email>jang@linux.vnet.ibm.com</email>
</author>
<published>2009-06-22T10:08:12+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=36e3e72120e27939233e4bd88a8d74b3a2377428'/>
<id>36e3e72120e27939233e4bd88a8d74b3a2377428</id>
<content type='text'>
For devices without QIOASSIST primed SBALS were extracted in a loop.
Remove the loop since get_buf_states can already return more than
one primed SBAL.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
For devices without QIOASSIST primed SBALS were extracted in a loop.
Remove the loop since get_buf_states can already return more than
one primed SBAL.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] qdio: fix check for running under z/VM</title>
<updated>2009-06-22T10:08:20+00:00</updated>
<author>
<name>Jan Glauber</name>
<email>jang@linux.vnet.ibm.com</email>
</author>
<published>2009-06-22T10:08:11+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=9a2c160a8cbd5b3253672b3bac462c64d0d2eef7'/>
<id>9a2c160a8cbd5b3253672b3bac462c64d0d2eef7</id>
<content type='text'>
The check whether qdio runs under z/VM was incorrect since SIGA-Sync is not
set if the device runs with QIOASSIST. Use MACHINE_IS_VM instead to prevent
polling under z/VM.

Merge qdio_inbound_q_done and tiqdio_is_inbound_q_done.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
The check whether qdio runs under z/VM was incorrect since SIGA-Sync is not
set if the device runs with QIOASSIST. Use MACHINE_IS_VM instead to prevent
polling under z/VM.

Merge qdio_inbound_q_done and tiqdio_is_inbound_q_done.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>[S390] qdio: move adapter interrupt tasklet code</title>
<updated>2009-06-22T10:08:19+00:00</updated>
<author>
<name>Jan Glauber</name>
<email>jang@linux.vnet.ibm.com</email>
</author>
<published>2009-06-22T10:08:10+00:00</published>
<link rel='alternate' type='text/html' href='http://rtsrv.cs.unc.edu/cgit/cgit.cgi/litmus-rt.git/commit/?id=60b5df2f12f2ab54bfa7c1f0f0ce3f5953e73c0b'/>
<id>60b5df2f12f2ab54bfa7c1f0f0ce3f5953e73c0b</id>
<content type='text'>
Move the adapter interrupt tasklet function to the qdio main code
since all the functions used by the tasklet are located there.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Move the adapter interrupt tasklet function to the qdio main code
since all the functions used by the tasklet are located there.

Signed-off-by: Jan Glauber &lt;jang@linux.vnet.ibm.com&gt;
Signed-off-by: Martin Schwidefsky &lt;schwidefsky@de.ibm.com&gt;
</pre>
</div>
</content>
</entry>
</feed>
