6,534
edits
Changes
created
== Changes ==
Since {{kernel link|rhel5|028stab060.2}}:
* Audit in CT0 re-enabled back ({{bug|1157}})
* UBC: files precharging/uncharging algorithm corrected
* Fixed crash while using NFS volume with non-hidden submounts
=== Configs ===
Same as in {{kernel link|rhel5|028stab060.2}}, plus:
* +<code>CONFIG_TICK_DIVIDER=y</code>
=== Compatibility ===
No new issues.
<includeonly>[[{{PAGENAME}}/changes#Patches|{{Long changelog message}}]]</includeonly><noinclude>
=== Patches ===
==== diff-ve-close-audit-in-ct-20090202 ====
<div class=change>
Patch from Pavel Emelianov (xemul@):<br/>
put audit back but make it work in ve0 only.
Bug #269595<br/>
{{Bug|1157}}
</div>
==== diff-ubc-file-precharge-correct-uncharge-20080120 ====
<div class=change>
Patch from Marat Stanichenko (mstanichenko@):<br/>
ubc: Make ubc uncharge files in case of abundance precharged files
There is a chance when we do not start uncharging
because ub_barrier_farnr() is not hit for UB_NUMFILE
and ub_barrier_farsz() is not hit for UB_KMEMSIZE (SLM
for example set ubc barrier to a huge value).
This fact can lead us to the situation when two tasks
are able to consume all of UB_NUMFILE
and UB_KMEMSIZE in despite that they close opened files.
Bug #267453.
</div>
==== diff-vzdq-nfs-submount-sops-20090119 ====
<div class=change>
Patch from Vasily Averin (vvs@):<br/>
nfs submount copies super_operations from parent NFS, where it can be replaced by vzquota.
As result on unmount of NFS submounts kernel calls vzquota callback and frees resources
used by parent NFS that lead to crash on access to super_operations.
We are not going to use vzquota on NFS submounts and replace copy of super_operations
for NFS submount by direct set up of required operations.
Bug #267448.
</div>
==== diff-ve-allow-write-loginuid-in-ct-20090209 ====
<div class=change>
Patch from Marat Stanichenko (mstanichenko@):<br/>
Allow CTs write to /proc/pid/loginuid
We should give an ability to write to /proc/$pid/loginuid
inside CT and therefore abolish the first hunk of
diff-ve-close-audit-in-ct-20090202 patch. Otherwise it
affects container's ssh connectivity. Pam complains about
impossibility to write to /proc/self/loginuid.
Bug #270484.
</div>
</noinclude>
Since {{kernel link|rhel5|028stab060.2}}:
* Audit in CT0 re-enabled back ({{bug|1157}})
* UBC: files precharging/uncharging algorithm corrected
* Fixed crash while using NFS volume with non-hidden submounts
=== Configs ===
Same as in {{kernel link|rhel5|028stab060.2}}, plus:
* +<code>CONFIG_TICK_DIVIDER=y</code>
=== Compatibility ===
No new issues.
<includeonly>[[{{PAGENAME}}/changes#Patches|{{Long changelog message}}]]</includeonly><noinclude>
=== Patches ===
==== diff-ve-close-audit-in-ct-20090202 ====
<div class=change>
Patch from Pavel Emelianov (xemul@):<br/>
put audit back but make it work in ve0 only.
Bug #269595<br/>
{{Bug|1157}}
</div>
==== diff-ubc-file-precharge-correct-uncharge-20080120 ====
<div class=change>
Patch from Marat Stanichenko (mstanichenko@):<br/>
ubc: Make ubc uncharge files in case of abundance precharged files
There is a chance when we do not start uncharging
because ub_barrier_farnr() is not hit for UB_NUMFILE
and ub_barrier_farsz() is not hit for UB_KMEMSIZE (SLM
for example set ubc barrier to a huge value).
This fact can lead us to the situation when two tasks
are able to consume all of UB_NUMFILE
and UB_KMEMSIZE in despite that they close opened files.
Bug #267453.
</div>
==== diff-vzdq-nfs-submount-sops-20090119 ====
<div class=change>
Patch from Vasily Averin (vvs@):<br/>
nfs submount copies super_operations from parent NFS, where it can be replaced by vzquota.
As result on unmount of NFS submounts kernel calls vzquota callback and frees resources
used by parent NFS that lead to crash on access to super_operations.
We are not going to use vzquota on NFS submounts and replace copy of super_operations
for NFS submount by direct set up of required operations.
Bug #267448.
</div>
==== diff-ve-allow-write-loginuid-in-ct-20090209 ====
<div class=change>
Patch from Marat Stanichenko (mstanichenko@):<br/>
Allow CTs write to /proc/pid/loginuid
We should give an ability to write to /proc/$pid/loginuid
inside CT and therefore abolish the first hunk of
diff-ve-close-audit-in-ct-20090202 patch. Otherwise it
affects container's ssh connectivity. Pam complains about
impossibility to write to /proc/self/loginuid.
Bug #270484.
</div>
</noinclude>