Recently we did dom0 Xen RAS rebase work. Basically it rebased from Jeremy''s pvops (2.6.32), to Konrad''s pvops (3.1.0, branch "devel/acpi-cpufreq.v3") patch 1: for mcelog logic patch 2: for dom0 vmce patch 3: for cpu sysfs and cpu online/offline patch 4: for cpu hotadd patch 5: for memory hotadd prepare work patch 6: for memory hotadd patch 7: for cpu hotadd prepare work patch 8: complement cpu hotadd lacked at Konrad''s tree patch 9: update PM logic patch 10: upate cpu online/offline logic Thanks, Jinsong
On Thu, Dec 22, 2011 at 09:57:55AM +0000, Liu, Jinsong wrote:> Recently we did dom0 Xen RAS rebase work.Hey Liu, Thank you for posting this. Couple of questions: 1). How does one test these patches? 2). Please CC LKML and also the ACPI maintainer - since a majority of this also touch their tree. And the memory hotadd maintainer as well. 3). Can any of this code (like the MCE) leverage the existing generic code? 4). Have you guys talked with Len Brown (the ACPI maintainer) about some of this code and how to make it fit within his goals? Thanks!> > Basically it rebased from Jeremy''s pvops (2.6.32), to Konrad''s pvops (3.1.0, branch "devel/acpi-cpufreq.v3") > patch 1: for mcelog logic > patch 2: for dom0 vmce > patch 3: for cpu sysfs and cpu online/offline > patch 4: for cpu hotadd > patch 5: for memory hotadd prepare work > patch 6: for memory hotadd > patch 7: for cpu hotadd prepare work > patch 8: complement cpu hotadd lacked at Konrad''s tree > patch 9: update PM logic > patch 10: upate cpu online/offline logic > > > Thanks, > Jinsong > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel
Konrad Rzeszutek Wilk wrote:> On Thu, Dec 22, 2011 at 09:57:55AM +0000, Liu, Jinsong wrote: >> Recently we did dom0 Xen RAS rebase work. > > Hey Liu, > > Thank you for posting this. Couple of questions: > 1). How does one test these patches?These patches include mce, cpu online/offline, cpu hotadd, memory hotadd. Testing method of these features are 1. mce: a normal way is to inject error through EINJ user interface, and getting mcelog at dom0 side 2. cpu online/offline: by echoing sysfs/.../xen_pcpu user interface, and getting result at sysfs/.../xen_pcpu 3. cpu hotadd: Intel has some new server, by pressing a button can add cpu. We can verify cpu hotadd by sysfs/.../xen_pcpu interface 4. memory hotadd: also by pressing a button to add memory, and can verify memory hotadd by ''xl info''> 2). Please CC LKML and also the ACPI maintainer - since a majority > of this also touch their tree. And the memory hotadd maintainer > as well.OK, cc LKML ACPI maintainer: len.brown@intel.com memory hotadd maintianer: I''m not sure who is maintainer, but I will cc yakui.zhao@intel.com and len.brown@intel.com I''ll re-send the patches later, cc above people.> 3). Can any of this code (like the MCE) leverage the existing > generic code?Basically no, for mce, xen implement mce logic at hypervisor, at dom0 side, it only add mcelog interface to get log from hypervisor and transfer to kernel mcelog format.> 4). Have you guys talked with Len Brown (the ACPI maintainer) > about some of this code and how to make it fit within his > goals?Not yet, but will. Thanks, Jinsong> > Thanks! >> >> Basically it rebased from Jeremy''s pvops (2.6.32), to Konrad''s pvops >> (3.1.0, branch "devel/acpi-cpufreq.v3") patch 1: for mcelog logic >> patch 2: for dom0 vmce >> patch 3: for cpu sysfs and cpu online/offline >> patch 4: for cpu hotadd >> patch 5: for memory hotadd prepare work >> patch 6: for memory hotadd >> patch 7: for cpu hotadd prepare work >> patch 8: complement cpu hotadd lacked at Konrad''s tree >> patch 9: update PM logic >> patch 10: upate cpu online/offline logic >> >> >> Thanks, >> Jinsong >> _______________________________________________ >> Xen-devel mailing list >> Xen-devel@lists.xensource.com >> http://lists.xensource.com/xen-devel
Liu, Jinsong wrote:> Konrad Rzeszutek Wilk wrote: >> On Thu, Dec 22, 2011 at 09:57:55AM +0000, Liu, Jinsong wrote: >>> Recently we did dom0 Xen RAS rebase work. >> >> Hey Liu, >> >> Thank you for posting this. Couple of questions: >> 1). How does one test these patches? > > These patches include mce, cpu online/offline, cpu hotadd, memory > hotadd. > Testing method of these features are > 1. mce: a normal way is to inject error through EINJ user interface, > and getting mcelog at dom0 side > 2. cpu online/offline: by echoing sysfs/.../xen_pcpu user interface, > and getting result at sysfs/.../xen_pcpu > 3. cpu hotadd: Intel has some new server, by pressing a button can > add cpu. We can verify cpu hotadd by sysfs/.../xen_pcpu interface > 4. memory hotadd: also by pressing a button to add memory, and can > verify memory hotadd by ''xl info'' > >> 2). Please CC LKML and also the ACPI maintainer - since a majority >> of this also touch their tree. And the memory hotadd maintainer >> as well. > > OK, cc > LKML > ACPI maintainer: len.brown@intel.com > memory hotadd maintianer: I''m not sure who is maintainer, but I will > cc yakui.zhao@intel.com and len.brown@intel.comWill also cc Tony Luck and Andi Kleen, mce maintainers.> > I''ll re-send the patches later, cc above people. > >> 3). Can any of this code (like the MCE) leverage the existing >> generic code? > > Basically no, for mce, xen implement mce logic at hypervisor, > at dom0 side, it only add mcelog interface to get log from hypervisor > and transfer to kernel mcelog format. > >> 4). Have you guys talked with Len Brown (the ACPI maintainer) >> about some of this code and how to make it fit within his >> goals? > > Not yet, but will. > > Thanks, > Jinsong > >> >> Thanks! >>> >>> Basically it rebased from Jeremy''s pvops (2.6.32), to Konrad''s pvops >>> (3.1.0, branch "devel/acpi-cpufreq.v3") patch 1: for mcelog logic >>> patch 2: for dom0 vmce patch 3: for cpu sysfs and cpu online/offline >>> patch 4: for cpu hotadd >>> patch 5: for memory hotadd prepare work >>> patch 6: for memory hotadd >>> patch 7: for cpu hotadd prepare work >>> patch 8: complement cpu hotadd lacked at Konrad''s tree >>> patch 9: update PM logic >>> patch 10: upate cpu online/offline logic >>> >>> >>> Thanks, >>> Jinsong >>> _______________________________________________ >>> Xen-devel mailing list >>> Xen-devel@lists.xensource.com >>> http://lists.xensource.com/xen-devel