nvdimm: support nvdimm on arm64 kernel by Pennyzct · Pull Request #377 · kata-containers/packaging

Pennyzct · 2019-03-05T07:41:53Z

let's open nvdimm-related kernel config parameters on arm64, such as
CONFIG_ACPI_NFIT, etc. and we also need to backport patch kvm:arm64:Dynamic IPA and 52bit IPA and related dependency into v4.19.X to fully support nvdimm from guest kernel.
Former patch has already been merged into v4.20.X.

Fixes: #376

Signed-off-by: Penny Zheng penny.zheng@arm.com

Pennyzct · 2019-03-05T08:28:31Z

Hi~ @grahamwhaley @chavafg Is there any way to trigger ARM CI here? ;)

grahamwhaley

A couple of questions, but generally
lgtm

grahamwhaley · 2019-03-05T10:52:06Z

kernel/configs/arm64_kata_kvm_4.19.x

 # CONFIG_OF_OVERLAY is not set
 # CONFIG_PARPORT is not set
+CONFIG_PNP=y
+CONFIG_PNP_DEBUG_MESSAGES=y


not sure if you set this deliberately? Payoff of size vs debuggability...

disable it. ;)

grahamwhaley · 2019-03-05T10:52:52Z

kernel/configs/arm64_kata_kvm_4.19.x

+# CONFIG_THERMAL_GOV_FAIR_SHARE is not set
+# CONFIG_THERMAL_GOV_STEP_WISE is not set
+# CONFIG_THERMAL_GOV_BANG_BANG is not set
+CONFIG_THERMAL_GOV_USER_SPACE=y


On purpose? Maybe you gained some default =y items during a new config generation. If you have not already, you might want to scrutinize the diff for any stray new additions you didn't intend.

noooo, my bad, I just quickly compared the new diff with x86 config(x86_64_kata_kvm_4.19.X) to make two configs as same as possible, not scrutinizing what these diff represent.
I will turn off the CONFIG_THERMAL, since VM may not be involved in thermal managment. ;)

grahamwhaley · 2019-03-05T10:55:07Z

@Pennyzct - no, no way to trigger the ARM CI here - in fact, we tried to add the Power8 CI here last week iirc (or maybe that was osbuilder), and it didn't work as it went to pull OBS images that did not exist I think.
But, @chavafg , might there be a way to trigger a PR on another repo that has this PR as a dependency, to get it pulled in? I'm not sure..

grahamwhaley · 2019-03-05T10:56:35Z

ooh, 'leaky pods' on the 18.04 CI....
@nitkon - did you get any further on that, or are you now needing direction/assistance?

bash sanity/check_sanity.sh
ERROR: 3 pods left and found at /var/lib/vc/sbs
Makefile:45: recipe for target 'functional' failed
make: *** [functional] Error 1
Build step 'Execute shell' marked build as failure

nitkon · 2019-03-05T11:46:58Z

Yes, @jodh-intel suggested to get a kernel backtrace of such VM Panics. We can probably get that by enabling serial console and then redirecting the guest serial to some host file.
However reproducing this issue locally would be the first challenge as running these tests individually don't result in panic(In my testing atleast)

jodh-intel · 2019-03-05T14:25:17Z

@nitkon - you should get something in the proxy log without the need for a serial console. Certainly oops's are logged an I've seen panics when the rootfs is not mountable.

Pennyzct · 2019-03-06T07:23:41Z

Hi~ @grahamwhaley I had another relevant PR in runtime repo, and i will try to make this pr as dependency to see if ARM CI on runtime repo could test these two PRs together. ;)

Pennyzct · 2019-03-07T05:38:10Z

Hi~@grahamwhaley I have addressed the comments and done the fix, ptal. ;)
Also, I have already done the re-base and update the kata_config_version.
For now, I'm struggling to add this PR as dependency for another related PR runtime/#1323.
So I could trigger the ARM CI under runtime/ repo to test this two PRs together. I definitely once saw how to make this happen, like dependency: packaging/#377, I just couldn't find the reference, would anyone like to lend a hand here? ;) @jodh-intel @chavafg

jodh-intel · 2019-03-07T09:23:59Z

Hi @Pennyzct - It's actually documented here:

https://github.com/kata-containers/tests#breaking-compatibility

And here's an example of a PR that used it:

virtcontainers: set private propagation in rootfs runtime#980

Original guest image was reprensented as block device in qemu-aarch64, and it will bring up write lock error when running multiple containers. Thanks to the new expanded IPA_SIZE feature in kernel 4.20 and Eric Auger's related patch set in qemu(which are still under upstream review), we could fully support nvdimm on arm64. Depends-on: github.com/kata-containers/packaging#377 Fixes: kata-containers#843 Signed-off-by: Penny Zheng <penny.zheng@arm.com>

Since we overrided the func appendImage for aarch64, we should also provide related unit test. Depends-on: github.com/kata-containers/packaging#377 Fixes: kata-containers#843 Signed-off-by: Penny Zheng <penny.zheng@arm.com>

Pennyzct · 2019-03-07T10:12:20Z

Hi~ @jodh-intel tons of thanks!!!
I have already add dependencies in PR runtime/#1323, let's see how it works~~~~;)

amshinde · 2019-03-11T17:27:01Z

/test

amshinde · 2019-03-11T18:42:12Z

Still seeing failures:

Kernel directory has changes check kernel/kata_config_version changed
ERROR: Please bump version in kernel/kata_config_version

In spite of the version being changed. Retriggering one more time...

/test

let's open nvdimm-related kernel config parameters on arm64, such as CONFIG_ACPI_NFIT, etc. and we also need to backport patch 'kvm:arm64:Dynamic IPA and 52bit IPA'(https://patchwork.kernel.org/cover/10616271/) and related dependency into v4.19.X to fully support nvdimm from guest kernel. Former patch has already been merged into v4.20.X. Fixes: kata-containers#376 Signed-off-by: Penny Zheng <penny.zheng@arm.com>

Pennyzct · 2019-03-12T07:09:43Z

Hi~ @amshinde re-based and updated the version number. ;)

jodh-intel · 2019-03-12T09:13:29Z

Thanks @Pennyzct. Let's retry the CI...

/retest

Pennyzct mentioned this pull request Mar 5, 2019

arm64: support NVDIMM kata-containers/runtime#1323

Merged

jodh-intel approved these changes Mar 5, 2019

View reviewed changes

grahamwhaley approved these changes Mar 5, 2019

View reviewed changes

devimc approved these changes Mar 5, 2019

View reviewed changes

Pennyzct force-pushed the nvdimm branch from f07b83f to 2b5bdfd Compare March 6, 2019 06:59

Pennyzct force-pushed the nvdimm branch 2 times, most recently from efa99dd to 8570097 Compare March 7, 2019 03:10

Pennyzct force-pushed the nvdimm branch from 8570097 to 5dd1d7c Compare March 12, 2019 06:47

jodh-intel merged commit 9a8553f into kata-containers:master Mar 12, 2019

Conversation

Pennyzct commented Mar 5, 2019

Uh oh!

Pennyzct commented Mar 5, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

grahamwhaley left a comment

Choose a reason for hiding this comment

Uh oh!

grahamwhaley Mar 5, 2019

Choose a reason for hiding this comment

Uh oh!

Pennyzct Mar 6, 2019

Choose a reason for hiding this comment

Uh oh!

grahamwhaley Mar 5, 2019

Choose a reason for hiding this comment

Uh oh!

Pennyzct Mar 6, 2019

Choose a reason for hiding this comment

Uh oh!

grahamwhaley commented Mar 5, 2019

Uh oh!

grahamwhaley commented Mar 5, 2019

Uh oh!

nitkon commented Mar 5, 2019

Uh oh!

jodh-intel commented Mar 5, 2019

Uh oh!

Pennyzct commented Mar 6, 2019

Uh oh!

Pennyzct commented Mar 7, 2019

Uh oh!

jodh-intel commented Mar 7, 2019

Uh oh!

Pennyzct commented Mar 7, 2019

Uh oh!

amshinde commented Mar 11, 2019

Uh oh!

amshinde commented Mar 11, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Pennyzct commented Mar 12, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jodh-intel commented Mar 12, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Pennyzct commented Mar 5, 2019 •

edited

Loading

amshinde commented Mar 11, 2019 •

edited

Loading

Pennyzct commented Mar 12, 2019 •

edited

Loading