kvmarm.lists.cs.columbia.edu archive mirror
 help / color / mirror / Atom feed
From: Auger Eric <eric.auger@redhat.com>
To: Alexandru Elisei <alexandru.elisei@arm.com>,
	kvm@vger.kernel.org, kvmarm@lists.cs.columbia.edu
Cc: pbonzini@redhat.com
Subject: Re: [kvm-unit-tests RFC PATCH v2 0/5] arm64: Statistical Profiling Extension Tests
Date: Fri, 30 Oct 2020 19:17:41 +0100	[thread overview]
Message-ID: <88f5068a-7a1c-4870-ebc4-e2c45616e905@redhat.com> (raw)
In-Reply-To: <20201027171944.13933-1-alexandru.elisei@arm.com>

Hi Alexandru,

[+ Drew]

On 10/27/20 6:19 PM, Alexandru Elisei wrote:
> This series implements two basic tests for KVM SPE: one that checks that
> the reported features match what is specified in the architecture,
> implemented in patch #3 ("arm64: spe: Add introspection test"), and another
> test that checks that the buffer management interrupt is asserted
> correctly, implemented in patch #5 ("am64: spe: Add buffer test"). The rest
> of the patches are either preparatory patches, or a fix, in the case of
> patch #2 ("lib/{bitops,alloc_page}.h: Add missing headers").
> 
> This series builds on Eric's initial version [1], to which I've added the
> buffer tests that I used while developing SPE support for KVM.

As you respin my series, with my prior agreement, I expected to find
most of the code I wrote, obviously with some potential fixes and
adaptations to fit your needs for additional tests.

However, in this v2, two significant v1 patches have disappeared:

1) [3/4] spe: Add profiling buffer test (170 LOC diffstat)
2) [4/4] spe: Test Profiling Buffer Events (150 LOC diffstat)

They were actually the crux of the original series (the introspection
test being required as a prerequisite but not testing much really ;-).

1) consists in a "spe-buffer" test starting the profiling on a mastered
sequence of instructions (as done for PMU event counters). It introduces
the infra to start the profiling, prepare SPE reset config, the macro
definitions, start/stop/drain, the code under profiling and basically
checks that the buffer was effectively written. We also check we do not
get any spurious event as it is not expected.

=> This test has disappeared and the infra now is diluted in
[kvm-unit-tests RFC PATCH v2 5/5] am64: spe: Add buffer test. However no
credit is given to my work as my S-o-b has disappeared.

2) consists in a "spe-events" test checking we effectively get the
buffer full event when duly expected. This introduces the infra to
handle interrupts, check the occurence of events by analyzing the
syndrome registers, compare occurences against expected ones. This
largely mimics what we have with PMU tests.

=> This test is part of [kvm-unit-tests RFC PATCH v2 5/5], relying on a
different stop condition, and again the infra is diluted in the same
patch, with large arbitrary changes, without any credit given to my
work. Those changes may explain why you removed my S-o-b but given the
anteriority of my series, this does not look normal to me, in a
community environment.

As discussed privately, this gives me the impression that those two
patches were totally ignored while respinning.

Please could you restructure the series at least to keep the buffer-full
test + infra separate from the new tests and reset a collaborative S-o-b.

Then if you think there are issues wrt the 1st test, "spe-buffer", not
included in this series, please let's discuss and fix/improve but not
simply trash it as is (in an everyone growing spirit).

An alternative is I can take back the ownership of my series and push it
upstream in a standard way. Then either you rebase your new tests on top
of it or I will be happy to do it for you after discussions on the
technical comments.

Thanks

Eric

> 
> KVM SPE support is current in RFC phase, hence why this series is also sent
> as RFC. The KVM patches needed for the tests to work can be found at [2].
> Userspace support is also necessary, which I've implemented for kvmtool;
> this can be found at [3]. This series is also available in a repo [4] to make
> testing easier.
> 
> [1] https://www.spinics.net/lists/kvm/msg223792.html
> [2] https://gitlab.arm.com/linux-arm/linux-ae/-/tree/kvm-spe-v3
> [3] https://gitlab.arm.com/linux-arm/kvmtool-ae/-/tree/kvm-spe-v3
> [4] https://gitlab.arm.com/linux-arm/kvm-unit-tests-ae/-/tree/kvm-spe-v2
> 
> Alexandru Elisei (3):
>   lib/{bitops,alloc_page}.h: Add missing headers
>   lib: arm/arm64: Add function to unmap a page
>   am64: spe: Add buffer test
> 
> Eric Auger (2):
>   arm64: Move get_id_aa64dfr0() in processor.h
>   arm64: spe: Add introspection test
> 
>  arm/Makefile.arm64        |   1 +
>  lib/arm/asm/mmu-api.h     |   1 +
>  lib/arm64/asm/processor.h |   5 +
>  lib/alloc_page.h          |   2 +
>  lib/bitops.h              |   2 +
>  lib/libcflat.h            |   1 +
>  lib/arm/mmu.c             |  32 +++
>  arm/pmu.c                 |   1 -
>  arm/spe.c                 | 506 ++++++++++++++++++++++++++++++++++++++
>  arm/unittests.cfg         |  15 ++
>  10 files changed, 565 insertions(+), 1 deletion(-)
>  create mode 100644 arm/spe.c
> 

_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm

  parent reply	other threads:[~2020-10-30 18:17 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-27 17:19 [kvm-unit-tests RFC PATCH v2 0/5] arm64: Statistical Profiling Extension Tests Alexandru Elisei
2020-10-27 17:19 ` [kvm-unit-tests RFC PATCH v2 1/5] arm64: Move get_id_aa64dfr0() in processor.h Alexandru Elisei
2020-10-27 17:19 ` [kvm-unit-tests RFC PATCH v2 2/5] lib/{bitops, alloc_page}.h: Add missing headers Alexandru Elisei
2020-10-30 15:29   ` Auger Eric
2020-10-30 15:58     ` Alexandru Elisei
2020-10-27 17:19 ` [kvm-unit-tests RFC PATCH v2 3/5] arm64: spe: Add introspection test Alexandru Elisei
2020-10-30 15:29   ` Auger Eric
2020-10-30 15:59     ` Alexandru Elisei
2020-10-30 17:09       ` Auger Eric
2020-10-30 17:50         ` Alexandru Elisei
2020-10-30 17:52           ` Auger Eric
2020-11-02 10:47           ` Andrew Jones
2020-10-27 17:19 ` [kvm-unit-tests RFC PATCH v2 4/5] lib: arm/arm64: Add function to unmap a page Alexandru Elisei
2020-10-30 15:46   ` Auger Eric
2020-10-30 16:00     ` Alexandru Elisei
2020-10-27 17:19 ` [kvm-unit-tests RFC PATCH v2 5/5] am64: spe: Add buffer test Alexandru Elisei
2020-10-30 17:59   ` Auger Eric
2020-10-30 16:02 ` [kvm-unit-tests RFC PATCH v2 0/5] arm64: Statistical Profiling Extension Tests Alexandru Elisei
2020-10-30 18:17 ` Auger Eric [this message]
2020-10-30 22:31   ` Alexandru Elisei
2020-11-02 12:19     ` Auger Eric

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=88f5068a-7a1c-4870-ebc4-e2c45616e905@redhat.com \
    --to=eric.auger@redhat.com \
    --cc=alexandru.elisei@arm.com \
    --cc=kvm@vger.kernel.org \
    --cc=kvmarm@lists.cs.columbia.edu \
    --cc=pbonzini@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).