diff mbox series

[5/6] CHANGELOG.md: Add entries for emulation

Message ID 20210324164407.302062-5-george.dunlap@citrix.com (mailing list archive)
State Superseded
Headers show
Series [1/6] Intel Processor Trace Support: Add CHANGELOG.md and SUPPORT.md entries | expand

Commit Message

George Dunlap March 24, 2021, 4:44 p.m. UTC
Signed-off-by: George Dunlap <george.dunlap@citrix.com>
---
CC: Jan Beulich <jbeulich@suse.com>
CC: Ian Jackson <ian.jackson@citrix.com>
---
 CHANGELOG.md | 2 ++
 1 file changed, 2 insertions(+)

Comments

Jan Beulich March 25, 2021, 7:57 a.m. UTC | #1
On 24.03.2021 17:44, George Dunlap wrote:
> --- a/CHANGELOG.md
> +++ b/CHANGELOG.md
> @@ -14,6 +14,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
>   - Named PCI devices for xl/libxl
>   - Support for zstd-compressed dom0 kernels
>   - Library improvements from NetBSD ports upstreamed
> + - x86_emulate: Support AVX-VNNI instructions
> + - x86_emulate: Expanded testing for several instruction classes

TBH both seem too minor to me to be mentioned here. If I was to pick
just one, I'd keep the former and drop the latter.

Jan
George Dunlap March 25, 2021, 9:04 a.m. UTC | #2
> On Mar 25, 2021, at 7:57 AM, Jan Beulich <jbeulich@suse.com> wrote:
> 
> On 24.03.2021 17:44, George Dunlap wrote:
>> --- a/CHANGELOG.md
>> +++ b/CHANGELOG.md
>> @@ -14,6 +14,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
>>  - Named PCI devices for xl/libxl
>>  - Support for zstd-compressed dom0 kernels
>>  - Library improvements from NetBSD ports upstreamed
>> + - x86_emulate: Support AVX-VNNI instructions
>> + - x86_emulate: Expanded testing for several instruction classes
> 
> TBH both seem too minor to me to be mentioned here. If I was to pick
> just one, I'd keep the former and drop the latter.

It’s always quite difficult in these releases to help express to people exactly what’s happened.  We’ve got over 1000 changesets — what have we been doing?  If my star chart is correct, the latter represents nearly 1700 extra lines of codes in tools/tests/x86_emulator/predicates.c.  It’s actually a reasonable chunk of code churn, which I think is worth highlighting.

Re the former, I think letting someone know that a new instruction is emulated is worthwhile.

Ian is the maintainer, so ultimately I think it’s his call?

 -George
Andrew Cooper March 25, 2021, 9:37 a.m. UTC | #3
On 25/03/2021 09:04, George Dunlap wrote:
>
>> On Mar 25, 2021, at 7:57 AM, Jan Beulich <jbeulich@suse.com> wrote:
>>
>> On 24.03.2021 17:44, George Dunlap wrote:
>>> --- a/CHANGELOG.md
>>> +++ b/CHANGELOG.md
>>> @@ -14,6 +14,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
>>>  - Named PCI devices for xl/libxl
>>>  - Support for zstd-compressed dom0 kernels
>>>  - Library improvements from NetBSD ports upstreamed
>>> + - x86_emulate: Support AVX-VNNI instructions
>>> + - x86_emulate: Expanded testing for several instruction classes
>> TBH both seem too minor to me to be mentioned here. If I was to pick
>> just one, I'd keep the former and drop the latter.
> It’s always quite difficult in these releases to help express to people exactly what’s happened.  We’ve got over 1000 changesets — what have we been doing?  If my star chart is correct, the latter represents nearly 1700 extra lines of codes in tools/tests/x86_emulator/predicates.c.  It’s actually a reasonable chunk of code churn, which I think is worth highlighting.

I agree with Jan.  New concrete instruction groups are useful for an end
user to read.  "We refactored some internal of a test harness" isn't,
especially when it would most likely be repeated every release.

I'd drop the latter line and just keep the former.

~Andrew
Jan Beulich March 25, 2021, 1:58 p.m. UTC | #4
On 25.03.2021 10:37, Andrew Cooper wrote:
> On 25/03/2021 09:04, George Dunlap wrote:
>>> On Mar 25, 2021, at 7:57 AM, Jan Beulich <jbeulich@suse.com> wrote:
>>> On 24.03.2021 17:44, George Dunlap wrote:
>>>> --- a/CHANGELOG.md
>>>> +++ b/CHANGELOG.md
>>>> @@ -14,6 +14,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
>>>>  - Named PCI devices for xl/libxl
>>>>  - Support for zstd-compressed dom0 kernels
>>>>  - Library improvements from NetBSD ports upstreamed
>>>> + - x86_emulate: Support AVX-VNNI instructions
>>>> + - x86_emulate: Expanded testing for several instruction classes
>>> TBH both seem too minor to me to be mentioned here. If I was to pick
>>> just one, I'd keep the former and drop the latter.
>> It’s always quite difficult in these releases to help express to people exactly what’s happened.  We’ve got over 1000 changesets — what have we been doing?  If my star chart is correct, the latter represents nearly 1700 extra lines of codes in tools/tests/x86_emulator/predicates.c.  It’s actually a reasonable chunk of code churn, which I think is worth highlighting.
> 
> I agree with Jan.  New concrete instruction groups are useful for an end
> user to read.  "We refactored some internal of a test harness" isn't,
> especially when it would most likely be repeated every release.
> 
> I'd drop the latter line and just keep the former.

And then perhaps also not focus on the emulator, but merely on the
fact that guests now can use that ISA extension (for which the
emulator change is a [the only] prereq)?

Jan
George Dunlap March 29, 2021, 4:06 p.m. UTC | #5
> On Mar 25, 2021, at 1:58 PM, Jan Beulich <JBeulich@suse.com> wrote:
> 
> On 25.03.2021 10:37, Andrew Cooper wrote:
>> On 25/03/2021 09:04, George Dunlap wrote:
>>>> On Mar 25, 2021, at 7:57 AM, Jan Beulich <jbeulich@suse.com> wrote:
>>>> On 24.03.2021 17:44, George Dunlap wrote:
>>>>> --- a/CHANGELOG.md
>>>>> +++ b/CHANGELOG.md
>>>>> @@ -14,6 +14,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
>>>>> - Named PCI devices for xl/libxl
>>>>> - Support for zstd-compressed dom0 kernels
>>>>> - Library improvements from NetBSD ports upstreamed
>>>>> + - x86_emulate: Support AVX-VNNI instructions
>>>>> + - x86_emulate: Expanded testing for several instruction classes
>>>> TBH both seem too minor to me to be mentioned here. If I was to pick
>>>> just one, I'd keep the former and drop the latter.
>>> It’s always quite difficult in these releases to help express to people exactly what’s happened.  We’ve got over 1000 changesets — what have we been doing?  If my star chart is correct, the latter represents nearly 1700 extra lines of codes in tools/tests/x86_emulator/predicates.c.  It’s actually a reasonable chunk of code churn, which I think is worth highlighting.
>> 
>> I agree with Jan.  New concrete instruction groups are useful for an end
>> user to read.  "We refactored some internal of a test harness" isn't,
>> especially when it would most likely be repeated every release.
>> 
>> I'd drop the latter line and just keep the former.
> 
> And then perhaps also not focus on the emulator, but merely on the
> fact that guests now can use that ISA extension (for which the
> emulator change is a [the only] prereq)?

Would you mind giving a concrete suggestion?

Thanks,
 -George
Jan Beulich March 29, 2021, 4:24 p.m. UTC | #6
On 29.03.2021 18:06, George Dunlap wrote:
> 
> 
>> On Mar 25, 2021, at 1:58 PM, Jan Beulich <JBeulich@suse.com> wrote:
>>
>> On 25.03.2021 10:37, Andrew Cooper wrote:
>>> On 25/03/2021 09:04, George Dunlap wrote:
>>>>> On Mar 25, 2021, at 7:57 AM, Jan Beulich <jbeulich@suse.com> wrote:
>>>>> On 24.03.2021 17:44, George Dunlap wrote:
>>>>>> --- a/CHANGELOG.md
>>>>>> +++ b/CHANGELOG.md
>>>>>> @@ -14,6 +14,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
>>>>>> - Named PCI devices for xl/libxl
>>>>>> - Support for zstd-compressed dom0 kernels
>>>>>> - Library improvements from NetBSD ports upstreamed
>>>>>> + - x86_emulate: Support AVX-VNNI instructions
>>>>>> + - x86_emulate: Expanded testing for several instruction classes
>>>>> TBH both seem too minor to me to be mentioned here. If I was to pick
>>>>> just one, I'd keep the former and drop the latter.
>>>> It’s always quite difficult in these releases to help express to people exactly what’s happened.  We’ve got over 1000 changesets — what have we been doing?  If my star chart is correct, the latter represents nearly 1700 extra lines of codes in tools/tests/x86_emulator/predicates.c.  It’s actually a reasonable chunk of code churn, which I think is worth highlighting.
>>>
>>> I agree with Jan.  New concrete instruction groups are useful for an end
>>> user to read.  "We refactored some internal of a test harness" isn't,
>>> especially when it would most likely be repeated every release.
>>>
>>> I'd drop the latter line and just keep the former.
>>
>> And then perhaps also not focus on the emulator, but merely on the
>> fact that guests now can use that ISA extension (for which the
>> emulator change is a [the only] prereq)?
> 
> Would you mind giving a concrete suggestion?

x86: Allow domains to use AVX-VNNI instructions

Jan
diff mbox series

Patch

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 3f4eed1d98..df6c638152 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -14,6 +14,8 @@  The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
  - Named PCI devices for xl/libxl
  - Support for zstd-compressed dom0 kernels
  - Library improvements from NetBSD ports upstreamed
+ - x86_emulate: Support AVX-VNNI instructions
+ - x86_emulate: Expanded testing for several instruction classes
 
 ## Removed / support downgraded