[PATCH] docs/system/arm/virt.rst: Add note on CPU features off by default

[PATCH] docs/system/arm/virt.rst: Add note on CPU features off by default

[Gustavo Romero]

Add a note on CPU features off by default on `virt` machines. These CPU
features will remain off even if `-cpu max` is given so the user must
turn on them explicitly on `virt` machines.

Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
---
 docs/system/arm/virt.rst | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/docs/system/arm/virt.rst b/docs/system/arm/virt.rst
index c245c52b7a..0e9ad526b4 100644
--- a/docs/system/arm/virt.rst
+++ b/docs/system/arm/virt.rst
@@ -69,6 +69,17 @@ Supported guest CPU types:
 Note that the default is ``cortex-a15``, so for an AArch64 guest you must
 specify a CPU type.
 
+Also, please note that passing ``max`` CPU (i.e. ``-cpu max``) won't
+enable all the CPU features for a given ``virt`` machine. Some CPU
+features are ``off`` by default (e.g. MTE) -- see machine-specific list
+below or check them for a given machine passing ``help``. For instance,
+for the ``virt`` machine pass ``-machine virt,help`` to QEMU.
+
+For example, because MTE CPU feature is ``off`` by default it's
+necessary to turn  MTE ``on`` explicitly to have this CPU feature
+available on the ``virt`` machine, even when ``max`` CPU is specified:
+``-machine virt,mte=on -cpu max`` .
+
 Graphics output is available, but unlike the x86 PC machine types
 there is no default display device enabled: you should select one from
 the Display devices section of "-device help". The recommended option
-- 
2.34.1

Re: [PATCH] docs/system/arm/virt.rst: Add note on CPU features off by default

[Peter Maydell]

On Mon, 22 Jan 2024 at 13:43, Gustavo Romero <gustavo.romero@linaro.org> wrote:
>
> Add a note on CPU features off by default on `virt` machines. These CPU
> features will remain off even if `-cpu max` is given so the user must
> turn on them explicitly on `virt` machines.
>
> Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>

Hi -- it looks like you forgot to cc qemu-devel on this;
could you do that on the next version, please? (All our
automated patch tooling looks at qemu-devel only. The
idea is that for other mailing lists you cc the other
list plus qemu-devel.)

> --- a/docs/system/arm/virt.rst
> +++ b/docs/system/arm/virt.rst
> @@ -69,6 +69,17 @@ Supported guest CPU types:
>  Note that the default is ``cortex-a15``, so for an AArch64 guest you must
>  specify a CPU type.
>
> +Also, please note that passing ``max`` CPU (i.e. ``-cpu max``) won't
> +enable all the CPU features for a given ``virt`` machine. Some CPU
> +features are ``off`` by default (e.g. MTE) -- see machine-specific list
> +below or check them for a given machine passing ``help``. For instance,
> +for the ``virt`` machine pass ``-machine virt,help`` to QEMU.
> +
> +For example, because MTE CPU feature is ``off`` by default it's
> +necessary to turn  MTE ``on`` explicitly to have this CPU feature
> +available on the ``virt`` machine, even when ``max`` CPU is specified:
> +``-machine virt,mte=on -cpu max`` .

I think I would phrase this as:

Where a CPU architectural feature requires support in
both the CPU itself and in the wider system, it may
not be enabled by default, but instead require a machine
option to enable it. For instance MTE support must be
enabled with ``-machine virt,mte=on``, as well as by
selecting an MTE-capable CPU with the ``-cpu`` option.
See the list of machine-specific options below.

> +
>  Graphics output is available, but unlike the x86 PC machine types
>  there is no default display device enabled: you should select one from
>  the Display devices section of "-device help". The recommended option

thanks
-- PMM

Re: [PATCH] docs/system/arm/virt.rst: Add note on CPU features off by default

[Gustavo Romero]

On 1/22/24 1:31 PM, Peter Maydell wrote:
> On Mon, 22 Jan 2024 at 13:43, Gustavo Romero <gustavo.romero@linaro.org> wrote:
>>
>> Add a note on CPU features off by default on `virt` machines. These CPU
>> features will remain off even if `-cpu max` is given so the user must
>> turn on them explicitly on `virt` machines.
>>
>> Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
> 
> Hi -- it looks like you forgot to cc qemu-devel on this;
> could you do that on the next version, please? (All our
> automated patch tooling looks at qemu-devel only. The
> idea is that for other mailing lists you cc the other
> list plus qemu-devel.)

Hi Peter! Oh, I missed it indeed :-) Did that for the v2.
Thanks!


>> --- a/docs/system/arm/virt.rst
>> +++ b/docs/system/arm/virt.rst
>> @@ -69,6 +69,17 @@ Supported guest CPU types:
>>   Note that the default is ``cortex-a15``, so for an AArch64 guest you must
>>   specify a CPU type.
>>
>> +Also, please note that passing ``max`` CPU (i.e. ``-cpu max``) won't
>> +enable all the CPU features for a given ``virt`` machine. Some CPU
>> +features are ``off`` by default (e.g. MTE) -- see machine-specific list
>> +below or check them for a given machine passing ``help``. For instance,
>> +for the ``virt`` machine pass ``-machine virt,help`` to QEMU.
>> +
>> +For example, because MTE CPU feature is ``off`` by default it's
>> +necessary to turn  MTE ``on`` explicitly to have this CPU feature
>> +available on the ``virt`` machine, even when ``max`` CPU is specified:
>> +``-machine virt,mte=on -cpu max`` .
> 
> I think I would phrase this as:
> 
> Where a CPU architectural feature requires support in
> both the CPU itself and in the wider system, it may
> not be enabled by default, but instead require a machine
> option to enable it. For instance MTE support must be
> enabled with ``-machine virt,mte=on``, as well as by
> selecting an MTE-capable CPU with the ``-cpu`` option.
> See the list of machine-specific options below.

Got it. Sure, that version is much better due to the
explanation of why some feature are off. I kept the initial
part about -cpu max not being sufficient to select the feature
because I believe some users might get confused by selecting the
max cpu and not getting all the CPU features, like it happened
to me :)


Thanks!


Cheers,
Gustavo