[v2,05/19] riscv: add ISA extension parsing for vector crypto extensions

Message ID 20231017131456.2053396-6-cleger@rivosinc.com
State New
Headers
Series riscv: report more ISA extensions through hwprobe |

Commit Message

Clément Léger Oct. 17, 2023, 1:14 p.m. UTC
  Add parsing of some Zv* vector crypto ISA extensions that are mentioned
in "RISC-V Cryptography Extensions Volume II" [1]. These ISA extensions
are the following:

- Zvbb: Vector Basic Bit-manipulation
- Zvbc: Vector Carryless Multiplication
- Zvkb: Vector Cryptography Bit-manipulation
- Zvkg: Vector GCM/GMAC.
- Zvkned: NIST Suite: Vector AES Block Cipher
- Zvknh[ab]: NIST Suite: Vector SHA-2 Secure Hash
- Zvksed: ShangMi Suite: SM4 Block Cipher
- Zvksh: ShangMi Suite: SM3 Secure Hash
- Zvkn: NIST Algorithm Suite
- Zvknc: NIST Algorithm Suite with carryless multiply
- Zvkng: NIST Algorithm Suite with GCM.
- Zvks: ShangMi Algorithm Suite
- Zvksc: ShangMi Algorithm Suite with carryless multiplication
- Zvksg: ShangMi Algorithm Suite with GCM.
- Zvkt: Vector Data-Independent Execution Latency.

Link: https://drive.google.com/file/d/1gb9OLH-DhbCgWp7VwpPOVrrY6f3oSJLL/view [1]
Signed-off-by: Clément Léger <cleger@rivosinc.com>
---
 arch/riscv/include/asm/hwcap.h | 10 ++++++
 arch/riscv/kernel/cpufeature.c | 56 ++++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+)
  

Comments

Jerry Shih Oct. 18, 2023, 1:45 a.m. UTC | #1
On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> 	__RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
> 	__RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
> 	__RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
> +	__RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
> +	__RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
> +	__RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),

The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.

+	__RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
+	__RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVKB),
+	__RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),

or

+	__RISCV_ISA_EXT_BUNDLE(zvbb, riscv_zvbb_bundled_exts),
+	__RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),

[1]
https://github.com/riscv/riscv-crypto/blob/main/doc/vector/riscv-crypto-vector-zvkb.adoc

-Jerry
  
Clément Léger Oct. 18, 2023, 12:52 p.m. UTC | #2
On 18/10/2023 03:45, Jerry Shih wrote:
> On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
>> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
>> 	__RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
>> 	__RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
>> 	__RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
>> +	__RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
>> +	__RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
>> +	__RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
> 
> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.

Hi Jerry,

Thanks for catching this, I think some other extensions will fall in
this category as well then (Zvknha/Zvknhb). I will verify that.

Clément

> 
> +	__RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
> +	__RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVKB),
> +	__RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
> 
> or
> 
> +	__RISCV_ISA_EXT_BUNDLE(zvbb, riscv_zvbb_bundled_exts),
> +	__RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
> 
> [1]
> https://github.com/riscv/riscv-crypto/blob/main/doc/vector/riscv-crypto-vector-zvkb.adoc
> 
> -Jerry
  
Evan Green Oct. 18, 2023, 5:26 p.m. UTC | #3
On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
>
>
>
> On 18/10/2023 03:45, Jerry Shih wrote:
> > On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
> >> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> >>      __RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
> >>      __RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
> >>      __RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
> >> +    __RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
> >> +    __RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
> >> +    __RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
> >
> > The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
>
> Hi Jerry,
>
> Thanks for catching this, I think some other extensions will fall in
> this category as well then (Zvknha/Zvknhb). I will verify that.

The bundling mechanism works well when an extension is a pure lasso
around other extensions. We'd have to tweak that code if we wanted to
support cases like this, where the extension is a superset of others,
but also contains loose change not present anywhere else (and
therefore also needs to stand as a separate bit).

IMO, decomposing "pure" bundles makes sense since otherwise usermode
would have to query multiple distinct bitmaps that meant the same
thing (eg check the Zk bit, or maybe check the Zkn/Zkr/Zkt bits, or
maybe check the Zbkb/Zbkc... bits, and they're all equivalent). But
when an extension is a superset that also contains loose change, there
really aren't two equivalent bitmasks, each bit adds something new.

There's an argument to be made for still turning on the containing
extensions to cover for silly ISA strings (eg ISA strings that
advertise the superset but fail to advertise the containing
extensions). We can decide if we want to work that hard to cover
hypothetical broken ISA strings now, or wait until they show up.
Personally I would wait until something broken shows up. But others
may feel differently.

-Evan
  
Clément Léger Oct. 19, 2023, 9:35 a.m. UTC | #4
On 18/10/2023 19:26, Evan Green wrote:
> On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
>>
>>
>>
>> On 18/10/2023 03:45, Jerry Shih wrote:
>>> On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
>>>> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
>>>>      __RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
>>>>      __RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
>>>>      __RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
>>>> +    __RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
>>>> +    __RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
>>>> +    __RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
>>>
>>> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
>>
>> Hi Jerry,
>>
>> Thanks for catching this, I think some other extensions will fall in
>> this category as well then (Zvknha/Zvknhb). I will verify that.
> 
> The bundling mechanism works well when an extension is a pure lasso
> around other extensions. We'd have to tweak that code if we wanted to
> support cases like this, where the extension is a superset of others,
> but also contains loose change not present anywhere else (and
> therefore also needs to stand as a separate bit).

For Zvbb and Zvknhb, I used the following code:

static const unsigned int riscv_zvbb_bundled_exts[] = {
	RISCV_ISA_EXT_ZVKB,
	RISCV_ISA_EXT_ZVBB
};

static const unsigned int riscv_zvknhb_bundled_exts[] = {
	RISCV_ISA_EXT_ZVKNHA,
	RISCV_ISA_EXT_ZVKNHB
};

Which correctly results in both extension (superset + base set) being
enabled when only one is set. Is there something that I'm missing ?

> 
> IMO, decomposing "pure" bundles makes sense since otherwise usermode
> would have to query multiple distinct bitmaps that meant the same
> thing (eg check the Zk bit, or maybe check the Zkn/Zkr/Zkt bits, or
> maybe check the Zbkb/Zbkc... bits, and they're all equivalent). But
> when an extension is a superset that also contains loose change, there
> really aren't two equivalent bitmasks, each bit adds something new.

Agreed but if a system only report ZVBB for instance and the user wants
ZVKB, then it is clear that ZVKB should be reported as well I guess. So
in the end, it works much like "bundle" extension, just that the bundle
is actually a "real" ISA extension by itself.

Clément

> 
> There's an argument to be made for still turning on the containing
> extensions to cover for silly ISA strings (eg ISA strings that
> advertise the superset but fail to advertise the containing
> extensions). We can decide if we want to work that hard to cover
> hypothetical broken ISA strings now, or wait until they show up.
> Personally I would wait until something broken shows up. But others
> may feel differently.
> 
> -Evan
  
Conor Dooley Oct. 19, 2023, 3:33 p.m. UTC | #5
On Thu, Oct 19, 2023 at 11:35:59AM +0200, Clément Léger wrote:
> 
> 
> On 18/10/2023 19:26, Evan Green wrote:
> > On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
> >>
> >>
> >>
> >> On 18/10/2023 03:45, Jerry Shih wrote:
> >>> On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
> >>>> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> >>>>      __RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
> >>>>      __RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
> >>>>      __RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
> >>>> +    __RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
> >>>> +    __RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
> >>>> +    __RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
> >>>
> >>> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
> >>
> >> Hi Jerry,
> >>
> >> Thanks for catching this, I think some other extensions will fall in
> >> this category as well then (Zvknha/Zvknhb). I will verify that.
> > 
> > The bundling mechanism works well when an extension is a pure lasso
> > around other extensions. We'd have to tweak that code if we wanted to
> > support cases like this, where the extension is a superset of others,
> > but also contains loose change not present anywhere else (and
> > therefore also needs to stand as a separate bit).
> 
> For Zvbb and Zvknhb, I used the following code:
> 
> static const unsigned int riscv_zvbb_bundled_exts[] = {
> 	RISCV_ISA_EXT_ZVKB,
> 	RISCV_ISA_EXT_ZVBB
> };
> 
> static const unsigned int riscv_zvknhb_bundled_exts[] = {
> 	RISCV_ISA_EXT_ZVKNHA,
> 	RISCV_ISA_EXT_ZVKNHB
> };
> 
> Which correctly results in both extension (superset + base set) being
> enabled when only one is set. Is there something that I'm missing ?
> 
> > 
> > IMO, decomposing "pure" bundles makes sense since otherwise usermode
> > would have to query multiple distinct bitmaps that meant the same
> > thing (eg check the Zk bit, or maybe check the Zkn/Zkr/Zkt bits, or
> > maybe check the Zbkb/Zbkc... bits, and they're all equivalent). But
> > when an extension is a superset that also contains loose change, there
> > really aren't two equivalent bitmasks, each bit adds something new.
> 
> Agreed but if a system only report ZVBB for instance and the user wants
> ZVKB, then it is clear that ZVKB should be reported as well I guess. So
> in the end, it works much like "bundle" extension, just that the bundle
> is actually a "real" ISA extension by itself.
> 
> Clément
> 
> > 
> > There's an argument to be made for still turning on the containing
> > extensions to cover for silly ISA strings (eg ISA strings that
> > advertise the superset but fail to advertise the containing
> > extensions). We can decide if we want to work that hard to cover
> > hypothetical broken ISA strings now, or wait until they show up.
> > Personally I would wait until something broken shows up. But others
> > may feel differently.

I'm not really sure that those are "silly" ISA strings. People are going
to do it that way because it is much easier than spelling out 5 dozen
sub-components, and it is pretty inevitable that subsets will be
introduced in the future for extensions we currently have.

IMO, it's perfectly valid to say you have the supersets and not spell
out all the subcomponents.
  
Evan Green Oct. 19, 2023, 4:19 p.m. UTC | #6
On Thu, Oct 19, 2023 at 8:33 AM Conor Dooley <conor@kernel.org> wrote:
>
> On Thu, Oct 19, 2023 at 11:35:59AM +0200, Clément Léger wrote:
> >
> >
> > On 18/10/2023 19:26, Evan Green wrote:
> > > On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
> > >>
> > >>
> > >>
> > >> On 18/10/2023 03:45, Jerry Shih wrote:
> > >>> On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
> > >>>> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> > >>>>      __RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
> > >>>>      __RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
> > >>>>      __RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
> > >>>> +    __RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
> > >>>> +    __RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
> > >>>> +    __RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
> > >>>
> > >>> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
> > >>
> > >> Hi Jerry,
> > >>
> > >> Thanks for catching this, I think some other extensions will fall in
> > >> this category as well then (Zvknha/Zvknhb). I will verify that.
> > >
> > > The bundling mechanism works well when an extension is a pure lasso
> > > around other extensions. We'd have to tweak that code if we wanted to
> > > support cases like this, where the extension is a superset of others,
> > > but also contains loose change not present anywhere else (and
> > > therefore also needs to stand as a separate bit).
> >
> > For Zvbb and Zvknhb, I used the following code:
> >
> > static const unsigned int riscv_zvbb_bundled_exts[] = {
> >       RISCV_ISA_EXT_ZVKB,
> >       RISCV_ISA_EXT_ZVBB
> > };
> >
> > static const unsigned int riscv_zvknhb_bundled_exts[] = {
> >       RISCV_ISA_EXT_ZVKNHA,
> >       RISCV_ISA_EXT_ZVKNHB
> > };
> >
> > Which correctly results in both extension (superset + base set) being
> > enabled when only one is set. Is there something that I'm missing ?
> >
> > >
> > > IMO, decomposing "pure" bundles makes sense since otherwise usermode
> > > would have to query multiple distinct bitmaps that meant the same
> > > thing (eg check the Zk bit, or maybe check the Zkn/Zkr/Zkt bits, or
> > > maybe check the Zbkb/Zbkc... bits, and they're all equivalent). But
> > > when an extension is a superset that also contains loose change, there
> > > really aren't two equivalent bitmasks, each bit adds something new.
> >
> > Agreed but if a system only report ZVBB for instance and the user wants
> > ZVKB, then it is clear that ZVKB should be reported as well I guess. So
> > in the end, it works much like "bundle" extension, just that the bundle
> > is actually a "real" ISA extension by itself.
> >
> > Clément
> >
> > >
> > > There's an argument to be made for still turning on the containing
> > > extensions to cover for silly ISA strings (eg ISA strings that
> > > advertise the superset but fail to advertise the containing
> > > extensions). We can decide if we want to work that hard to cover
> > > hypothetical broken ISA strings now, or wait until they show up.
> > > Personally I would wait until something broken shows up. But others
> > > may feel differently.
>
> I'm not really sure that those are "silly" ISA strings. People are going
> to do it that way because it is much easier than spelling out 5 dozen
> sub-components, and it is pretty inevitable that subsets will be
> introduced in the future for extensions we currently have.
>
> IMO, it's perfectly valid to say you have the supersets and not spell
> out all the subcomponents.

Hm, ok. If ISA strings are likely to be written that way, then I agree
having the kernel flip on all the contained extensions is a good idea.
We can tweak patch 2 to support the parsing of struct
riscv_isa_ext_data with both .id and .bundle_size set (instead of only
one or the other as it is now). Looking back at that patch, it looks
quite doable. Alright!

-Evan
  
Jerry Shih Oct. 20, 2023, 2:43 a.m. UTC | #7
On Oct 19, 2023, at 17:35, Clément Léger <cleger@rivosinc.com> wrote:
> On 18/10/2023 19:26, Evan Green wrote:
>> On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
>>> 
>>> On 18/10/2023 03:45, Jerry Shih wrote:
>>>> 
>>>> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
>>> 
>>> Hi Jerry,
>>> 
>>> Thanks for catching this, I think some other extensions will fall in
>>> this category as well then (Zvknha/Zvknhb). I will verify that.
>> 
>> The bundling mechanism works well when an extension is a pure lasso
>> around other extensions. We'd have to tweak that code if we wanted to
>> support cases like this, where the extension is a superset of others,
>> but also contains loose change not present anywhere else (and
>> therefore also needs to stand as a separate bit).
> 
> For Zvbb and Zvknhb, I used the following code:
> 
> static const unsigned int riscv_zvbb_bundled_exts[] = {
> 	RISCV_ISA_EXT_ZVKB,
> 	RISCV_ISA_EXT_ZVBB
> };
> 
> static const unsigned int riscv_zvknhb_bundled_exts[] = {
> 	RISCV_ISA_EXT_ZVKNHA,
> 	RISCV_ISA_EXT_ZVKNHB
> };
> 
> Which correctly results in both extension (superset + base set) being
> enabled when only one is set. Is there something that I'm missing ?

We should not bundle zvknha and zvknhb together. They are exclusive.
Please check:
https://github.com/riscv/riscv-crypto/issues/364#issuecomment-1726782096

-Jerry
  
Clément Léger Oct. 23, 2023, 7:14 a.m. UTC | #8
On 20/10/2023 04:43, Jerry Shih wrote:
> On Oct 19, 2023, at 17:35, Clément Léger <cleger@rivosinc.com> wrote:
>> On 18/10/2023 19:26, Evan Green wrote:
>>> On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
>>>>
>>>> On 18/10/2023 03:45, Jerry Shih wrote:
>>>>>
>>>>> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
>>>>
>>>> Hi Jerry,
>>>>
>>>> Thanks for catching this, I think some other extensions will fall in
>>>> this category as well then (Zvknha/Zvknhb). I will verify that.
>>>
>>> The bundling mechanism works well when an extension is a pure lasso
>>> around other extensions. We'd have to tweak that code if we wanted to
>>> support cases like this, where the extension is a superset of others,
>>> but also contains loose change not present anywhere else (and
>>> therefore also needs to stand as a separate bit).
>>
>> For Zvbb and Zvknhb, I used the following code:
>>
>> static const unsigned int riscv_zvbb_bundled_exts[] = {
>> 	RISCV_ISA_EXT_ZVKB,
>> 	RISCV_ISA_EXT_ZVBB
>> };
>>
>> static const unsigned int riscv_zvknhb_bundled_exts[] = {
>> 	RISCV_ISA_EXT_ZVKNHA,
>> 	RISCV_ISA_EXT_ZVKNHB
>> };
>>
>> Which correctly results in both extension (superset + base set) being
>> enabled when only one is set. Is there something that I'm missing ?
> 
> We should not bundle zvknha and zvknhb together. They are exclusive.

Yes, but for instance, what happens if the user query the zvknha (if it
only needs SHA256) but zvknhb is present. If we don't declare zvknha,
then it will fail but the support would actually be present due to
zvknhb being there.

Clément


> Please check:
> https://github.com/riscv/riscv-crypto/issues/364#issuecomment-1726782096
> 
> -Jerry
>
  
Clément Léger Oct. 23, 2023, 7:24 a.m. UTC | #9
On 19/10/2023 18:19, Evan Green wrote:
> On Thu, Oct 19, 2023 at 8:33 AM Conor Dooley <conor@kernel.org> wrote:
>>
>> On Thu, Oct 19, 2023 at 11:35:59AM +0200, Clément Léger wrote:
>>>
>>>
>>> On 18/10/2023 19:26, Evan Green wrote:
>>>> On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>> On 18/10/2023 03:45, Jerry Shih wrote:
>>>>>> On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
>>>>>>> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
>>>>>>>      __RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
>>>>>>>      __RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
>>>>>>>      __RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
>>>>>>> +    __RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
>>>>>>> +    __RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
>>>>>>> +    __RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
>>>>>>
>>>>>> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
>>>>>
>>>>> Hi Jerry,
>>>>>
>>>>> Thanks for catching this, I think some other extensions will fall in
>>>>> this category as well then (Zvknha/Zvknhb). I will verify that.
>>>>
>>>> The bundling mechanism works well when an extension is a pure lasso
>>>> around other extensions. We'd have to tweak that code if we wanted to
>>>> support cases like this, where the extension is a superset of others,
>>>> but also contains loose change not present anywhere else (and
>>>> therefore also needs to stand as a separate bit).
>>>
>>> For Zvbb and Zvknhb, I used the following code:
>>>
>>> static const unsigned int riscv_zvbb_bundled_exts[] = {
>>>       RISCV_ISA_EXT_ZVKB,
>>>       RISCV_ISA_EXT_ZVBB
>>> };
>>>
>>> static const unsigned int riscv_zvknhb_bundled_exts[] = {
>>>       RISCV_ISA_EXT_ZVKNHA,
>>>       RISCV_ISA_EXT_ZVKNHB
>>> };
>>>
>>> Which correctly results in both extension (superset + base set) being
>>> enabled when only one is set. Is there something that I'm missing ?
>>>
>>>>
>>>> IMO, decomposing "pure" bundles makes sense since otherwise usermode
>>>> would have to query multiple distinct bitmaps that meant the same
>>>> thing (eg check the Zk bit, or maybe check the Zkn/Zkr/Zkt bits, or
>>>> maybe check the Zbkb/Zbkc... bits, and they're all equivalent). But
>>>> when an extension is a superset that also contains loose change, there
>>>> really aren't two equivalent bitmasks, each bit adds something new.
>>>
>>> Agreed but if a system only report ZVBB for instance and the user wants
>>> ZVKB, then it is clear that ZVKB should be reported as well I guess. So
>>> in the end, it works much like "bundle" extension, just that the bundle
>>> is actually a "real" ISA extension by itself.
>>>
>>> Clément
>>>
>>>>
>>>> There's an argument to be made for still turning on the containing
>>>> extensions to cover for silly ISA strings (eg ISA strings that
>>>> advertise the superset but fail to advertise the containing
>>>> extensions). We can decide if we want to work that hard to cover
>>>> hypothetical broken ISA strings now, or wait until they show up.
>>>> Personally I would wait until something broken shows up. But others
>>>> may feel differently.
>>
>> I'm not really sure that those are "silly" ISA strings. People are going
>> to do it that way because it is much easier than spelling out 5 dozen
>> sub-components, and it is pretty inevitable that subsets will be
>> introduced in the future for extensions we currently have.
>>
>> IMO, it's perfectly valid to say you have the supersets and not spell
>> out all the subcomponents.
> 
> Hm, ok. If ISA strings are likely to be written that way, then I agree
> having the kernel flip on all the contained extensions is a good idea.
> We can tweak patch 2 to support the parsing of struct
> riscv_isa_ext_data with both .id and .bundle_size set (instead of only
> one or the other as it is now). Looking back at that patch, it looks
> quite doable. Alright!

Hey Evan,

do you have anything against using this code:

static const unsigned int riscv_zvbb_bundled_exts[] = {
	RISCV_ISA_EXT_ZVKB,
	RISCV_ISA_EXT_ZVBB
};

...

Then declaring zvbb like that:

__RISCV_ISA_EXT_BUNDLE(zvbb, riscv_zvbb_bundled_exts),

I agree that it is *not* a bundled extension but it actually already
works with Conor's code. Not sure that adding more code is needed to
handle that case.

Clément


> 
> -Evan
  
Evan Green Oct. 23, 2023, 4:18 p.m. UTC | #10
On Mon, Oct 23, 2023 at 12:24 AM Clément Léger <cleger@rivosinc.com> wrote:
>
>
>
> On 19/10/2023 18:19, Evan Green wrote:
> > On Thu, Oct 19, 2023 at 8:33 AM Conor Dooley <conor@kernel.org> wrote:
> >>
> >> On Thu, Oct 19, 2023 at 11:35:59AM +0200, Clément Léger wrote:
> >>>
> >>>
> >>> On 18/10/2023 19:26, Evan Green wrote:
> >>>> On Wed, Oct 18, 2023 at 5:53 AM Clément Léger <cleger@rivosinc.com> wrote:
> >>>>>
> >>>>>
> >>>>>
> >>>>> On 18/10/2023 03:45, Jerry Shih wrote:
> >>>>>> On Oct 17, 2023, at 21:14, Clément Léger <cleger@rivosinc.com> wrote:
> >>>>>>> @@ -221,6 +261,22 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> >>>>>>>      __RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
> >>>>>>>      __RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
> >>>>>>>      __RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
> >>>>>>> +    __RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
> >>>>>>> +    __RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
> >>>>>>> +    __RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
> >>>>>>
> >>>>>> The `Zvkb` is the subset of `Zvbb`[1]. So, the `Zvkb` should be bundled with `Zvbb`.
> >>>>>
> >>>>> Hi Jerry,
> >>>>>
> >>>>> Thanks for catching this, I think some other extensions will fall in
> >>>>> this category as well then (Zvknha/Zvknhb). I will verify that.
> >>>>
> >>>> The bundling mechanism works well when an extension is a pure lasso
> >>>> around other extensions. We'd have to tweak that code if we wanted to
> >>>> support cases like this, where the extension is a superset of others,
> >>>> but also contains loose change not present anywhere else (and
> >>>> therefore also needs to stand as a separate bit).
> >>>
> >>> For Zvbb and Zvknhb, I used the following code:
> >>>
> >>> static const unsigned int riscv_zvbb_bundled_exts[] = {
> >>>       RISCV_ISA_EXT_ZVKB,
> >>>       RISCV_ISA_EXT_ZVBB
> >>> };
> >>>
> >>> static const unsigned int riscv_zvknhb_bundled_exts[] = {
> >>>       RISCV_ISA_EXT_ZVKNHA,
> >>>       RISCV_ISA_EXT_ZVKNHB
> >>> };
> >>>
> >>> Which correctly results in both extension (superset + base set) being
> >>> enabled when only one is set. Is there something that I'm missing ?
> >>>
> >>>>
> >>>> IMO, decomposing "pure" bundles makes sense since otherwise usermode
> >>>> would have to query multiple distinct bitmaps that meant the same
> >>>> thing (eg check the Zk bit, or maybe check the Zkn/Zkr/Zkt bits, or
> >>>> maybe check the Zbkb/Zbkc... bits, and they're all equivalent). But
> >>>> when an extension is a superset that also contains loose change, there
> >>>> really aren't two equivalent bitmasks, each bit adds something new.
> >>>
> >>> Agreed but if a system only report ZVBB for instance and the user wants
> >>> ZVKB, then it is clear that ZVKB should be reported as well I guess. So
> >>> in the end, it works much like "bundle" extension, just that the bundle
> >>> is actually a "real" ISA extension by itself.
> >>>
> >>> Clément
> >>>
> >>>>
> >>>> There's an argument to be made for still turning on the containing
> >>>> extensions to cover for silly ISA strings (eg ISA strings that
> >>>> advertise the superset but fail to advertise the containing
> >>>> extensions). We can decide if we want to work that hard to cover
> >>>> hypothetical broken ISA strings now, or wait until they show up.
> >>>> Personally I would wait until something broken shows up. But others
> >>>> may feel differently.
> >>
> >> I'm not really sure that those are "silly" ISA strings. People are going
> >> to do it that way because it is much easier than spelling out 5 dozen
> >> sub-components, and it is pretty inevitable that subsets will be
> >> introduced in the future for extensions we currently have.
> >>
> >> IMO, it's perfectly valid to say you have the supersets and not spell
> >> out all the subcomponents.
> >
> > Hm, ok. If ISA strings are likely to be written that way, then I agree
> > having the kernel flip on all the contained extensions is a good idea.
> > We can tweak patch 2 to support the parsing of struct
> > riscv_isa_ext_data with both .id and .bundle_size set (instead of only
> > one or the other as it is now). Looking back at that patch, it looks
> > quite doable. Alright!
>
> Hey Evan,
>
> do you have anything against using this code:
>
> static const unsigned int riscv_zvbb_bundled_exts[] = {
>         RISCV_ISA_EXT_ZVKB,
>         RISCV_ISA_EXT_ZVBB
> };
>
> ...
>
> Then declaring zvbb like that:
>
> __RISCV_ISA_EXT_BUNDLE(zvbb, riscv_zvbb_bundled_exts),
>
> I agree that it is *not* a bundled extension but it actually already
> works with Conor's code. Not sure that adding more code is needed to
> handle that case.

Ah, I had missed that Zvbb was in Zvbb's own bundle. I see now that it
works, but it also feels a bit like we're working around our own code.

An alternate way, which you can decide if you like better, would be a
new macro (something like __RISCV_ISA_EXT_DATA_BUNDLE(), but better
names welcome) that allows setting both .id and .bundle_size. Then the
else-if in match_isa_ext() could just turn into two independent ifs.
You'd have to define an "invalid" value for .id, since 0 is 'a', but
that should be straightforward. Or maybe jiggle things around a bit so
0 is invalid and 'a' is 1.

-Evan
  

Patch

diff --git a/arch/riscv/include/asm/hwcap.h b/arch/riscv/include/asm/hwcap.h
index ab80d822c847..a2fac23b0cc0 100644
--- a/arch/riscv/include/asm/hwcap.h
+++ b/arch/riscv/include/asm/hwcap.h
@@ -69,6 +69,16 @@ 
 #define RISCV_ISA_EXT_ZKSED		51
 #define RISCV_ISA_EXT_ZKSH		52
 #define RISCV_ISA_EXT_ZKT		53
+#define RISCV_ISA_EXT_ZVBB		54
+#define RISCV_ISA_EXT_ZVBC		55
+#define RISCV_ISA_EXT_ZVKB		56
+#define RISCV_ISA_EXT_ZVKG		57
+#define RISCV_ISA_EXT_ZVKNED		58
+#define RISCV_ISA_EXT_ZVKNHA		59
+#define RISCV_ISA_EXT_ZVKNHB		60
+#define RISCV_ISA_EXT_ZVKSED		61
+#define RISCV_ISA_EXT_ZVKSH		62
+#define RISCV_ISA_EXT_ZVKT		63
 
 #define RISCV_ISA_EXT_MAX		64
 
diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
index d3682fdfd9f1..8cf0b8b442ae 100644
--- a/arch/riscv/kernel/cpufeature.c
+++ b/arch/riscv/kernel/cpufeature.c
@@ -144,6 +144,46 @@  static const unsigned int riscv_zks_bundled_exts[] = {
 	RISCV_ISA_EXT_ZKSH
 };
 
+#define RISCV_ISA_EXT_ZVKN	\
+	RISCV_ISA_EXT_ZVKNED,	\
+	RISCV_ISA_EXT_ZVKNHB,	\
+	RISCV_ISA_EXT_ZVKB,	\
+	RISCV_ISA_EXT_ZVKT
+
+static const unsigned int riscv_zvkn_bundled_exts[] = {
+	RISCV_ISA_EXT_ZVKN
+};
+
+static const unsigned int riscv_zvknc_bundled_exts[] = {
+	RISCV_ISA_EXT_ZVKN,
+	RISCV_ISA_EXT_ZVBC,
+};
+
+static const unsigned int riscv_zvkng_bundled_exts[] = {
+	RISCV_ISA_EXT_ZVKN,
+	RISCV_ISA_EXT_ZVKG,
+};
+
+#define RISCV_ISA_EXT_ZVKS	\
+	RISCV_ISA_EXT_ZVKSED,	\
+	RISCV_ISA_EXT_ZVKSH,	\
+	RISCV_ISA_EXT_ZVKB,	\
+	RISCV_ISA_EXT_ZVKT
+
+static const unsigned int riscv_zvks_bundled_exts[] = {
+	RISCV_ISA_EXT_ZVKS
+};
+
+static const unsigned int riscv_zvksc_bundled_exts[] = {
+	RISCV_ISA_EXT_ZVKS,
+	RISCV_ISA_EXT_ZVBC,
+};
+
+static const unsigned int riscv_zvksg_bundled_exts[] = {
+	RISCV_ISA_EXT_ZVKS,
+	RISCV_ISA_EXT_ZVKG,
+};
+
 /*
  * The canonical order of ISA extension names in the ISA string is defined in
  * chapter 27 of the unprivileged specification.
@@ -221,6 +261,22 @@  const struct riscv_isa_ext_data riscv_isa_ext[] = {
 	__RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
 	__RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
 	__RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
+	__RISCV_ISA_EXT_DATA(zvbb, RISCV_ISA_EXT_ZVBB),
+	__RISCV_ISA_EXT_DATA(zvbc, RISCV_ISA_EXT_ZVBC),
+	__RISCV_ISA_EXT_DATA(zvkb, RISCV_ISA_EXT_ZVKB),
+	__RISCV_ISA_EXT_DATA(zvkg, RISCV_ISA_EXT_ZVKG),
+	__RISCV_ISA_EXT_BUNDLE(zvkn, riscv_zvkn_bundled_exts),
+	__RISCV_ISA_EXT_BUNDLE(zvknc, riscv_zvknc_bundled_exts),
+	__RISCV_ISA_EXT_DATA(zvkned, RISCV_ISA_EXT_ZVKNED),
+	__RISCV_ISA_EXT_BUNDLE(zvkng, riscv_zvkng_bundled_exts),
+	__RISCV_ISA_EXT_DATA(zvknha, RISCV_ISA_EXT_ZVKNHA),
+	__RISCV_ISA_EXT_DATA(zvknhb, RISCV_ISA_EXT_ZVKNHB),
+	__RISCV_ISA_EXT_BUNDLE(zvks, riscv_zvks_bundled_exts),
+	__RISCV_ISA_EXT_BUNDLE(zvksc, riscv_zvksc_bundled_exts),
+	__RISCV_ISA_EXT_DATA(zvksed, RISCV_ISA_EXT_ZVKSED),
+	__RISCV_ISA_EXT_DATA(zvksh, RISCV_ISA_EXT_ZVKSH),
+	__RISCV_ISA_EXT_BUNDLE(zvksg, riscv_zvksg_bundled_exts),
+	__RISCV_ISA_EXT_DATA(zvkt, RISCV_ISA_EXT_ZVKT),
 	__RISCV_ISA_EXT_DATA(smaia, RISCV_ISA_EXT_SMAIA),
 	__RISCV_ISA_EXT_DATA(ssaia, RISCV_ISA_EXT_SSAIA),
 	__RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),