Re: [PATCH v4 1/4] RISC-V: Add Bitmanip/Scalar Crypto parsing from DT
From: Andrew Jones
Date: Thu Jul 13 2023 - 08:46:17 EST
On Thu, Jul 13, 2023 at 12:27:24PM +0100, Conor Dooley wrote:
> On Thu, Jul 13, 2023 at 10:46:14AM +0200, Andrew Jones wrote:
> > On Wed, Jul 12, 2023 at 10:43:33AM -0700, Evan Green wrote:
> > > On Wed, Jul 12, 2023 at 3:39 AM Conor Dooley <conor.dooley@xxxxxxxxxxxxx> wrote:
> > > > On Wed, Jul 12, 2023 at 10:41:17AM +0200, Samuel Ortiz wrote:
> > > > > From: "Hongren (Zenithal) Zheng" <i@xxxxxxxxxxx>
> > > > >
> > > > > Parse Zb/Zk related string from DT and output them to cpuinfo.
> > > >
> > > > One thing that has sprung to mind is that this is not limited to DT
> > > > anymore, since the information could in theory come from ACPI too.
> > > > Ditto the title I guess.
> > > >
> > > > > It is worth noting that the Scalar Crypto extension defines "zk" as a
> > > > > shorthand for the Zkn, Zkr and Zkt extensions. Since the Zkn one also
> > > > > implies the Zbkb, Zbkc and Zbkx extensions, simply passing the valid
> > > > > "zk" extension name through a DT will enable all of the Zbkb, Zbkc,
> > > > > Zbkx, Zkn, Zkr and Zkt extensions.
> > > > >
> > > > > Also, since there currently is no mechanism to merge all enabled
> > > > > extensions, the generated cpuinfo output could be relatively large.
> > > > > For example, setting the "riscv,isa" DT property to "rv64imafdc_zk_zks"
> > > > > will generate the following cpuinfo output:
> > > > > "rv64imafdc_zbkb_zbkc_zbkx_zknd_zkne_zknh_zkr_zksed_zksh_zkt".
> > > >
> > > > On that note, I've created another version of what checking for
> > > > supersets could look like, since it'll be needed either by my series or
> > > > this one, depending on what gets merged first. I've yet to test the
> > > > dedicated extensions part of it, but I wanted to get this out before I
> > > > went looking at other fixes in the area.
> > > >
> > > > Evan, since it was you that commented on this stuff last time around,
> > > > could you take another look? I'm still not keen on the "subset_of"
> > > > arrays, but they're an improvement on what I had last time around for
> > > > sure.
> > > >
> > >
> > > This looks alright to me. At the risk of getting into bikeshedding
> > > territory, the only awkward bit of it is it composes the extensions in
> > > sort of the opposite way you'd expect. I tend to think of Zks as being
> > > comprised of {zbkb, zbkc, zksed, zksh},
> >
> > This is also the way I think of it, so, FWIW, I prefer the approach below,
> > where bundles are expanded.
>
> I took the patch Evan sent me off-list & have spun it into a "real"
> patch. I did some minor changes (c99 loop, __init marking) to match what
> the code was already doing and expanded it to work for the new property
> stuff.
> I'm not sure that match_isa_ext() needs to be a function, given the
> single user - it might be better refactored to drop the outer if & return
> whether a match was found.
> If done that would allow sharing the same function between isa string and
> dedicated property parsing. I left it intact for now. Might also be
> worth adding kerneldoc for that struct.
>
> -- >8 --
> From 8300b98e487e57d192bc9581c6a3639c40c52623 Mon Sep 17 00:00:00 2001
> From: Evan Green <evan@xxxxxxxxxxxx>
> Date: Wed, 12 Jul 2023 10:36:15 -0700
> Subject: [PATCH] RISC-V: Add support Scalar Crypto using "bundled extensions"
>
> The Scalar Crypto specification defines Zk as a shorthand for the
> Zkn, Zkr and Zkt extensions. The same follows for both Zkn, Zks and Zbk,
> which are all shorthands for various other extensions. The detailed
> breakdown can be found in their dt-binding entries.
>
> Since Zkn also implies the Zbkb, Zbkc and Zbkx extensions, simply passing
> "zk" through a DT should enable all of Zbkb, Zbkc, Zbkx, Zkn, Zkr and Zkt.
> For example, setting the "riscv,isa" DT property to "rv64imafdc_zk"
> should generate the following cpuinfo output:
> "rv64imafdc_zicntr_zicsr_zifencei_zihpm_zbkb_zbkc_zbkx_zknd_zkne_zknh_zkr_zkt"
>
> riscv_isa_ext_data grows a pair of new members, to permit setting the
> relevant bits for "bundled" extensions, both while parsing the ISA string
> and the new dedicated extension properties
>
> Co-developed-by: Conor Dooley <conor.dooley@xxxxxxxxxxxxx>
> Signed-off-by: Conor Dooley <conor.dooley@xxxxxxxxxxxxx>
> ---
> There's basically none of the original patch left, so I have dropped the
> co-developed-bys and signed-off-bys from before. It does need one from
> you now though Evan!
> ---
> arch/riscv/include/asm/hwcap.h | 13 +++++
> arch/riscv/kernel/cpufeature.c | 94 ++++++++++++++++++++++++++++------
> 2 files changed, 92 insertions(+), 15 deletions(-)
>
> diff --git a/arch/riscv/include/asm/hwcap.h b/arch/riscv/include/asm/hwcap.h
> index b7b58258f6c7..f74308633e45 100644
> --- a/arch/riscv/include/asm/hwcap.h
> +++ b/arch/riscv/include/asm/hwcap.h
> @@ -58,6 +58,17 @@
> #define RISCV_ISA_EXT_ZICSR 40
> #define RISCV_ISA_EXT_ZIFENCEI 41
> #define RISCV_ISA_EXT_ZIHPM 42
> +#define RISCV_ISA_EXT_ZBC 43
> +#define RISCV_ISA_EXT_ZBKB 44
> +#define RISCV_ISA_EXT_ZBKC 45
> +#define RISCV_ISA_EXT_ZBKX 46
> +#define RISCV_ISA_EXT_ZKND 47
> +#define RISCV_ISA_EXT_ZKNE 48
> +#define RISCV_ISA_EXT_ZKNH 49
> +#define RISCV_ISA_EXT_ZKR 50
> +#define RISCV_ISA_EXT_ZKSED 51
> +#define RISCV_ISA_EXT_ZKSH 52
> +#define RISCV_ISA_EXT_ZKT 53
>
> #define RISCV_ISA_EXT_MAX 64
>
> @@ -77,6 +88,8 @@ struct riscv_isa_ext_data {
> const unsigned int id;
> const char *name;
> const char *property;
> + const unsigned int *bundle_ids;
> + const unsigned int bundle_size;
> };
>
> extern const struct riscv_isa_ext_data riscv_isa_ext[];
> diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
> index ccbe27b3060b..d999d73554f0 100644
> --- a/arch/riscv/kernel/cpufeature.c
> +++ b/arch/riscv/kernel/cpufeature.c
> @@ -105,6 +105,39 @@ static bool riscv_isa_extension_check(int id)
> .id = _id, \
> }
>
> +#define __RISCV_ISA_EXT_BUNDLE(_name, _bundled_exts) { \
> + .name = #_name, \
> + .property = #_name, \
> + .bundle_ids = _bundled_exts, \
> + .bundle_size = ARRAY_SIZE(_bundled_exts) \
> +}
> +
> +static const unsigned int riscv_zk_bundled_exts[] = {
> + RISCV_ISA_EXT_ZBKB,
> + RISCV_ISA_EXT_ZBKC,
> + RISCV_ISA_EXT_ZBKX,
> + RISCV_ISA_EXT_ZKND,
> + RISCV_ISA_EXT_ZKNE,
> + RISCV_ISA_EXT_ZKR,
> + RISCV_ISA_EXT_ZKT,
I think RISCV_ISA_EXT_ZKNH also belongs in this bundle,
since the spec says zk is the zkn bundle plus zkr and zkt.
> +};
> +
> +static const unsigned int riscv_zkn_bundled_exts[] = {
> + RISCV_ISA_EXT_ZBKB,
> + RISCV_ISA_EXT_ZBKC,
> + RISCV_ISA_EXT_ZBKX,
> + RISCV_ISA_EXT_ZKND,
> + RISCV_ISA_EXT_ZKNE,
> + RISCV_ISA_EXT_ZKNH,
> +};
> +
> +static const unsigned int riscv_zks_bundled_exts[] = {
> + RISCV_ISA_EXT_ZBKB,
> + RISCV_ISA_EXT_ZBKC,
> + RISCV_ISA_EXT_ZKSED,
> + RISCV_ISA_EXT_ZKSH
And, per the spec, this one appears to be missing RISCV_ISA_EXT_ZBKX.
I found [1] which calls these shorthands "group names", so maybe we should
use the term "group" instead of "bundle"? I'm tempted to try to directly
code that graphic in [1] with something like...
#define Zks_group1 \
RISCV_ISA_EXT_ZKSED, \
RISCV_ISA_EXT_ZKSH
#define Zks_group2 \
RISCV_ISA_EXT_ZBKB, \
RISCV_ISA_EXT_ZBKC, \
RISCV_ISA_EXT_ZBKX
#define Zks_group \
Zks_group1, \
Zks_group2
#define Zkn_group1 \
RISCV_ISA_EXT_ZKND, \
RISCV_ISA_EXT_ZKNE, \
RISCV_ISA_EXT_ZKNH
#define Zkn_group2 \
Zks_group2
#define Zkn_group \
Zkn_group1, \
Zkn_group2
static const unsigned int riscv_zks_group[] = {
Zks_group,
};
static const unsigned int riscv_zkn_group[] = {
Zkn_group,
};
static const unsigned int riscv_zk_group[] = {
Zks_group,
Zkn_group,
RISCV_ISA_EXT_ZKR,
RISCV_ISA_EXT_ZKT,
};
...but now that I have, I'm not sure I like the looks of it...
[1] https://wiki.riscv.org/display/HOME/Scalar+Cryptography+Instruction+Set+Extension+Group+Names+Diagram
Thanks,
drew
> +};
> +
> /*
> * The canonical order of ISA extension names in the ISA string is defined in
> * chapter 27 of the unprivileged specification.
> @@ -167,7 +200,20 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> __RISCV_ISA_EXT_DATA(zihpm, RISCV_ISA_EXT_ZIHPM),
> __RISCV_ISA_EXT_DATA(zba, RISCV_ISA_EXT_ZBA),
> __RISCV_ISA_EXT_DATA(zbb, RISCV_ISA_EXT_ZBB),
> + __RISCV_ISA_EXT_DATA(zbc, RISCV_ISA_EXT_ZBC),
> + __RISCV_ISA_EXT_DATA(zbkb, RISCV_ISA_EXT_ZBKB),
> + __RISCV_ISA_EXT_DATA(zbkc, RISCV_ISA_EXT_ZBKC),
> + __RISCV_ISA_EXT_DATA(zbkx, RISCV_ISA_EXT_ZBKX),
> __RISCV_ISA_EXT_DATA(zbs, RISCV_ISA_EXT_ZBS),
> + __RISCV_ISA_EXT_BUNDLE(zk, riscv_zk_bundled_exts),
> + __RISCV_ISA_EXT_BUNDLE(zkn, riscv_zkn_bundled_exts),
> + __RISCV_ISA_EXT_DATA(zknd, RISCV_ISA_EXT_ZKND),
> + __RISCV_ISA_EXT_DATA(zkne, RISCV_ISA_EXT_ZKNE),
> + __RISCV_ISA_EXT_DATA(zknh, RISCV_ISA_EXT_ZKNH),
> + __RISCV_ISA_EXT_DATA(zkr, RISCV_ISA_EXT_ZKR),
> + __RISCV_ISA_EXT_BUNDLE(zks, riscv_zks_bundled_exts),
> + __RISCV_ISA_EXT_DATA(zksed, RISCV_ISA_EXT_ZKSED),
> + __RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
> __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),
> @@ -179,6 +225,26 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
>
> const size_t riscv_isa_ext_count = ARRAY_SIZE(riscv_isa_ext);
>
> +static void inline __init match_isa_ext(const struct riscv_isa_ext_data *ext, const char *name,
> + const char *name_end, struct riscv_isainfo *isainfo)
> +{
> + if ((name_end - name == strlen(ext->name)) &&
> + !strncasecmp(name, ext->name, name_end - name)) {
> + /*
> + * If this is a bundle, enable all the ISA extensions that
> + * comprise the bundle.
> + */
> + if (ext->bundle_size) {
> + for (int i = 0; i < ext->bundle_size; i++) {
> + if (riscv_isa_extension_check(ext->bundle_ids[i]))
> + set_bit(ext->bundle_ids[i], isainfo->isa);
> + }
> + } else if (riscv_isa_extension_check(ext->id)) {
> + set_bit(ext->id, isainfo->isa);
> + }
> + }
> +}
> +
> static void __init riscv_parse_isa_string(unsigned long *this_hwcap, struct riscv_isainfo *isainfo,
> unsigned long *isa2hwcap, const char *isa)
> {
> @@ -310,14 +376,6 @@ static void __init riscv_parse_isa_string(unsigned long *this_hwcap, struct risc
> if (*isa == '_')
> ++isa;
>
> -#define SET_ISA_EXT_MAP(name, bit) \
> - do { \
> - if ((ext_end - ext == strlen(name)) && \
> - !strncasecmp(ext, name, strlen(name)) && \
> - riscv_isa_extension_check(bit)) \
> - set_bit(bit, isainfo->isa); \
> - } while (false) \
> -
> if (unlikely(ext_err))
> continue;
> if (!ext_long) {
> @@ -329,10 +387,8 @@ static void __init riscv_parse_isa_string(unsigned long *this_hwcap, struct risc
> }
> } else {
> for (int i = 0; i < riscv_isa_ext_count; i++)
> - SET_ISA_EXT_MAP(riscv_isa_ext[i].name,
> - riscv_isa_ext[i].id);
> + match_isa_ext(&riscv_isa_ext[i], ext, ext_end, isainfo);
> }
> -#undef SET_ISA_EXT_MAP
> }
> }
>
> @@ -436,18 +492,26 @@ static int __init riscv_fill_hwcap_from_ext_list(unsigned long *isa2hwcap)
> }
>
> for (int i = 0; i < riscv_isa_ext_count; i++) {
> + const struct riscv_isa_ext_data ext = riscv_isa_ext[i];
> +
> if (of_property_match_string(cpu_node, "riscv,isa-extensions",
> - riscv_isa_ext[i].property) < 0)
> + ext.property) < 0)
> continue;
>
> - if (!riscv_isa_extension_check(riscv_isa_ext[i].id))
> + if (ext.bundle_size) {
> + for (int j = 0; j < ext.bundle_size; j++) {
> + if (riscv_isa_extension_check(ext.bundle_ids[i]))
> + set_bit(ext.bundle_ids[j], this_isa);
> + }
> + } else if (riscv_isa_extension_check(ext.id)) {
> + set_bit(ext.id, this_isa);
> + } else {
> continue;
> + }
>
> /* Only single letter extensions get set in hwcap */
> if (strnlen(riscv_isa_ext[i].name, 2) == 1)
> this_hwcap |= isa2hwcap[riscv_isa_ext[i].id];
> -
> - set_bit(riscv_isa_ext[i].id, this_isa);
> }
>
> of_node_put(cpu_node);
> --
> 2.40.1
>
>