Skip to contentSkip to conversation

RISC-V: Add Bitmanip/Scalar Crypto HWCAP

PATCH In review Hongren (Zenithal) Zheng 4 participants 15 messages Jun 12, 2022 – Nov 24, 2022

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

v3 cover RISC-V: Add Bitmanip/Scalar Crypto HWCAP

This patchset proposes a currently viable and forward compatible way to expose the bitmanip/scalar crypto capability of the platform to the userspace.

Currently viable refers to the property that hardware platforms can easily modify the riscv,isa field in DT to tell the kernel it has the capability. Note that QEMU has already done so in its device tree.

Forward compatible refers to the property that userspace can still detect the capability of the environment by using HWCAP regardless of how the mechanism changes below kernel in the future. I do know that it has not been settled how to discover a capability, but I think kernel has to offer some API after all, and HWCAP is the preferred way among other mechanisms for now.

A note here is that the draft riscv-platform spec considers DT as mandatory discovery mechanism for embedded platform For other discovery mechanism like ACPI/SMBIOS, similar code path can be added but the HWCAP interface could remain unchanged

More discussion on userspace discovering
can be found on my PR to openssl
https://github.com/openssl/openssl/pull/18197

---
v2:
* Fixed checkpatch problem found by Heiko Stuebner
* rebased on riscv/for-next as that branch has merged the support of Svpbmt extension
* Changed the order of logical ids in riscv_isa_ext_id to the order specified in riscv-isa-manual
* added note on riscv-platform-spec

v3:
* rebased on master as riscv/for-next has been merged
* fix commit message as suggested by Conor Dooley

Hongren (Zenithal) Zheng (3):
RISC-V: add Bitmanip/Scalar Crypto parsing from DT
RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto
RISC-V: HWCAP: parse Bitmanip/Scalar Crypto HWCAP from DT

 arch/riscv/include/asm/elf.h        |  2 +
 arch/riscv/include/asm/hwcap.h      | 22 +++++++-
 arch/riscv/include/uapi/asm/hwcap.h | 22 ++++++++
 arch/riscv/kernel/cpu.c             | 14 +++++
 arch/riscv/kernel/cpufeature.c      | 81 +++++++++++++++++++++++++----
 5 files changed, 129 insertions(+), 12 deletions(-)
Signature · 1 line
2.35.1
#1 lore raw

PATCH v3 1/3 RISC-V: add Bitmanip/Scalar Crypto parsing from DT

This patch parses Zb/Zk related string from DT and output them in cpuinfo

One thing worth noting is that if DT provides zk, all zbkb, zbkc, zbkx and zkn, zkr, zkt would be enabled.

Note that zk is a valid extension name and the current DT binding spec allows this.

This patch also changes the logical id of existing multi-letter extensions and adds a statement that instead of logical id compatibility, the order is needed.

There currently lacks a mechanism to merge them when producing cpuinfo. Namely if you provide a riscv,isa "rv64imafdc_zk_zks", the cpuinfo output would be "rv64imafdc_zbkb_zbkc_zbkx_zknd_zkne_zknh_zkr_zksed _zksh_zkt"

---
 arch/riscv/include/asm/hwcap.h | 20 +++++++++++++++++++-
 arch/riscv/kernel/cpu.c        | 14 ++++++++++++++
 arch/riscv/kernel/cpufeature.c | 33 +++++++++++++++++++++++++++++++++
 3 files changed, 66 insertions(+), 1 deletion(-)
arch/riscv/include/asm/hwcap.h +19 -1
index 4e2486881840..02c454a12683 100644
@@ -49,9 +49,27 @@ extern unsigned long elf_hwcap;
  * RISCV_ISA_EXT_MAX. 0-25 range is reserved for single letter
  * extensions while all the multi-letter extensions should define the next
  * available logical extension id.
+ *
+ * The order of them should be maintained according to the riscv-isa-manual.
+ * As this is an internal API, changing the id of one extension does
+ * not affect compatibility.
  */
 enum riscv_isa_ext_id {
-	RISCV_ISA_EXT_SSCOFPMF = RISCV_ISA_EXT_BASE,
+	RISCV_ISA_EXT_ZBA = RISCV_ISA_EXT_BASE,
+	RISCV_ISA_EXT_ZBB,
+	RISCV_ISA_EXT_ZBC,
+	RISCV_ISA_EXT_ZBKB,
+	RISCV_ISA_EXT_ZBKC,
+	RISCV_ISA_EXT_ZBKX,
+	RISCV_ISA_EXT_ZBS,
+	RISCV_ISA_EXT_ZKND,
+	RISCV_ISA_EXT_ZKNE,
+	RISCV_ISA_EXT_ZKNH,
+	RISCV_ISA_EXT_ZKR,
+	RISCV_ISA_EXT_ZKSED,
+	RISCV_ISA_EXT_ZKSH,
+	RISCV_ISA_EXT_ZKT,
+	RISCV_ISA_EXT_SSCOFPMF,
 	RISCV_ISA_EXT_SVPBMT,
 	RISCV_ISA_EXT_ID_MAX = RISCV_ISA_EXT_MAX,
 };
arch/riscv/kernel/cpu.c +14 -0
index fba9e9f46a8c..d9ff9bff3d45 100644
@@ -87,6 +87,20 @@ int riscv_of_parent_hartid(struct device_node *node)
  *    extensions by an underscore.
  */
 static struct riscv_isa_ext_data isa_ext_arr[] = {
+	__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_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_DATA(zksed, RISCV_ISA_EXT_ZKSED),
+	__RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
+	__RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
 	__RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),
 	__RISCV_ISA_EXT_DATA(svpbmt, RISCV_ISA_EXT_SVPBMT),
 	__RISCV_ISA_EXT_DATA("", RISCV_ISA_EXT_MAX),
arch/riscv/kernel/cpufeature.c +33 -0
index a6f62a6d1edd..0c2638365ec2 100644
@@ -199,6 +199,39 @@ void __init riscv_fill_hwcap(void)
 			} else {
 				SET_ISA_EXT_MAP("sscofpmf", RISCV_ISA_EXT_SSCOFPMF);
 				SET_ISA_EXT_MAP("svpbmt", RISCV_ISA_EXT_SVPBMT);
+				SET_ISA_EXT_MAP("zba", RISCV_ISA_EXT_ZBA);
+				SET_ISA_EXT_MAP("zbb", RISCV_ISA_EXT_ZBB);
+				SET_ISA_EXT_MAP("zbc", RISCV_ISA_EXT_ZBC);
+				SET_ISA_EXT_MAP("zbs", RISCV_ISA_EXT_ZBS);
+				SET_ISA_EXT_MAP("zbkb", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zbkc", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zbks", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zknd", RISCV_ISA_EXT_ZKND);
+				SET_ISA_EXT_MAP("zkne", RISCV_ISA_EXT_ZKNE);
+				SET_ISA_EXT_MAP("zknh", RISCV_ISA_EXT_ZKNH);
+				SET_ISA_EXT_MAP("zksed", RISCV_ISA_EXT_ZKSED);
+				SET_ISA_EXT_MAP("zksh", RISCV_ISA_EXT_ZKSH);
+				SET_ISA_EXT_MAP("zkr", RISCV_ISA_EXT_ZKR);
+				SET_ISA_EXT_MAP("zkt", RISCV_ISA_EXT_ZKT);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZKND);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZKNE);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZKNH);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZKSED);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZKSH);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKND);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKNE);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKNH);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKR);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKT);
 			}
 #undef SET_ISA_EXT_MAP
 		}
Signature · 1 line
2.35.1
Tested-by on v3 1/3
#2 lore raw

Re: [PATCH v3 1/3] RISC-V: add Bitmanip/Scalar Crypto parsing from DT

Hi,

Quoted 14 lines · Hongren (Zenithal) Zheng

This patch parses Zb/Zk related string from DT and output them in cpuinfo

One thing worth noting is that if DT provides zk, all zbkb, zbkc, zbkx and zkn, zkr, zkt would be enabled.

Note that zk is a valid extension name and the current DT binding spec allows this.

This patch also changes the logical id of existing multi-letter extensions and adds a statement that instead of logical id compatibility, the order is needed.

Quoted 8 lines

There currently lacks a mechanism to merge them when producing cpuinfo. Namely if you provide a riscv,isa "rv64imafdc_zk_zks", the cpuinfo output would be "rv64imafdc_zbkb_zbkc_zbkx_zknd_zkne_zknh_zkr_zksed _zksh_zkt"

FWIW, after some rebasing:

Cheers,
Samuel.

#3 lore raw

Re: [PATCH v3 1/3] RISC-V: add Bitmanip/Scalar Crypto parsing from DT

In reply to Hongren (Zenithal) Zheng

Hey Hongren,

Quoted 2 lines · Hongren (Zenithal) Zheng

This patch parses Zb/Zk related string from DT and output them in cpuinfo

Could you please wrap changelogs at 72 character rather than 50?

Quoted 6 lines

One thing worth noting is that if DT provides zk, all zbkb, zbkc, zbkx and zkn, zkr, zkt would be enabled.

Note that zk is a valid extension name and the current DT binding spec allows this.

Where is the "DT binding spec" for this?
At the time, IIRC, putting any multiletter extensions in a DT was a violation of the dt-binding, so I am interested in seeing what spec had covered this.

Quoted 24 lines

This patch also changes the logical id of existing multi-letter extensions and adds a statement that instead of logical id compatibility, the order is needed.

There currently lacks a mechanism to merge them when producing cpuinfo. Namely if you provide a riscv,isa "rv64imafdc_zk_zks", the cpuinfo output would be "rv64imafdc_zbkb_zbkc_zbkx_zknd_zkne_zknh_zkr_zksed _zksh_zkt"

---
 arch/riscv/include/asm/hwcap.h | 20 +++++++++++++++++++-
 arch/riscv/kernel/cpu.c        | 14 ++++++++++++++
 arch/riscv/kernel/cpufeature.c | 33 +++++++++++++++++++++++++++++++++
 3 files changed, 66 insertions(+), 1 deletion(-)
arch/riscv/include/asm/hwcap.h +0 -0
index 4e2486881840..02c454a12683 100644

arch/riscv/include/asm/hwcap.h extern unsigned long elf_hwcap; · line 55

Quoted 7 lines
arch/riscv/include/asm/hwcap.h +4 -0
@@ -49,9 +49,27 @@ extern unsigned long elf_hwcap;
  * RISCV_ISA_EXT_MAX. 0-25 range is reserved for single letter
  * extensions while all the multi-letter extensions should define the next
  * available logical extension id.
+ *
+ * The order of them should be maintained according to the riscv-isa-manual.
+ * As this is an internal API, changing the id of one extension does
+ * not affect compatibility.

Hmm, not too sure about this. See 61a41d16ad20 ("RISC-V: Print SSTC in canonical order"), where Palmer re-sorted another one of these internal APIs (isa_ext_arr) in a different order to what it would appear you're proposing here.

I am really really bad when it comes to specs, manuals etc - is there a link to the specific part of the riscv-isa-manual that we can use here for reference? Otherwise I'd much rather stick with what I **thought** was the "canonical order", _Sfoo_Zfoo_Xfoo.

Or are these slightly different - and fit in the "Additional Standard Extension Names" category rather than "Machine-level Instruction-Set Extensions"?

The only ones I see pointed out in the current table [0] are Ztso, Zicsr, Zifencei & Zam which is what I (maybe wrongly!!) assumed to be the exceptions to the S before Z rule.

I tried to have a look in the "bitmanip-1.0.0" pdf at [2], but I don't see it use the word "Standard" anywhere. As I already said, this stuff confuses me no end so some clarification would be really helpful :)

The same applies to the crypto stuff.

In other news, since Palmer did do a re-order, and its been several months - you'll need to rebase & re-submit anyway.

Thanks,
Conor.

0 - https://github.com/riscv/riscv-isa-manual/blob/master/src/naming.tex#L132
1 - https://github.com/riscv/riscv-isa-manual/blob/master/src/naming.tex#L182
2 - https://github.com/riscv/riscv-bitmanip/releases/tag/1.0.0

Quoted 93 lines
arch/riscv/include/asm/hwcap.h +15 -1
  */
 enum riscv_isa_ext_id {
-	RISCV_ISA_EXT_SSCOFPMF = RISCV_ISA_EXT_BASE,
+	RISCV_ISA_EXT_ZBA = RISCV_ISA_EXT_BASE,
+	RISCV_ISA_EXT_ZBB,
+	RISCV_ISA_EXT_ZBC,
+	RISCV_ISA_EXT_ZBKB,
+	RISCV_ISA_EXT_ZBKC,
+	RISCV_ISA_EXT_ZBKX,
+	RISCV_ISA_EXT_ZBS,
+	RISCV_ISA_EXT_ZKND,
+	RISCV_ISA_EXT_ZKNE,
+	RISCV_ISA_EXT_ZKNH,
+	RISCV_ISA_EXT_ZKR,
+	RISCV_ISA_EXT_ZKSED,
+	RISCV_ISA_EXT_ZKSH,
+	RISCV_ISA_EXT_ZKT,
+	RISCV_ISA_EXT_SSCOFPMF,
 	RISCV_ISA_EXT_SVPBMT,
 	RISCV_ISA_EXT_ID_MAX = RISCV_ISA_EXT_MAX,
 };
arch/riscv/kernel/cpu.c +14 -0
index fba9e9f46a8c..d9ff9bff3d45 100644
@@ -87,6 +87,20 @@ int riscv_of_parent_hartid(struct device_node *node)
  *    extensions by an underscore.
  */
 static struct riscv_isa_ext_data isa_ext_arr[] = {
+	__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_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_DATA(zksed, RISCV_ISA_EXT_ZKSED),
+	__RISCV_ISA_EXT_DATA(zksh, RISCV_ISA_EXT_ZKSH),
+	__RISCV_ISA_EXT_DATA(zkt, RISCV_ISA_EXT_ZKT),
 	__RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),
 	__RISCV_ISA_EXT_DATA(svpbmt, RISCV_ISA_EXT_SVPBMT),
 	__RISCV_ISA_EXT_DATA("", RISCV_ISA_EXT_MAX),
arch/riscv/kernel/cpufeature.c +33 -0
index a6f62a6d1edd..0c2638365ec2 100644
@@ -199,6 +199,39 @@ void __init riscv_fill_hwcap(void)
 			} else {
 				SET_ISA_EXT_MAP("sscofpmf", RISCV_ISA_EXT_SSCOFPMF);
 				SET_ISA_EXT_MAP("svpbmt", RISCV_ISA_EXT_SVPBMT);
+				SET_ISA_EXT_MAP("zba", RISCV_ISA_EXT_ZBA);
+				SET_ISA_EXT_MAP("zbb", RISCV_ISA_EXT_ZBB);
+				SET_ISA_EXT_MAP("zbc", RISCV_ISA_EXT_ZBC);
+				SET_ISA_EXT_MAP("zbs", RISCV_ISA_EXT_ZBS);
+				SET_ISA_EXT_MAP("zbkb", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zbkc", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zbks", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zknd", RISCV_ISA_EXT_ZKND);
+				SET_ISA_EXT_MAP("zkne", RISCV_ISA_EXT_ZKNE);
+				SET_ISA_EXT_MAP("zknh", RISCV_ISA_EXT_ZKNH);
+				SET_ISA_EXT_MAP("zksed", RISCV_ISA_EXT_ZKSED);
+				SET_ISA_EXT_MAP("zksh", RISCV_ISA_EXT_ZKSH);
+				SET_ISA_EXT_MAP("zkr", RISCV_ISA_EXT_ZKR);
+				SET_ISA_EXT_MAP("zkt", RISCV_ISA_EXT_ZKT);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZKND);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZKNE);
+				SET_ISA_EXT_MAP("zkn", RISCV_ISA_EXT_ZKNH);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZKSED);
+				SET_ISA_EXT_MAP("zks", RISCV_ISA_EXT_ZKSH);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZBKB);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZBKC);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZBKX);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKND);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKNE);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKNH);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKR);
+				SET_ISA_EXT_MAP("zk", RISCV_ISA_EXT_ZKT);
 			}
 #undef SET_ISA_EXT_MAP
 		}
Signature · 1 line
2.35.1
#4 lore raw

PATCH v3 2/3 RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

In reply to Hongren (Zenithal) Zheng

userspace currently lacks a way to detect whether the platform has Bitmanip/Scalar Crypto capability, this patch adds an interface such that the userspace can detect it.

RISC-V currently still has no mature mechanism, but no matter how things in the spec changes, (no matter how "M" mode things change), the kernel still needs to offer some API to the userspace.

More discussion can be found at
https://github.com/openssl/openssl/pull/18197
In short, userspace currently has to use env var to detect them.

This interface does not make any assumptions about the underlying hardware

---
 arch/riscv/include/uapi/asm/hwcap.h | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
arch/riscv/include/uapi/asm/hwcap.h +22 -0
index 46dc3f5ee99f..bfed3e5c338c 100644
@@ -22,4 +22,26 @@
 #define COMPAT_HWCAP_ISA_D	(1 << ('D' - 'A'))
 #define COMPAT_HWCAP_ISA_C	(1 << ('C' - 'A'))
 
+/*
+ * HWCAP2 flags - for elf_hwcap2 (in kernel) and AT_HWCAP2
+ *
+ * As only 32 bits of elf_hwcap (in kernel) could be used
+ * and RISC-V has reserved 26 bits of it, other caps like
+ * bitmanip and crypto can not be placed in AT_HWCAP
+ */
+#define COMPAT_HWCAP2_ISA_ZBA   (1 <<  0)
+#define COMPAT_HWCAP2_ISA_ZBB   (1 <<  1)
+#define COMPAT_HWCAP2_ISA_ZBC   (1 <<  2)
+#define COMPAT_HWCAP2_ISA_ZBS   (1 <<  3)
+#define COMPAT_HWCAP2_ISA_ZBKB  (1 <<  4)
+#define COMPAT_HWCAP2_ISA_ZBKC  (1 <<  5)
+#define COMPAT_HWCAP2_ISA_ZBKX  (1 <<  6)
+#define COMPAT_HWCAP2_ISA_ZKND  (1 <<  7)
+#define COMPAT_HWCAP2_ISA_ZKNE  (1 <<  8)
+#define COMPAT_HWCAP2_ISA_ZKNH  (1 <<  9)
+#define COMPAT_HWCAP2_ISA_ZKSED (1 << 10)
+#define COMPAT_HWCAP2_ISA_ZKSH  (1 << 11)
+#define COMPAT_HWCAP2_ISA_ZKR   (1 << 12)
+#define COMPAT_HWCAP2_ISA_ZKT   (1 << 13)
+
 #endif /* _UAPI_ASM_RISCV_HWCAP_H */
Signature · 1 line
2.35.1
#5 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 5 lines · Hongren (Zenithal) Zheng
arch/riscv/include/uapi/asm/hwcap.h +0 -0
index 46dc3f5ee99f..bfed3e5c338c 100644

arch/riscv/include/uapi/asm/hwcap.h line 31

Quoted 10 lines
arch/riscv/include/uapi/asm/hwcap.h +7 -0
@@ -22,4 +22,26 @@
 #define COMPAT_HWCAP_ISA_D	(1 << ('D' - 'A'))
 #define COMPAT_HWCAP_ISA_C	(1 << ('C' - 'A'))
 
+/*
+ * HWCAP2 flags - for elf_hwcap2 (in kernel) and AT_HWCAP2
+ *
+ * As only 32 bits of elf_hwcap (in kernel) could be used
+ * and RISC-V has reserved 26 bits of it, other caps like
+ * bitmanip and crypto can not be placed in AT_HWCAP
+ */

Have we agreed that multi-letter ISA extensions would use hwcap to be exposed to userspace? With so many potential extensions, we could quickly run out of space on AT_HWCAP2 as well.

Cheers,
Samuel.

#6 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 9 lines · Samuel Ortiz
Quoted 8 lines · Hongren (Zenithal) Zheng
arch/riscv/include/uapi/asm/hwcap.h +0 -0
index 46dc3f5ee99f..bfed3e5c338c 100644
@@ -22,4 +22,26 @@
 #define COMPAT_HWCAP_ISA_D	(1 << ('D' - 'A'))
 #define COMPAT_HWCAP_ISA_C	(1 << ('C' - 'A'))
 
Quoted 11 lines

arch/riscv/include/uapi/asm/hwcap.h line 31

Quoted 7 lines
arch/riscv/include/uapi/asm/hwcap.h +7 -0
+/*
+ * HWCAP2 flags - for elf_hwcap2 (in kernel) and AT_HWCAP2
+ *
+ * As only 32 bits of elf_hwcap (in kernel) could be used
+ * and RISC-V has reserved 26 bits of it, other caps like
+ * bitmanip and crypto can not be placed in AT_HWCAP
+ */

Have we agreed that multi-letter ISA extensions would use hwcap to be exposed to userspace? With so many potential extensions, we could quickly run out of space on AT_HWCAP2 as well.

Palmer whipped up a PoC hwprobe interface (during Plumbers I think) that Heiko is currently looking into - I think his motivation is misaligned access performance. There's a branch but I have no idea if it even compiles... I'm mostly waiting for whatever Heiko comes up with ;)

https://git.kernel.org/pub/scm/linux/kernel/git/palmer/linux.git/log/?h=riscv-hwprobe-v1

This patchset seems to need a rebase anyway per your other reply, but I guess that the new proposed interface would be preferable?

Thanks,
Conor.

#7 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 27 lines · Conor Dooley
Quoted 9 lines · Samuel Ortiz
Quoted 8 lines · Hongren (Zenithal) Zheng
arch/riscv/include/uapi/asm/hwcap.h +0 -0
index 46dc3f5ee99f..bfed3e5c338c 100644
@@ -22,4 +22,26 @@
 #define COMPAT_HWCAP_ISA_D	(1 << ('D' - 'A'))
 #define COMPAT_HWCAP_ISA_C	(1 << ('C' - 'A'))
 
Quoted 11 lines

arch/riscv/include/uapi/asm/hwcap.h line 31

Quoted 7 lines
arch/riscv/include/uapi/asm/hwcap.h +7 -0
+/*
+ * HWCAP2 flags - for elf_hwcap2 (in kernel) and AT_HWCAP2
+ *
+ * As only 32 bits of elf_hwcap (in kernel) could be used
+ * and RISC-V has reserved 26 bits of it, other caps like
+ * bitmanip and crypto can not be placed in AT_HWCAP
+ */

Have we agreed that multi-letter ISA extensions would use hwcap to be exposed to userspace? With so many potential extensions, we could quickly run out of space on AT_HWCAP2 as well.

Palmer whipped up a PoC hwprobe interface (during Plumbers I think) that Heiko is currently looking into - I think his motivation is misaligned access performance. There's a branch but I have no idea if it even compiles... I'm mostly waiting for whatever Heiko comes up with ;)

Quoted 4 lines

https://git.kernel.org/pub/scm/linux/kernel/git/palmer/linux.git/log/?h=riscv-hwprobe-v1

This patchset seems to need a rebase anyway per your other reply, but I guess that the new proposed interface would be preferable?

I think so, yes.
Patch #1 is definitely needed regardless of which interface we pick for exposing the ISA strings to userspace.

Cheers,
Samuel.

#8 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 2 lines · Samuel Ortiz

Patch #1 is definitely needed regardless of which interface we pick for exposing the ISA strings to userspace.

I took another look at #1, and I feel more confused about what constitutes canonical order than I did before! If you know better than I, and you probably do since you're interested in these 6 month old patches, some insight would be appreciated!

Thanks,
Conor.

#9 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 9 lines · Conor Dooley
Quoted 2 lines · Samuel Ortiz

Patch #1 is definitely needed regardless of which interface we pick for exposing the ISA strings to userspace.

I took another look at #1, and I feel more confused about what constitutes canonical order than I did before! If you know better than I, and you probably do since you're interested in these 6 month old patches, some insight would be appreciated!

Assuming we don't go with hwcap, I dont think the order of the riscv_isa_ext_id enum matters that much?

iiuc we're building the cpuinfo string from the riscv_isa_ext_data array, and I think the current code is incorrect:

static struct riscv_isa_ext_data isa_ext_arr[] = {
    __RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),
    __RISCV_ISA_EXT_DATA(sstc, RISCV_ISA_EXT_SSTC),
    __RISCV_ISA_EXT_DATA(svinval, RISCV_ISA_EXT_SVINVAL),
    __RISCV_ISA_EXT_DATA(svpbmt, RISCV_ISA_EXT_SVPBMT),
    __RISCV_ISA_EXT_DATA(zicbom, RISCV_ISA_EXT_ZICBOM),
    __RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
    __RISCV_ISA_EXT_DATA("", RISCV_ISA_EXT_MAX),
};

zicbom and zihintpause should come before supervisor level extensions. I'm going to send a patch for that.

And the Zb/Zk ones should come after the Zi ones, and before the supervisor level ones (The I category comes before the B or the K one). So we should check that when patch #1 is rebased.

Cheers,
Samuel.

#10 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 10 lines · Samuel Ortiz

[You don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe

Quoted 5 lines · Conor Dooley
Quoted 2 lines · Samuel Ortiz

Patch #1 is definitely needed regardless of which interface we pick for exposing the ISA strings to userspace.

Quoted 7 lines
Quoted 4 lines

I took another look at #1, and I feel more confused about what constitutes canonical order than I did before! If you know better than I, and you probably do since you're interested in these 6 month old patches, some insight would be appreciated!

Assuming we don't go with hwcap, I dont think the order of the riscv_isa_ext_id enum matters that much?

The chief put it in canonical order so that's good enough for me!

Quoted 14 lines

iiuc we're building the cpuinfo string from the riscv_isa_ext_data array, and I think the current code is incorrect:

static struct riscv_isa_ext_data isa_ext_arr[] = {
    __RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),
    __RISCV_ISA_EXT_DATA(sstc, RISCV_ISA_EXT_SSTC),
    __RISCV_ISA_EXT_DATA(svinval, RISCV_ISA_EXT_SVINVAL),
    __RISCV_ISA_EXT_DATA(svpbmt, RISCV_ISA_EXT_SVPBMT),
    __RISCV_ISA_EXT_DATA(zicbom, RISCV_ISA_EXT_ZICBOM),
    __RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
    __RISCV_ISA_EXT_DATA("", RISCV_ISA_EXT_MAX),
};
Quoted 2 lines

zicbom and zihintpause should come before supervisor level extensions. I'm going to send a patch for that.

idk, Palmer explicitly re-ordered this:
https://lore.kernel.org/linux-riscv/[email protected]/

By my reading of the isa manual, what Palmer did is correct as those are not "Additional Standard Extensions". /shrug

Quoted 3 lines

And the Zb/Zk ones should come after the Zi ones, and before the supervisor level ones (The I category comes before the B or the K one).

This I agree with though.

#11 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 35 lines · Conor Dooley
Quoted 10 lines · Samuel Ortiz

[You don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe

Quoted 5 lines · Conor Dooley
Quoted 2 lines · Samuel Ortiz

Patch #1 is definitely needed regardless of which interface we pick for exposing the ISA strings to userspace.

Quoted 7 lines
Quoted 4 lines

I took another look at #1, and I feel more confused about what constitutes canonical order than I did before! If you know better than I, and you probably do since you're interested in these 6 month old patches, some insight would be appreciated!

Assuming we don't go with hwcap, I dont think the order of the riscv_isa_ext_id enum matters that much?

The chief put it in canonical order so that's good enough for me!

Quoted 14 lines

iiuc we're building the cpuinfo string from the riscv_isa_ext_data array, and I think the current code is incorrect:

static struct riscv_isa_ext_data isa_ext_arr[] = {
    __RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),
    __RISCV_ISA_EXT_DATA(sstc, RISCV_ISA_EXT_SSTC),
    __RISCV_ISA_EXT_DATA(svinval, RISCV_ISA_EXT_SVINVAL),
    __RISCV_ISA_EXT_DATA(svpbmt, RISCV_ISA_EXT_SVPBMT),
    __RISCV_ISA_EXT_DATA(zicbom, RISCV_ISA_EXT_ZICBOM),
    __RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
    __RISCV_ISA_EXT_DATA("", RISCV_ISA_EXT_MAX),
};
Quoted 8 lines
Quoted 2 lines

zicbom and zihintpause should come before supervisor level extensions. I'm going to send a patch for that.

idk, Palmer explicitly re-ordered this:
https://lore.kernel.org/linux-riscv/[email protected]/

By my reading of the isa manual, what Palmer did is correct as those are not "Additional Standard Extensions". /shrug

Hmm, by their name (Z[a-b]+) they are Additional Standard Extensions. What am I missing?

Cheers,
Samuel.

Quoted 1 line
#12 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 39 lines · Samuel Ortiz
Quoted 38 lines · Conor Dooley
Quoted 10 lines · Samuel Ortiz

[You don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe

Quoted 5 lines · Conor Dooley
Quoted 2 lines · Samuel Ortiz

Patch #1 is definitely needed regardless of which interface we pick for exposing the ISA strings to userspace.

Quoted 7 lines
Quoted 4 lines

I took another look at #1, and I feel more confused about what constitutes canonical order than I did before! If you know better than I, and you probably do since you're interested in these 6 month old patches, some insight would be appreciated!

Assuming we don't go with hwcap, I dont think the order of the riscv_isa_ext_id enum matters that much?

The chief put it in canonical order so that's good enough for me!

Quoted 16 lines

iiuc we're building the cpuinfo string from the riscv_isa_ext_data array, and I think the current code is incorrect:

static struct riscv_isa_ext_data isa_ext_arr[] = {
    __RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),
    __RISCV_ISA_EXT_DATA(sstc, RISCV_ISA_EXT_SSTC),
    __RISCV_ISA_EXT_DATA(svinval, RISCV_ISA_EXT_SVINVAL),
    __RISCV_ISA_EXT_DATA(svpbmt, RISCV_ISA_EXT_SVPBMT),
    __RISCV_ISA_EXT_DATA(zicbom, RISCV_ISA_EXT_ZICBOM),
    __RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
    __RISCV_ISA_EXT_DATA("", RISCV_ISA_EXT_MAX),
};

zicbom and zihintpause should come before supervisor level extensions. I'm going to send a patch for that.

Quoted 8 lines
Quoted 5 lines

idk, Palmer explicitly re-ordered this:
https://lore.kernel.org/linux-riscv/[email protected]/

By my reading of the isa manual, what Palmer did is correct as those are not "Additional Standard Extensions". /shrug

Hmm, by their name (Z[a-b]+) they are Additional Standard Extensions. What am I missing?

Right, and this is where I get confused. Zam and Ztso *are* Additional Standard Extensions, I think we can agree on that one? For those extensions:
\chapter{``Ztso'' Standard Extension for Total Store Ordering, v0.1} \chapter{``Zam'' Standard Extension for Misaligned Atomics, v0.1}

They're also called out specifically in the table:
https://github.com/riscv/riscv-isa-manual/blob/master/src/naming.tex#L147

For Zihintpause however:
\chapter{``Zihintpause'' Pause Hint, Version 2.0}

See what I mean? I looked at the specs for the bitmanip stuff and for crypto, which both never mention being standard.

That table has the caption:

Quoted 3 lines

The table also defines the canonical order in which extension names must appear in the name string, with top-to-bottom in table indicating first-to-last in the name string.

It only calls out Zicsr, Zifencei, Zam and Ztso are being permitted before Sdef, but as I said I am not a specs person, so perhaps some of the extensions in question are intended to go there but have not yet been merged into the isa manual doc. Zihintpause *is* in the isa manual though but not specifically called out.

Anyways, hopefully that at least helps with my line of thinking!

Conor.

#13 lore raw

Re: [PATCH v3 2/3] RISC-V: uapi: add HWCAP for Bitmanip/Scalar Crypto

Quoted 55 lines · Conor Dooley
Quoted 39 lines · Samuel Ortiz
Quoted 38 lines · Conor Dooley
Quoted 10 lines · Samuel Ortiz

[You don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe

Quoted 5 lines · Conor Dooley
Quoted 2 lines · Samuel Ortiz

Patch #1 is definitely needed regardless of which interface we pick for exposing the ISA strings to userspace.

Quoted 7 lines
Quoted 4 lines

I took another look at #1, and I feel more confused about what constitutes canonical order than I did before! If you know better than I, and you probably do since you're interested in these 6 month old patches, some insight would be appreciated!

Assuming we don't go with hwcap, I dont think the order of the riscv_isa_ext_id enum matters that much?

The chief put it in canonical order so that's good enough for me!

Quoted 16 lines

iiuc we're building the cpuinfo string from the riscv_isa_ext_data array, and I think the current code is incorrect:

static struct riscv_isa_ext_data isa_ext_arr[] = {
    __RISCV_ISA_EXT_DATA(sscofpmf, RISCV_ISA_EXT_SSCOFPMF),
    __RISCV_ISA_EXT_DATA(sstc, RISCV_ISA_EXT_SSTC),
    __RISCV_ISA_EXT_DATA(svinval, RISCV_ISA_EXT_SVINVAL),
    __RISCV_ISA_EXT_DATA(svpbmt, RISCV_ISA_EXT_SVPBMT),
    __RISCV_ISA_EXT_DATA(zicbom, RISCV_ISA_EXT_ZICBOM),
    __RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
    __RISCV_ISA_EXT_DATA("", RISCV_ISA_EXT_MAX),
};

zicbom and zihintpause should come before supervisor level extensions. I'm going to send a patch for that.

Quoted 8 lines
Quoted 5 lines

idk, Palmer explicitly re-ordered this:
https://lore.kernel.org/linux-riscv/[email protected]/

By my reading of the isa manual, what Palmer did is correct as those are not "Additional Standard Extensions". /shrug

Hmm, by their name (Z[a-b]+) they are Additional Standard Extensions. What am I missing?

Right, and this is where I get confused. Zam and Ztso *are* Additional Standard Extensions, I think we can agree on that one? For those extensions:
\chapter{``Ztso'' Standard Extension for Total Store Ordering, v0.1} \chapter{``Zam'' Standard Extension for Misaligned Atomics, v0.1}

Quoted 8 lines

They're also called out specifically in the table:
https://github.com/riscv/riscv-isa-manual/blob/master/src/naming.tex#L147

For Zihintpause however:
\chapter{``Zihintpause'' Pause Hint, Version 2.0}

See what I mean? I looked at the specs for the bitmanip stuff and for crypto, which both never mention being standard.

I *think* this is because Zihintpause, bitmap and crypto are ratified but not yet part of an official spec (non-draft) release?

Quoted 5 lines

That table has the caption:

Quoted 3 lines

The table also defines the canonical order in which extension names must appear in the name string, with top-to-bottom in table indicating first-to-last in the name string.

Quoted 7 lines

It only calls out Zicsr, Zifencei, Zam and Ztso are being permitted before Sdef, but as I said I am not a specs person, so perhaps some of the extensions in question are intended to go there but have not yet been merged into the isa manual doc. Zihintpause *is* in the isa manual though but not specifically called out.

Anyways, hopefully that at least helps with my line of thinking!

It does, thanks. It's a little confusing, I agree.

Cheers,
Samuel.

#14 lore raw

PATCH v3 3/3 RISC-V: HWCAP: parse Bitmanip/Scalar Crypto HWCAP from DT

In reply to Hongren (Zenithal) Zheng

One viable way to detect Zb/Zk HWCAP is from the DT binding. No matter how the "M" mode things change, this way can always be an auxiliary way to detect it.

Note that QEMU currently has "zba/zbb/zbc/zbs" in their device tree riscv,isa

This also fixes the isa2hwcap way as using unsigned char for long extension is not viable. Note that the tolower function ensures functionality. For other no-hwcap extension (e.g. h, s, u), or ("|") with 0 has no effect on hwcap.

---
 arch/riscv/include/asm/elf.h   |  2 ++
 arch/riscv/include/asm/hwcap.h |  2 ++
 arch/riscv/kernel/cpufeature.c | 48 ++++++++++++++++++++++++++--------
 3 files changed, 41 insertions(+), 11 deletions(-)
arch/riscv/include/asm/elf.h +2 -0
index 14fc7342490b..cbf70c5ac1a4 100644
@@ -65,7 +65,9 @@ extern bool compat_elf_check_arch(Elf32_Ehdr *hdr);
  * but it's not easy, and we've already done it here.
  */
 #define ELF_HWCAP	(elf_hwcap)
+#define ELF_HWCAP2	(elf_hwcap2)
 extern unsigned long elf_hwcap;
+extern unsigned long elf_hwcap2;
 
 /*
  * This yields a string that ld.so will use to load implementation
arch/riscv/include/asm/hwcap.h +2 -0
index 02c454a12683..ef0349c5d303 100644
@@ -17,12 +17,14 @@
  * instruction set this cpu supports.
  */
 #define ELF_HWCAP		(elf_hwcap)
+#define ELF_HWCAP2		(elf_hwcap2)
 
 enum {
 	CAP_HWCAP = 1,
 };
 
 extern unsigned long elf_hwcap;
+extern unsigned long elf_hwcap2;
 
 #define RISCV_ISA_EXT_a		('a' - 'a')
 #define RISCV_ISA_EXT_c		('c' - 'a')
arch/riscv/kernel/cpufeature.c +37 -11

85 lines not on this page · raw message

Signature · 1 line
2.35.1