[PATCH v7 0/6] RISC-V: ifunced memcpy using new kernel hwprobe interface

[PATCH v7 0/6] RISC-V: ifunced memcpy using new kernel hwprobe interface

[Evan Green]

This series illustrates the use of a recently accepted Linux syscall that
enumerates architectural information about the RISC-V cores the system
is running on. In this series we expose a small wrapper function around
the syscall. An ifunc selector for memcpy queries it to see if unaligned
access is "fast" on this hardware. If it is, it selects a newly provided
implementation of memcpy that doesn't work hard at aligning the src and
destination buffers.

For applications and libraries outside of glibc that want to use
__riscv_hwprobe() in ifunc selectors, this series also sends a pointer
to the riscv_hwprobe() function in as the second argument to ifunc
selectors. A new inline convenience function can help application and
library callers to check for validity and quickly probe a single key.

The memcpy implementation is independent enough from the rest of the
series that it can be omitted safely if desired.

Performance numbers were compared using a small test program [1], run on
a D1 Nezha board, which supports fast unaligned access. "Fast" here
means copying unaligned words is faster than copying byte-wise, but
still slower than copying aligned words. Here's the speed of various
memcpy()s with the generic implementation. The numbers before are using
v4's memcpy implementation, with the "copy last byte via overlapping
misaligned word" fix this should get slightly better.

memcpy size 1 count 1000000 offset 0 took 109564 us
memcpy size 3 count 1000000 offset 0 took 138425 us
memcpy size 4 count 1000000 offset 0 took 148374 us
memcpy size 7 count 1000000 offset 0 took 178433 us
memcpy size 8 count 1000000 offset 0 took 188430 us
memcpy size f count 1000000 offset 0 took 266118 us
memcpy size f count 1000000 offset 1 took 265940 us
memcpy size f count 1000000 offset 3 took 265934 us
memcpy size f count 1000000 offset 7 took 266215 us
memcpy size f count 1000000 offset 8 took 265954 us
memcpy size f count 1000000 offset 9 took 265886 us
memcpy size 10 count 1000000 offset 0 took 195308 us
memcpy size 11 count 1000000 offset 0 took 205161 us
memcpy size 17 count 1000000 offset 0 took 274376 us
memcpy size 18 count 1000000 offset 0 took 199188 us
memcpy size 19 count 1000000 offset 0 took 209258 us
memcpy size 1f count 1000000 offset 0 took 278263 us
memcpy size 20 count 1000000 offset 0 took 207364 us
memcpy size 21 count 1000000 offset 0 took 217143 us
memcpy size 3f count 1000000 offset 0 took 300023 us
memcpy size 40 count 1000000 offset 0 took 231063 us
memcpy size 41 count 1000000 offset 0 took 241259 us
memcpy size 7c count 100000 offset 0 took 32807 us
memcpy size 7f count 100000 offset 0 took 36274 us
memcpy size ff count 100000 offset 0 took 47818 us
memcpy size ff count 100000 offset 0 took 47932 us
memcpy size 100 count 100000 offset 0 took 40468 us
memcpy size 200 count 100000 offset 0 took 64245 us
memcpy size 27f count 100000 offset 0 took 82549 us
memcpy size 400 count 100000 offset 0 took 111254 us
memcpy size 407 count 100000 offset 0 took 119364 us
memcpy size 800 count 100000 offset 0 took 203899 us
memcpy size 87f count 100000 offset 0 took 222465 us
memcpy size 87f count 100000 offset 3 took 222289 us
memcpy size 1000 count 100000 offset 0 took 388846 us
memcpy size 1000 count 100000 offset 1 took 468827 us
memcpy size 1000 count 100000 offset 3 took 397098 us
memcpy size 1000 count 100000 offset 4 took 397379 us
memcpy size 1000 count 100000 offset 5 took 397368 us
memcpy size 1000 count 100000 offset 7 took 396867 us
memcpy size 1000 count 100000 offset 8 took 389227 us
memcpy size 1000 count 100000 offset 9 took 395949 us
memcpy size 3000 count 50000 offset 0 took 674837 us
memcpy size 3000 count 50000 offset 1 took 676944 us
memcpy size 3000 count 50000 offset 3 took 679709 us
memcpy size 3000 count 50000 offset 4 took 680829 us
memcpy size 3000 count 50000 offset 5 took 678024 us
memcpy size 3000 count 50000 offset 7 took 681097 us
memcpy size 3000 count 50000 offset 8 took 670004 us
memcpy size 3000 count 50000 offset 9 took 674553 us

Here is that same test run with the assembly memcpy() in this series:
memcpy size 1 count 1000000 offset 0 took 92703 us
memcpy size 3 count 1000000 offset 0 took 112527 us
memcpy size 4 count 1000000 offset 0 took 120481 us
memcpy size 7 count 1000000 offset 0 took 149558 us
memcpy size 8 count 1000000 offset 0 took 90617 us
memcpy size f count 1000000 offset 0 took 174373 us
memcpy size f count 1000000 offset 1 took 178615 us
memcpy size f count 1000000 offset 3 took 178845 us
memcpy size f count 1000000 offset 7 took 178636 us
memcpy size f count 1000000 offset 8 took 174442 us
memcpy size f count 1000000 offset 9 took 178660 us
memcpy size 10 count 1000000 offset 0 took 99845 us
memcpy size 11 count 1000000 offset 0 took 112522 us
memcpy size 17 count 1000000 offset 0 took 179735 us
memcpy size 18 count 1000000 offset 0 took 110870 us
memcpy size 19 count 1000000 offset 0 took 121472 us
memcpy size 1f count 1000000 offset 0 took 188231 us
memcpy size 20 count 1000000 offset 0 took 119571 us
memcpy size 21 count 1000000 offset 0 took 132429 us
memcpy size 3f count 1000000 offset 0 took 227021 us
memcpy size 40 count 1000000 offset 0 took 166416 us
memcpy size 41 count 1000000 offset 0 took 180206 us
memcpy size 7c count 100000 offset 0 took 28602 us
memcpy size 7f count 100000 offset 0 took 31676 us
memcpy size ff count 100000 offset 0 took 39257 us
memcpy size ff count 100000 offset 0 took 39176 us
memcpy size 100 count 100000 offset 0 took 21928 us
memcpy size 200 count 100000 offset 0 took 35814 us
memcpy size 27f count 100000 offset 0 took 60315 us
memcpy size 400 count 100000 offset 0 took 63652 us
memcpy size 407 count 100000 offset 0 took 73160 us
memcpy size 800 count 100000 offset 0 took 121532 us
memcpy size 87f count 100000 offset 0 took 147269 us
memcpy size 87f count 100000 offset 3 took 144744 us
memcpy size 1000 count 100000 offset 0 took 232057 us
memcpy size 1000 count 100000 offset 1 took 254319 us
memcpy size 1000 count 100000 offset 3 took 256973 us
memcpy size 1000 count 100000 offset 4 took 257655 us
memcpy size 1000 count 100000 offset 5 took 259456 us
memcpy size 1000 count 100000 offset 7 took 260849 us
memcpy size 1000 count 100000 offset 8 took 232347 us
memcpy size 1000 count 100000 offset 9 took 254330 us
memcpy size 3000 count 50000 offset 0 took 382376 us
memcpy size 3000 count 50000 offset 1 took 389872 us
memcpy size 3000 count 50000 offset 3 took 385310 us
memcpy size 3000 count 50000 offset 4 took 389748 us
memcpy size 3000 count 50000 offset 5 took 391707 us
memcpy size 3000 count 50000 offset 7 took 386778 us
memcpy size 3000 count 50000 offset 8 took 385691 us
memcpy size 3000 count 50000 offset 9 took 392030 us

The assembly routine is measurably better.

[1] https://pastebin.com/DRyECNQW


Changes in v7:
 - Bumped Versions up to 2.39 (Joseph)
 - Used INTERNAL_SYSCALL_CALL, and return positive errno to match
   pthreads API (Florian).
 - Remove __THROW since it creates a warning in combination with the
   fortified access decorators.
 - Use INTERNAL_VSYSCALL_CALL (Florian)
 - Remove __THROW from function pointer type, as it creates warnings
   together with __fortified_attr_access.
 - Introduced static inline helper (Richard)
 - Use new helper function in memcpy ifunc selector (Richard)

Changes in v6:
 - Prefixed __riscv_hwprobe() parameters names with __ to avoid user
   macro namespace pollution (Joseph)
 - Introduced riscv-ifunc.h for multi-arg ifunc selectors.
 - Fix a couple regressions in the assembly from v5 :/
 - Use passed hwprobe pointer in memcpy ifunc selector.

Changes in v5:
 - Do unaligned word access for final trailing bytes (Richard)

Changes in v4:
 - Remove __USE_GNU (Florian)
 - __nonnull, __wur, __THROW, and  __fortified_attr_access decorations
  (Florian)
 - change long to long int (Florian)
 - Fix comment formatting (Florian)
 - Update backup kernel header content copy.
 - Fix function declaration formatting (Florian)
 - Changed export versions to 2.38
 - Fixed comment style (Florian)

Changes in v3:
 - Update argument types to match v4 kernel interface
 - Add the "return" to the vsyscall
 - Fix up vdso arg types to match kernel v4 version
 - Remove ifdef around INLINE_VSYSCALL (Adhemerval)
 - Word align dest for large memcpy()s.
 - Add tags
 - Remove spurious blank line from sysdeps/riscv/memcpy.c

Changes in v2:
 - hwprobe.h: Use __has_include and duplicate Linux content to make
   compilation work when Linux headers are absent (Adhemerval)
 - hwprobe.h: Put declaration under __USE_GNU (Adhemerval)
 - Use INLINE_SYSCALL_CALL (Adhemerval)
 - Update versions
 - Update UNALIGNED_MASK to match kernel v3 series.
 - Add vDSO interface
 - Used _MASK instead of _FAST value itself.

Evan Green (6):
  riscv: Add Linux hwprobe syscall support
  riscv: Add hwprobe vdso call support
  riscv: Add __riscv_hwprobe pointer to ifunc calls
  riscv: Enable multi-arg ifunc resolvers
  riscv: Add ifunc helper method to hwprobe.h
  riscv: Add and use alignment-ignorant memcpy

 include/libc-symbols.h                        |  28 ++--
 sysdeps/riscv/dl-irel.h                       |   8 +-
 sysdeps/riscv/memcopy.h                       |  26 ++++
 sysdeps/riscv/memcpy.c                        |  63 ++++++++
 sysdeps/riscv/memcpy_noalignment.S            | 138 ++++++++++++++++++
 sysdeps/riscv/riscv-ifunc.h                   |  27 ++++
 sysdeps/unix/sysv/linux/dl-vdso-setup.c       |  10 ++
 sysdeps/unix/sysv/linux/dl-vdso-setup.h       |   3 +
 sysdeps/unix/sysv/linux/riscv/Makefile        |   8 +-
 sysdeps/unix/sysv/linux/riscv/Versions        |   3 +
 sysdeps/unix/sysv/linux/riscv/hwprobe.c       |  50 +++++++
 .../unix/sysv/linux/riscv/memcpy-generic.c    |  24 +++
 .../unix/sysv/linux/riscv/rv32/libc.abilist   |   1 +
 .../unix/sysv/linux/riscv/rv64/libc.abilist   |   1 +
 sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h   | 106 ++++++++++++++
 sysdeps/unix/sysv/linux/riscv/sysdep.h        |   1 +
 16 files changed, 480 insertions(+), 17 deletions(-)
 create mode 100644 sysdeps/riscv/memcopy.h
 create mode 100644 sysdeps/riscv/memcpy.c
 create mode 100644 sysdeps/riscv/memcpy_noalignment.S
 create mode 100644 sysdeps/riscv/riscv-ifunc.h
 create mode 100644 sysdeps/unix/sysv/linux/riscv/hwprobe.c
 create mode 100644 sysdeps/unix/sysv/linux/riscv/memcpy-generic.c
 create mode 100644 sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h

-- 
2.34.1

[PATCH v7 1/6] riscv: Add Linux hwprobe syscall support

[Evan Green]

Add awareness and a thin wrapper function around a new Linux system call
that allows callers to get architecture and microarchitecture
information about the CPUs from the kernel. This can be used to
do things like dynamically choose a memcpy implementation.

Signed-off-by: Evan Green <evan@rivosinc.com>
Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>
---

Changes in v7:
 - Bumped Versions up to 2.39 (Joseph)
 - Used INTERNAL_SYSCALL_CALL, and return positive errno to match
   pthreads API (Florian).
 - Remove __THROW since it creates a warning in combination with the
   fortified access decorators.

Changes in v6:
 - Prefixed __riscv_hwprobe() parameters names with __ to avoid user
   macro namespace pollution (Joseph)

Changes in v4:
 - Remove __USE_GNU (Florian)
 - __nonnull, __wur, __THROW, and  __fortified_attr_access decorations
  (Florian)
 - change long to long int (Florian)
 - Fix comment formatting (Florian)
 - Update backup kernel header content copy.
 - Fix function declaration formatting (Florian)
 - Changed export versions to 2.38

Changes in v3:
 - Update argument types to match v4 kernel interface

Changes in v2:
 - hwprobe.h: Use __has_include and duplicate Linux content to make
   compilation work when Linux headers are absent (Adhemerval)
 - hwprobe.h: Put declaration under __USE_GNU (Adhemerval)
 - Use INLINE_SYSCALL_CALL (Adhemerval)
 - Update versions
 - Update UNALIGNED_MASK to match kernel v3 series.

 sysdeps/unix/sysv/linux/riscv/Makefile        |  4 +-
 sysdeps/unix/sysv/linux/riscv/Versions        |  3 +
 sysdeps/unix/sysv/linux/riscv/hwprobe.c       | 39 ++++++++++
 .../unix/sysv/linux/riscv/rv32/libc.abilist   |  1 +
 .../unix/sysv/linux/riscv/rv64/libc.abilist   |  1 +
 sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h   | 72 +++++++++++++++++++
 6 files changed, 118 insertions(+), 2 deletions(-)
 create mode 100644 sysdeps/unix/sysv/linux/riscv/hwprobe.c
 create mode 100644 sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h

diff --git a/sysdeps/unix/sysv/linux/riscv/Makefile b/sysdeps/unix/sysv/linux/riscv/Makefile
index 4b6eacb32f..45cc29e40d 100644
--- a/sysdeps/unix/sysv/linux/riscv/Makefile
+++ b/sysdeps/unix/sysv/linux/riscv/Makefile
@@ -1,6 +1,6 @@
 ifeq ($(subdir),misc)
-sysdep_headers += sys/cachectl.h
-sysdep_routines += flush-icache
+sysdep_headers += sys/cachectl.h sys/hwprobe.h
+sysdep_routines += flush-icache hwprobe
 endif
 
 ifeq ($(subdir),stdlib)
diff --git a/sysdeps/unix/sysv/linux/riscv/Versions b/sysdeps/unix/sysv/linux/riscv/Versions
index 5625d2a0b8..8717b62a4a 100644
--- a/sysdeps/unix/sysv/linux/riscv/Versions
+++ b/sysdeps/unix/sysv/linux/riscv/Versions
@@ -8,4 +8,7 @@ libc {
   GLIBC_2.27 {
     __riscv_flush_icache;
   }
+  GLIBC_2.39 {
+    __riscv_hwprobe;
+  }
 }
diff --git a/sysdeps/unix/sysv/linux/riscv/hwprobe.c b/sysdeps/unix/sysv/linux/riscv/hwprobe.c
new file mode 100644
index 0000000000..f2dc46cfd7
--- /dev/null
+++ b/sysdeps/unix/sysv/linux/riscv/hwprobe.c
@@ -0,0 +1,39 @@
+/* RISC-V hardware feature probing support on Linux
+   Copyright (C) 2023 Free Software Foundation, Inc.
+
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public License as
+   published by the Free Software Foundation; either version 2.1 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sys/syscall.h>
+#include <sys/hwprobe.h>
+#include <sysdep.h>
+#include <sysdep-vdso.h>
+
+int __riscv_hwprobe (struct riscv_hwprobe *__pairs, size_t __pair_count,
+		     size_t __cpu_count, unsigned long int *__cpus,
+		     unsigned int __flags)
+{
+  int r;
+
+  r = INTERNAL_SYSCALL_CALL (riscv_hwprobe, 5, __pairs, __pair_count,
+                             __cpu_count, __cpus, __flags);
+
+  /* Negate negative errno values to match pthreads API. */
+  if (r < 0)
+    r = -r;
+
+  return r;
+}
diff --git a/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist b/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist
index b9740a1afc..8fab4a606f 100644
--- a/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist
+++ b/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist
@@ -2436,3 +2436,4 @@ GLIBC_2.38 strlcat F
 GLIBC_2.38 strlcpy F
 GLIBC_2.38 wcslcat F
 GLIBC_2.38 wcslcpy F
+GLIBC_2.38 __riscv_hwprobe F
diff --git a/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist b/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist
index e3b4656aa2..1ebb91deed 100644
--- a/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist
+++ b/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist
@@ -2636,3 +2636,4 @@ GLIBC_2.38 strlcat F
 GLIBC_2.38 strlcpy F
 GLIBC_2.38 wcslcat F
 GLIBC_2.38 wcslcpy F
+GLIBC_2.38 __riscv_hwprobe F
diff --git a/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h b/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
new file mode 100644
index 0000000000..aa189a4818
--- /dev/null
+++ b/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
@@ -0,0 +1,72 @@
+/* RISC-V architecture probe interface
+   Copyright (C) 2023 Free Software Foundation, Inc.
+
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library.  If not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#ifndef _SYS_HWPROBE_H
+#define _SYS_HWPROBE_H 1
+
+#include <features.h>
+#include <stddef.h>
+#ifdef __has_include
+# if __has_include (<asm/hwprobe.h>)
+#  include <asm/hwprobe.h>
+# endif
+#endif
+
+/* Define a (probably stale) version of the interface if the Linux headers
+   aren't present.  */
+#ifndef RISCV_HWPROBE_KEY_MVENDORID
+struct riscv_hwprobe {
+	signed long long int key;
+	unsigned long long int value;
+};
+
+#define RISCV_HWPROBE_KEY_MVENDORID	0
+#define RISCV_HWPROBE_KEY_MARCHID	1
+#define RISCV_HWPROBE_KEY_MIMPID	2
+#define RISCV_HWPROBE_KEY_BASE_BEHAVIOR	3
+#define		RISCV_HWPROBE_BASE_BEHAVIOR_IMA	(1 << 0)
+#define RISCV_HWPROBE_KEY_IMA_EXT_0	4
+#define		RISCV_HWPROBE_IMA_FD		(1 << 0)
+#define		RISCV_HWPROBE_IMA_C		(1 << 1)
+#define		RISCV_HWPROBE_IMA_V		(1 << 2)
+#define		RISCV_HWPROBE_EXT_ZBA		(1 << 3)
+#define		RISCV_HWPROBE_EXT_ZBB		(1 << 4)
+#define		RISCV_HWPROBE_EXT_ZBS		(1 << 5)
+#define RISCV_HWPROBE_KEY_CPUPERF_0	5
+#define		RISCV_HWPROBE_MISALIGNED_UNKNOWN	(0 << 0)
+#define		RISCV_HWPROBE_MISALIGNED_EMULATED	(1 << 0)
+#define		RISCV_HWPROBE_MISALIGNED_SLOW		(2 << 0)
+#define		RISCV_HWPROBE_MISALIGNED_FAST		(3 << 0)
+#define		RISCV_HWPROBE_MISALIGNED_UNSUPPORTED	(4 << 0)
+#define		RISCV_HWPROBE_MISALIGNED_MASK		(7 << 0)
+
+#endif /* RISCV_HWPROBE_KEY_MVENDORID */
+
+__BEGIN_DECLS
+
+extern int __riscv_hwprobe (struct riscv_hwprobe *__pairs, size_t __pair_count,
+			    size_t __cpu_count, unsigned long int *__cpus,
+			    unsigned int __flags)
+     __nonnull ((1)) __wur
+     __fortified_attr_access (__read_write__, 1, 2)
+     __fortified_attr_access (__read_only__, 4, 3);
+
+__END_DECLS
+
+#endif /* sys/hwprobe.h */
-- 
2.34.1

[PATCH v7 2/6] riscv: Add hwprobe vdso call support

[Evan Green]

The new riscv_hwprobe syscall also comes with a vDSO for faster answers
to your most common questions. Call in today to speak with a kernel
representative near you!

Signed-off-by: Evan Green <evan@rivosinc.com>
Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>
---

Changes in v7:
 - Use INTERNAL_VSYSCALL_CALL (Florian)

Changes in v3:
 - Add the "return" to the vsyscall
 - Fix up vdso arg types to match kernel v4 version
 - Remove ifdef around INLINE_VSYSCALL (Adhemerval)

Changes in v2:
 - Add vDSO interface

 sysdeps/unix/sysv/linux/dl-vdso-setup.c | 10 ++++++++++
 sysdeps/unix/sysv/linux/dl-vdso-setup.h |  3 +++
 sysdeps/unix/sysv/linux/riscv/hwprobe.c | 17 ++++++++++++++---
 sysdeps/unix/sysv/linux/riscv/sysdep.h  |  1 +
 4 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/sysdeps/unix/sysv/linux/dl-vdso-setup.c b/sysdeps/unix/sysv/linux/dl-vdso-setup.c
index 97eaaeac37..ed8b1ef426 100644
--- a/sysdeps/unix/sysv/linux/dl-vdso-setup.c
+++ b/sysdeps/unix/sysv/linux/dl-vdso-setup.c
@@ -71,6 +71,16 @@ PROCINFO_CLASS int (*_dl_vdso_clock_getres_time64) (clockid_t,
 # ifdef HAVE_GET_TBFREQ
 PROCINFO_CLASS uint64_t (*_dl_vdso_get_tbfreq)(void) RELRO;
 # endif
+
+/* RISC-V specific ones.  */
+# ifdef HAVE_RISCV_HWPROBE
+PROCINFO_CLASS int (*_dl_vdso_riscv_hwprobe)(void *,
+                                             size_t,
+                                             size_t,
+                                             unsigned long *,
+                                             unsigned int) RELRO;
+# endif
+
 #endif
 
 #undef RELRO
diff --git a/sysdeps/unix/sysv/linux/dl-vdso-setup.h b/sysdeps/unix/sysv/linux/dl-vdso-setup.h
index 867072b897..39eafd5316 100644
--- a/sysdeps/unix/sysv/linux/dl-vdso-setup.h
+++ b/sysdeps/unix/sysv/linux/dl-vdso-setup.h
@@ -47,6 +47,9 @@ setup_vdso_pointers (void)
 #ifdef HAVE_GET_TBFREQ
   GLRO(dl_vdso_get_tbfreq) = dl_vdso_vsym (HAVE_GET_TBFREQ);
 #endif
+#ifdef HAVE_RISCV_HWPROBE
+  GLRO(dl_vdso_riscv_hwprobe) = dl_vdso_vsym (HAVE_RISCV_HWPROBE);
+#endif
 }
 
 #endif
diff --git a/sysdeps/unix/sysv/linux/riscv/hwprobe.c b/sysdeps/unix/sysv/linux/riscv/hwprobe.c
index f2dc46cfd7..57a4b8a0a8 100644
--- a/sysdeps/unix/sysv/linux/riscv/hwprobe.c
+++ b/sysdeps/unix/sysv/linux/riscv/hwprobe.c
@@ -27,9 +27,20 @@ int __riscv_hwprobe (struct riscv_hwprobe *__pairs, size_t __pair_count,
 		     unsigned int __flags)
 {
   int r;
-
-  r = INTERNAL_SYSCALL_CALL (riscv_hwprobe, 5, __pairs, __pair_count,
-                             __cpu_count, __cpus, __flags);
+  __riscv_hwprobe_t vdso_hwprobe =
+    (__riscv_hwprobe_t)GLRO(dl_vdso_riscv_hwprobe);
+
+  if (vdso_hwprobe != NULL)
+    {
+      r = INTERNAL_VSYSCALL_CALL (vdso_hwprobe, 5, __pairs, __pair_count,
+                                  __cpu_count, __cpus, __flags);
+    }
+  else
+    {
+      r = INTERNAL_SYSCALL_CALL (riscv_hwprobe, 5, __pairs, __pair_count,
+                                 __cpu_count, __cpus, __flags);
+
+    }
 
   /* Negate negative errno values to match pthreads API. */
   if (r < 0)
diff --git a/sysdeps/unix/sysv/linux/riscv/sysdep.h b/sysdeps/unix/sysv/linux/riscv/sysdep.h
index 5583b96d23..ee015dfeb6 100644
--- a/sysdeps/unix/sysv/linux/riscv/sysdep.h
+++ b/sysdeps/unix/sysv/linux/riscv/sysdep.h
@@ -156,6 +156,7 @@
 /* List of system calls which are supported as vsyscalls (for RV32 and
    RV64).  */
 # define HAVE_GETCPU_VSYSCALL		"__vdso_getcpu"
+# define HAVE_RISCV_HWPROBE		"__vdso_riscv_hwprobe"
 
 # undef HAVE_INTERNAL_BRK_ADDR_SYMBOL
 # define HAVE_INTERNAL_BRK_ADDR_SYMBOL 1
-- 
2.34.1

[PATCH v7 3/6] riscv: Add __riscv_hwprobe pointer to ifunc calls

[Evan Green]

The new __riscv_hwprobe() function is designed to be used by ifunc
selector functions. This presents a challenge for applications and
libraries, as ifunc selectors are invoked before all relocations have
been performed, so an external call to __riscv_hwprobe() from an ifunc
selector won't work. To address this, pass a pointer to the
__riscv_hwprobe() vDSO function into ifunc selectors as the second
argument (alongside dl_hwcap, which was already being passed).

Include a typedef as well for convenience, so that ifunc users don't
have to go through contortions to call this routine. Users will need to
remember to check the second argument for NULL, both to account for
older glibcs that don't pass the function, and older kernels that don't
have the vDSO pointer.

Signed-off-by: Evan Green <evan@rivosinc.com>

---

Changes in v7:
 - Remove __THROW from function pointer type, as it creates warnings
   together with __fortified_attr_access.

 sysdeps/riscv/dl-irel.h                     |  8 ++++----
 sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h | 10 ++++++++++
 2 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/sysdeps/riscv/dl-irel.h b/sysdeps/riscv/dl-irel.h
index eaeec5467c..2147504458 100644
--- a/sysdeps/riscv/dl-irel.h
+++ b/sysdeps/riscv/dl-irel.h
@@ -31,10 +31,10 @@ static inline ElfW(Addr)
 __attribute ((always_inline))
 elf_ifunc_invoke (ElfW(Addr) addr)
 {
-  /* The second argument is a void pointer to preserve the extension
-     fexibility.  */
-  return ((ElfW(Addr) (*) (uint64_t, void *)) (addr))
-	 (GLRO(dl_hwcap), NULL);
+  /* The third argument is a void pointer to preserve the extension
+     flexibility.  */
+  return ((ElfW(Addr) (*) (uint64_t, void *, void *)) (addr))
+	 (GLRO(dl_hwcap), GLRO(dl_vdso_riscv_hwprobe), NULL);
 }
 
 static inline void
diff --git a/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h b/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
index aa189a4818..fd3be5a411 100644
--- a/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
+++ b/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
@@ -67,6 +67,16 @@ extern int __riscv_hwprobe (struct riscv_hwprobe *__pairs, size_t __pair_count,
      __fortified_attr_access (__read_write__, 1, 2)
      __fortified_attr_access (__read_only__, 4, 3);
 
+/* A pointer to the __riscv_hwprobe vDSO function is passed as the second
+   argument to ifunc selector routines. Include a function pointer type for
+   convenience in calling the function in those settings. */
+typedef int (*__riscv_hwprobe_t) (struct riscv_hwprobe *__pairs, size_t __pair_count,
+				  size_t __cpu_count, unsigned long int *__cpus,
+				  unsigned int __flags)
+     __nonnull ((1)) __wur
+     __fortified_attr_access (__read_write__, 1, 2)
+     __fortified_attr_access (__read_only__, 4, 3);
+
 __END_DECLS
 
 #endif /* sys/hwprobe.h */
-- 
2.34.1

[PATCH v7 4/6] riscv: Enable multi-arg ifunc resolvers

[Evan Green]

RISC-V is apparently the first architecture to pass more than one
argument to ifunc resolvers. The helper macros in libc-symbols.h,
__ifunc_resolver(), __ifunc(), and __ifunc_hidden(), are incompatible
with this. These macros have an "arg" (non-final) parameter that
represents the parameter signature of the ifunc resolver. The result is
an inability to pass the required comma through in a single preprocessor
argument.

Rearrange the __ifunc_resolver() macro to be variadic, and pass the
types as those variable parameters. Move the guts of __ifunc() and
__ifunc_hidden() into new macros, __ifunc_args(), and
__ifunc_args_hidden(), that pass the variable arguments down through to
__ifunc_resolver(). Then redefine __ifunc() and __ifunc_hidden(), which
are used in a bunch of places, to simply shuffle the arguments down into
__ifunc_args[_hidden]. Finally, define a riscv-ifunc.h header, which
provides convenience macros to those looking to write ifunc selectors
that use both arguments.

Signed-off-by: Evan Green <evan@rivosinc.com>

---

(no changes since v6)

Changes in v6:
 - Introduced riscv-ifunc.h for multi-arg ifunc selectors.

Note: I opted to create another layer of macros (__ifunc_args()) rather
than doing the treewide change to rearrange the signature of __ifunc()
and __ifunc_hidden(). If folks like the overall approach but would
prefer the treewide change, I can do that too.

---
 include/libc-symbols.h      | 28 +++++++++++++++++-----------
 sysdeps/riscv/riscv-ifunc.h | 27 +++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 11 deletions(-)
 create mode 100644 sysdeps/riscv/riscv-ifunc.h

diff --git a/include/libc-symbols.h b/include/libc-symbols.h
index 5794614488..36b92039c5 100644
--- a/include/libc-symbols.h
+++ b/include/libc-symbols.h
@@ -665,9 +665,9 @@ for linking")
 #endif
 
 /* Helper / base  macros for indirect function symbols.  */
-#define __ifunc_resolver(type_name, name, expr, arg, init, classifier)	\
+#define __ifunc_resolver(type_name, name, expr, init, classifier, ...)	\
   classifier inhibit_stack_protector					\
-  __typeof (type_name) *name##_ifunc (arg)				\
+  __typeof (type_name) *name##_ifunc (__VA_ARGS__)			\
   {									\
     init ();								\
     __typeof (type_name) *res = expr;					\
@@ -675,13 +675,13 @@ for linking")
   }
 
 #ifdef HAVE_GCC_IFUNC
-# define __ifunc(type_name, name, expr, arg, init)			\
+# define __ifunc_args(type_name, name, expr, init, ...)			\
   extern __typeof (type_name) name __attribute__			\
 			      ((ifunc (#name "_ifunc")));		\
-  __ifunc_resolver (type_name, name, expr, arg, init, static)
+  __ifunc_resolver (type_name, name, expr, init, static, __VA_ARGS__)
 
-# define __ifunc_hidden(type_name, name, expr, arg, init)	\
-  __ifunc (type_name, name, expr, arg, init)
+# define __ifunc_args_hidden(type_name, name, expr, init, ...)		\
+  __ifunc (type_name, name, expr, init, __VA_ARGS__)
 #else
 /* Gcc does not support __attribute__ ((ifunc (...))).  Use the old behaviour
    as fallback.  But keep in mind that the debug information for the ifunc
@@ -692,18 +692,24 @@ for linking")
    different signatures.  (Gcc support is disabled at least on a ppc64le
    Ubuntu 14.04 system.)  */
 
-# define __ifunc(type_name, name, expr, arg, init)			\
+# define __ifunc_args(type_name, name, expr, init, ...)			\
   extern __typeof (type_name) name;					\
-  __typeof (type_name) *name##_ifunc (arg) __asm__ (#name);		\
-  __ifunc_resolver (type_name, name, expr, arg, init,)			\
+  __typeof (type_name) *name##_ifunc (__VA_ARGS__) __asm__ (#name);	\
+  __ifunc_resolver (type_name, name, expr, init, , __VA_ARGS__)		\
  __asm__ (".type " #name ", %gnu_indirect_function");
 
-# define __ifunc_hidden(type_name, name, expr, arg, init)		\
+# define __ifunc_args_hidden(type_name, name, expr, init, ...)		\
   extern __typeof (type_name) __libc_##name;				\
-  __ifunc (type_name, __libc_##name, expr, arg, init)			\
+  __ifunc (type_name, __libc_##name, expr, __VA_INIT__, init)		\
   strong_alias (__libc_##name, name);
 #endif /* !HAVE_GCC_IFUNC  */
 
+#define __ifunc(type_name, name, expr, arg, init)			\
+  __ifunc_args (type_name, name, expr, init, arg)
+
+#define __ifunc_hidden(type_name, name, expr, arg, init)		\
+  __ifunc_args_hidden (type_name, expr, init, arg)
+
 /* The following macros are used for indirect function symbols in libc.so.
    First of all, you need to have the function prototyped somewhere,
    say in foo.h:
diff --git a/sysdeps/riscv/riscv-ifunc.h b/sysdeps/riscv/riscv-ifunc.h
new file mode 100644
index 0000000000..7bff591d1e
--- /dev/null
+++ b/sysdeps/riscv/riscv-ifunc.h
@@ -0,0 +1,27 @@
+/* Common definition for ifunc resolvers.  Linux/RISC-V version.
+   This file is part of the GNU C Library.
+   Copyright (C) 2023 Free Software Foundation, Inc.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include <ifunc-init.h>
+#include <sys/hwprobe.h>
+
+#define INIT_ARCH()
+
+#define riscv_libc_ifunc(name, expr)				\
+  __ifunc_args (name, name, expr(hwcap, hwprobe), INIT_ARCH,	\
+                uint64_t hwcap, __riscv_hwprobe_t hwprobe)
-- 
2.34.1

[PATCH v7 5/6] riscv: Add ifunc helper method to hwprobe.h

[Evan Green]

Add a little helper method so it's easier to fetch a single value from
the hwprobe function when used within an ifunc selector.

Signed-off-by: Evan Green <evan@rivosinc.com>

---

Changes in v7:
 - Introduced static inline helper (Richard)

 sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h | 24 +++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h b/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
index fd3be5a411..7378b1baa2 100644
--- a/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
+++ b/sysdeps/unix/sysv/linux/riscv/sys/hwprobe.h
@@ -22,6 +22,7 @@
 
 #include <features.h>
 #include <stddef.h>
+#include <errno.h>
 #ifdef __has_include
 # if __has_include (<asm/hwprobe.h>)
 #  include <asm/hwprobe.h>
@@ -79,4 +80,27 @@ typedef int (*__riscv_hwprobe_t) (struct riscv_hwprobe *__pairs, size_t __pair_c
 
 __END_DECLS
 
+/* Helper function usable from ifunc selectors that probes a single key. */
+static inline int __riscv_hwprobe_one(__riscv_hwprobe_t hwprobe_func,
+                                      signed long long int key,
+                                      unsigned long long int *value)
+{
+  struct riscv_hwprobe pair;
+  int rc;
+
+  if (!hwprobe_func)
+    return ENOSYS;
+
+  pair.key = key;
+  rc = hwprobe_func(&pair, 1, 0, NULL, 0);
+  if (rc)
+    return rc;
+
+  if (pair.key < 0)
+    return ENOENT;
+
+  *value = pair.value;
+  return 0;
+}
+
 #endif /* sys/hwprobe.h */
-- 
2.34.1

[PATCH v7 6/6] riscv: Add and use alignment-ignorant memcpy

[Evan Green]

For CPU implementations that can perform unaligned accesses with little
or no performance penalty, create a memcpy implementation that does not
bother aligning buffers. It will use a block of integer registers, a
single integer register, and fall back to bytewise copy for the
remainder.

Signed-off-by: Evan Green <evan@rivosinc.com>
Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>

---

Changes in v7:
 - Use new helper function in memcpy ifunc selector (Richard)

Changes in v6:
 - Fix a couple regressions in the assembly from v5 :/
 - Use passed hwprobe pointer in memcpy ifunc selector.

Changes in v5:
 - Do unaligned word access for final trailing bytes (Richard)

Changes in v4:
 - Fixed comment style (Florian)

Changes in v3:
 - Word align dest for large memcpy()s.
 - Add tags
 - Remove spurious blank line from sysdeps/riscv/memcpy.c

Changes in v2:
 - Used _MASK instead of _FAST value itself.


---
 sysdeps/riscv/memcopy.h                       |  26 ++++
 sysdeps/riscv/memcpy.c                        |  63 ++++++++
 sysdeps/riscv/memcpy_noalignment.S            | 138 ++++++++++++++++++
 sysdeps/unix/sysv/linux/riscv/Makefile        |   4 +
 .../unix/sysv/linux/riscv/memcpy-generic.c    |  24 +++
 5 files changed, 255 insertions(+)
 create mode 100644 sysdeps/riscv/memcopy.h
 create mode 100644 sysdeps/riscv/memcpy.c
 create mode 100644 sysdeps/riscv/memcpy_noalignment.S
 create mode 100644 sysdeps/unix/sysv/linux/riscv/memcpy-generic.c

diff --git a/sysdeps/riscv/memcopy.h b/sysdeps/riscv/memcopy.h
new file mode 100644
index 0000000000..2b685c8aa0
--- /dev/null
+++ b/sysdeps/riscv/memcopy.h
@@ -0,0 +1,26 @@
+/* memcopy.h -- definitions for memory copy functions. RISC-V version.
+   Copyright (C) 2023 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdeps/generic/memcopy.h>
+
+/* Redefine the generic memcpy implementation to __memcpy_generic, so
+   the memcpy ifunc can select between generic and special versions.
+   In rtld, don't bother with all the ifunciness. */
+#if IS_IN (libc)
+#define MEMCPY __memcpy_generic
+#endif
diff --git a/sysdeps/riscv/memcpy.c b/sysdeps/riscv/memcpy.c
new file mode 100644
index 0000000000..285ca85f8e
--- /dev/null
+++ b/sysdeps/riscv/memcpy.c
@@ -0,0 +1,63 @@
+/* Multiple versions of memcpy.
+   All versions must be listed in ifunc-impl-list.c.
+   Copyright (C) 2017-2023 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#if IS_IN (libc)
+/* Redefine memcpy so that the compiler won't complain about the type
+   mismatch with the IFUNC selector in strong_alias, below.  */
+# undef memcpy
+# define memcpy __redirect_memcpy
+# include <stdint.h>
+# include <string.h>
+# include <ifunc-init.h>
+# include <riscv-ifunc.h>
+# include <sys/hwprobe.h>
+
+# define INIT_ARCH()
+
+extern __typeof (__redirect_memcpy) __libc_memcpy;
+
+extern __typeof (__redirect_memcpy) __memcpy_generic attribute_hidden;
+extern __typeof (__redirect_memcpy) __memcpy_noalignment attribute_hidden;
+
+static inline __typeof (__redirect_memcpy) *
+select_memcpy_ifunc (uint64_t dl_hwcap, __riscv_hwprobe_t hwprobe_func)
+{
+  unsigned long long int value;
+
+  INIT_ARCH ();
+
+  if (__riscv_hwprobe_one(hwprobe_func, RISCV_HWPROBE_KEY_CPUPERF_0, &value))
+    return __memcpy_generic;
+
+  if ((value & RISCV_HWPROBE_MISALIGNED_MASK) == RISCV_HWPROBE_MISALIGNED_FAST)
+    return __memcpy_noalignment;
+
+  return __memcpy_generic;
+}
+
+riscv_libc_ifunc (__libc_memcpy, select_memcpy_ifunc);
+
+# undef memcpy
+strong_alias (__libc_memcpy, memcpy);
+# ifdef SHARED
+__hidden_ver1 (memcpy, __GI_memcpy, __redirect_memcpy)
+  __attribute__ ((visibility ("hidden"))) __attribute_copy__ (memcpy);
+# endif
+
+#endif
diff --git a/sysdeps/riscv/memcpy_noalignment.S b/sysdeps/riscv/memcpy_noalignment.S
new file mode 100644
index 0000000000..f3bf8e5867
--- /dev/null
+++ b/sysdeps/riscv/memcpy_noalignment.S
@@ -0,0 +1,138 @@
+/* memcpy for RISC-V, ignoring buffer alignment
+   Copyright (C) 2023 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library.  If not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include <sys/asm.h>
+
+/* void *memcpy(void *, const void *, size_t) */
+ENTRY (__memcpy_noalignment)
+	move t6, a0  /* Preserve return value */
+
+	/* Bail if 0 */
+	beqz a2, 7f
+
+	/* Jump to byte copy if size < SZREG */
+	li a4, SZREG
+	bltu a2, a4, 5f
+
+	/* Round down to the nearest "page" size */
+	andi a4, a2, ~((16*SZREG)-1)
+	beqz a4, 2f
+	add a3, a1, a4
+
+	/* Copy the first word to get dest word aligned */
+	andi a5, t6, SZREG-1
+	beqz a5, 1f
+	REG_L a6, (a1)
+	REG_S a6, (t6)
+
+	/* Align dst up to a word, move src and size as well. */
+	addi t6, t6, SZREG-1
+	andi t6, t6, ~(SZREG-1)
+	sub a5, t6, a0
+	add a1, a1, a5
+	sub a2, a2, a5
+
+	/* Recompute page count */
+	andi a4, a2, ~((16*SZREG)-1)
+	beqz a4, 2f
+
+1:
+	/* Copy "pages" (chunks of 16 registers) */
+	REG_L a4,       0(a1)
+	REG_L a5,   SZREG(a1)
+	REG_L a6, 2*SZREG(a1)
+	REG_L a7, 3*SZREG(a1)
+	REG_L t0, 4*SZREG(a1)
+	REG_L t1, 5*SZREG(a1)
+	REG_L t2, 6*SZREG(a1)
+	REG_L t3, 7*SZREG(a1)
+	REG_L t4, 8*SZREG(a1)
+	REG_L t5, 9*SZREG(a1)
+	REG_S a4,       0(t6)
+	REG_S a5,   SZREG(t6)
+	REG_S a6, 2*SZREG(t6)
+	REG_S a7, 3*SZREG(t6)
+	REG_S t0, 4*SZREG(t6)
+	REG_S t1, 5*SZREG(t6)
+	REG_S t2, 6*SZREG(t6)
+	REG_S t3, 7*SZREG(t6)
+	REG_S t4, 8*SZREG(t6)
+	REG_S t5, 9*SZREG(t6)
+	REG_L a4, 10*SZREG(a1)
+	REG_L a5, 11*SZREG(a1)
+	REG_L a6, 12*SZREG(a1)
+	REG_L a7, 13*SZREG(a1)
+	REG_L t0, 14*SZREG(a1)
+	REG_L t1, 15*SZREG(a1)
+	addi a1, a1, 16*SZREG
+	REG_S a4, 10*SZREG(t6)
+	REG_S a5, 11*SZREG(t6)
+	REG_S a6, 12*SZREG(t6)
+	REG_S a7, 13*SZREG(t6)
+	REG_S t0, 14*SZREG(t6)
+	REG_S t1, 15*SZREG(t6)
+	addi t6, t6, 16*SZREG
+	bltu a1, a3, 1b
+	andi a2, a2, (16*SZREG)-1  /* Update count */
+
+2:
+	/* Remainder is smaller than a page, compute native word count */
+	beqz a2, 7f
+	andi a5, a2, ~(SZREG-1)
+	andi a2, a2, (SZREG-1)
+	add a3, a1, a5
+	/* Jump directly to last word if no words. */
+	beqz a5, 4f
+
+3:
+	/* Use single native register copy */
+	REG_L a4, 0(a1)
+	addi a1, a1, SZREG
+	REG_S a4, 0(t6)
+	addi t6, t6, SZREG
+	bltu a1, a3, 3b
+
+	/* Jump directly out if no more bytes */
+	beqz a2, 7f
+
+4:
+	/* Copy the last word unaligned */
+	add a3, a1, a2
+	add a4, t6, a2
+	REG_L a5, -SZREG(a3)
+	REG_S a5, -SZREG(a4)
+	ret
+
+5:
+	/* Copy bytes when the total copy is <SZREG */
+	add a3, a1, a2
+
+6:
+	lb a4, 0(a1)
+	addi a1, a1, 1
+	sb a4, 0(t6)
+	addi t6, t6, 1
+	bltu a1, a3, 6b
+
+7:
+	ret
+
+END (__memcpy_noalignment)
+
+hidden_def (__memcpy_noalignment)
diff --git a/sysdeps/unix/sysv/linux/riscv/Makefile b/sysdeps/unix/sysv/linux/riscv/Makefile
index 45cc29e40d..aa9ea443d6 100644
--- a/sysdeps/unix/sysv/linux/riscv/Makefile
+++ b/sysdeps/unix/sysv/linux/riscv/Makefile
@@ -7,6 +7,10 @@ ifeq ($(subdir),stdlib)
 gen-as-const-headers += ucontext_i.sym
 endif
 
+ifeq ($(subdir),string)
+sysdep_routines += memcpy memcpy-generic memcpy_noalignment
+endif
+
 abi-variants := ilp32 ilp32d lp64 lp64d
 
 ifeq (,$(filter $(default-abi),$(abi-variants)))
diff --git a/sysdeps/unix/sysv/linux/riscv/memcpy-generic.c b/sysdeps/unix/sysv/linux/riscv/memcpy-generic.c
new file mode 100644
index 0000000000..0abe03f7f5
--- /dev/null
+++ b/sysdeps/unix/sysv/linux/riscv/memcpy-generic.c
@@ -0,0 +1,24 @@
+/* Re-include the default memcpy implementation.
+   Copyright (C) 2023 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <string.h>
+
+extern __typeof (memcpy) __memcpy_generic;
+hidden_proto(__memcpy_generic)
+
+#include <string/memcpy.c>
-- 
2.34.1

Re: [PATCH v7 4/6] riscv: Enable multi-arg ifunc resolvers

[enh]

[-- Attachment #1: Type: text/plain, Size: 6762 bytes --]

On Wed, Aug 23, 2023, 15:32 Evan Green <evan@rivosinc.com> wrote:

> RISC-V is apparently the first architecture to pass more than one
> argument to ifunc resolvers.


how does arm64 work? that passes hwcap and a pointer to a struct containing
hwcap2, so presumably they had to solve this already?

The helper macros in libc-symbols.h,
> __ifunc_resolver(), __ifunc(), and __ifunc_hidden(), are incompatible
> with this. These macros have an "arg" (non-final) parameter that
> represents the parameter signature of the ifunc resolver. The result is
> an inability to pass the required comma through in a single preprocessor
> argument.
>
> Rearrange the __ifunc_resolver() macro to be variadic, and pass the
> types as those variable parameters. Move the guts of __ifunc() and
> __ifunc_hidden() into new macros, __ifunc_args(), and
> __ifunc_args_hidden(), that pass the variable arguments down through to
> __ifunc_resolver(). Then redefine __ifunc() and __ifunc_hidden(), which
> are used in a bunch of places, to simply shuffle the arguments down into
> __ifunc_args[_hidden]. Finally, define a riscv-ifunc.h header, which
> provides convenience macros to those looking to write ifunc selectors
> that use both arguments.
>
> Signed-off-by: Evan Green <evan@rivosinc.com>
>
> ---
>
> (no changes since v6)
>
> Changes in v6:
>  - Introduced riscv-ifunc.h for multi-arg ifunc selectors.
>
> Note: I opted to create another layer of macros (__ifunc_args()) rather
> than doing the treewide change to rearrange the signature of __ifunc()
> and __ifunc_hidden(). If folks like the overall approach but would
> prefer the treewide change, I can do that too.
>
> ---
>  include/libc-symbols.h      | 28 +++++++++++++++++-----------
>  sysdeps/riscv/riscv-ifunc.h | 27 +++++++++++++++++++++++++++
>  2 files changed, 44 insertions(+), 11 deletions(-)
>  create mode 100644 sysdeps/riscv/riscv-ifunc.h
>
> diff --git a/include/libc-symbols.h b/include/libc-symbols.h
> index 5794614488..36b92039c5 100644
> --- a/include/libc-symbols.h
> +++ b/include/libc-symbols.h
> @@ -665,9 +665,9 @@ for linking")
>  #endif
>
>  /* Helper / base  macros for indirect function symbols.  */
> -#define __ifunc_resolver(type_name, name, expr, arg, init, classifier) \
> +#define __ifunc_resolver(type_name, name, expr, init, classifier, ...) \
>    classifier inhibit_stack_protector                                   \
> -  __typeof (type_name) *name##_ifunc (arg)                             \
> +  __typeof (type_name) *name##_ifunc (__VA_ARGS__)                     \
>    {                                                                    \
>      init ();                                                           \
>      __typeof (type_name) *res = expr;                                  \
> @@ -675,13 +675,13 @@ for linking")
>    }
>
>  #ifdef HAVE_GCC_IFUNC
> -# define __ifunc(type_name, name, expr, arg, init)                     \
> +# define __ifunc_args(type_name, name, expr, init, ...)
>       \
>    extern __typeof (type_name) name __attribute__                       \
>                               ((ifunc (#name "_ifunc")));               \
> -  __ifunc_resolver (type_name, name, expr, arg, init, static)
> +  __ifunc_resolver (type_name, name, expr, init, static, __VA_ARGS__)
>
> -# define __ifunc_hidden(type_name, name, expr, arg, init)      \
> -  __ifunc (type_name, name, expr, arg, init)
> +# define __ifunc_args_hidden(type_name, name, expr, init, ...)         \
> +  __ifunc (type_name, name, expr, init, __VA_ARGS__)
>  #else
>  /* Gcc does not support __attribute__ ((ifunc (...))).  Use the old
> behaviour
>     as fallback.  But keep in mind that the debug information for the ifunc
> @@ -692,18 +692,24 @@ for linking")
>     different signatures.  (Gcc support is disabled at least on a ppc64le
>     Ubuntu 14.04 system.)  */
>
> -# define __ifunc(type_name, name, expr, arg, init)                     \
> +# define __ifunc_args(type_name, name, expr, init, ...)
>       \
>    extern __typeof (type_name) name;                                    \
> -  __typeof (type_name) *name##_ifunc (arg) __asm__ (#name);            \
> -  __ifunc_resolver (type_name, name, expr, arg, init,)                 \
> +  __typeof (type_name) *name##_ifunc (__VA_ARGS__) __asm__ (#name);    \
> +  __ifunc_resolver (type_name, name, expr, init, , __VA_ARGS__)
>       \
>   __asm__ (".type " #name ", %gnu_indirect_function");
>
> -# define __ifunc_hidden(type_name, name, expr, arg, init)              \
> +# define __ifunc_args_hidden(type_name, name, expr, init, ...)         \
>    extern __typeof (type_name) __libc_##name;                           \
> -  __ifunc (type_name, __libc_##name, expr, arg, init)                  \
> +  __ifunc (type_name, __libc_##name, expr, __VA_INIT__, init)          \
>    strong_alias (__libc_##name, name);
>  #endif /* !HAVE_GCC_IFUNC  */
>
> +#define __ifunc(type_name, name, expr, arg, init)                      \
> +  __ifunc_args (type_name, name, expr, init, arg)
> +
> +#define __ifunc_hidden(type_name, name, expr, arg, init)               \
> +  __ifunc_args_hidden (type_name, expr, init, arg)
> +
>  /* The following macros are used for indirect function symbols in libc.so.
>     First of all, you need to have the function prototyped somewhere,
>     say in foo.h:
> diff --git a/sysdeps/riscv/riscv-ifunc.h b/sysdeps/riscv/riscv-ifunc.h
> new file mode 100644
> index 0000000000..7bff591d1e
> --- /dev/null
> +++ b/sysdeps/riscv/riscv-ifunc.h
> @@ -0,0 +1,27 @@
> +/* Common definition for ifunc resolvers.  Linux/RISC-V version.
> +   This file is part of the GNU C Library.
> +   Copyright (C) 2023 Free Software Foundation, Inc.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library; if not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +#include <sysdep.h>
> +#include <ifunc-init.h>
> +#include <sys/hwprobe.h>
> +
> +#define INIT_ARCH()
> +
> +#define riscv_libc_ifunc(name, expr)                           \
> +  __ifunc_args (name, name, expr(hwcap, hwprobe), INIT_ARCH,   \
> +                uint64_t hwcap, __riscv_hwprobe_t hwprobe)
> --
> 2.34.1
>
>

Re: [PATCH v7 4/6] riscv: Enable multi-arg ifunc resolvers

[Evan Green]

On Wed, Aug 23, 2023 at 4:20 PM enh <enh@google.com> wrote:
>
>
>
> On Wed, Aug 23, 2023, 15:32 Evan Green <evan@rivosinc.com> wrote:
>>
>> RISC-V is apparently the first architecture to pass more than one
>> argument to ifunc resolvers.
>
>
> how does arm64 work? that passes hwcap and a pointer to a struct containing hwcap2, so presumably they had to solve this already?

This macro is a glibc internal thing, so I think they only had to
solve it if they needed to use that second arg in any of the ifunc
selectors written within glibc. When I grep around for libc_ifunc in
aarch64 files, they seem to use none of the parameters passed (and can
pretend the signature is void). Instead they peer at midr or other
globals set up in INIT_ARCH() to do vendor-y things (eg return
__memcpy_thunderx if the vendor/partnum match).

I suppose I could cheat too (Florian mentioned this in the last spin),
skip this patch, and just reach for the internal __riscv_hwprobe
symbol directly in the memcpy ifunc. I kind of liked that this memcpy
change served as a nice example, but I suppose there's an argument to
be made for aiming for brevity.

-Evan

Re: [PATCH v7 1/6] riscv: Add Linux hwprobe syscall support

[Florian Weimer]

* Evan Green:

> diff --git a/sysdeps/unix/sysv/linux/riscv/hwprobe.c b/sysdeps/unix/sysv/linux/riscv/hwprobe.c
> new file mode 100644
> index 0000000000..f2dc46cfd7
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/riscv/hwprobe.c
> @@ -0,0 +1,39 @@

> +int __riscv_hwprobe (struct riscv_hwprobe *__pairs, size_t __pair_count,
> +		     size_t __cpu_count, unsigned long int *__cpus,
> +		     unsigned int __flags)
> +{
> +  int r;
> +
> +  r = INTERNAL_SYSCALL_CALL (riscv_hwprobe, 5, __pairs, __pair_count,
> +                             __cpu_count, __cpus, __flags);
> +
> +  /* Negate negative errno values to match pthreads API. */
> +  if (r < 0)
> +    r = -r;
> +
> +  return r;
> +}

I would expect:

+  /* Positive errno values to match pthreads API. */
+  return -r;

If the kernel ever returns positive values, the if condition won't save
us.

Thanks,
Florian

Re: [PATCH v7 1/6] riscv: Add Linux hwprobe syscall support

[Joseph Myers]

On Wed, 23 Aug 2023, Evan Green wrote:

> diff --git a/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist b/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist
> index b9740a1afc..8fab4a606f 100644
> --- a/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist
> +++ b/sysdeps/unix/sysv/linux/riscv/rv32/libc.abilist
> @@ -2436,3 +2436,4 @@ GLIBC_2.38 strlcat F
>  GLIBC_2.38 strlcpy F
>  GLIBC_2.38 wcslcat F
>  GLIBC_2.38 wcslcpy F
> +GLIBC_2.38 __riscv_hwprobe F
> diff --git a/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist b/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist
> index e3b4656aa2..1ebb91deed 100644
> --- a/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist
> +++ b/sysdeps/unix/sysv/linux/riscv/rv64/libc.abilist
> @@ -2636,3 +2636,4 @@ GLIBC_2.38 strlcat F
>  GLIBC_2.38 strlcpy F
>  GLIBC_2.38 wcslcat F
>  GLIBC_2.38 wcslcpy F
> +GLIBC_2.38 __riscv_hwprobe F

These are missing the update to 2.39 to match the Versions file, so I'd 
have expected the ABI tests to fail.

-- 
Joseph S. Myers
joseph@codesourcery.com