* [PATCH v3 01/18] x86/vdso: Fix latent bug in vclock_pages calculation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 02/18] parisc: Remove unused symbol vdso_data Thomas Weißschuh
` (17 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The vclock pages are *after* the non-vclock pages.
Currently there are both two vclock and two non-vclock pages so the
existing logic works by accident.
As soon as the number of pages changes it will break however.
This will be the case with the introduction of the generic vDSO data
storage.
Use a macro to keep the calculation understandable and in sync between
the linker script and mapping code.
Fixes: e93d2521b27f ("x86/vdso: Split virtual clock pages into dedicated mapping")
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/x86/entry/vdso/vdso-layout.lds.S | 2 +-
arch/x86/entry/vdso/vma.c | 2 +-
arch/x86/include/asm/vdso/vsyscall.h | 1 +
3 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/arch/x86/entry/vdso/vdso-layout.lds.S b/arch/x86/entry/vdso/vdso-layout.lds.S
index 872947c1004c35c006f7508eac7dff251c286aeb..918606ff92a988b14f5e64f984750ae04b3b6ede 100644
--- a/arch/x86/entry/vdso/vdso-layout.lds.S
+++ b/arch/x86/entry/vdso/vdso-layout.lds.S
@@ -24,7 +24,7 @@ SECTIONS
timens_page = vvar_start + PAGE_SIZE;
- vclock_pages = vvar_start + VDSO_NR_VCLOCK_PAGES * PAGE_SIZE;
+ vclock_pages = VDSO_VCLOCK_PAGES_START(vvar_start);
pvclock_page = vclock_pages + VDSO_PAGE_PVCLOCK_OFFSET * PAGE_SIZE;
hvclock_page = vclock_pages + VDSO_PAGE_HVCLOCK_OFFSET * PAGE_SIZE;
diff --git a/arch/x86/entry/vdso/vma.c b/arch/x86/entry/vdso/vma.c
index 39e6efc1a9cab760b8a65e4b02c624e8c75244b5..aa62949335ecec3765d3b46eac7f7b83be5efdda 100644
--- a/arch/x86/entry/vdso/vma.c
+++ b/arch/x86/entry/vdso/vma.c
@@ -290,7 +290,7 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
}
vma = _install_special_mapping(mm,
- addr + (__VVAR_PAGES - VDSO_NR_VCLOCK_PAGES) * PAGE_SIZE,
+ VDSO_VCLOCK_PAGES_START(addr),
VDSO_NR_VCLOCK_PAGES * PAGE_SIZE,
VM_READ|VM_MAYREAD|VM_IO|VM_DONTDUMP|
VM_PFNMAP,
diff --git a/arch/x86/include/asm/vdso/vsyscall.h b/arch/x86/include/asm/vdso/vsyscall.h
index 37b4a70559a8228601203fe7b99b9ddfc3d94f1b..88b31d4cdfaf331d2d597981d3f8ee0c5a339085 100644
--- a/arch/x86/include/asm/vdso/vsyscall.h
+++ b/arch/x86/include/asm/vdso/vsyscall.h
@@ -6,6 +6,7 @@
#define __VVAR_PAGES 4
#define VDSO_NR_VCLOCK_PAGES 2
+#define VDSO_VCLOCK_PAGES_START(_b) ((_b) + (__VVAR_PAGES - VDSO_NR_VCLOCK_PAGES) * PAGE_SIZE)
#define VDSO_PAGE_PVCLOCK_OFFSET 0
#define VDSO_PAGE_HVCLOCK_OFFSET 1
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 02/18] parisc: Remove unused symbol vdso_data
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 01/18] x86/vdso: Fix latent bug in vclock_pages calculation Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-05 18:06 ` Helge Deller
2025-02-04 12:05 ` [PATCH v3 03/18] vdso: Introduce vdso/align.h Thomas Weißschuh
` (16 subsequent siblings)
18 siblings, 1 reply; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The symbol vdso_data is and has been unused.
Remove it.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/parisc/include/asm/vdso.h | 2 --
1 file changed, 2 deletions(-)
diff --git a/arch/parisc/include/asm/vdso.h b/arch/parisc/include/asm/vdso.h
index 2a2dc11b5545fc642a7ae4596dc174111433e948..5f581c1d64606b3ec8f9967efe31dd3d551adf76 100644
--- a/arch/parisc/include/asm/vdso.h
+++ b/arch/parisc/include/asm/vdso.h
@@ -12,8 +12,6 @@
#define VDSO64_SYMBOL(tsk, name) ((tsk)->mm->context.vdso_base + (vdso64_offset_##name))
#define VDSO32_SYMBOL(tsk, name) ((tsk)->mm->context.vdso_base + (vdso32_offset_##name))
-extern struct vdso_data *vdso_data;
-
#endif /* __ASSEMBLY __ */
/* Default link addresses for the vDSOs */
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 03/18] vdso: Introduce vdso/align.h
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 01/18] x86/vdso: Fix latent bug in vclock_pages calculation Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 02/18] parisc: Remove unused symbol vdso_data Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 04/18] vdso: Rename included Makefile Thomas Weißschuh
` (15 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The vDSO implementation can only include headers from the vdso/
namespace. To enable the usage of the ALIGN() macro from the vDSO, move
linux/align.h to vdso/align.h wholly.
As the only dependency linux/const.h is only a wrapper around
vdso/const.h anyways adapt that dependency.
Also provide a compatibility wrapper linux/align.h.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
include/linux/align.h | 10 +---------
include/vdso/align.h | 15 +++++++++++++++
2 files changed, 16 insertions(+), 9 deletions(-)
diff --git a/include/linux/align.h b/include/linux/align.h
index 2b4acec7b95a27b6768d48f46519abc584c94d5d..55debf105a5d610c592e5c53a4153018ea1ae2c6 100644
--- a/include/linux/align.h
+++ b/include/linux/align.h
@@ -2,14 +2,6 @@
#ifndef _LINUX_ALIGN_H
#define _LINUX_ALIGN_H
-#include <linux/const.h>
-
-/* @a is a power of 2 value */
-#define ALIGN(x, a) __ALIGN_KERNEL((x), (a))
-#define ALIGN_DOWN(x, a) __ALIGN_KERNEL((x) - ((a) - 1), (a))
-#define __ALIGN_MASK(x, mask) __ALIGN_KERNEL_MASK((x), (mask))
-#define PTR_ALIGN(p, a) ((typeof(p))ALIGN((unsigned long)(p), (a)))
-#define PTR_ALIGN_DOWN(p, a) ((typeof(p))ALIGN_DOWN((unsigned long)(p), (a)))
-#define IS_ALIGNED(x, a) (((x) & ((typeof(x))(a) - 1)) == 0)
+#include <vdso/align.h>
#endif /* _LINUX_ALIGN_H */
diff --git a/include/vdso/align.h b/include/vdso/align.h
new file mode 100644
index 0000000000000000000000000000000000000000..02dd8626b5c5aa59d2ee0b15799b6e9adb343f65
--- /dev/null
+++ b/include/vdso/align.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __VDSO_ALIGN_H
+#define __VDSO_ALIGN_H
+
+#include <vdso/const.h>
+
+/* @a is a power of 2 value */
+#define ALIGN(x, a) __ALIGN_KERNEL((x), (a))
+#define ALIGN_DOWN(x, a) __ALIGN_KERNEL((x) - ((a) - 1), (a))
+#define __ALIGN_MASK(x, mask) __ALIGN_KERNEL_MASK((x), (mask))
+#define PTR_ALIGN(p, a) ((typeof(p))ALIGN((unsigned long)(p), (a)))
+#define PTR_ALIGN_DOWN(p, a) ((typeof(p))ALIGN_DOWN((unsigned long)(p), (a)))
+#define IS_ALIGNED(x, a) (((x) & ((typeof(x))(a) - 1)) == 0)
+
+#endif /* __VDSO_ALIGN_H */
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 04/18] vdso: Rename included Makefile
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (2 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 03/18] vdso: Introduce vdso/align.h Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 05/18] vdso: Add generic time data storage Thomas Weißschuh
` (14 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
As the Makefile is included into other Makefiles it can not be used to
define objects to be built from the current source directory.
However the generic datastore will introduce such a local source file.
Rename the included Makefile so it is clear how it is to be used and to
make room for a regular Makefile in lib/vdso/.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/arm/vdso/Makefile | 2 +-
arch/arm64/kernel/vdso/Makefile | 2 +-
arch/arm64/kernel/vdso32/Makefile | 2 +-
arch/csky/kernel/vdso/Makefile | 2 +-
arch/loongarch/vdso/Makefile | 2 +-
arch/mips/vdso/Makefile | 2 +-
arch/parisc/kernel/vdso32/Makefile | 2 +-
arch/parisc/kernel/vdso64/Makefile | 2 +-
arch/powerpc/kernel/vdso/Makefile | 2 +-
arch/riscv/kernel/vdso/Makefile | 2 +-
arch/s390/kernel/vdso32/Makefile | 2 +-
arch/s390/kernel/vdso64/Makefile | 2 +-
arch/x86/entry/vdso/Makefile | 2 +-
lib/vdso/{Makefile => Makefile.include} | 0
14 files changed, 13 insertions(+), 13 deletions(-)
diff --git a/arch/arm/vdso/Makefile b/arch/arm/vdso/Makefile
index 8a306bbec4a0c4d4b6580fe88187faf9f5422eed..cb044bfd145d10b5bf840d31900aefc8a187e5bd 100644
--- a/arch/arm/vdso/Makefile
+++ b/arch/arm/vdso/Makefile
@@ -1,7 +1,7 @@
# SPDX-License-Identifier: GPL-2.0
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
hostprogs := vdsomunge
diff --git a/arch/arm64/kernel/vdso/Makefile b/arch/arm64/kernel/vdso/Makefile
index 35685c0360441ddb0e549ff0abe39cf4fce64071..5e27e46aa49674bb4d2186bcba548aa841116fa9 100644
--- a/arch/arm64/kernel/vdso/Makefile
+++ b/arch/arm64/kernel/vdso/Makefile
@@ -7,7 +7,7 @@
#
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
obj-vdso := vgettimeofday.o note.o sigreturn.o vgetrandom.o vgetrandom-chacha.o
diff --git a/arch/arm64/kernel/vdso32/Makefile b/arch/arm64/kernel/vdso32/Makefile
index 25a2cb6317f3592179dded21218e81036a8f67bc..f2dfdc7dc8185bc045907283b68ab18fed980312 100644
--- a/arch/arm64/kernel/vdso32/Makefile
+++ b/arch/arm64/kernel/vdso32/Makefile
@@ -3,7 +3,7 @@
# Makefile for vdso32
#
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
# Same as cc-*option, but using CC_COMPAT instead of CC
ifeq ($(CONFIG_CC_IS_CLANG), y)
diff --git a/arch/csky/kernel/vdso/Makefile b/arch/csky/kernel/vdso/Makefile
index 069ef0b17fe5235918a6aa13aa120857bbbf3faf..a3042287a0707a9aa5d512441311c3deffff2ceb 100644
--- a/arch/csky/kernel/vdso/Makefile
+++ b/arch/csky/kernel/vdso/Makefile
@@ -1,7 +1,7 @@
# SPDX-License-Identifier: GPL-2.0-only
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
# Symbols present in the vdso
vdso-syms += rt_sigreturn
diff --git a/arch/loongarch/vdso/Makefile b/arch/loongarch/vdso/Makefile
index fdde1bcd4e2663bd400dcc6becc4261b7d5dce3a..1c26147aff7018d190c49aebf6012f6780770dd2 100644
--- a/arch/loongarch/vdso/Makefile
+++ b/arch/loongarch/vdso/Makefile
@@ -2,7 +2,7 @@
# Objects to go into the VDSO.
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
obj-vdso-y := elf.o vgetcpu.o vgettimeofday.o vgetrandom.o \
vgetrandom-chacha.o sigreturn.o
diff --git a/arch/mips/vdso/Makefile b/arch/mips/vdso/Makefile
index b289b2c1b2946057c29cde8ee456b311fa25d448..fb4c493aaffa904d51f68b483ab83256a2e358e4 100644
--- a/arch/mips/vdso/Makefile
+++ b/arch/mips/vdso/Makefile
@@ -2,7 +2,7 @@
# Objects to go into the VDSO.
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
obj-vdso-y := elf.o vgettimeofday.o sigreturn.o
diff --git a/arch/parisc/kernel/vdso32/Makefile b/arch/parisc/kernel/vdso32/Makefile
index 288f8b85978fb20acd5d95ca9f44edf3e6f93d6c..4ee8d17da229b2d768ec8d8633197c91526f0f94 100644
--- a/arch/parisc/kernel/vdso32/Makefile
+++ b/arch/parisc/kernel/vdso32/Makefile
@@ -1,5 +1,5 @@
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
KCOV_INSTRUMENT := n
diff --git a/arch/parisc/kernel/vdso64/Makefile b/arch/parisc/kernel/vdso64/Makefile
index bc5d9553f3112a4bcb218d6e351159b55feea17f..c63f4069170f465079e5eed7208f9f64e866b245 100644
--- a/arch/parisc/kernel/vdso64/Makefile
+++ b/arch/parisc/kernel/vdso64/Makefile
@@ -1,5 +1,5 @@
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
KCOV_INSTRUMENT := n
diff --git a/arch/powerpc/kernel/vdso/Makefile b/arch/powerpc/kernel/vdso/Makefile
index 0e3ed6fb199fff7ad25aadca76e3a49a283b3f94..e8824f93332610db259b303c63957436bfef4191 100644
--- a/arch/powerpc/kernel/vdso/Makefile
+++ b/arch/powerpc/kernel/vdso/Makefile
@@ -3,7 +3,7 @@
# List of files in the vdso, has to be asm only for now
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
obj-vdso32 = sigtramp32-32.o gettimeofday-32.o datapage-32.o cacheflush-32.o note-32.o getcpu-32.o
obj-vdso64 = sigtramp64-64.o gettimeofday-64.o datapage-64.o cacheflush-64.o note-64.o getcpu-64.o
diff --git a/arch/riscv/kernel/vdso/Makefile b/arch/riscv/kernel/vdso/Makefile
index 9a1b555e87331fb288eff12470ad498199d7cf24..ad73607abc2808af2cd1aaf839b227c78d7a1769 100644
--- a/arch/riscv/kernel/vdso/Makefile
+++ b/arch/riscv/kernel/vdso/Makefile
@@ -2,7 +2,7 @@
# Copied from arch/tile/kernel/vdso/Makefile
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
# Symbols present in the vdso
vdso-syms = rt_sigreturn
ifdef CONFIG_64BIT
diff --git a/arch/s390/kernel/vdso32/Makefile b/arch/s390/kernel/vdso32/Makefile
index 2c5afb88d298263a70abbe1e4f903a95c0389225..1e4ddd1a683ff84492f0f4b48d0efa00688129c2 100644
--- a/arch/s390/kernel/vdso32/Makefile
+++ b/arch/s390/kernel/vdso32/Makefile
@@ -2,7 +2,7 @@
# List of files in the vdso
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
obj-vdso32 = vdso_user_wrapper-32.o note-32.o
# Build rules
diff --git a/arch/s390/kernel/vdso64/Makefile b/arch/s390/kernel/vdso64/Makefile
index ad206f2068d8c0fa1d3319cc1f4dfd27f61d0882..d8f0df74280960cb351154a8a73b4f7fe83a9125 100644
--- a/arch/s390/kernel/vdso64/Makefile
+++ b/arch/s390/kernel/vdso64/Makefile
@@ -2,7 +2,7 @@
# List of files in the vdso
# Include the generic Makefile to check the built vdso.
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
obj-vdso64 = vdso_user_wrapper.o note.o vgetrandom-chacha.o
obj-cvdso64 = vdso64_generic.o getcpu.o vgetrandom.o
VDSO_CFLAGS_REMOVE := -pg $(CC_FLAGS_FTRACE) $(CC_FLAGS_EXPOLINE)
diff --git a/arch/x86/entry/vdso/Makefile b/arch/x86/entry/vdso/Makefile
index c9216ac4fb1eb8c1e5bc5e33b41e9e40c7924cbf..1c0072336e66177453f8bbad743c6a2ee1f1849d 100644
--- a/arch/x86/entry/vdso/Makefile
+++ b/arch/x86/entry/vdso/Makefile
@@ -4,7 +4,7 @@
#
# Include the generic Makefile to check the built vDSO:
-include $(srctree)/lib/vdso/Makefile
+include $(srctree)/lib/vdso/Makefile.include
# Files to link into the vDSO:
vobjs-y := vdso-note.o vclock_gettime.o vgetcpu.o vgetrandom.o vgetrandom-chacha.o
diff --git a/lib/vdso/Makefile b/lib/vdso/Makefile.include
similarity index 100%
rename from lib/vdso/Makefile
rename to lib/vdso/Makefile.include
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 05/18] vdso: Add generic time data storage
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (3 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 04/18] vdso: Rename included Makefile Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 06/18] vdso: Add generic random " Thomas Weißschuh
` (13 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
Historically each architecture defined their own way to store the vDSO
data page. Add a generic mechanism to provide storage for that page.
Furthermore this generic storage will be extended to also provide
uniform storage for *non*-time-related data, like the random state or
architecture-specific data. These will have their own pages and data
structures, so rename 'vdso_data' into 'vdso_time_data' to make that
split clear from the name.
Also introduce a new consistent naming scheme for the symbols related to
the vDSO, which makes it clear if the symbol is accessible from
userspace or kernel space and the type of data behind the symbol.
The generic fault handler contains an optimization to prefault the vvar
page when the timens page is accessed. This was lifted from s390 and x86.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
include/asm-generic/vdso/vsyscall.h | 16 ++++++
include/linux/time_namespace.h | 1 +
include/linux/vdso_datastore.h | 10 ++++
include/vdso/datapage.h | 41 +++++++++++---
kernel/time/vsyscall.c | 8 +--
lib/Makefile | 2 +-
lib/vdso/Kconfig | 5 ++
lib/vdso/Makefile | 3 ++
lib/vdso/datastore.c | 103 ++++++++++++++++++++++++++++++++++++
lib/vdso/gettimeofday.c | 25 ++++++---
10 files changed, 195 insertions(+), 19 deletions(-)
diff --git a/include/asm-generic/vdso/vsyscall.h b/include/asm-generic/vdso/vsyscall.h
index 01dafd604188fb0512d21c4ce4b027f7da54f5a0..ac5b93b91993224da245e80031b9f51e0c083f3c 100644
--- a/include/asm-generic/vdso/vsyscall.h
+++ b/include/asm-generic/vdso/vsyscall.h
@@ -4,12 +4,28 @@
#ifndef __ASSEMBLY__
+#ifdef CONFIG_GENERIC_VDSO_DATA_STORE
+
+#ifndef __arch_get_vdso_u_time_data
+static __always_inline const struct vdso_time_data *__arch_get_vdso_u_time_data(void)
+{
+ return vdso_u_time_data;
+}
+#endif
+
+#else /* !CONFIG_GENERIC_VDSO_DATA_STORE */
+
#ifndef __arch_get_k_vdso_data
static __always_inline struct vdso_data *__arch_get_k_vdso_data(void)
{
return NULL;
}
#endif /* __arch_get_k_vdso_data */
+#define vdso_k_time_data __arch_get_k_vdso_data()
+
+#define __arch_get_vdso_u_time_data __arch_get_vdso_data
+
+#endif /* CONFIG_GENERIC_VDSO_DATA_STORE */
#ifndef __arch_update_vsyscall
static __always_inline void __arch_update_vsyscall(struct vdso_data *vdata)
diff --git a/include/linux/time_namespace.h b/include/linux/time_namespace.h
index 876e31b4461d0ee01fe2bd3d136acdea2611789f..4b81db223f5450218dfaf553b24195be9ba97c08 100644
--- a/include/linux/time_namespace.h
+++ b/include/linux/time_namespace.h
@@ -8,6 +8,7 @@
#include <linux/ns_common.h>
#include <linux/err.h>
#include <linux/time64.h>
+#include <vdso/datapage.h>
struct user_namespace;
extern struct user_namespace init_user_ns;
diff --git a/include/linux/vdso_datastore.h b/include/linux/vdso_datastore.h
new file mode 100644
index 0000000000000000000000000000000000000000..a91fa24b06e09321fdff8c2c7bdfbc1b206db574
--- /dev/null
+++ b/include/linux/vdso_datastore.h
@@ -0,0 +1,10 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_VDSO_DATASTORE_H
+#define _LINUX_VDSO_DATASTORE_H
+
+#include <linux/mm_types.h>
+
+extern const struct vm_special_mapping vdso_vvar_mapping;
+struct vm_area_struct *vdso_install_vvar_mapping(struct mm_struct *mm, unsigned long addr);
+
+#endif /* _LINUX_VDSO_DATASTORE_H */
diff --git a/include/vdso/datapage.h b/include/vdso/datapage.h
index d967baa0cd0c65784e38dc4fcd7b9e8273923947..b3d8087488ff35fbbc4d5058ae21e2c6cc58ed9c 100644
--- a/include/vdso/datapage.h
+++ b/include/vdso/datapage.h
@@ -45,11 +45,11 @@ struct arch_vdso_time_data {};
*
* There is one vdso_timestamp object in vvar for each vDSO-accelerated
* clock_id. For high-resolution clocks, this encodes the time
- * corresponding to vdso_data.cycle_last. For coarse clocks this encodes
+ * corresponding to vdso_time_data.cycle_last. For coarse clocks this encodes
* the actual time.
*
* To be noticed that for highres clocks nsec is left-shifted by
- * vdso_data.cs[x].shift.
+ * vdso_time_data[x].shift.
*/
struct vdso_timestamp {
u64 sec;
@@ -57,7 +57,7 @@ struct vdso_timestamp {
};
/**
- * struct vdso_data - vdso datapage representation
+ * struct vdso_time_data - vdso datapage representation
* @seq: timebase sequence counter
* @clock_mode: clock mode
* @cycle_last: timebase at clocksource init
@@ -74,7 +74,7 @@ struct vdso_timestamp {
* @arch_data: architecture specific data (optional, defaults
* to an empty struct)
*
- * vdso_data will be accessed by 64 bit and compat code at the same time
+ * vdso_time_data will be accessed by 64 bit and compat code at the same time
* so we should be careful before modifying this structure.
*
* The ordering of the struct members is optimized to have fast access to the
@@ -92,7 +92,7 @@ struct vdso_timestamp {
* For clocks which are not affected by time namespace adjustment the
* offset must be zero.
*/
-struct vdso_data {
+struct vdso_time_data {
u32 seq;
s32 clock_mode;
@@ -117,6 +117,8 @@ struct vdso_data {
struct arch_vdso_time_data arch_data;
};
+#define vdso_data vdso_time_data
+
/**
* struct vdso_rng_data - vdso RNG state information
* @generation: counter representing the number of RNG reseeds
@@ -136,18 +138,34 @@ struct vdso_rng_data {
* With the hidden visibility, the compiler simply generates a PC-relative
* relocation, and this is what we need.
*/
-extern struct vdso_data _vdso_data[CS_BASES] __attribute__((visibility("hidden")));
-extern struct vdso_data _timens_data[CS_BASES] __attribute__((visibility("hidden")));
+#ifndef CONFIG_GENERIC_VDSO_DATA_STORE
+extern struct vdso_time_data _vdso_data[CS_BASES] __attribute__((visibility("hidden")));
+extern struct vdso_time_data _timens_data[CS_BASES] __attribute__((visibility("hidden")));
extern struct vdso_rng_data _vdso_rng_data __attribute__((visibility("hidden")));
+#else
+extern struct vdso_time_data vdso_u_time_data[CS_BASES] __attribute__((visibility("hidden")));
+
+extern struct vdso_time_data *vdso_k_time_data;
+#endif
/**
* union vdso_data_store - Generic vDSO data page
*/
union vdso_data_store {
- struct vdso_data data[CS_BASES];
+ struct vdso_time_data data[CS_BASES];
u8 page[1U << CONFIG_PAGE_SHIFT];
};
+#ifdef CONFIG_GENERIC_VDSO_DATA_STORE
+
+enum vdso_pages {
+ VDSO_TIME_PAGE_OFFSET,
+ VDSO_TIMENS_PAGE_OFFSET,
+ VDSO_NR_PAGES
+};
+
+#endif /* CONFIG_GENERIC_VDSO_DATA_STORE */
+
/*
* The generic vDSO implementation requires that gettimeofday.h
* provides:
@@ -164,6 +182,13 @@ union vdso_data_store {
#include <asm/vdso/gettimeofday.h>
#endif /* ENABLE_COMPAT_VDSO */
+#else /* !__ASSEMBLY__ */
+
+#define VDSO_VVAR_SYMS \
+ PROVIDE(vdso_u_data = . - __VDSO_PAGES * PAGE_SIZE); \
+ PROVIDE(vdso_u_time_data = vdso_u_data); \
+
+
#endif /* !__ASSEMBLY__ */
#endif /* __VDSO_DATAPAGE_H */
diff --git a/kernel/time/vsyscall.c b/kernel/time/vsyscall.c
index 05d3831431658227c080a89202f45e7a0af88895..09c1e39a6dd8e24aa161982c04078e78d52fa737 100644
--- a/kernel/time/vsyscall.c
+++ b/kernel/time/vsyscall.c
@@ -77,7 +77,7 @@ static inline void update_vdso_data(struct vdso_data *vdata,
void update_vsyscall(struct timekeeper *tk)
{
- struct vdso_data *vdata = __arch_get_k_vdso_data();
+ struct vdso_data *vdata = vdso_k_time_data;
struct vdso_timestamp *vdso_ts;
s32 clock_mode;
u64 nsec;
@@ -128,7 +128,7 @@ void update_vsyscall(struct timekeeper *tk)
void update_vsyscall_tz(void)
{
- struct vdso_data *vdata = __arch_get_k_vdso_data();
+ struct vdso_data *vdata = vdso_k_time_data;
vdata[CS_HRES_COARSE].tz_minuteswest = sys_tz.tz_minuteswest;
vdata[CS_HRES_COARSE].tz_dsttime = sys_tz.tz_dsttime;
@@ -150,7 +150,7 @@ void update_vsyscall_tz(void)
*/
unsigned long vdso_update_begin(void)
{
- struct vdso_data *vdata = __arch_get_k_vdso_data();
+ struct vdso_data *vdata = vdso_k_time_data;
unsigned long flags = timekeeper_lock_irqsave();
vdso_write_begin(vdata);
@@ -167,7 +167,7 @@ unsigned long vdso_update_begin(void)
*/
void vdso_update_end(unsigned long flags)
{
- struct vdso_data *vdata = __arch_get_k_vdso_data();
+ struct vdso_data *vdata = vdso_k_time_data;
vdso_write_end(vdata);
__arch_sync_vdso_data(vdata);
diff --git a/lib/Makefile b/lib/Makefile
index d5cfc7afbbb8215cdb8d03c26fe7b7be13d75d24..fcdee83deb5c6390451a1634ab3b776cf149681f 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -132,7 +132,7 @@ endif
obj-$(CONFIG_DEBUG_INFO_REDUCED) += debug_info.o
CFLAGS_debug_info.o += $(call cc-option, -femit-struct-debug-detailed=any)
-obj-y += math/ crypto/
+obj-y += math/ crypto/ vdso/
obj-$(CONFIG_GENERIC_IOMAP) += iomap.o
obj-$(CONFIG_HAS_IOMEM) += iomap_copy.o devres.o
diff --git a/lib/vdso/Kconfig b/lib/vdso/Kconfig
index 82fe827af5426d8b644bff00e9097e3228d0ebaa..45df764b49ad62479e6456e00c053e46131936a3 100644
--- a/lib/vdso/Kconfig
+++ b/lib/vdso/Kconfig
@@ -43,3 +43,8 @@ config VDSO_GETRANDOM
bool
help
Selected by architectures that support vDSO getrandom().
+
+config GENERIC_VDSO_DATA_STORE
+ bool
+ help
+ Selected by architectures that use the generic vDSO data store.
diff --git a/lib/vdso/Makefile b/lib/vdso/Makefile
new file mode 100644
index 0000000000000000000000000000000000000000..aedd40aaa950c86f1454d095d9d46992b0cc0abd
--- /dev/null
+++ b/lib/vdso/Makefile
@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+obj-$(CONFIG_GENERIC_VDSO_DATA_STORE) += datastore.o
diff --git a/lib/vdso/datastore.c b/lib/vdso/datastore.c
new file mode 100644
index 0000000000000000000000000000000000000000..d3181768424cc3ee2e980bd537ac5ca14dfcb304
--- /dev/null
+++ b/lib/vdso/datastore.c
@@ -0,0 +1,103 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/linkage.h>
+#include <linux/mmap_lock.h>
+#include <linux/mm.h>
+#include <linux/time_namespace.h>
+#include <linux/types.h>
+#include <linux/vdso_datastore.h>
+#include <vdso/datapage.h>
+
+/*
+ * The vDSO data page.
+ */
+#ifdef CONFIG_HAVE_GENERIC_VDSO
+static union vdso_data_store vdso_time_data_store __page_aligned_data;
+struct vdso_time_data *vdso_k_time_data = vdso_time_data_store.data;
+static_assert(sizeof(vdso_time_data_store) == PAGE_SIZE);
+#endif /* CONFIG_HAVE_GENERIC_VDSO */
+
+static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
+ struct vm_area_struct *vma, struct vm_fault *vmf)
+{
+ struct page *timens_page = find_timens_vvar_page(vma);
+ unsigned long addr, pfn;
+ vm_fault_t err;
+
+ switch (vmf->pgoff) {
+ case VDSO_TIME_PAGE_OFFSET:
+ if (!IS_ENABLED(CONFIG_HAVE_GENERIC_VDSO))
+ return VM_FAULT_SIGBUS;
+ pfn = __phys_to_pfn(__pa_symbol(vdso_k_time_data));
+ if (timens_page) {
+ /*
+ * Fault in VVAR page too, since it will be accessed
+ * to get clock data anyway.
+ */
+ addr = vmf->address + VDSO_TIMENS_PAGE_OFFSET * PAGE_SIZE;
+ err = vmf_insert_pfn(vma, addr, pfn);
+ if (unlikely(err & VM_FAULT_ERROR))
+ return err;
+ pfn = page_to_pfn(timens_page);
+ }
+ break;
+ case VDSO_TIMENS_PAGE_OFFSET:
+ /*
+ * If a task belongs to a time namespace then a namespace
+ * specific VVAR is mapped with the VVAR_DATA_PAGE_OFFSET and
+ * the real VVAR page is mapped with the VVAR_TIMENS_PAGE_OFFSET
+ * offset.
+ * See also the comment near timens_setup_vdso_data().
+ */
+ if (!IS_ENABLED(CONFIG_TIME_NS) || !timens_page)
+ return VM_FAULT_SIGBUS;
+ pfn = __phys_to_pfn(__pa_symbol(vdso_k_time_data));
+ break;
+ default:
+ return VM_FAULT_SIGBUS;
+ }
+
+ return vmf_insert_pfn(vma, vmf->address, pfn);
+}
+
+const struct vm_special_mapping vdso_vvar_mapping = {
+ .name = "[vvar]",
+ .fault = vvar_fault,
+};
+
+struct vm_area_struct *vdso_install_vvar_mapping(struct mm_struct *mm, unsigned long addr)
+{
+ return _install_special_mapping(mm, addr, VDSO_NR_PAGES * PAGE_SIZE,
+ VM_READ | VM_MAYREAD | VM_IO | VM_DONTDUMP | VM_PFNMAP,
+ &vdso_vvar_mapping);
+}
+
+#ifdef CONFIG_TIME_NS
+/*
+ * The vvar page layout depends on whether a task belongs to the root or
+ * non-root time namespace. Whenever a task changes its namespace, the VVAR
+ * page tables are cleared and then they will be re-faulted with a
+ * corresponding layout.
+ * See also the comment near timens_setup_vdso_data() for details.
+ */
+int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
+{
+ struct mm_struct *mm = task->mm;
+ struct vm_area_struct *vma;
+ VMA_ITERATOR(vmi, mm, 0);
+
+ mmap_read_lock(mm);
+ for_each_vma(vmi, vma) {
+ if (vma_is_special_mapping(vma, &vdso_vvar_mapping))
+ zap_vma_pages(vma);
+ }
+ mmap_read_unlock(mm);
+
+ return 0;
+}
+
+struct vdso_time_data *arch_get_vdso_data(void *vvar_page)
+{
+ return (struct vdso_time_data *)vvar_page;
+}
+#endif
diff --git a/lib/vdso/gettimeofday.c b/lib/vdso/gettimeofday.c
index c01eaafd8041916878f3f407f7f4fc3e0a0611ca..20c5b8752fcc81d9a044169b9a3ae534d840ef91 100644
--- a/lib/vdso/gettimeofday.c
+++ b/lib/vdso/gettimeofday.c
@@ -5,6 +5,9 @@
#include <vdso/datapage.h>
#include <vdso/helpers.h>
+/* Bring in default accessors */
+#include <vdso/vsyscall.h>
+
#ifndef vdso_calc_ns
#ifdef VDSO_DELTA_NOMASK
@@ -69,6 +72,16 @@ static inline bool vdso_cycles_ok(u64 cycles)
#endif
#ifdef CONFIG_TIME_NS
+
+#ifdef CONFIG_GENERIC_VDSO_DATA_STORE
+static __always_inline
+const struct vdso_time_data *__arch_get_vdso_u_timens_data(const struct vdso_time_data *vd)
+{
+ return (void *)vd + PAGE_SIZE;
+}
+#define __arch_get_timens_vdso_data(vd) __arch_get_vdso_u_timens_data(vd)
+#endif /* CONFIG_GENERIC_VDSO_DATA_STORE */
+
static __always_inline int do_hres_timens(const struct vdso_data *vdns, clockid_t clk,
struct __kernel_timespec *ts)
{
@@ -282,7 +295,7 @@ __cvdso_clock_gettime_data(const struct vdso_data *vd, clockid_t clock,
static __maybe_unused int
__cvdso_clock_gettime(clockid_t clock, struct __kernel_timespec *ts)
{
- return __cvdso_clock_gettime_data(__arch_get_vdso_data(), clock, ts);
+ return __cvdso_clock_gettime_data(__arch_get_vdso_u_time_data(), clock, ts);
}
#ifdef BUILD_VDSO32
@@ -308,7 +321,7 @@ __cvdso_clock_gettime32_data(const struct vdso_data *vd, clockid_t clock,
static __maybe_unused int
__cvdso_clock_gettime32(clockid_t clock, struct old_timespec32 *res)
{
- return __cvdso_clock_gettime32_data(__arch_get_vdso_data(), clock, res);
+ return __cvdso_clock_gettime32_data(__arch_get_vdso_u_time_data(), clock, res);
}
#endif /* BUILD_VDSO32 */
@@ -342,7 +355,7 @@ __cvdso_gettimeofday_data(const struct vdso_data *vd,
static __maybe_unused int
__cvdso_gettimeofday(struct __kernel_old_timeval *tv, struct timezone *tz)
{
- return __cvdso_gettimeofday_data(__arch_get_vdso_data(), tv, tz);
+ return __cvdso_gettimeofday_data(__arch_get_vdso_u_time_data(), tv, tz);
}
#ifdef VDSO_HAS_TIME
@@ -365,7 +378,7 @@ __cvdso_time_data(const struct vdso_data *vd, __kernel_old_time_t *time)
static __maybe_unused __kernel_old_time_t __cvdso_time(__kernel_old_time_t *time)
{
- return __cvdso_time_data(__arch_get_vdso_data(), time);
+ return __cvdso_time_data(__arch_get_vdso_u_time_data(), time);
}
#endif /* VDSO_HAS_TIME */
@@ -425,7 +438,7 @@ int __cvdso_clock_getres_data(const struct vdso_data *vd, clockid_t clock,
static __maybe_unused
int __cvdso_clock_getres(clockid_t clock, struct __kernel_timespec *res)
{
- return __cvdso_clock_getres_data(__arch_get_vdso_data(), clock, res);
+ return __cvdso_clock_getres_data(__arch_get_vdso_u_time_data(), clock, res);
}
#ifdef BUILD_VDSO32
@@ -451,7 +464,7 @@ __cvdso_clock_getres_time32_data(const struct vdso_data *vd, clockid_t clock,
static __maybe_unused int
__cvdso_clock_getres_time32(clockid_t clock, struct old_timespec32 *res)
{
- return __cvdso_clock_getres_time32_data(__arch_get_vdso_data(),
+ return __cvdso_clock_getres_time32_data(__arch_get_vdso_u_time_data(),
clock, res);
}
#endif /* BUILD_VDSO32 */
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 06/18] vdso: Add generic random data storage
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (4 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 05/18] vdso: Add generic time data storage Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 07/18] vdso: Add generic architecture-specific " Thomas Weißschuh
` (12 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
Extend the generic vDSO data storage with a page for the random state data.
The random state data is stored in a dedicated page, as the existing
storage page is only meant for time-related, time-namespace-aware data.
This simplifies to access logic to not need to handle time namespaces
anymore and also frees up more space in the time-related page.
In case further generic vDSO data store is required it can be added to
the random state page.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
drivers/char/random.c | 6 +++---
include/asm-generic/vdso/vsyscall.h | 12 ++++++++++++
include/vdso/datapage.h | 10 ++++++++++
lib/vdso/datastore.c | 14 ++++++++++++++
lib/vdso/getrandom.c | 8 ++++++--
5 files changed, 45 insertions(+), 5 deletions(-)
diff --git a/drivers/char/random.c b/drivers/char/random.c
index 2581186fa61b7b278916a27f99ee22b4851dc37a..92cbd24a36d8c73ac6a17e8e787eeae3f6af1922 100644
--- a/drivers/char/random.c
+++ b/drivers/char/random.c
@@ -278,7 +278,7 @@ static void crng_reseed(struct work_struct *work)
WRITE_ONCE(base_crng.generation, next_gen);
#ifdef CONFIG_VDSO_GETRANDOM
/* base_crng.generation's invalid value is ULONG_MAX, while
- * _vdso_rng_data.generation's invalid value is 0, so add one to the
+ * vdso_k_rng_data->generation's invalid value is 0, so add one to the
* former to arrive at the latter. Use smp_store_release so that this
* is ordered with the write above to base_crng.generation. Pairs with
* the smp_rmb() before the syscall in the vDSO code.
@@ -290,7 +290,7 @@ static void crng_reseed(struct work_struct *work)
* because the vDSO side only checks whether the value changed, without
* actually using or interpreting the value.
*/
- smp_store_release((unsigned long *)&__arch_get_k_vdso_rng_data()->generation, next_gen + 1);
+ smp_store_release((unsigned long *)&vdso_k_rng_data->generation, next_gen + 1);
#endif
if (!static_branch_likely(&crng_is_ready))
crng_init = CRNG_READY;
@@ -743,7 +743,7 @@ static void __cold _credit_init_bits(size_t bits)
queue_work(system_unbound_wq, &set_ready);
atomic_notifier_call_chain(&random_ready_notifier, 0, NULL);
#ifdef CONFIG_VDSO_GETRANDOM
- WRITE_ONCE(__arch_get_k_vdso_rng_data()->is_ready, true);
+ WRITE_ONCE(vdso_k_rng_data->is_ready, true);
#endif
wake_up_interruptible(&crng_init_wait);
kill_fasync(&fasync, SIGIO, POLL_IN);
diff --git a/include/asm-generic/vdso/vsyscall.h b/include/asm-generic/vdso/vsyscall.h
index ac5b93b91993224da245e80031b9f51e0c083f3c..a5f973e4e2723ef1815c88a7846f7c477e3febd9 100644
--- a/include/asm-generic/vdso/vsyscall.h
+++ b/include/asm-generic/vdso/vsyscall.h
@@ -13,6 +13,13 @@ static __always_inline const struct vdso_time_data *__arch_get_vdso_u_time_data(
}
#endif
+#ifndef __arch_get_vdso_u_rng_data
+static __always_inline const struct vdso_rng_data *__arch_get_vdso_u_rng_data(void)
+{
+ return &vdso_u_rng_data;
+}
+#endif
+
#else /* !CONFIG_GENERIC_VDSO_DATA_STORE */
#ifndef __arch_get_k_vdso_data
@@ -25,6 +32,11 @@ static __always_inline struct vdso_data *__arch_get_k_vdso_data(void)
#define __arch_get_vdso_u_time_data __arch_get_vdso_data
+#ifndef __arch_get_vdso_u_rng_data
+#define __arch_get_vdso_u_rng_data() __arch_get_vdso_rng_data()
+#endif
+#define vdso_k_rng_data __arch_get_k_vdso_rng_data()
+
#endif /* CONFIG_GENERIC_VDSO_DATA_STORE */
#ifndef __arch_update_vsyscall
diff --git a/include/vdso/datapage.h b/include/vdso/datapage.h
index b3d8087488ff35fbbc4d5058ae21e2c6cc58ed9c..7dbc8797e3b8923c06f4b6b34790766e68b2abd2 100644
--- a/include/vdso/datapage.h
+++ b/include/vdso/datapage.h
@@ -144,8 +144,10 @@ extern struct vdso_time_data _timens_data[CS_BASES] __attribute__((visibility("h
extern struct vdso_rng_data _vdso_rng_data __attribute__((visibility("hidden")));
#else
extern struct vdso_time_data vdso_u_time_data[CS_BASES] __attribute__((visibility("hidden")));
+extern struct vdso_rng_data vdso_u_rng_data __attribute__((visibility("hidden")));
extern struct vdso_time_data *vdso_k_time_data;
+extern struct vdso_rng_data *vdso_k_rng_data;
#endif
/**
@@ -161,6 +163,7 @@ union vdso_data_store {
enum vdso_pages {
VDSO_TIME_PAGE_OFFSET,
VDSO_TIMENS_PAGE_OFFSET,
+ VDSO_RNG_PAGE_OFFSET,
VDSO_NR_PAGES
};
@@ -184,9 +187,16 @@ enum vdso_pages {
#else /* !__ASSEMBLY__ */
+#ifdef CONFIG_VDSO_GETRANDOM
+#define __vdso_u_rng_data PROVIDE(vdso_u_rng_data = vdso_u_data + 2 * PAGE_SIZE);
+#else
+#define __vdso_u_rng_data
+#endif
+
#define VDSO_VVAR_SYMS \
PROVIDE(vdso_u_data = . - __VDSO_PAGES * PAGE_SIZE); \
PROVIDE(vdso_u_time_data = vdso_u_data); \
+ __vdso_u_rng_data \
#endif /* !__ASSEMBLY__ */
diff --git a/lib/vdso/datastore.c b/lib/vdso/datastore.c
index d3181768424cc3ee2e980bd537ac5ca14dfcb304..9260b00dc8520af109bd9dd05ac7f7504eafbea0 100644
--- a/lib/vdso/datastore.c
+++ b/lib/vdso/datastore.c
@@ -17,6 +17,15 @@ struct vdso_time_data *vdso_k_time_data = vdso_time_data_store.data;
static_assert(sizeof(vdso_time_data_store) == PAGE_SIZE);
#endif /* CONFIG_HAVE_GENERIC_VDSO */
+#ifdef CONFIG_VDSO_GETRANDOM
+static union {
+ struct vdso_rng_data data;
+ u8 page[PAGE_SIZE];
+} vdso_rng_data_store __page_aligned_data;
+struct vdso_rng_data *vdso_k_rng_data = &vdso_rng_data_store.data;
+static_assert(sizeof(vdso_rng_data_store) == PAGE_SIZE);
+#endif /* CONFIG_VDSO_GETRANDOM */
+
static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
struct vm_area_struct *vma, struct vm_fault *vmf)
{
@@ -53,6 +62,11 @@ static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
return VM_FAULT_SIGBUS;
pfn = __phys_to_pfn(__pa_symbol(vdso_k_time_data));
break;
+ case VDSO_RNG_PAGE_OFFSET:
+ if (!IS_ENABLED(CONFIG_VDSO_GETRANDOM))
+ return VM_FAULT_SIGBUS;
+ pfn = __phys_to_pfn(__pa_symbol(vdso_k_rng_data));
+ break;
default:
return VM_FAULT_SIGBUS;
}
diff --git a/lib/vdso/getrandom.c b/lib/vdso/getrandom.c
index 938ca539aaa64bc46280ef6dd17aa661126699eb..440f8a6203a69a6462aafee4ad8d5670cef6a353 100644
--- a/lib/vdso/getrandom.c
+++ b/lib/vdso/getrandom.c
@@ -12,6 +12,9 @@
#include <uapi/linux/mman.h>
#include <uapi/linux/random.h>
+/* Bring in default accessors */
+#include <vdso/vsyscall.h>
+
#undef PAGE_SIZE
#undef PAGE_MASK
#define PAGE_SIZE (1UL << CONFIG_PAGE_SHIFT)
@@ -152,7 +155,7 @@ __cvdso_getrandom_data(const struct vdso_rng_data *rng_info, void *buffer, size_
/*
* Prevent the syscall from being reordered wrt current_generation. Pairs with the
- * smp_store_release(&_vdso_rng_data.generation) in random.c.
+ * smp_store_release(&vdso_k_rng_data->generation) in random.c.
*/
smp_rmb();
@@ -256,5 +259,6 @@ __cvdso_getrandom_data(const struct vdso_rng_data *rng_info, void *buffer, size_
static __always_inline ssize_t
__cvdso_getrandom(void *buffer, size_t len, unsigned int flags, void *opaque_state, size_t opaque_len)
{
- return __cvdso_getrandom_data(__arch_get_vdso_rng_data(), buffer, len, flags, opaque_state, opaque_len);
+ return __cvdso_getrandom_data(__arch_get_vdso_u_rng_data(), buffer, len, flags,
+ opaque_state, opaque_len);
}
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 07/18] vdso: Add generic architecture-specific data storage
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (5 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 06/18] vdso: Add generic random " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 08/18] arm64: vdso: Switch to generic storage implementation Thomas Weißschuh
` (11 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
Some architectures need to architecture-specific data to the vDSO.
Enable the generic vDSO storage mechanism to both store and map this
data. Some architectures require more than a single page, like
LoongArch, so prepare for that usecase, too.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/Kconfig | 4 ++++
include/vdso/datapage.h | 25 +++++++++++++++++++++++++
lib/vdso/datastore.c | 14 ++++++++++++++
3 files changed, 43 insertions(+)
diff --git a/arch/Kconfig b/arch/Kconfig
index b8a4ff36558228240080a5677f702d37f4f8d547..9f6eb09ef12d76e755c0fd2df71ba703c251052f 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -1584,6 +1584,10 @@ config HAVE_SPARSE_SYSCALL_NR
entries at 4000, 5000 and 6000 locations. This option turns on syscall
related optimizations for a given architecture.
+config ARCH_HAS_VDSO_ARCH_DATA
+ depends on GENERIC_VDSO_DATA_STORE
+ bool
+
config ARCH_HAS_VDSO_TIME_DATA
bool
diff --git a/include/vdso/datapage.h b/include/vdso/datapage.h
index 7dbc8797e3b8923c06f4b6b34790766e68b2abd2..46658b39c28250e977f5a3454224c3ed0fb4c81d 100644
--- a/include/vdso/datapage.h
+++ b/include/vdso/datapage.h
@@ -9,11 +9,13 @@
#include <uapi/linux/types.h>
#include <uapi/asm-generic/errno-base.h>
+#include <vdso/align.h>
#include <vdso/bits.h>
#include <vdso/clocksource.h>
#include <vdso/ktime.h>
#include <vdso/limits.h>
#include <vdso/math64.h>
+#include <vdso/page.h>
#include <vdso/processor.h>
#include <vdso/time.h>
#include <vdso/time32.h>
@@ -25,6 +27,15 @@
struct arch_vdso_time_data {};
#endif
+#if defined(CONFIG_ARCH_HAS_VDSO_ARCH_DATA)
+#include <asm/vdso/arch_data.h>
+#elif defined(CONFIG_GENERIC_VDSO_DATA_STORE)
+struct vdso_arch_data {
+ /* Needed for the generic code, never actually used at runtime */
+ char __unused;
+};
+#endif
+
#define VDSO_BASES (CLOCK_TAI + 1)
#define VDSO_HRES (BIT(CLOCK_REALTIME) | \
BIT(CLOCK_MONOTONIC) | \
@@ -145,9 +156,11 @@ extern struct vdso_rng_data _vdso_rng_data __attribute__((visibility("hidden")))
#else
extern struct vdso_time_data vdso_u_time_data[CS_BASES] __attribute__((visibility("hidden")));
extern struct vdso_rng_data vdso_u_rng_data __attribute__((visibility("hidden")));
+extern struct vdso_arch_data vdso_u_arch_data __attribute__((visibility("hidden")));
extern struct vdso_time_data *vdso_k_time_data;
extern struct vdso_rng_data *vdso_k_rng_data;
+extern struct vdso_arch_data *vdso_k_arch_data;
#endif
/**
@@ -160,10 +173,15 @@ union vdso_data_store {
#ifdef CONFIG_GENERIC_VDSO_DATA_STORE
+#define VDSO_ARCH_DATA_SIZE ALIGN(sizeof(struct vdso_arch_data), PAGE_SIZE)
+#define VDSO_ARCH_DATA_PAGES (VDSO_ARCH_DATA_SIZE >> PAGE_SHIFT)
+
enum vdso_pages {
VDSO_TIME_PAGE_OFFSET,
VDSO_TIMENS_PAGE_OFFSET,
VDSO_RNG_PAGE_OFFSET,
+ VDSO_ARCH_PAGES_START,
+ VDSO_ARCH_PAGES_END = VDSO_ARCH_PAGES_START + VDSO_ARCH_DATA_PAGES - 1,
VDSO_NR_PAGES
};
@@ -193,10 +211,17 @@ enum vdso_pages {
#define __vdso_u_rng_data
#endif
+#ifdef CONFIG_ARCH_HAS_VDSO_ARCH_DATA
+#define __vdso_u_arch_data PROVIDE(vdso_u_arch_data = vdso_u_data + 3 * PAGE_SIZE);
+#else
+#define __vdso_u_arch_data
+#endif
+
#define VDSO_VVAR_SYMS \
PROVIDE(vdso_u_data = . - __VDSO_PAGES * PAGE_SIZE); \
PROVIDE(vdso_u_time_data = vdso_u_data); \
__vdso_u_rng_data \
+ __vdso_u_arch_data \
#endif /* !__ASSEMBLY__ */
diff --git a/lib/vdso/datastore.c b/lib/vdso/datastore.c
index 9260b00dc8520af109bd9dd05ac7f7504eafbea0..0959d62d78586ef458eb5f7c1e152f6b5d4cbf85 100644
--- a/lib/vdso/datastore.c
+++ b/lib/vdso/datastore.c
@@ -26,6 +26,14 @@ struct vdso_rng_data *vdso_k_rng_data = &vdso_rng_data_store.data;
static_assert(sizeof(vdso_rng_data_store) == PAGE_SIZE);
#endif /* CONFIG_VDSO_GETRANDOM */
+#ifdef CONFIG_ARCH_HAS_VDSO_ARCH_DATA
+static union {
+ struct vdso_arch_data data;
+ u8 page[VDSO_ARCH_DATA_SIZE];
+} vdso_arch_data_store __page_aligned_data;
+struct vdso_arch_data *vdso_k_arch_data = &vdso_arch_data_store.data;
+#endif /* CONFIG_ARCH_HAS_VDSO_ARCH_DATA */
+
static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
struct vm_area_struct *vma, struct vm_fault *vmf)
{
@@ -67,6 +75,12 @@ static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
return VM_FAULT_SIGBUS;
pfn = __phys_to_pfn(__pa_symbol(vdso_k_rng_data));
break;
+ case VDSO_ARCH_PAGES_START ... VDSO_ARCH_PAGES_END:
+ if (!IS_ENABLED(CONFIG_ARCH_HAS_VDSO_ARCH_DATA))
+ return VM_FAULT_SIGBUS;
+ pfn = __phys_to_pfn(__pa_symbol(vdso_k_arch_data)) +
+ vmf->pgoff - VDSO_ARCH_PAGES_START;
+ break;
default:
return VM_FAULT_SIGBUS;
}
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 08/18] arm64: vdso: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (6 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 07/18] vdso: Add generic architecture-specific " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 09/18] riscv: " Thomas Weißschuh
` (10 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
This switch also moves the random state data out of the time data page.
The currently used hardcoded __VDSO_RND_DATA_OFFSET does not take into
account changes to the time data page layout.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/arm64/Kconfig | 1 +
arch/arm64/include/asm/vdso.h | 2 +-
arch/arm64/include/asm/vdso/compat_gettimeofday.h | 36 +++------
arch/arm64/include/asm/vdso/getrandom.h | 12 ---
arch/arm64/include/asm/vdso/gettimeofday.h | 16 +---
arch/arm64/include/asm/vdso/vsyscall.h | 25 +------
arch/arm64/kernel/vdso.c | 90 ++---------------------
arch/arm64/kernel/vdso/vdso.lds.S | 7 +-
arch/arm64/kernel/vdso32/vdso.lds.S | 7 +-
9 files changed, 26 insertions(+), 170 deletions(-)
diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index fcdd0ed3eca8909aee56bf7bcce5f8135254f0d7..3770b73f270bbf3e2f48d77e8dd231861ac4e8d7 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -162,6 +162,7 @@ config ARM64
select GENERIC_SMP_IDLE_THREAD
select GENERIC_TIME_VSYSCALL
select GENERIC_GETTIMEOFDAY
+ select GENERIC_VDSO_DATA_STORE
select GENERIC_VDSO_TIME_NS
select HARDIRQS_SW_RESEND
select HAS_IOPORT
diff --git a/arch/arm64/include/asm/vdso.h b/arch/arm64/include/asm/vdso.h
index 3e3c3fdb184274abd20647335b19e81e709506db..61679070f595cd25fb2f516b5b077599a9570689 100644
--- a/arch/arm64/include/asm/vdso.h
+++ b/arch/arm64/include/asm/vdso.h
@@ -5,7 +5,7 @@
#ifndef __ASM_VDSO_H
#define __ASM_VDSO_H
-#define __VVAR_PAGES 2
+#define __VDSO_PAGES 4
#ifndef __ASSEMBLY__
diff --git a/arch/arm64/include/asm/vdso/compat_gettimeofday.h b/arch/arm64/include/asm/vdso/compat_gettimeofday.h
index 778c1202bbbf9f5db6bfced62a833d50e1675b08..957ee12fcc54bd7f978fbcd8945bce62327b037a 100644
--- a/arch/arm64/include/asm/vdso/compat_gettimeofday.h
+++ b/arch/arm64/include/asm/vdso/compat_gettimeofday.h
@@ -104,7 +104,7 @@ int clock_getres32_fallback(clockid_t _clkid, struct old_timespec32 *_ts)
}
static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
u64 res;
@@ -131,43 +131,31 @@ static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
return res;
}
-static __always_inline const struct vdso_data *__arch_get_vdso_data(void)
+static __always_inline const struct vdso_time_data *__arch_get_vdso_u_time_data(void)
{
- const struct vdso_data *ret;
+ const struct vdso_time_data *ret;
/*
- * This simply puts &_vdso_data into ret. The reason why we don't use
- * `ret = _vdso_data` is that the compiler tends to optimise this in a
- * very suboptimal way: instead of keeping &_vdso_data in a register,
- * it goes through a relocation almost every time _vdso_data must be
+ * This simply puts &_vdso_time_data into ret. The reason why we don't use
+ * `ret = _vdso_time_data` is that the compiler tends to optimise this in a
+ * very suboptimal way: instead of keeping &_vdso_time_data in a register,
+ * it goes through a relocation almost every time _vdso_time_data must be
* accessed (even in subfunctions). This is both time and space
* consuming: each relocation uses a word in the code section, and it
* has to be loaded at runtime.
*
* This trick hides the assignment from the compiler. Since it cannot
* track where the pointer comes from, it will only use one relocation
- * where __arch_get_vdso_data() is called, and then keep the result in
- * a register.
+ * where __aarch64_get_vdso_u_time_data() is called, and then keep the
+ * result in a register.
*/
- asm volatile("mov %0, %1" : "=r"(ret) : "r"(_vdso_data));
+ asm volatile("mov %0, %1" : "=r"(ret) : "r"(vdso_u_time_data));
return ret;
}
+#define __arch_get_vdso_u_time_data __arch_get_vdso_u_time_data
-#ifdef CONFIG_TIME_NS
-static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
-{
- const struct vdso_data *ret;
-
- /* See __arch_get_vdso_data(). */
- asm volatile("mov %0, %1" : "=r"(ret) : "r"(_timens_data));
-
- return ret;
-}
-#endif
-
-static inline bool vdso_clocksource_ok(const struct vdso_data *vd)
+static inline bool vdso_clocksource_ok(const struct vdso_time_data *vd)
{
return vd->clock_mode == VDSO_CLOCKMODE_ARCHTIMER;
}
diff --git a/arch/arm64/include/asm/vdso/getrandom.h b/arch/arm64/include/asm/vdso/getrandom.h
index 342f807e204442c95cb9e9a849da38effbe50334..a2197da1951b021baff4eb08b3f1c314df2f5f23 100644
--- a/arch/arm64/include/asm/vdso/getrandom.h
+++ b/arch/arm64/include/asm/vdso/getrandom.h
@@ -33,18 +33,6 @@ static __always_inline ssize_t getrandom_syscall(void *_buffer, size_t _len, uns
return ret;
}
-static __always_inline const struct vdso_rng_data *__arch_get_vdso_rng_data(void)
-{
- /*
- * The RNG data is in the real VVAR data page, but if a task belongs to a time namespace
- * then VVAR_DATA_PAGE_OFFSET points to the namespace-specific VVAR page and VVAR_TIMENS_
- * PAGE_OFFSET points to the real VVAR page.
- */
- if (IS_ENABLED(CONFIG_TIME_NS) && _vdso_data->clock_mode == VDSO_CLOCKMODE_TIMENS)
- return (void *)&_vdso_rng_data + VVAR_TIMENS_PAGE_OFFSET * (1UL << CONFIG_PAGE_SHIFT);
- return &_vdso_rng_data;
-}
-
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETRANDOM_H */
diff --git a/arch/arm64/include/asm/vdso/gettimeofday.h b/arch/arm64/include/asm/vdso/gettimeofday.h
index 764d13e2916c559bc8fd7efa032b6f468ac0d4a2..92a2b59a9f3df4d20feb483e6d8ebd1d813b7932 100644
--- a/arch/arm64/include/asm/vdso/gettimeofday.h
+++ b/arch/arm64/include/asm/vdso/gettimeofday.h
@@ -67,7 +67,7 @@ int clock_getres_fallback(clockid_t _clkid, struct __kernel_timespec *_ts)
}
static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
u64 res;
@@ -99,20 +99,6 @@ static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
return res;
}
-static __always_inline
-const struct vdso_data *__arch_get_vdso_data(void)
-{
- return _vdso_data;
-}
-
-#ifdef CONFIG_TIME_NS
-static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
-{
- return _timens_data;
-}
-#endif
-
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETTIMEOFDAY_H */
diff --git a/arch/arm64/include/asm/vdso/vsyscall.h b/arch/arm64/include/asm/vdso/vsyscall.h
index eea51946d45a2f8c7eebfff971d74878be53a798..3f65cbd00635aab50a4e0c6058d38b39fd6d43a9 100644
--- a/arch/arm64/include/asm/vdso/vsyscall.h
+++ b/arch/arm64/include/asm/vdso/vsyscall.h
@@ -2,41 +2,18 @@
#ifndef __ASM_VDSO_VSYSCALL_H
#define __ASM_VDSO_VSYSCALL_H
-#define __VDSO_RND_DATA_OFFSET 480
-
#ifndef __ASSEMBLY__
#include <vdso/datapage.h>
-enum vvar_pages {
- VVAR_DATA_PAGE_OFFSET,
- VVAR_TIMENS_PAGE_OFFSET,
- VVAR_NR_PAGES,
-};
-
#define VDSO_PRECISION_MASK ~(0xFF00ULL<<48)
-extern struct vdso_data *vdso_data;
/*
* Update the vDSO data page to keep in sync with kernel timekeeping.
*/
static __always_inline
-struct vdso_data *__arm64_get_k_vdso_data(void)
-{
- return vdso_data;
-}
-#define __arch_get_k_vdso_data __arm64_get_k_vdso_data
-
-static __always_inline
-struct vdso_rng_data *__arm64_get_k_vdso_rnd_data(void)
-{
- return (void *)vdso_data + __VDSO_RND_DATA_OFFSET;
-}
-#define __arch_get_k_vdso_rng_data __arm64_get_k_vdso_rnd_data
-
-static __always_inline
-void __arm64_update_vsyscall(struct vdso_data *vdata)
+void __arm64_update_vsyscall(struct vdso_time_data *vdata)
{
vdata[CS_HRES_COARSE].mask = VDSO_PRECISION_MASK;
vdata[CS_RAW].mask = VDSO_PRECISION_MASK;
diff --git a/arch/arm64/kernel/vdso.c b/arch/arm64/kernel/vdso.c
index e8ed8e5b713b525abac828b1a17ab9e6d974a3e9..887ac0b05961eb83afe649a0afcaf7deda4b186b 100644
--- a/arch/arm64/kernel/vdso.c
+++ b/arch/arm64/kernel/vdso.c
@@ -18,7 +18,7 @@
#include <linux/sched.h>
#include <linux/signal.h>
#include <linux/slab.h>
-#include <linux/time_namespace.h>
+#include <linux/vdso_datastore.h>
#include <linux/vmalloc.h>
#include <vdso/datapage.h>
#include <vdso/helpers.h>
@@ -57,12 +57,6 @@ static struct vdso_abi_info vdso_info[] __ro_after_init = {
#endif /* CONFIG_COMPAT_VDSO */
};
-/*
- * The vDSO data page.
- */
-static union vdso_data_store vdso_data_store __page_aligned_data;
-struct vdso_data *vdso_data = vdso_data_store.data;
-
static int vdso_mremap(const struct vm_special_mapping *sm,
struct vm_area_struct *new_vma)
{
@@ -104,78 +98,6 @@ static int __init __vdso_init(enum vdso_abi abi)
return 0;
}
-#ifdef CONFIG_TIME_NS
-struct vdso_data *arch_get_vdso_data(void *vvar_page)
-{
- return (struct vdso_data *)(vvar_page);
-}
-
-static const struct vm_special_mapping vvar_map;
-
-/*
- * The vvar mapping contains data for a specific time namespace, so when a task
- * changes namespace we must unmap its vvar data for the old namespace.
- * Subsequent faults will map in data for the new namespace.
- *
- * For more details see timens_setup_vdso_data().
- */
-int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
-{
- struct mm_struct *mm = task->mm;
- struct vm_area_struct *vma;
- VMA_ITERATOR(vmi, mm, 0);
-
- mmap_read_lock(mm);
-
- for_each_vma(vmi, vma) {
- if (vma_is_special_mapping(vma, &vvar_map))
- zap_vma_pages(vma);
- }
-
- mmap_read_unlock(mm);
- return 0;
-}
-#endif
-
-static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
- struct vm_area_struct *vma, struct vm_fault *vmf)
-{
- struct page *timens_page = find_timens_vvar_page(vma);
- unsigned long pfn;
-
- switch (vmf->pgoff) {
- case VVAR_DATA_PAGE_OFFSET:
- if (timens_page)
- pfn = page_to_pfn(timens_page);
- else
- pfn = sym_to_pfn(vdso_data);
- break;
-#ifdef CONFIG_TIME_NS
- case VVAR_TIMENS_PAGE_OFFSET:
- /*
- * If a task belongs to a time namespace then a namespace
- * specific VVAR is mapped with the VVAR_DATA_PAGE_OFFSET and
- * the real VVAR page is mapped with the VVAR_TIMENS_PAGE_OFFSET
- * offset.
- * See also the comment near timens_setup_vdso_data().
- */
- if (!timens_page)
- return VM_FAULT_SIGBUS;
- pfn = sym_to_pfn(vdso_data);
- break;
-#endif /* CONFIG_TIME_NS */
- default:
- return VM_FAULT_SIGBUS;
- }
-
- return vmf_insert_pfn(vma, vmf->address, pfn);
-}
-
-static const struct vm_special_mapping vvar_map = {
- .name = "[vvar]",
- .fault = vvar_fault,
-};
-
static int __setup_additional_pages(enum vdso_abi abi,
struct mm_struct *mm,
struct linux_binprm *bprm,
@@ -185,11 +107,11 @@ static int __setup_additional_pages(enum vdso_abi abi,
unsigned long gp_flags = 0;
void *ret;
- BUILD_BUG_ON(VVAR_NR_PAGES != __VVAR_PAGES);
+ BUILD_BUG_ON(VDSO_NR_PAGES != __VDSO_PAGES);
vdso_text_len = vdso_info[abi].vdso_pages << PAGE_SHIFT;
/* Be sure to map the data page */
- vdso_mapping_len = vdso_text_len + VVAR_NR_PAGES * PAGE_SIZE;
+ vdso_mapping_len = vdso_text_len + VDSO_NR_PAGES * PAGE_SIZE;
vdso_base = get_unmapped_area(NULL, 0, vdso_mapping_len, 0, 0);
if (IS_ERR_VALUE(vdso_base)) {
@@ -197,16 +119,14 @@ static int __setup_additional_pages(enum vdso_abi abi,
goto up_fail;
}
- ret = _install_special_mapping(mm, vdso_base, VVAR_NR_PAGES * PAGE_SIZE,
- VM_READ|VM_MAYREAD|VM_PFNMAP,
- &vvar_map);
+ ret = vdso_install_vvar_mapping(mm, vdso_base);
if (IS_ERR(ret))
goto up_fail;
if (system_supports_bti_kernel())
gp_flags = VM_ARM64_BTI;
- vdso_base += VVAR_NR_PAGES * PAGE_SIZE;
+ vdso_base += VDSO_NR_PAGES * PAGE_SIZE;
mm->context.vdso = (void *)vdso_base;
ret = _install_special_mapping(mm, vdso_base, vdso_text_len,
VM_READ|VM_EXEC|gp_flags|
diff --git a/arch/arm64/kernel/vdso/vdso.lds.S b/arch/arm64/kernel/vdso/vdso.lds.S
index 4ec32e86a8da22d5e2315e55ae2e86ec8e7f5f9a..2e20d0593f182bfefcf39aebb1400f32af0866dc 100644
--- a/arch/arm64/kernel/vdso/vdso.lds.S
+++ b/arch/arm64/kernel/vdso/vdso.lds.S
@@ -20,11 +20,8 @@ OUTPUT_ARCH(aarch64)
SECTIONS
{
- PROVIDE(_vdso_data = . - __VVAR_PAGES * PAGE_SIZE);
- PROVIDE(_vdso_rng_data = _vdso_data + __VDSO_RND_DATA_OFFSET);
-#ifdef CONFIG_TIME_NS
- PROVIDE(_timens_data = _vdso_data + PAGE_SIZE);
-#endif
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
diff --git a/arch/arm64/kernel/vdso32/vdso.lds.S b/arch/arm64/kernel/vdso32/vdso.lds.S
index 732702a187e9e8b62972ee0aad75c23568e5e779..e02b27487ce80497663dd042357558fea201fcce 100644
--- a/arch/arm64/kernel/vdso32/vdso.lds.S
+++ b/arch/arm64/kernel/vdso32/vdso.lds.S
@@ -12,16 +12,15 @@
#include <asm/page.h>
#include <asm/vdso.h>
#include <asm-generic/vmlinux.lds.h>
+#include <vdso/datapage.h>
OUTPUT_FORMAT("elf32-littlearm", "elf32-bigarm", "elf32-littlearm")
OUTPUT_ARCH(arm)
SECTIONS
{
- PROVIDE_HIDDEN(_vdso_data = . - __VVAR_PAGES * PAGE_SIZE);
-#ifdef CONFIG_TIME_NS
- PROVIDE_HIDDEN(_timens_data = _vdso_data + PAGE_SIZE);
-#endif
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 09/18] riscv: vdso: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (7 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 08/18] arm64: vdso: Switch to generic storage implementation Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-22 8:17 ` Xi Ruoyao
2025-02-04 12:05 ` [PATCH v3 10/18] LoongArch: vDSO: " Thomas Weißschuh
` (9 subsequent siblings)
18 siblings, 1 reply; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/riscv/Kconfig | 3 +-
arch/riscv/include/asm/vdso.h | 2 +-
.../include/asm/vdso/{time_data.h => arch_data.h} | 8 +-
arch/riscv/include/asm/vdso/gettimeofday.h | 14 +---
arch/riscv/include/asm/vdso/vsyscall.h | 9 ---
arch/riscv/kernel/sys_hwprobe.c | 3 +-
arch/riscv/kernel/vdso.c | 90 +---------------------
arch/riscv/kernel/vdso/hwprobe.c | 6 +-
arch/riscv/kernel/vdso/vdso.lds.S | 7 +-
9 files changed, 18 insertions(+), 124 deletions(-)
diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
index 7612c52e9b1e35607f1dd4603a596416d3357a71..aa8ea53186c04ad68582255f74b09a0605fe8368 100644
--- a/arch/riscv/Kconfig
+++ b/arch/riscv/Kconfig
@@ -53,7 +53,7 @@ config RISCV
select ARCH_HAS_SYSCALL_WRAPPER
select ARCH_HAS_TICK_BROADCAST if GENERIC_CLOCKEVENTS_BROADCAST
select ARCH_HAS_UBSAN
- select ARCH_HAS_VDSO_TIME_DATA
+ select ARCH_HAS_VDSO_ARCH_DATA if GENERIC_VDSO_DATA_STORE
select ARCH_KEEP_MEMBLOCK if ACPI
select ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE if 64BIT && MMU
select ARCH_OPTIONAL_KERNEL_RWX if ARCH_HAS_STRICT_KERNEL_RWX
@@ -116,6 +116,7 @@ config RISCV
select GENERIC_SCHED_CLOCK
select GENERIC_SMP_IDLE_THREAD
select GENERIC_TIME_VSYSCALL if MMU && 64BIT
+ select GENERIC_VDSO_DATA_STORE if MMU
select GENERIC_VDSO_TIME_NS if HAVE_GENERIC_VDSO
select HARDIRQS_SW_RESEND
select HAS_IOPORT if MMU
diff --git a/arch/riscv/include/asm/vdso.h b/arch/riscv/include/asm/vdso.h
index f891478829a52c41e06240f67611694cc28197d9..c130d8100232cbe50e52e35eb418e354bd114cb7 100644
--- a/arch/riscv/include/asm/vdso.h
+++ b/arch/riscv/include/asm/vdso.h
@@ -14,7 +14,7 @@
*/
#ifdef CONFIG_MMU
-#define __VVAR_PAGES 2
+#define __VDSO_PAGES 4
#ifndef __ASSEMBLY__
#include <generated/vdso-offsets.h>
diff --git a/arch/riscv/include/asm/vdso/time_data.h b/arch/riscv/include/asm/vdso/arch_data.h
similarity index 71%
rename from arch/riscv/include/asm/vdso/time_data.h
rename to arch/riscv/include/asm/vdso/arch_data.h
index dfa65228999bed41dfd6c5e36cb678e1e055eec8..da57a3786f7a53c866fc00948826b4a2d839940f 100644
--- a/arch/riscv/include/asm/vdso/time_data.h
+++ b/arch/riscv/include/asm/vdso/arch_data.h
@@ -1,12 +1,12 @@
/* SPDX-License-Identifier: GPL-2.0 */
-#ifndef __RISCV_ASM_VDSO_TIME_DATA_H
-#define __RISCV_ASM_VDSO_TIME_DATA_H
+#ifndef __RISCV_ASM_VDSO_ARCH_DATA_H
+#define __RISCV_ASM_VDSO_ARCH_DATA_H
#include <linux/types.h>
#include <vdso/datapage.h>
#include <asm/hwprobe.h>
-struct arch_vdso_time_data {
+struct vdso_arch_data {
/* Stash static answers to the hwprobe queries when all CPUs are selected. */
__u64 all_cpu_hwprobe_values[RISCV_HWPROBE_MAX_KEY + 1];
@@ -14,4 +14,4 @@ struct arch_vdso_time_data {
__u8 homogeneous_cpus;
};
-#endif /* __RISCV_ASM_VDSO_TIME_DATA_H */
+#endif /* __RISCV_ASM_VDSO_ARCH_DATA_H */
diff --git a/arch/riscv/include/asm/vdso/gettimeofday.h b/arch/riscv/include/asm/vdso/gettimeofday.h
index ba3283cf7accaa93a38512d2c17eda0eefde0612..29164f84f93cec6e28251e6a0adfbc341ac88241 100644
--- a/arch/riscv/include/asm/vdso/gettimeofday.h
+++ b/arch/riscv/include/asm/vdso/gettimeofday.h
@@ -69,7 +69,7 @@ int clock_getres_fallback(clockid_t _clkid, struct __kernel_timespec *_ts)
#endif /* CONFIG_GENERIC_TIME_VSYSCALL */
static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
/*
* The purpose of csr_read(CSR_TIME) is to trap the system into
@@ -79,18 +79,6 @@ static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
return csr_read(CSR_TIME);
}
-static __always_inline const struct vdso_data *__arch_get_vdso_data(void)
-{
- return _vdso_data;
-}
-
-#ifdef CONFIG_TIME_NS
-static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
-{
- return _timens_data;
-}
-#endif
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETTIMEOFDAY_H */
diff --git a/arch/riscv/include/asm/vdso/vsyscall.h b/arch/riscv/include/asm/vdso/vsyscall.h
index e8a9c4b53c0c9f4744196eed800b21f3918d1040..1140b54b4bc8278d7a322036cd9f84f71258f246 100644
--- a/arch/riscv/include/asm/vdso/vsyscall.h
+++ b/arch/riscv/include/asm/vdso/vsyscall.h
@@ -6,15 +6,6 @@
#include <vdso/datapage.h>
-extern struct vdso_data *vdso_data;
-
-static __always_inline struct vdso_data *__riscv_get_k_vdso_data(void)
-{
- return vdso_data;
-}
-
-#define __arch_get_k_vdso_data __riscv_get_k_vdso_data
-
/* The asm-generic header needs to be included after the definitions above */
#include <asm-generic/vdso/vsyscall.h>
diff --git a/arch/riscv/kernel/sys_hwprobe.c b/arch/riscv/kernel/sys_hwprobe.c
index bcd3b816306c22df62f60ad044f4ae58f7dad4d1..04a4e549551284bb3340673eb76a2e7bd457025e 100644
--- a/arch/riscv/kernel/sys_hwprobe.c
+++ b/arch/riscv/kernel/sys_hwprobe.c
@@ -450,8 +450,7 @@ static int do_riscv_hwprobe(struct riscv_hwprobe __user *pairs,
static int __init init_hwprobe_vdso_data(void)
{
- struct vdso_data *vd = __arch_get_k_vdso_data();
- struct arch_vdso_time_data *avd = &vd->arch_data;
+ struct vdso_arch_data *avd = vdso_k_arch_data;
u64 id_bitsmash = 0;
struct riscv_hwprobe pair;
int key;
diff --git a/arch/riscv/kernel/vdso.c b/arch/riscv/kernel/vdso.c
index 3ca3ae4277e187e790a8bf513a9e80d8b6290bb2..cc2895d1fbc2fe752b3edc94f4e28a6a8fca7a3b 100644
--- a/arch/riscv/kernel/vdso.c
+++ b/arch/riscv/kernel/vdso.c
@@ -13,20 +13,11 @@
#include <linux/err.h>
#include <asm/page.h>
#include <asm/vdso.h>
-#include <linux/time_namespace.h>
+#include <linux/vdso_datastore.h>
#include <vdso/datapage.h>
#include <vdso/vsyscall.h>
-enum vvar_pages {
- VVAR_DATA_PAGE_OFFSET,
- VVAR_TIMENS_PAGE_OFFSET,
- VVAR_NR_PAGES,
-};
-
-#define VVAR_SIZE (VVAR_NR_PAGES << PAGE_SHIFT)
-
-static union vdso_data_store vdso_data_store __page_aligned_data;
-struct vdso_data *vdso_data = vdso_data_store.data;
+#define VVAR_SIZE (VDSO_NR_PAGES << PAGE_SHIFT)
struct __vdso_info {
const char *name;
@@ -79,78 +70,6 @@ static void __init __vdso_init(struct __vdso_info *vdso_info)
vdso_info->cm->pages = vdso_pagelist;
}
-#ifdef CONFIG_TIME_NS
-struct vdso_data *arch_get_vdso_data(void *vvar_page)
-{
- return (struct vdso_data *)(vvar_page);
-}
-
-static const struct vm_special_mapping rv_vvar_map;
-
-/*
- * The vvar mapping contains data for a specific time namespace, so when a task
- * changes namespace we must unmap its vvar data for the old namespace.
- * Subsequent faults will map in data for the new namespace.
- *
- * For more details see timens_setup_vdso_data().
- */
-int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
-{
- struct mm_struct *mm = task->mm;
- struct vm_area_struct *vma;
- VMA_ITERATOR(vmi, mm, 0);
-
- mmap_read_lock(mm);
-
- for_each_vma(vmi, vma) {
- if (vma_is_special_mapping(vma, &rv_vvar_map))
- zap_vma_pages(vma);
- }
-
- mmap_read_unlock(mm);
- return 0;
-}
-#endif
-
-static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
- struct vm_area_struct *vma, struct vm_fault *vmf)
-{
- struct page *timens_page = find_timens_vvar_page(vma);
- unsigned long pfn;
-
- switch (vmf->pgoff) {
- case VVAR_DATA_PAGE_OFFSET:
- if (timens_page)
- pfn = page_to_pfn(timens_page);
- else
- pfn = sym_to_pfn(vdso_data);
- break;
-#ifdef CONFIG_TIME_NS
- case VVAR_TIMENS_PAGE_OFFSET:
- /*
- * If a task belongs to a time namespace then a namespace
- * specific VVAR is mapped with the VVAR_DATA_PAGE_OFFSET and
- * the real VVAR page is mapped with the VVAR_TIMENS_PAGE_OFFSET
- * offset.
- * See also the comment near timens_setup_vdso_data().
- */
- if (!timens_page)
- return VM_FAULT_SIGBUS;
- pfn = sym_to_pfn(vdso_data);
- break;
-#endif /* CONFIG_TIME_NS */
- default:
- return VM_FAULT_SIGBUS;
- }
-
- return vmf_insert_pfn(vma, vmf->address, pfn);
-}
-
-static const struct vm_special_mapping rv_vvar_map = {
- .name = "[vvar]",
- .fault = vvar_fault,
-};
-
static struct vm_special_mapping rv_vdso_map __ro_after_init = {
.name = "[vdso]",
.mremap = vdso_mremap,
@@ -196,7 +115,7 @@ static int __setup_additional_pages(struct mm_struct *mm,
unsigned long vdso_base, vdso_text_len, vdso_mapping_len;
void *ret;
- BUILD_BUG_ON(VVAR_NR_PAGES != __VVAR_PAGES);
+ BUILD_BUG_ON(VDSO_NR_PAGES != __VDSO_PAGES);
vdso_text_len = vdso_info->vdso_pages << PAGE_SHIFT;
/* Be sure to map the data page */
@@ -208,8 +127,7 @@ static int __setup_additional_pages(struct mm_struct *mm,
goto up_fail;
}
- ret = _install_special_mapping(mm, vdso_base, VVAR_SIZE,
- (VM_READ | VM_MAYREAD | VM_PFNMAP), &rv_vvar_map);
+ ret = vdso_install_vvar_mapping(mm, vdso_base);
if (IS_ERR(ret))
goto up_fail;
diff --git a/arch/riscv/kernel/vdso/hwprobe.c b/arch/riscv/kernel/vdso/hwprobe.c
index a158c029344f60c022e7565757ff44df7e3d89e5..2ddeba6c68dda09b0249117fd06a5d249f3b0abd 100644
--- a/arch/riscv/kernel/vdso/hwprobe.c
+++ b/arch/riscv/kernel/vdso/hwprobe.c
@@ -16,8 +16,7 @@ static int riscv_vdso_get_values(struct riscv_hwprobe *pairs, size_t pair_count,
size_t cpusetsize, unsigned long *cpus,
unsigned int flags)
{
- const struct vdso_data *vd = __arch_get_vdso_data();
- const struct arch_vdso_time_data *avd = &vd->arch_data;
+ const struct vdso_arch_data *avd = &vdso_u_arch_data;
bool all_cpus = !cpusetsize && !cpus;
struct riscv_hwprobe *p = pairs;
struct riscv_hwprobe *end = pairs + pair_count;
@@ -51,8 +50,7 @@ static int riscv_vdso_get_cpus(struct riscv_hwprobe *pairs, size_t pair_count,
size_t cpusetsize, unsigned long *cpus,
unsigned int flags)
{
- const struct vdso_data *vd = __arch_get_vdso_data();
- const struct arch_vdso_time_data *avd = &vd->arch_data;
+ const struct vdso_arch_data *avd = &vdso_u_arch_data;
struct riscv_hwprobe *p = pairs;
struct riscv_hwprobe *end = pairs + pair_count;
unsigned char *c = (unsigned char *)cpus;
diff --git a/arch/riscv/kernel/vdso/vdso.lds.S b/arch/riscv/kernel/vdso/vdso.lds.S
index cbe2a179331d2511a8b4a26c06383e46131661b1..8e86965a8aae4d7c5a36d0f26026cd1c8680b339 100644
--- a/arch/riscv/kernel/vdso/vdso.lds.S
+++ b/arch/riscv/kernel/vdso/vdso.lds.S
@@ -4,15 +4,14 @@
*/
#include <asm/page.h>
#include <asm/vdso.h>
+#include <vdso/datapage.h>
OUTPUT_ARCH(riscv)
SECTIONS
{
- PROVIDE(_vdso_data = . - __VVAR_PAGES * PAGE_SIZE);
-#ifdef CONFIG_TIME_NS
- PROVIDE(_timens_data = _vdso_data + PAGE_SIZE);
-#endif
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* Re: [PATCH v3 09/18] riscv: vdso: Switch to generic storage implementation
2025-02-04 12:05 ` [PATCH v3 09/18] riscv: " Thomas Weißschuh
@ 2025-02-22 8:17 ` Xi Ruoyao
2025-02-22 10:14 ` Thomas Gleixner
0 siblings, 1 reply; 29+ messages in thread
From: Xi Ruoyao @ 2025-02-22 8:17 UTC (permalink / raw)
To: Thomas Weißschuh, James E.J. Bottomley, Helge Deller,
Andy Lutomirski, Thomas Gleixner, Vincenzo Frascino,
Anna-Maria Behnsen, Frederic Weisbecker, Andrew Morton,
Catalin Marinas, Will Deacon, Theodore Ts'o,
Jason A. Donenfeld, Paul Walmsley, Palmer Dabbelt, Albert Ou,
Huacai Chen, WANG Xuerui, Russell King, Heiko Carstens,
Vasily Gorbik, Alexander Gordeev, Christian Borntraeger,
Sven Schnelle, Thomas Bogendoerfer, Michael Ellerman,
Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, linux-csky
On Tue, 2025-02-04 at 13:05 +0100, Thomas Weißschuh wrote:
> The generic storage implementation provides the same features as the
> custom one. However it can be shared between architectures, making
> maintenance easier.
>
> Co-developed-by: Nam Cao <namcao@linutronix.de>
> Signed-off-by: Nam Cao <namcao@linutronix.de>
> Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
I made a RISC-V vDSO getrandom implementation on top of this and it
works fine. I'll submit it when this is merged.
Tested-by: Xi Ruoyao <xry111@xry111.site>
> ---
> arch/riscv/Kconfig | 3 +-
> arch/riscv/include/asm/vdso.h | 2 +-
> .../include/asm/vdso/{time_data.h => arch_data.h} | 8 +-
> arch/riscv/include/asm/vdso/gettimeofday.h | 14 +---
> arch/riscv/include/asm/vdso/vsyscall.h | 9 ---
> arch/riscv/kernel/sys_hwprobe.c | 3 +-
> arch/riscv/kernel/vdso.c | 90 +------------
> ---------
> arch/riscv/kernel/vdso/hwprobe.c | 6 +-
> arch/riscv/kernel/vdso/vdso.lds.S | 7 +-
> 9 files changed, 18 insertions(+), 124 deletions(-)
>
> diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> index
> 7612c52e9b1e35607f1dd4603a596416d3357a71..aa8ea53186c04ad68582255f74b0
> 9a0605fe8368 100644
> --- a/arch/riscv/Kconfig
> +++ b/arch/riscv/Kconfig
> @@ -53,7 +53,7 @@ config RISCV
> select ARCH_HAS_SYSCALL_WRAPPER
> select ARCH_HAS_TICK_BROADCAST if
> GENERIC_CLOCKEVENTS_BROADCAST
> select ARCH_HAS_UBSAN
> - select ARCH_HAS_VDSO_TIME_DATA
> + select ARCH_HAS_VDSO_ARCH_DATA if GENERIC_VDSO_DATA_STORE
> select ARCH_KEEP_MEMBLOCK if ACPI
> select ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE if 64BIT &&
> MMU
> select ARCH_OPTIONAL_KERNEL_RWX if ARCH_HAS_STRICT_KERNEL_RWX
> @@ -116,6 +116,7 @@ config RISCV
> select GENERIC_SCHED_CLOCK
> select GENERIC_SMP_IDLE_THREAD
> select GENERIC_TIME_VSYSCALL if MMU && 64BIT
> + select GENERIC_VDSO_DATA_STORE if MMU
> select GENERIC_VDSO_TIME_NS if HAVE_GENERIC_VDSO
> select HARDIRQS_SW_RESEND
> select HAS_IOPORT if MMU
> diff --git a/arch/riscv/include/asm/vdso.h
> b/arch/riscv/include/asm/vdso.h
> index
> f891478829a52c41e06240f67611694cc28197d9..c130d8100232cbe50e52e35eb418
> e354bd114cb7 100644
> --- a/arch/riscv/include/asm/vdso.h
> +++ b/arch/riscv/include/asm/vdso.h
> @@ -14,7 +14,7 @@
> */
> #ifdef CONFIG_MMU
>
> -#define __VVAR_PAGES 2
> +#define __VDSO_PAGES 4
>
> #ifndef __ASSEMBLY__
> #include <generated/vdso-offsets.h>
> diff --git a/arch/riscv/include/asm/vdso/time_data.h
> b/arch/riscv/include/asm/vdso/arch_data.h
> similarity index 71%
> rename from arch/riscv/include/asm/vdso/time_data.h
> rename to arch/riscv/include/asm/vdso/arch_data.h
> index
> dfa65228999bed41dfd6c5e36cb678e1e055eec8..da57a3786f7a53c866fc00948826
> b4a2d839940f 100644
> --- a/arch/riscv/include/asm/vdso/time_data.h
> +++ b/arch/riscv/include/asm/vdso/arch_data.h
> @@ -1,12 +1,12 @@
> /* SPDX-License-Identifier: GPL-2.0 */
> -#ifndef __RISCV_ASM_VDSO_TIME_DATA_H
> -#define __RISCV_ASM_VDSO_TIME_DATA_H
> +#ifndef __RISCV_ASM_VDSO_ARCH_DATA_H
> +#define __RISCV_ASM_VDSO_ARCH_DATA_H
>
> #include <linux/types.h>
> #include <vdso/datapage.h>
> #include <asm/hwprobe.h>
>
> -struct arch_vdso_time_data {
> +struct vdso_arch_data {
> /* Stash static answers to the hwprobe queries when all CPUs
> are selected. */
> __u64 all_cpu_hwprobe_values[RISCV_HWPROBE_MAX_KEY + 1];
>
> @@ -14,4 +14,4 @@ struct arch_vdso_time_data {
> __u8 homogeneous_cpus;
> };
>
> -#endif /* __RISCV_ASM_VDSO_TIME_DATA_H */
> +#endif /* __RISCV_ASM_VDSO_ARCH_DATA_H */
> diff --git a/arch/riscv/include/asm/vdso/gettimeofday.h
> b/arch/riscv/include/asm/vdso/gettimeofday.h
> index
> ba3283cf7accaa93a38512d2c17eda0eefde0612..29164f84f93cec6e28251e6a0adf
> bc341ac88241 100644
> --- a/arch/riscv/include/asm/vdso/gettimeofday.h
> +++ b/arch/riscv/include/asm/vdso/gettimeofday.h
> @@ -69,7 +69,7 @@ int clock_getres_fallback(clockid_t _clkid, struct
> __kernel_timespec *_ts)
> #endif /* CONFIG_GENERIC_TIME_VSYSCALL */
>
> static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
> - const struct
> vdso_data *vd)
> + const struct
> vdso_time_data *vd)
> {
> /*
> * The purpose of csr_read(CSR_TIME) is to trap the system
> into
> @@ -79,18 +79,6 @@ static __always_inline u64
> __arch_get_hw_counter(s32 clock_mode,
> return csr_read(CSR_TIME);
> }
>
> -static __always_inline const struct vdso_data
> *__arch_get_vdso_data(void)
> -{
> - return _vdso_data;
> -}
> -
> -#ifdef CONFIG_TIME_NS
> -static __always_inline
> -const struct vdso_data *__arch_get_timens_vdso_data(const struct
> vdso_data *vd)
> -{
> - return _timens_data;
> -}
> -#endif
> #endif /* !__ASSEMBLY__ */
>
> #endif /* __ASM_VDSO_GETTIMEOFDAY_H */
> diff --git a/arch/riscv/include/asm/vdso/vsyscall.h
> b/arch/riscv/include/asm/vdso/vsyscall.h
> index
> e8a9c4b53c0c9f4744196eed800b21f3918d1040..1140b54b4bc8278d7a322036cd9f
> 84f71258f246 100644
> --- a/arch/riscv/include/asm/vdso/vsyscall.h
> +++ b/arch/riscv/include/asm/vdso/vsyscall.h
> @@ -6,15 +6,6 @@
>
> #include <vdso/datapage.h>
>
> -extern struct vdso_data *vdso_data;
> -
> -static __always_inline struct vdso_data
> *__riscv_get_k_vdso_data(void)
> -{
> - return vdso_data;
> -}
> -
> -#define __arch_get_k_vdso_data __riscv_get_k_vdso_data
> -
> /* The asm-generic header needs to be included after the definitions
> above */
> #include <asm-generic/vdso/vsyscall.h>
>
> diff --git a/arch/riscv/kernel/sys_hwprobe.c
> b/arch/riscv/kernel/sys_hwprobe.c
> index
> bcd3b816306c22df62f60ad044f4ae58f7dad4d1..04a4e549551284bb3340673eb76a
> 2e7bd457025e 100644
> --- a/arch/riscv/kernel/sys_hwprobe.c
> +++ b/arch/riscv/kernel/sys_hwprobe.c
> @@ -450,8 +450,7 @@ static int do_riscv_hwprobe(struct riscv_hwprobe
> __user *pairs,
>
> static int __init init_hwprobe_vdso_data(void)
> {
> - struct vdso_data *vd = __arch_get_k_vdso_data();
> - struct arch_vdso_time_data *avd = &vd->arch_data;
> + struct vdso_arch_data *avd = vdso_k_arch_data;
> u64 id_bitsmash = 0;
> struct riscv_hwprobe pair;
> int key;
> diff --git a/arch/riscv/kernel/vdso.c b/arch/riscv/kernel/vdso.c
> index
> 3ca3ae4277e187e790a8bf513a9e80d8b6290bb2..cc2895d1fbc2fe752b3edc94f4e2
> 8a6a8fca7a3b 100644
> --- a/arch/riscv/kernel/vdso.c
> +++ b/arch/riscv/kernel/vdso.c
> @@ -13,20 +13,11 @@
> #include <linux/err.h>
> #include <asm/page.h>
> #include <asm/vdso.h>
> -#include <linux/time_namespace.h>
> +#include <linux/vdso_datastore.h>
> #include <vdso/datapage.h>
> #include <vdso/vsyscall.h>
>
> -enum vvar_pages {
> - VVAR_DATA_PAGE_OFFSET,
> - VVAR_TIMENS_PAGE_OFFSET,
> - VVAR_NR_PAGES,
> -};
> -
> -#define VVAR_SIZE (VVAR_NR_PAGES << PAGE_SHIFT)
> -
> -static union vdso_data_store vdso_data_store __page_aligned_data;
> -struct vdso_data *vdso_data = vdso_data_store.data;
> +#define VVAR_SIZE (VDSO_NR_PAGES << PAGE_SHIFT)
>
> struct __vdso_info {
> const char *name;
> @@ -79,78 +70,6 @@ static void __init __vdso_init(struct __vdso_info
> *vdso_info)
> vdso_info->cm->pages = vdso_pagelist;
> }
>
> -#ifdef CONFIG_TIME_NS
> -struct vdso_data *arch_get_vdso_data(void *vvar_page)
> -{
> - return (struct vdso_data *)(vvar_page);
> -}
> -
> -static const struct vm_special_mapping rv_vvar_map;
> -
> -/*
> - * The vvar mapping contains data for a specific time namespace, so
> when a task
> - * changes namespace we must unmap its vvar data for the old
> namespace.
> - * Subsequent faults will map in data for the new namespace.
> - *
> - * For more details see timens_setup_vdso_data().
> - */
> -int vdso_join_timens(struct task_struct *task, struct time_namespace
> *ns)
> -{
> - struct mm_struct *mm = task->mm;
> - struct vm_area_struct *vma;
> - VMA_ITERATOR(vmi, mm, 0);
> -
> - mmap_read_lock(mm);
> -
> - for_each_vma(vmi, vma) {
> - if (vma_is_special_mapping(vma, &rv_vvar_map))
> - zap_vma_pages(vma);
> - }
> -
> - mmap_read_unlock(mm);
> - return 0;
> -}
> -#endif
> -
> -static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
> - struct vm_area_struct *vma, struct
> vm_fault *vmf)
> -{
> - struct page *timens_page = find_timens_vvar_page(vma);
> - unsigned long pfn;
> -
> - switch (vmf->pgoff) {
> - case VVAR_DATA_PAGE_OFFSET:
> - if (timens_page)
> - pfn = page_to_pfn(timens_page);
> - else
> - pfn = sym_to_pfn(vdso_data);
> - break;
> -#ifdef CONFIG_TIME_NS
> - case VVAR_TIMENS_PAGE_OFFSET:
> - /*
> - * If a task belongs to a time namespace then a
> namespace
> - * specific VVAR is mapped with the
> VVAR_DATA_PAGE_OFFSET and
> - * the real VVAR page is mapped with the
> VVAR_TIMENS_PAGE_OFFSET
> - * offset.
> - * See also the comment near
> timens_setup_vdso_data().
> - */
> - if (!timens_page)
> - return VM_FAULT_SIGBUS;
> - pfn = sym_to_pfn(vdso_data);
> - break;
> -#endif /* CONFIG_TIME_NS */
> - default:
> - return VM_FAULT_SIGBUS;
> - }
> -
> - return vmf_insert_pfn(vma, vmf->address, pfn);
> -}
> -
> -static const struct vm_special_mapping rv_vvar_map = {
> - .name = "[vvar]",
> - .fault = vvar_fault,
> -};
> -
> static struct vm_special_mapping rv_vdso_map __ro_after_init = {
> .name = "[vdso]",
> .mremap = vdso_mremap,
> @@ -196,7 +115,7 @@ static int __setup_additional_pages(struct
> mm_struct *mm,
> unsigned long vdso_base, vdso_text_len, vdso_mapping_len;
> void *ret;
>
> - BUILD_BUG_ON(VVAR_NR_PAGES != __VVAR_PAGES);
> + BUILD_BUG_ON(VDSO_NR_PAGES != __VDSO_PAGES);
>
> vdso_text_len = vdso_info->vdso_pages << PAGE_SHIFT;
> /* Be sure to map the data page */
> @@ -208,8 +127,7 @@ static int __setup_additional_pages(struct
> mm_struct *mm,
> goto up_fail;
> }
>
> - ret = _install_special_mapping(mm, vdso_base, VVAR_SIZE,
> - (VM_READ | VM_MAYREAD | VM_PFNMAP), &rv_vvar_map);
> + ret = vdso_install_vvar_mapping(mm, vdso_base);
> if (IS_ERR(ret))
> goto up_fail;
>
> diff --git a/arch/riscv/kernel/vdso/hwprobe.c
> b/arch/riscv/kernel/vdso/hwprobe.c
> index
> a158c029344f60c022e7565757ff44df7e3d89e5..2ddeba6c68dda09b0249117fd06a
> 5d249f3b0abd 100644
> --- a/arch/riscv/kernel/vdso/hwprobe.c
> +++ b/arch/riscv/kernel/vdso/hwprobe.c
> @@ -16,8 +16,7 @@ static int riscv_vdso_get_values(struct
> riscv_hwprobe *pairs, size_t pair_count,
> size_t cpusetsize, unsigned long
> *cpus,
> unsigned int flags)
> {
> - const struct vdso_data *vd = __arch_get_vdso_data();
> - const struct arch_vdso_time_data *avd = &vd->arch_data;
> + const struct vdso_arch_data *avd = &vdso_u_arch_data;
> bool all_cpus = !cpusetsize && !cpus;
> struct riscv_hwprobe *p = pairs;
> struct riscv_hwprobe *end = pairs + pair_count;
> @@ -51,8 +50,7 @@ static int riscv_vdso_get_cpus(struct riscv_hwprobe
> *pairs, size_t pair_count,
> size_t cpusetsize, unsigned long
> *cpus,
> unsigned int flags)
> {
> - const struct vdso_data *vd = __arch_get_vdso_data();
> - const struct arch_vdso_time_data *avd = &vd->arch_data;
> + const struct vdso_arch_data *avd = &vdso_u_arch_data;
> struct riscv_hwprobe *p = pairs;
> struct riscv_hwprobe *end = pairs + pair_count;
> unsigned char *c = (unsigned char *)cpus;
> diff --git a/arch/riscv/kernel/vdso/vdso.lds.S
> b/arch/riscv/kernel/vdso/vdso.lds.S
> index
> cbe2a179331d2511a8b4a26c06383e46131661b1..8e86965a8aae4d7c5a36d0f26026
> cd1c8680b339 100644
> --- a/arch/riscv/kernel/vdso/vdso.lds.S
> +++ b/arch/riscv/kernel/vdso/vdso.lds.S
> @@ -4,15 +4,14 @@
> */
> #include <asm/page.h>
> #include <asm/vdso.h>
> +#include <vdso/datapage.h>
>
> OUTPUT_ARCH(riscv)
>
> SECTIONS
> {
> - PROVIDE(_vdso_data = . - __VVAR_PAGES * PAGE_SIZE);
> -#ifdef CONFIG_TIME_NS
> - PROVIDE(_timens_data = _vdso_data + PAGE_SIZE);
> -#endif
> + VDSO_VVAR_SYMS
> +
> . = SIZEOF_HEADERS;
>
> .hash : { *(.hash) } :text
>
--
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University
^ permalink raw reply [flat|nested] 29+ messages in thread* Re: [PATCH v3 09/18] riscv: vdso: Switch to generic storage implementation
2025-02-22 8:17 ` Xi Ruoyao
@ 2025-02-22 10:14 ` Thomas Gleixner
0 siblings, 0 replies; 29+ messages in thread
From: Thomas Gleixner @ 2025-02-22 10:14 UTC (permalink / raw)
To: Xi Ruoyao, Thomas Weißschuh, James E.J. Bottomley,
Helge Deller, Andy Lutomirski, Vincenzo Frascino,
Anna-Maria Behnsen, Frederic Weisbecker, Andrew Morton,
Catalin Marinas, Will Deacon, Theodore Ts'o,
Jason A. Donenfeld, Paul Walmsley, Palmer Dabbelt, Albert Ou,
Huacai Chen, WANG Xuerui, Russell King, Heiko Carstens,
Vasily Gorbik, Alexander Gordeev, Christian Borntraeger,
Sven Schnelle, Thomas Bogendoerfer, Michael Ellerman,
Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, linux-csky
On Sat, Feb 22 2025 at 16:17, Xi Ruoyao wrote:
> On Tue, 2025-02-04 at 13:05 +0100, Thomas Weißschuh wrote:
>> The generic storage implementation provides the same features as the
>> custom one. However it can be shared between architectures, making
>> maintenance easier.
>>
>> Co-developed-by: Nam Cao <namcao@linutronix.de>
>> Signed-off-by: Nam Cao <namcao@linutronix.de>
>> Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
>
> I made a RISC-V vDSO getrandom implementation on top of this and it
> works fine. I'll submit it when this is merged.
You can submit it right now. If it's reviewed and if Jason agrees, this
can be merged through the timers/vdso branch on top of the pending
changes.
Thanks for testing it!
tglx
^ permalink raw reply [flat|nested] 29+ messages in thread
* [PATCH v3 10/18] LoongArch: vDSO: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (8 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 09/18] riscv: " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-22 8:20 ` Xi Ruoyao
2025-02-04 12:05 ` [PATCH v3 11/18] arm: vdso: " Thomas Weißschuh
` (8 subsequent siblings)
18 siblings, 1 reply; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/loongarch/Kconfig | 2 +
arch/loongarch/include/asm/vdso.h | 1 -
arch/loongarch/include/asm/vdso/arch_data.h | 25 +++++++
arch/loongarch/include/asm/vdso/getrandom.h | 5 --
arch/loongarch/include/asm/vdso/gettimeofday.h | 14 +---
arch/loongarch/include/asm/vdso/vdso.h | 38 +----------
arch/loongarch/include/asm/vdso/vsyscall.h | 17 -----
arch/loongarch/kernel/asm-offsets.c | 2 +-
arch/loongarch/kernel/vdso.c | 92 +-------------------------
arch/loongarch/vdso/vdso.lds.S | 8 +--
arch/loongarch/vdso/vgetcpu.c | 12 +---
11 files changed, 39 insertions(+), 177 deletions(-)
diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
index 2b8bd27a852fee1f4418c7b28133a85060bea26a..d7ddf2a43e63b290c23ca64c15e33e2af00fc6ae 100644
--- a/arch/loongarch/Kconfig
+++ b/arch/loongarch/Kconfig
@@ -30,6 +30,7 @@ config LOONGARCH
select ARCH_HAS_SET_MEMORY
select ARCH_HAS_SET_DIRECT_MAP
select ARCH_HAS_TICK_BROADCAST if GENERIC_CLOCKEVENTS_BROADCAST
+ select ARCH_HAS_VDSO_ARCH_DATA
select ARCH_INLINE_READ_LOCK if !PREEMPTION
select ARCH_INLINE_READ_LOCK_BH if !PREEMPTION
select ARCH_INLINE_READ_LOCK_IRQ if !PREEMPTION
@@ -106,6 +107,7 @@ config LOONGARCH
select GENERIC_SCHED_CLOCK
select GENERIC_SMP_IDLE_THREAD
select GENERIC_TIME_VSYSCALL
+ select GENERIC_VDSO_DATA_STORE
select GENERIC_VDSO_TIME_NS
select GPIOLIB
select HAS_IOPORT
diff --git a/arch/loongarch/include/asm/vdso.h b/arch/loongarch/include/asm/vdso.h
index d3ba35eb23e77082ea8ed564fe7378ad80df81b2..f72ec79e2dde52f8888750a53c4af85d6ab77b45 100644
--- a/arch/loongarch/include/asm/vdso.h
+++ b/arch/loongarch/include/asm/vdso.h
@@ -31,7 +31,6 @@ struct loongarch_vdso_info {
unsigned long size;
unsigned long offset_sigreturn;
struct vm_special_mapping code_mapping;
- struct vm_special_mapping data_mapping;
};
extern struct loongarch_vdso_info vdso_info;
diff --git a/arch/loongarch/include/asm/vdso/arch_data.h b/arch/loongarch/include/asm/vdso/arch_data.h
new file mode 100644
index 0000000000000000000000000000000000000000..322d0a5f1c844f02de9b074d6021e8e15116f8a2
--- /dev/null
+++ b/arch/loongarch/include/asm/vdso/arch_data.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Author: Huacai Chen <chenhuacai@loongson.cn>
+ * Copyright (C) 2020-2022 Loongson Technology Corporation Limited
+ */
+
+#ifndef _VDSO_ARCH_DATA_H
+#define _VDSO_ARCH_DATA_H
+
+#ifndef __ASSEMBLY__
+
+#include <asm/asm.h>
+#include <asm/vdso.h>
+
+struct vdso_pcpu_data {
+ u32 node;
+} ____cacheline_aligned_in_smp;
+
+struct vdso_arch_data {
+ struct vdso_pcpu_data pdata[NR_CPUS];
+};
+
+#endif /* __ASSEMBLY__ */
+
+#endif
diff --git a/arch/loongarch/include/asm/vdso/getrandom.h b/arch/loongarch/include/asm/vdso/getrandom.h
index e80f3c4ac7481ba7f9f5d9210fefa78c3293243b..48c43f55b039b42168698614d0479b7a872d20f3 100644
--- a/arch/loongarch/include/asm/vdso/getrandom.h
+++ b/arch/loongarch/include/asm/vdso/getrandom.h
@@ -28,11 +28,6 @@ static __always_inline ssize_t getrandom_syscall(void *_buffer, size_t _len, uns
return ret;
}
-static __always_inline const struct vdso_rng_data *__arch_get_vdso_rng_data(void)
-{
- return &_loongarch_data.rng_data;
-}
-
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETRANDOM_H */
diff --git a/arch/loongarch/include/asm/vdso/gettimeofday.h b/arch/loongarch/include/asm/vdso/gettimeofday.h
index 7eb3f041af764d141b005f821593a358096874ba..88cfcf13311630ed5f1a734d23a2bc3f65d79a88 100644
--- a/arch/loongarch/include/asm/vdso/gettimeofday.h
+++ b/arch/loongarch/include/asm/vdso/gettimeofday.h
@@ -72,7 +72,7 @@ static __always_inline int clock_getres_fallback(
}
static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
uint64_t count;
@@ -89,18 +89,6 @@ static inline bool loongarch_vdso_hres_capable(void)
}
#define __arch_vdso_hres_capable loongarch_vdso_hres_capable
-static __always_inline const struct vdso_data *__arch_get_vdso_data(void)
-{
- return _vdso_data;
-}
-
-#ifdef CONFIG_TIME_NS
-static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
-{
- return _timens_data;
-}
-#endif
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETTIMEOFDAY_H */
diff --git a/arch/loongarch/include/asm/vdso/vdso.h b/arch/loongarch/include/asm/vdso/vdso.h
index 1c183a9b2115a29a997ec8db0e788d87fb191dce..50c65fb29dafb464f36615f49db8954ee0a775e3 100644
--- a/arch/loongarch/include/asm/vdso/vdso.h
+++ b/arch/loongarch/include/asm/vdso/vdso.h
@@ -12,43 +12,9 @@
#include <asm/asm.h>
#include <asm/page.h>
#include <asm/vdso.h>
+#include <vdso/datapage.h>
-struct vdso_pcpu_data {
- u32 node;
-} ____cacheline_aligned_in_smp;
-
-struct loongarch_vdso_data {
- struct vdso_pcpu_data pdata[NR_CPUS];
- struct vdso_rng_data rng_data;
-};
-
-/*
- * The layout of vvar:
- *
- * high
- * +---------------------+--------------------------+
- * | loongarch vdso data | LOONGARCH_VDSO_DATA_SIZE |
- * +---------------------+--------------------------+
- * | time-ns vdso data | PAGE_SIZE |
- * +---------------------+--------------------------+
- * | generic vdso data | PAGE_SIZE |
- * +---------------------+--------------------------+
- * low
- */
-#define LOONGARCH_VDSO_DATA_SIZE PAGE_ALIGN(sizeof(struct loongarch_vdso_data))
-#define LOONGARCH_VDSO_DATA_PAGES (LOONGARCH_VDSO_DATA_SIZE >> PAGE_SHIFT)
-
-enum vvar_pages {
- VVAR_GENERIC_PAGE_OFFSET,
- VVAR_TIMENS_PAGE_OFFSET,
- VVAR_LOONGARCH_PAGES_START,
- VVAR_LOONGARCH_PAGES_END = VVAR_LOONGARCH_PAGES_START + LOONGARCH_VDSO_DATA_PAGES - 1,
- VVAR_NR_PAGES,
-};
-
-#define VVAR_SIZE (VVAR_NR_PAGES << PAGE_SHIFT)
-
-extern struct loongarch_vdso_data _loongarch_data __attribute__((visibility("hidden")));
+#define VVAR_SIZE (VDSO_NR_PAGES << PAGE_SHIFT)
#endif /* __ASSEMBLY__ */
diff --git a/arch/loongarch/include/asm/vdso/vsyscall.h b/arch/loongarch/include/asm/vdso/vsyscall.h
index 8987e951d0a93c34ca75de676fb9c191ff4ef3c2..1140b54b4bc8278d7a322036cd9f84f71258f246 100644
--- a/arch/loongarch/include/asm/vdso/vsyscall.h
+++ b/arch/loongarch/include/asm/vdso/vsyscall.h
@@ -6,23 +6,6 @@
#include <vdso/datapage.h>
-extern struct vdso_data *vdso_data;
-extern struct vdso_rng_data *vdso_rng_data;
-
-static __always_inline
-struct vdso_data *__loongarch_get_k_vdso_data(void)
-{
- return vdso_data;
-}
-#define __arch_get_k_vdso_data __loongarch_get_k_vdso_data
-
-static __always_inline
-struct vdso_rng_data *__loongarch_get_k_vdso_rng_data(void)
-{
- return vdso_rng_data;
-}
-#define __arch_get_k_vdso_rng_data __loongarch_get_k_vdso_rng_data
-
/* The asm-generic header needs to be included after the definitions above */
#include <asm-generic/vdso/vsyscall.h>
diff --git a/arch/loongarch/kernel/asm-offsets.c b/arch/loongarch/kernel/asm-offsets.c
index 8be1c38ad8eb26cc3d901c4a0c0b71bde715057a..6f1524bba957e0ef4760a1bd4d32a2d11bb3f11a 100644
--- a/arch/loongarch/kernel/asm-offsets.c
+++ b/arch/loongarch/kernel/asm-offsets.c
@@ -315,6 +315,6 @@ static void __used output_vdso_defines(void)
{
COMMENT("LoongArch vDSO offsets.");
- DEFINE(__VVAR_PAGES, VVAR_NR_PAGES);
+ DEFINE(__VDSO_PAGES, VDSO_NR_PAGES);
BLANK();
}
diff --git a/arch/loongarch/kernel/vdso.c b/arch/loongarch/kernel/vdso.c
index 05e5fbac102a902016e633db75d9aff7ed550c50..10cf1608c7b32f8e03bc2c942b51413cf8b9c505 100644
--- a/arch/loongarch/kernel/vdso.c
+++ b/arch/loongarch/kernel/vdso.c
@@ -14,7 +14,7 @@
#include <linux/random.h>
#include <linux/sched.h>
#include <linux/slab.h>
-#include <linux/time_namespace.h>
+#include <linux/vdso_datastore.h>
#include <asm/page.h>
#include <asm/vdso.h>
@@ -25,18 +25,6 @@
extern char vdso_start[], vdso_end[];
-/* Kernel-provided data used by the VDSO. */
-static union vdso_data_store generic_vdso_data __page_aligned_data;
-
-static union {
- u8 page[LOONGARCH_VDSO_DATA_SIZE];
- struct loongarch_vdso_data vdata;
-} loongarch_vdso_data __page_aligned_data;
-
-struct vdso_data *vdso_data = generic_vdso_data.data;
-struct vdso_pcpu_data *vdso_pdata = loongarch_vdso_data.vdata.pdata;
-struct vdso_rng_data *vdso_rng_data = &loongarch_vdso_data.vdata.rng_data;
-
static int vdso_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma)
{
current->mm->context.vdso = (void *)(new_vma->vm_start);
@@ -44,53 +32,12 @@ static int vdso_mremap(const struct vm_special_mapping *sm, struct vm_area_struc
return 0;
}
-static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
- struct vm_area_struct *vma, struct vm_fault *vmf)
-{
- unsigned long pfn;
- struct page *timens_page = find_timens_vvar_page(vma);
-
- switch (vmf->pgoff) {
- case VVAR_GENERIC_PAGE_OFFSET:
- if (!timens_page)
- pfn = sym_to_pfn(vdso_data);
- else
- pfn = page_to_pfn(timens_page);
- break;
-#ifdef CONFIG_TIME_NS
- case VVAR_TIMENS_PAGE_OFFSET:
- /*
- * If a task belongs to a time namespace then a namespace specific
- * VVAR is mapped with the VVAR_GENERIC_PAGE_OFFSET and the real
- * VVAR page is mapped with the VVAR_TIMENS_PAGE_OFFSET offset.
- * See also the comment near timens_setup_vdso_data().
- */
- if (!timens_page)
- return VM_FAULT_SIGBUS;
- else
- pfn = sym_to_pfn(vdso_data);
- break;
-#endif /* CONFIG_TIME_NS */
- case VVAR_LOONGARCH_PAGES_START ... VVAR_LOONGARCH_PAGES_END:
- pfn = sym_to_pfn(&loongarch_vdso_data) + vmf->pgoff - VVAR_LOONGARCH_PAGES_START;
- break;
- default:
- return VM_FAULT_SIGBUS;
- }
-
- return vmf_insert_pfn(vma, vmf->address, pfn);
-}
-
struct loongarch_vdso_info vdso_info = {
.vdso = vdso_start,
.code_mapping = {
.name = "[vdso]",
.mremap = vdso_mremap,
},
- .data_mapping = {
- .name = "[vvar]",
- .fault = vvar_fault,
- },
.offset_sigreturn = vdso_offset_sigreturn,
};
@@ -101,7 +48,7 @@ static int __init init_vdso(void)
BUG_ON(!PAGE_ALIGNED(vdso_info.vdso));
for_each_possible_cpu(cpu)
- vdso_pdata[cpu].node = cpu_to_node(cpu);
+ vdso_k_arch_data->pdata[cpu].node = cpu_to_node(cpu);
vdso_info.size = PAGE_ALIGN(vdso_end - vdso_start);
vdso_info.code_mapping.pages =
@@ -115,37 +62,6 @@ static int __init init_vdso(void)
}
subsys_initcall(init_vdso);
-#ifdef CONFIG_TIME_NS
-struct vdso_data *arch_get_vdso_data(void *vvar_page)
-{
- return (struct vdso_data *)(vvar_page);
-}
-
-/*
- * The vvar mapping contains data for a specific time namespace, so when a
- * task changes namespace we must unmap its vvar data for the old namespace.
- * Subsequent faults will map in data for the new namespace.
- *
- * For more details see timens_setup_vdso_data().
- */
-int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
-{
- struct mm_struct *mm = task->mm;
- struct vm_area_struct *vma;
-
- VMA_ITERATOR(vmi, mm, 0);
-
- mmap_read_lock(mm);
- for_each_vma(vmi, vma) {
- if (vma_is_special_mapping(vma, &vdso_info.data_mapping))
- zap_vma_pages(vma);
- }
- mmap_read_unlock(mm);
-
- return 0;
-}
-#endif
-
static unsigned long vdso_base(void)
{
unsigned long base = STACK_TOP;
@@ -181,9 +97,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
goto out;
}
- vma = _install_special_mapping(mm, data_addr, VVAR_SIZE,
- VM_READ | VM_MAYREAD | VM_PFNMAP,
- &info->data_mapping);
+ vma = vdso_install_vvar_mapping(mm, data_addr);
if (IS_ERR(vma)) {
ret = PTR_ERR(vma);
goto out;
diff --git a/arch/loongarch/vdso/vdso.lds.S b/arch/loongarch/vdso/vdso.lds.S
index 160cfaef2de45b1243502c7356f8a913658548fe..8ff98649994750e74b30a91270d9d9ea2ba751e8 100644
--- a/arch/loongarch/vdso/vdso.lds.S
+++ b/arch/loongarch/vdso/vdso.lds.S
@@ -5,6 +5,7 @@
*/
#include <asm/page.h>
#include <generated/asm-offsets.h>
+#include <vdso/datapage.h>
OUTPUT_FORMAT("elf64-loongarch", "elf64-loongarch", "elf64-loongarch")
@@ -12,11 +13,8 @@ OUTPUT_ARCH(loongarch)
SECTIONS
{
- PROVIDE(_vdso_data = . - __VVAR_PAGES * PAGE_SIZE);
-#ifdef CONFIG_TIME_NS
- PROVIDE(_timens_data = _vdso_data + PAGE_SIZE);
-#endif
- PROVIDE(_loongarch_data = _vdso_data + 2 * PAGE_SIZE);
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
diff --git a/arch/loongarch/vdso/vgetcpu.c b/arch/loongarch/vdso/vgetcpu.c
index 0db51258b2a7ca7e44d2eb68ea514face48393f7..5301cd9d0f839eb0fd7b73a1d36e80aaa75d5e76 100644
--- a/arch/loongarch/vdso/vgetcpu.c
+++ b/arch/loongarch/vdso/vgetcpu.c
@@ -19,27 +19,19 @@ static __always_inline int read_cpu_id(void)
return cpu_id;
}
-static __always_inline const struct vdso_pcpu_data *get_pcpu_data(void)
-{
- return _loongarch_data.pdata;
-}
-
extern
int __vdso_getcpu(unsigned int *cpu, unsigned int *node, struct getcpu_cache *unused);
int __vdso_getcpu(unsigned int *cpu, unsigned int *node, struct getcpu_cache *unused)
{
int cpu_id;
- const struct vdso_pcpu_data *data;
cpu_id = read_cpu_id();
if (cpu)
*cpu = cpu_id;
- if (node) {
- data = get_pcpu_data();
- *node = data[cpu_id].node;
- }
+ if (node)
+ *node = vdso_u_arch_data.pdata[cpu_id].node;
return 0;
}
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* Re: [PATCH v3 10/18] LoongArch: vDSO: Switch to generic storage implementation
2025-02-04 12:05 ` [PATCH v3 10/18] LoongArch: vDSO: " Thomas Weißschuh
@ 2025-02-22 8:20 ` Xi Ruoyao
0 siblings, 0 replies; 29+ messages in thread
From: Xi Ruoyao @ 2025-02-22 8:20 UTC (permalink / raw)
To: Thomas Weißschuh, James E.J. Bottomley, Helge Deller,
Andy Lutomirski, Thomas Gleixner, Vincenzo Frascino,
Anna-Maria Behnsen, Frederic Weisbecker, Andrew Morton,
Catalin Marinas, Will Deacon, Theodore Ts'o,
Jason A. Donenfeld, Paul Walmsley, Palmer Dabbelt, Albert Ou,
Huacai Chen, WANG Xuerui, Russell King, Heiko Carstens,
Vasily Gorbik, Alexander Gordeev, Christian Borntraeger,
Sven Schnelle, Thomas Bogendoerfer, Michael Ellerman,
Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, linux-csky
On Tue, 2025-02-04 at 13:05 +0100, Thomas Weißschuh wrote:
> The generic storage implementation provides the same features as the
> custom one. However it can be shared between architectures, making
> maintenance easier.
>
> Co-developed-by: Nam Cao <namcao@linutronix.de>
> Signed-off-by: Nam Cao <namcao@linutronix.de>
> Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Tested-by: Xi Ruoyao <xry111@xry111.site>
/* snip */
> diff --git a/arch/loongarch/include/asm/vdso/getrandom.h b/arch/loongarch/include/asm/vdso/getrandom.h
> index e80f3c4ac7481ba7f9f5d9210fefa78c3293243b..48c43f55b039b42168698614d0479b7a872d20f3 100644
> --- a/arch/loongarch/include/asm/vdso/getrandom.h
> +++ b/arch/loongarch/include/asm/vdso/getrandom.h
You can drop "#include <vdso/vdso.h>" in this file.
> @@ -28,11 +28,6 @@ static __always_inline ssize_t getrandom_syscall(void *_buffer, size_t _len, uns
> return ret;
> }
>
> -static __always_inline const struct vdso_rng_data *__arch_get_vdso_rng_data(void)
> -{
> - return &_loongarch_data.rng_data;
> -}
> -
> #endif /* !__ASSEMBLY__ */
>
> #endif /* __ASM_VDSO_GETRANDOM_H */
--
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University
^ permalink raw reply [flat|nested] 29+ messages in thread
* [PATCH v3 11/18] arm: vdso: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (9 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 10/18] LoongArch: vDSO: " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 12/18] s390/vdso: " Thomas Weißschuh
` (7 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/arm/include/asm/vdso.h | 2 ++
arch/arm/include/asm/vdso/gettimeofday.h | 7 +------
arch/arm/include/asm/vdso/vsyscall.h | 12 ++---------
arch/arm/kernel/asm-offsets.c | 4 ----
arch/arm/kernel/vdso.c | 34 +++++++-------------------------
arch/arm/mm/Kconfig | 1 +
arch/arm/vdso/vdso.lds.S | 4 ++--
7 files changed, 15 insertions(+), 49 deletions(-)
diff --git a/arch/arm/include/asm/vdso.h b/arch/arm/include/asm/vdso.h
index 5b85889f82eeb422400c039e8b5e8a2a6d4553b3..88364a6727ffce224edd35dd62db1693ae5fcb1c 100644
--- a/arch/arm/include/asm/vdso.h
+++ b/arch/arm/include/asm/vdso.h
@@ -4,6 +4,8 @@
#ifdef __KERNEL__
+#define __VDSO_PAGES 4
+
#ifndef __ASSEMBLY__
struct mm_struct;
diff --git a/arch/arm/include/asm/vdso/gettimeofday.h b/arch/arm/include/asm/vdso/gettimeofday.h
index 592d3d015ca7298b3bc35871387dbb3e7f819827..1e9f81639c88cc23cae7cf267bf4674c6d6acec0 100644
--- a/arch/arm/include/asm/vdso/gettimeofday.h
+++ b/arch/arm/include/asm/vdso/gettimeofday.h
@@ -112,7 +112,7 @@ static inline bool arm_vdso_hres_capable(void)
#define __arch_vdso_hres_capable arm_vdso_hres_capable
static __always_inline u64 __arch_get_hw_counter(int clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
#ifdef CONFIG_ARM_ARCH_TIMER
u64 cycle_now;
@@ -135,11 +135,6 @@ static __always_inline u64 __arch_get_hw_counter(int clock_mode,
#endif
}
-static __always_inline const struct vdso_data *__arch_get_vdso_data(void)
-{
- return _vdso_data;
-}
-
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETTIMEOFDAY_H */
diff --git a/arch/arm/include/asm/vdso/vsyscall.h b/arch/arm/include/asm/vdso/vsyscall.h
index 705414710dcdbfa9a97c344806bd079c08368801..4e7226ad02ec4dcf88203c9046e1b320a10e7373 100644
--- a/arch/arm/include/asm/vdso/vsyscall.h
+++ b/arch/arm/include/asm/vdso/vsyscall.h
@@ -7,22 +7,14 @@
#include <vdso/datapage.h>
#include <asm/cacheflush.h>
-extern struct vdso_data *vdso_data;
extern bool cntvct_ok;
static __always_inline
-struct vdso_data *__arm_get_k_vdso_data(void)
-{
- return vdso_data;
-}
-#define __arch_get_k_vdso_data __arm_get_k_vdso_data
-
-static __always_inline
-void __arm_sync_vdso_data(struct vdso_data *vdata)
+void __arch_sync_vdso_time_data(struct vdso_time_data *vdata)
{
flush_dcache_page(virt_to_page(vdata));
}
-#define __arch_sync_vdso_data __arm_sync_vdso_data
+#define __arch_sync_vdso_time_data __arch_sync_vdso_time_data
/* The asm-generic header needs to be included after the definitions above */
#include <asm-generic/vdso/vsyscall.h>
diff --git a/arch/arm/kernel/asm-offsets.c b/arch/arm/kernel/asm-offsets.c
index 4853875740d0fe61c6bbc32ddd9a16fa8d1fb530..123f4a8ef44660a39c4eff2a6e5fd86cb48fb572 100644
--- a/arch/arm/kernel/asm-offsets.c
+++ b/arch/arm/kernel/asm-offsets.c
@@ -153,10 +153,6 @@ int main(void)
DEFINE(CACHE_WRITEBACK_ORDER, __CACHE_WRITEBACK_ORDER);
DEFINE(CACHE_WRITEBACK_GRANULE, __CACHE_WRITEBACK_GRANULE);
BLANK();
-#ifdef CONFIG_VDSO
- DEFINE(VDSO_DATA_SIZE, sizeof(union vdso_data_store));
-#endif
- BLANK();
#ifdef CONFIG_ARM_MPU
DEFINE(MPU_RNG_INFO_RNGS, offsetof(struct mpu_rgn_info, rgns));
DEFINE(MPU_RNG_INFO_USED, offsetof(struct mpu_rgn_info, used));
diff --git a/arch/arm/kernel/vdso.c b/arch/arm/kernel/vdso.c
index 29dd2f3c62fec64c7f290468421bfad1e739c667..325448ffbba0c29895ea5d97e60d6f51e552cb2e 100644
--- a/arch/arm/kernel/vdso.c
+++ b/arch/arm/kernel/vdso.c
@@ -7,6 +7,7 @@
*/
#include <linux/cache.h>
+#include <linux/vdso_datastore.h>
#include <linux/elf.h>
#include <linux/err.h>
#include <linux/kernel.h>
@@ -33,15 +34,6 @@ extern char vdso_start[], vdso_end[];
/* Total number of pages needed for the data and text portions of the VDSO. */
unsigned int vdso_total_pages __ro_after_init;
-static union vdso_data_store vdso_data_store __page_aligned_data;
-struct vdso_data *vdso_data = vdso_data_store.data;
-
-static struct page *vdso_data_page __ro_after_init;
-static const struct vm_special_mapping vdso_data_mapping = {
- .name = "[vvar]",
- .pages = &vdso_data_page,
-};
-
static int vdso_mremap(const struct vm_special_mapping *sm,
struct vm_area_struct *new_vma)
{
@@ -192,9 +184,6 @@ static int __init vdso_init(void)
if (vdso_text_pagelist == NULL)
return -ENOMEM;
- /* Grab the VDSO data page. */
- vdso_data_page = virt_to_page(vdso_data);
-
/* Grab the VDSO text pages. */
for (i = 0; i < text_pages; i++) {
struct page *page;
@@ -205,7 +194,7 @@ static int __init vdso_init(void)
vdso_text_mapping.pages = vdso_text_pagelist;
- vdso_total_pages = 1; /* for the data/vvar page */
+ vdso_total_pages = VDSO_NR_PAGES; /* for the data/vvar pages */
vdso_total_pages += text_pages;
cntvct_ok = cntvct_functional();
@@ -216,16 +205,7 @@ static int __init vdso_init(void)
}
arch_initcall(vdso_init);
-static int install_vvar(struct mm_struct *mm, unsigned long addr)
-{
- struct vm_area_struct *vma;
-
- vma = _install_special_mapping(mm, addr, PAGE_SIZE,
- VM_READ | VM_MAYREAD,
- &vdso_data_mapping);
-
- return PTR_ERR_OR_ZERO(vma);
-}
+static_assert(__VDSO_PAGES == VDSO_NR_PAGES);
/* assumes mmap_lock is write-locked */
void arm_install_vdso(struct mm_struct *mm, unsigned long addr)
@@ -238,12 +218,12 @@ void arm_install_vdso(struct mm_struct *mm, unsigned long addr)
if (vdso_text_pagelist == NULL)
return;
- if (install_vvar(mm, addr))
+ if (IS_ERR(vdso_install_vvar_mapping(mm, addr)))
return;
- /* Account for vvar page. */
- addr += PAGE_SIZE;
- len = (vdso_total_pages - 1) << PAGE_SHIFT;
+ /* Account for vvar pages. */
+ addr += VDSO_NR_PAGES * PAGE_SIZE;
+ len = (vdso_total_pages - VDSO_NR_PAGES) << PAGE_SHIFT;
vma = _install_special_mapping(mm, addr, len,
VM_READ | VM_EXEC | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC,
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index 2b6f50dd547840adecbe08e684ed8f1a032cd7c2..5c1023a6d78c1b4db67b2d62b71af5a79b7e701f 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -928,6 +928,7 @@ config VDSO
select GENERIC_TIME_VSYSCALL
select GENERIC_VDSO_32
select GENERIC_GETTIMEOFDAY
+ select GENERIC_VDSO_DATA_STORE
help
Place in the process address space an ELF shared object
providing fast implementations of gettimeofday and
diff --git a/arch/arm/vdso/vdso.lds.S b/arch/arm/vdso/vdso.lds.S
index 9bfa0f52923c851b6b9e96d4dbb51eb3d5d0960a..7c08371f440027f9a52e9fffae530d7db05b7c43 100644
--- a/arch/arm/vdso/vdso.lds.S
+++ b/arch/arm/vdso/vdso.lds.S
@@ -11,16 +11,16 @@
*/
#include <linux/const.h>
-#include <asm/asm-offsets.h>
#include <asm/page.h>
#include <asm/vdso.h>
+#include <vdso/datapage.h>
OUTPUT_FORMAT("elf32-littlearm", "elf32-bigarm", "elf32-littlearm")
OUTPUT_ARCH(arm)
SECTIONS
{
- PROVIDE(_vdso_data = . - VDSO_DATA_SIZE);
+ VDSO_VVAR_SYMS
. = SIZEOF_HEADERS;
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 12/18] s390/vdso: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (10 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 11/18] arm: vdso: " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 13/18] MIPS: vdso: " Thomas Weißschuh
` (6 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Acked-by: Heiko Carstens <hca@linux.ibm.com>
---
arch/s390/Kconfig | 1 +
arch/s390/include/asm/vdso.h | 4 +-
arch/s390/include/asm/vdso/getrandom.h | 12 ----
arch/s390/include/asm/vdso/gettimeofday.h | 15 +----
arch/s390/include/asm/vdso/vsyscall.h | 20 -------
arch/s390/kernel/time.c | 6 +-
arch/s390/kernel/vdso.c | 97 ++-----------------------------
arch/s390/kernel/vdso32/vdso32.lds.S | 7 +--
arch/s390/kernel/vdso64/vdso64.lds.S | 8 +--
9 files changed, 17 insertions(+), 153 deletions(-)
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 9c9ec08d78c71b4d227beeafab1b82d6434cb5c7..7ed6f229250cf6c26efd4ce30e23123cdb030b2c 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -166,6 +166,7 @@ config S390
select GENERIC_PTDUMP
select GENERIC_SMP_IDLE_THREAD
select GENERIC_TIME_VSYSCALL
+ select GENERIC_VDSO_DATA_STORE
select GENERIC_VDSO_TIME_NS
select GENERIC_IOREMAP if PCI
select HAVE_ALIGNED_STRUCT_PAGE
diff --git a/arch/s390/include/asm/vdso.h b/arch/s390/include/asm/vdso.h
index 92c73e4d97a93a90cb640d44fca9856f93e0be14..420a073fdde5db92cfe84bb1f17a549f333f94e0 100644
--- a/arch/s390/include/asm/vdso.h
+++ b/arch/s390/include/asm/vdso.h
@@ -6,13 +6,11 @@
#ifndef __ASSEMBLY__
-extern struct vdso_data *vdso_data;
-
int vdso_getcpu_init(void);
#endif /* __ASSEMBLY__ */
-#define __VVAR_PAGES 2
+#define __VDSO_PAGES 4
#define VDSO_VERSION_STRING LINUX_2.6.29
diff --git a/arch/s390/include/asm/vdso/getrandom.h b/arch/s390/include/asm/vdso/getrandom.h
index 36355af7160bee4f04f5177cfa21283699bee006..f8713ce39bb2f2081c3bb5a42eddccc3cddc40c8 100644
--- a/arch/s390/include/asm/vdso/getrandom.h
+++ b/arch/s390/include/asm/vdso/getrandom.h
@@ -23,18 +23,6 @@ static __always_inline ssize_t getrandom_syscall(void *buffer, size_t len, unsig
return syscall3(__NR_getrandom, (long)buffer, (long)len, (long)flags);
}
-static __always_inline const struct vdso_rng_data *__arch_get_vdso_rng_data(void)
-{
- /*
- * The RNG data is in the real VVAR data page, but if a task belongs to a time namespace
- * then VVAR_DATA_PAGE_OFFSET points to the namespace-specific VVAR page and VVAR_TIMENS_
- * PAGE_OFFSET points to the real VVAR page.
- */
- if (IS_ENABLED(CONFIG_TIME_NS) && _vdso_data->clock_mode == VDSO_CLOCKMODE_TIMENS)
- return (void *)&_vdso_rng_data + VVAR_TIMENS_PAGE_OFFSET * PAGE_SIZE;
- return &_vdso_rng_data;
-}
-
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETRANDOM_H */
diff --git a/arch/s390/include/asm/vdso/gettimeofday.h b/arch/s390/include/asm/vdso/gettimeofday.h
index 7937765ccfa5b222931591cc5388827e5f75d1b5..fb4564308e9d77a40f8d5059503519e9f5961f14 100644
--- a/arch/s390/include/asm/vdso/gettimeofday.h
+++ b/arch/s390/include/asm/vdso/gettimeofday.h
@@ -14,12 +14,7 @@
#include <linux/compiler.h>
-static __always_inline const struct vdso_data *__arch_get_vdso_data(void)
-{
- return _vdso_data;
-}
-
-static inline u64 __arch_get_hw_counter(s32 clock_mode, const struct vdso_data *vd)
+static inline u64 __arch_get_hw_counter(s32 clock_mode, const struct vdso_time_data *vd)
{
u64 adj, now;
@@ -49,12 +44,4 @@ long clock_getres_fallback(clockid_t clkid, struct __kernel_timespec *ts)
return syscall2(__NR_clock_getres, (long)clkid, (long)ts);
}
-#ifdef CONFIG_TIME_NS
-static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
-{
- return _timens_data;
-}
-#endif
-
#endif
diff --git a/arch/s390/include/asm/vdso/vsyscall.h b/arch/s390/include/asm/vdso/vsyscall.h
index 3eb576ecd3bd998daf9372a52befa8b76d52f2bf..d346ebe513015e767405452cd69495ae497c91ee 100644
--- a/arch/s390/include/asm/vdso/vsyscall.h
+++ b/arch/s390/include/asm/vdso/vsyscall.h
@@ -2,32 +2,12 @@
#ifndef __ASM_VDSO_VSYSCALL_H
#define __ASM_VDSO_VSYSCALL_H
-#define __VDSO_RND_DATA_OFFSET 768
-
#ifndef __ASSEMBLY__
#include <linux/hrtimer.h>
#include <vdso/datapage.h>
#include <asm/vdso.h>
-enum vvar_pages {
- VVAR_DATA_PAGE_OFFSET,
- VVAR_TIMENS_PAGE_OFFSET,
- VVAR_NR_PAGES
-};
-
-static __always_inline struct vdso_data *__s390_get_k_vdso_data(void)
-{
- return vdso_data;
-}
-#define __arch_get_k_vdso_data __s390_get_k_vdso_data
-
-static __always_inline struct vdso_rng_data *__s390_get_k_vdso_rnd_data(void)
-{
- return (void *)vdso_data + __VDSO_RND_DATA_OFFSET;
-}
-#define __arch_get_k_vdso_rng_data __s390_get_k_vdso_rnd_data
-
/* The asm-generic header needs to be included after the definitions above */
#include <asm-generic/vdso/vsyscall.h>
diff --git a/arch/s390/kernel/time.c b/arch/s390/kernel/time.c
index e9f47c3a61978a45c72aee23bc44dcb128113c8c..41ca3586b19f6cac3753b52f0b99be62a33e1cb1 100644
--- a/arch/s390/kernel/time.c
+++ b/arch/s390/kernel/time.c
@@ -84,7 +84,7 @@ void __init time_early_init(void)
/* Initialize TOD steering parameters */
tod_steering_end = tod_clock_base.tod;
for (cs = 0; cs < CS_BASES; cs++)
- vdso_data[cs].arch_data.tod_steering_end = tod_steering_end;
+ vdso_k_time_data[cs].arch_data.tod_steering_end = tod_steering_end;
if (!test_facility(28))
return;
@@ -390,8 +390,8 @@ static void clock_sync_global(long delta)
tod_steering_delta);
tod_steering_end = now + (abs(tod_steering_delta) << 15);
for (cs = 0; cs < CS_BASES; cs++) {
- vdso_data[cs].arch_data.tod_steering_end = tod_steering_end;
- vdso_data[cs].arch_data.tod_steering_delta = tod_steering_delta;
+ vdso_k_time_data[cs].arch_data.tod_steering_end = tod_steering_end;
+ vdso_k_time_data[cs].arch_data.tod_steering_delta = tod_steering_delta;
}
/* Update LPAR offset. */
diff --git a/arch/s390/kernel/vdso.c b/arch/s390/kernel/vdso.c
index 598b512cde012b1857c748042ceeea65b54cb92f..70c8f9ad13cde96f0f307213bacb671b818f628b 100644
--- a/arch/s390/kernel/vdso.c
+++ b/arch/s390/kernel/vdso.c
@@ -16,8 +16,8 @@
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/smp.h>
-#include <linux/time_namespace.h>
#include <linux/random.h>
+#include <linux/vdso_datastore.h>
#include <vdso/datapage.h>
#include <asm/vdso/vsyscall.h>
#include <asm/alternative.h>
@@ -26,85 +26,6 @@
extern char vdso64_start[], vdso64_end[];
extern char vdso32_start[], vdso32_end[];
-static struct vm_special_mapping vvar_mapping;
-
-static union vdso_data_store vdso_data_store __page_aligned_data;
-
-struct vdso_data *vdso_data = vdso_data_store.data;
-
-#ifdef CONFIG_TIME_NS
-struct vdso_data *arch_get_vdso_data(void *vvar_page)
-{
- return (struct vdso_data *)(vvar_page);
-}
-
-/*
- * The VVAR page layout depends on whether a task belongs to the root or
- * non-root time namespace. Whenever a task changes its namespace, the VVAR
- * page tables are cleared and then they will be re-faulted with a
- * corresponding layout.
- * See also the comment near timens_setup_vdso_data() for details.
- */
-int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
-{
- struct mm_struct *mm = task->mm;
- VMA_ITERATOR(vmi, mm, 0);
- struct vm_area_struct *vma;
-
- mmap_read_lock(mm);
- for_each_vma(vmi, vma) {
- if (!vma_is_special_mapping(vma, &vvar_mapping))
- continue;
- zap_vma_pages(vma);
- break;
- }
- mmap_read_unlock(mm);
- return 0;
-}
-#endif
-
-static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
- struct vm_area_struct *vma, struct vm_fault *vmf)
-{
- struct page *timens_page = find_timens_vvar_page(vma);
- unsigned long addr, pfn;
- vm_fault_t err;
-
- switch (vmf->pgoff) {
- case VVAR_DATA_PAGE_OFFSET:
- pfn = virt_to_pfn(vdso_data);
- if (timens_page) {
- /*
- * Fault in VVAR page too, since it will be accessed
- * to get clock data anyway.
- */
- addr = vmf->address + VVAR_TIMENS_PAGE_OFFSET * PAGE_SIZE;
- err = vmf_insert_pfn(vma, addr, pfn);
- if (unlikely(err & VM_FAULT_ERROR))
- return err;
- pfn = page_to_pfn(timens_page);
- }
- break;
-#ifdef CONFIG_TIME_NS
- case VVAR_TIMENS_PAGE_OFFSET:
- /*
- * If a task belongs to a time namespace then a namespace
- * specific VVAR is mapped with the VVAR_DATA_PAGE_OFFSET and
- * the real VVAR page is mapped with the VVAR_TIMENS_PAGE_OFFSET
- * offset.
- * See also the comment near timens_setup_vdso_data().
- */
- if (!timens_page)
- return VM_FAULT_SIGBUS;
- pfn = virt_to_pfn(vdso_data);
- break;
-#endif /* CONFIG_TIME_NS */
- default:
- return VM_FAULT_SIGBUS;
- }
- return vmf_insert_pfn(vma, vmf->address, pfn);
-}
-
static int vdso_mremap(const struct vm_special_mapping *sm,
struct vm_area_struct *vma)
{
@@ -112,11 +33,6 @@ static int vdso_mremap(const struct vm_special_mapping *sm,
return 0;
}
-static struct vm_special_mapping vvar_mapping = {
- .name = "[vvar]",
- .fault = vvar_fault,
-};
-
static struct vm_special_mapping vdso64_mapping = {
.name = "[vdso]",
.mremap = vdso_mremap,
@@ -142,7 +58,7 @@ static int map_vdso(unsigned long addr, unsigned long vdso_mapping_len)
struct vm_area_struct *vma;
int rc;
- BUILD_BUG_ON(VVAR_NR_PAGES != __VVAR_PAGES);
+ BUILD_BUG_ON(VDSO_NR_PAGES != __VDSO_PAGES);
if (mmap_write_lock_killable(mm))
return -EINTR;
@@ -157,14 +73,11 @@ static int map_vdso(unsigned long addr, unsigned long vdso_mapping_len)
rc = vvar_start;
if (IS_ERR_VALUE(vvar_start))
goto out;
- vma = _install_special_mapping(mm, vvar_start, VVAR_NR_PAGES*PAGE_SIZE,
- VM_READ|VM_MAYREAD|VM_IO|VM_DONTDUMP|
- VM_PFNMAP,
- &vvar_mapping);
+ vma = vdso_install_vvar_mapping(mm, vvar_start);
rc = PTR_ERR(vma);
if (IS_ERR(vma))
goto out;
- vdso_text_start = vvar_start + VVAR_NR_PAGES * PAGE_SIZE;
+ vdso_text_start = vvar_start + VDSO_NR_PAGES * PAGE_SIZE;
/* VM_MAYWRITE for COW so gdb can set breakpoints */
vma = _install_special_mapping(mm, vdso_text_start, vdso_text_len,
VM_READ|VM_EXEC|
@@ -220,7 +133,7 @@ unsigned long vdso_text_size(void)
unsigned long vdso_size(void)
{
- return vdso_text_size() + VVAR_NR_PAGES * PAGE_SIZE;
+ return vdso_text_size() + VDSO_NR_PAGES * PAGE_SIZE;
}
int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
diff --git a/arch/s390/kernel/vdso32/vdso32.lds.S b/arch/s390/kernel/vdso32/vdso32.lds.S
index c916c4f73f766ebcc9328b189e46015e67393b0d..9630d58c20806b51f3a2f3b59a53409ad6d3508f 100644
--- a/arch/s390/kernel/vdso32/vdso32.lds.S
+++ b/arch/s390/kernel/vdso32/vdso32.lds.S
@@ -6,16 +6,15 @@
#include <asm/page.h>
#include <asm/vdso.h>
+#include <vdso/datapage.h>
OUTPUT_FORMAT("elf32-s390", "elf32-s390", "elf32-s390")
OUTPUT_ARCH(s390:31-bit)
SECTIONS
{
- PROVIDE(_vdso_data = . - __VVAR_PAGES * PAGE_SIZE);
-#ifdef CONFIG_TIME_NS
- PROVIDE(_timens_data = _vdso_data + PAGE_SIZE);
-#endif
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
diff --git a/arch/s390/kernel/vdso64/vdso64.lds.S b/arch/s390/kernel/vdso64/vdso64.lds.S
index ec42b7d9cb53094b783c6b0492962af7e5027831..e4f6551ae898881baff6d37701a9d8d9188a52cc 100644
--- a/arch/s390/kernel/vdso64/vdso64.lds.S
+++ b/arch/s390/kernel/vdso64/vdso64.lds.S
@@ -7,17 +7,15 @@
#include <asm/vdso/vsyscall.h>
#include <asm/page.h>
#include <asm/vdso.h>
+#include <vdso/datapage.h>
OUTPUT_FORMAT("elf64-s390", "elf64-s390", "elf64-s390")
OUTPUT_ARCH(s390:64-bit)
SECTIONS
{
- PROVIDE(_vdso_data = . - __VVAR_PAGES * PAGE_SIZE);
- PROVIDE(_vdso_rng_data = _vdso_data + __VDSO_RND_DATA_OFFSET);
-#ifdef CONFIG_TIME_NS
- PROVIDE(_timens_data = _vdso_data + PAGE_SIZE);
-#endif
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 13/18] MIPS: vdso: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (11 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 12/18] s390/vdso: " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 14/18] powerpc/vdso: " Thomas Weißschuh
` (5 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/mips/Kconfig | 1 +
arch/mips/include/asm/vdso/gettimeofday.h | 9 +++---
arch/mips/include/asm/vdso/vdso.h | 19 ++++++-------
arch/mips/include/asm/vdso/vsyscall.h | 14 ++-------
arch/mips/kernel/vdso.c | 47 ++++++++++++-------------------
arch/mips/vdso/vdso.lds.S | 5 +++-
6 files changed, 38 insertions(+), 57 deletions(-)
diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index 1924f2d839323243e378e32ef7d12640560a7be3..b51168e319ea3aa376e7016fe72e8db9a28fc979 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -51,6 +51,7 @@ config MIPS
select GENERIC_SMP_IDLE_THREAD
select GENERIC_IDLE_POLL_SETUP
select GENERIC_TIME_VSYSCALL
+ select GENERIC_VDSO_DATA_STORE
select GUP_GET_PXX_LOW_HIGH if CPU_MIPS32 && PHYS_ADDR_T_64BIT
select HAS_IOPORT if !NO_IOPORT_MAP || ISA
select HAVE_ARCH_COMPILER_H
diff --git a/arch/mips/include/asm/vdso/gettimeofday.h b/arch/mips/include/asm/vdso/gettimeofday.h
index 44a45f3fa4b01026e39efb7f8f51051ec464340c..fd32baa30e172cbabe2657d9debe67a5b9061f90 100644
--- a/arch/mips/include/asm/vdso/gettimeofday.h
+++ b/arch/mips/include/asm/vdso/gettimeofday.h
@@ -167,7 +167,7 @@ static __always_inline u64 read_r4k_count(void)
#ifdef CONFIG_CLKSRC_MIPS_GIC
-static __always_inline u64 read_gic_count(const struct vdso_data *data)
+static __always_inline u64 read_gic_count(const struct vdso_time_data *data)
{
void __iomem *gic = get_gic(data);
u32 hi, hi2, lo;
@@ -184,7 +184,7 @@ static __always_inline u64 read_gic_count(const struct vdso_data *data)
#endif
static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
#ifdef CONFIG_CSRC_R4K
if (clock_mode == VDSO_CLOCKMODE_R4K)
@@ -209,10 +209,11 @@ static inline bool mips_vdso_hres_capable(void)
}
#define __arch_vdso_hres_capable mips_vdso_hres_capable
-static __always_inline const struct vdso_data *__arch_get_vdso_data(void)
+static __always_inline const struct vdso_time_data *__arch_get_vdso_u_time_data(void)
{
- return get_vdso_data();
+ return get_vdso_time_data();
}
+#define __arch_get_vdso_u_time_data __arch_get_vdso_u_time_data
#endif /* !__ASSEMBLY__ */
diff --git a/arch/mips/include/asm/vdso/vdso.h b/arch/mips/include/asm/vdso/vdso.h
index 6cd88191fefa9ce6d8323dd70d45aa9a198ac9d0..acd0efcd3d93ef7e71107302a40d24b4b51aae99 100644
--- a/arch/mips/include/asm/vdso/vdso.h
+++ b/arch/mips/include/asm/vdso/vdso.h
@@ -5,16 +5,18 @@
*/
#include <asm/sgidefs.h>
+#include <vdso/page.h>
+
+#define __VDSO_PAGES 4
#ifndef __ASSEMBLY__
#include <asm/asm.h>
-#include <asm/page.h>
#include <asm/vdso.h>
-static inline unsigned long get_vdso_base(void)
+static inline const struct vdso_time_data *get_vdso_time_data(void)
{
- unsigned long addr;
+ const struct vdso_time_data *addr;
/*
* We can't use cpu_has_mips_r6 since it needs the cpu_data[]
@@ -27,7 +29,7 @@ static inline unsigned long get_vdso_base(void)
* We can't use addiupc because there is no label-label
* support for the addiupc reloc
*/
- __asm__("lapc %0, _start \n"
+ __asm__("lapc %0, vdso_u_time_data \n"
: "=r" (addr) : :);
#else
/*
@@ -46,7 +48,7 @@ static inline unsigned long get_vdso_base(void)
" .set noreorder \n"
" bal 1f \n"
" nop \n"
- " .word _start - . \n"
+ " .word vdso_u_time_data - . \n"
"1: lw %0, 0($31) \n"
" " STR(PTR_ADDU) " %0, $31, %0 \n"
" .set pop \n"
@@ -58,14 +60,9 @@ static inline unsigned long get_vdso_base(void)
return addr;
}
-static inline const struct vdso_data *get_vdso_data(void)
-{
- return (const struct vdso_data *)(get_vdso_base() - PAGE_SIZE);
-}
-
#ifdef CONFIG_CLKSRC_MIPS_GIC
-static inline void __iomem *get_gic(const struct vdso_data *data)
+static inline void __iomem *get_gic(const struct vdso_time_data *data)
{
return (void __iomem *)((unsigned long)data & PAGE_MASK) - PAGE_SIZE;
}
diff --git a/arch/mips/include/asm/vdso/vsyscall.h b/arch/mips/include/asm/vdso/vsyscall.h
index a4582870aaea49ac288d62ec4fa1338a98621918..2b1debb62dee07bb49ab64f755ff9120d512458a 100644
--- a/arch/mips/include/asm/vdso/vsyscall.h
+++ b/arch/mips/include/asm/vdso/vsyscall.h
@@ -2,22 +2,12 @@
#ifndef __ASM_VDSO_VSYSCALL_H
#define __ASM_VDSO_VSYSCALL_H
+#include <asm/page.h>
+
#ifndef __ASSEMBLY__
#include <vdso/datapage.h>
-extern struct vdso_data *vdso_data;
-
-/*
- * Update the vDSO data page to keep in sync with kernel timekeeping.
- */
-static __always_inline
-struct vdso_data *__mips_get_k_vdso_data(void)
-{
- return vdso_data;
-}
-#define __arch_get_k_vdso_data __mips_get_k_vdso_data
-
/* The asm-generic header needs to be included after the definitions above */
#include <asm-generic/vdso/vsyscall.h>
diff --git a/arch/mips/kernel/vdso.c b/arch/mips/kernel/vdso.c
index 75c9d3618f58856933ba25adb4933cda25d16cbf..de096777172f082b5b7847a5c14cae5ddaee0430 100644
--- a/arch/mips/kernel/vdso.c
+++ b/arch/mips/kernel/vdso.c
@@ -15,6 +15,7 @@
#include <linux/random.h>
#include <linux/sched.h>
#include <linux/slab.h>
+#include <linux/vdso_datastore.h>
#include <asm/abi.h>
#include <asm/mips-cps.h>
@@ -23,20 +24,7 @@
#include <vdso/helpers.h>
#include <vdso/vsyscall.h>
-/* Kernel-provided data used by the VDSO. */
-static union vdso_data_store mips_vdso_data __page_aligned_data;
-struct vdso_data *vdso_data = mips_vdso_data.data;
-
-/*
- * Mapping for the VDSO data/GIC pages. The real pages are mapped manually, as
- * what we map and where within the area they are mapped is determined at
- * runtime.
- */
-static struct page *no_pages[] = { NULL };
-static struct vm_special_mapping vdso_vvar_mapping = {
- .name = "[vvar]",
- .pages = no_pages,
-};
+static_assert(VDSO_NR_PAGES == __VDSO_PAGES);
static void __init init_vdso_image(struct mips_vdso_image *image)
{
@@ -90,7 +78,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
{
struct mips_vdso_image *image = current->thread.abi->vdso;
struct mm_struct *mm = current->mm;
- unsigned long gic_size, vvar_size, size, base, data_addr, vdso_addr, gic_pfn, gic_base;
+ unsigned long gic_size, size, base, data_addr, vdso_addr, gic_pfn, gic_base;
struct vm_area_struct *vma;
int ret;
@@ -119,8 +107,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
* the counter registers at the start.
*/
gic_size = mips_gic_present() ? PAGE_SIZE : 0;
- vvar_size = gic_size + PAGE_SIZE;
- size = vvar_size + image->size;
+ size = gic_size + VDSO_NR_PAGES * PAGE_SIZE + image->size;
/*
* Find a region that's large enough for us to perform the
@@ -143,15 +130,13 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
*/
if (cpu_has_dc_aliases) {
base = __ALIGN_MASK(base, shm_align_mask);
- base += ((unsigned long)vdso_data - gic_size) & shm_align_mask;
+ base += ((unsigned long)vdso_k_time_data - gic_size) & shm_align_mask;
}
data_addr = base + gic_size;
- vdso_addr = data_addr + PAGE_SIZE;
+ vdso_addr = data_addr + VDSO_NR_PAGES * PAGE_SIZE;
- vma = _install_special_mapping(mm, base, vvar_size,
- VM_READ | VM_MAYREAD,
- &vdso_vvar_mapping);
+ vma = vdso_install_vvar_mapping(mm, data_addr);
if (IS_ERR(vma)) {
ret = PTR_ERR(vma);
goto out;
@@ -161,6 +146,17 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
if (gic_size) {
gic_base = (unsigned long)mips_gic_base + MIPS_GIC_USER_OFS;
gic_pfn = PFN_DOWN(__pa(gic_base));
+ static const struct vm_special_mapping gic_mapping = {
+ .name = "[gic]",
+ .pages = (struct page **) { NULL },
+ };
+
+ vma = _install_special_mapping(mm, base, gic_size, VM_READ | VM_MAYREAD,
+ &gic_mapping);
+ if (IS_ERR(vma)) {
+ ret = PTR_ERR(vma);
+ goto out;
+ }
ret = io_remap_pfn_range(vma, base, gic_pfn, gic_size,
pgprot_noncached(vma->vm_page_prot));
@@ -168,13 +164,6 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
goto out;
}
- /* Map data page. */
- ret = remap_pfn_range(vma, data_addr,
- virt_to_phys(vdso_data) >> PAGE_SHIFT,
- PAGE_SIZE, vma->vm_page_prot);
- if (ret)
- goto out;
-
/* Map VDSO image. */
vma = _install_special_mapping(mm, vdso_addr, image->size,
VM_READ | VM_EXEC |
diff --git a/arch/mips/vdso/vdso.lds.S b/arch/mips/vdso/vdso.lds.S
index 836465e3bcb8a15cb3f0f96d852d2b252c9d5e2a..c8bbe56d89cb095da21ac1f56d9999914defc305 100644
--- a/arch/mips/vdso/vdso.lds.S
+++ b/arch/mips/vdso/vdso.lds.S
@@ -5,6 +5,8 @@
*/
#include <asm/sgidefs.h>
+#include <asm/vdso/vdso.h>
+#include <vdso/datapage.h>
#if _MIPS_SIM == _MIPS_SIM_ABI64
OUTPUT_FORMAT("elf64-tradlittlemips", "elf64-tradbigmips", "elf64-tradlittlemips")
@@ -18,7 +20,8 @@ OUTPUT_ARCH(mips)
SECTIONS
{
- PROVIDE(_start = .);
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
/*
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 14/18] powerpc/vdso: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (12 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 13/18] MIPS: vdso: " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-05 8:50 ` Christophe Leroy
2025-02-04 12:05 ` [PATCH v3 15/18] x86/vdso: " Thomas Weißschuh
` (4 subsequent siblings)
18 siblings, 1 reply; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/powerpc/Kconfig | 2 +
arch/powerpc/include/asm/vdso.h | 1 +
arch/powerpc/include/asm/vdso/arch_data.h | 37 +++++++++
arch/powerpc/include/asm/vdso/getrandom.h | 11 +--
arch/powerpc/include/asm/vdso/gettimeofday.h | 29 +++----
arch/powerpc/include/asm/vdso/vsyscall.h | 13 ---
arch/powerpc/include/asm/vdso_datapage.h | 44 +---------
arch/powerpc/kernel/asm-offsets.c | 1 -
arch/powerpc/kernel/time.c | 2 +-
arch/powerpc/kernel/vdso.c | 115 +++------------------------
arch/powerpc/kernel/vdso/cacheflush.S | 2 +-
arch/powerpc/kernel/vdso/datapage.S | 4 +-
arch/powerpc/kernel/vdso/gettimeofday.S | 4 +-
arch/powerpc/kernel/vdso/vdso32.lds.S | 4 +-
arch/powerpc/kernel/vdso/vdso64.lds.S | 4 +-
arch/powerpc/kernel/vdso/vgettimeofday.c | 14 ++--
16 files changed, 89 insertions(+), 198 deletions(-)
diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index 424f188e62d9886ee7f6d2531547f09a0606747d..8a2a6e403eb1e654254100cb7347477a3a82e3f7 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -159,6 +159,7 @@ config PPC
select ARCH_HAS_TICK_BROADCAST if GENERIC_CLOCKEVENTS_BROADCAST
select ARCH_HAS_UACCESS_FLUSHCACHE
select ARCH_HAS_UBSAN
+ select ARCH_HAS_VDSO_ARCH_DATA
select ARCH_HAVE_NMI_SAFE_CMPXCHG
select ARCH_HAVE_EXTRA_ELF_NOTES if SPU_BASE
select ARCH_KEEP_MEMBLOCK
@@ -209,6 +210,7 @@ config PPC
select GENERIC_PTDUMP
select GENERIC_SMP_IDLE_THREAD
select GENERIC_TIME_VSYSCALL
+ select GENERIC_VDSO_DATA_STORE
select GENERIC_VDSO_TIME_NS
select HAS_IOPORT if PCI
select HAVE_ARCH_AUDITSYSCALL
diff --git a/arch/powerpc/include/asm/vdso.h b/arch/powerpc/include/asm/vdso.h
index 8d972bc98b55fe916f23488ca9e2a5918046b9aa..1ca23fbfe087ae90b90c4286335f86d9f8121078 100644
--- a/arch/powerpc/include/asm/vdso.h
+++ b/arch/powerpc/include/asm/vdso.h
@@ -3,6 +3,7 @@
#define _ASM_POWERPC_VDSO_H
#define VDSO_VERSION_STRING LINUX_2.6.15
+#define __VDSO_PAGES 4
#ifndef __ASSEMBLY__
diff --git a/arch/powerpc/include/asm/vdso/arch_data.h b/arch/powerpc/include/asm/vdso/arch_data.h
new file mode 100644
index 0000000000000000000000000000000000000000..c240a6b875181ac4159f2e80b11f9bf214e22808
--- /dev/null
+++ b/arch/powerpc/include/asm/vdso/arch_data.h
@@ -0,0 +1,37 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2002 Peter Bergner <bergner@vnet.ibm.com>, IBM
+ * Copyright (C) 2005 Benjamin Herrenschmidy <benh@kernel.crashing.org>,
+ * IBM Corp.
+ */
+#ifndef _ASM_POWERPC_VDSO_ARCH_DATA_H
+#define _ASM_POWERPC_VDSO_ARCH_DATA_H
+
+#include <linux/unistd.h>
+#include <linux/types.h>
+
+#define SYSCALL_MAP_SIZE ((NR_syscalls + 31) / 32)
+
+#ifdef CONFIG_PPC64
+
+struct vdso_arch_data {
+ __u64 tb_ticks_per_sec; /* Timebase tics / sec */
+ __u32 dcache_block_size; /* L1 d-cache block size */
+ __u32 icache_block_size; /* L1 i-cache block size */
+ __u32 dcache_log_block_size; /* L1 d-cache log block size */
+ __u32 icache_log_block_size; /* L1 i-cache log block size */
+ __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
+ __u32 compat_syscall_map[SYSCALL_MAP_SIZE]; /* Map of compat syscalls */
+};
+
+#else /* CONFIG_PPC64 */
+
+struct vdso_arch_data {
+ __u64 tb_ticks_per_sec; /* Timebase tics / sec */
+ __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
+ __u32 compat_syscall_map[0]; /* No compat syscalls on PPC32 */
+};
+
+#endif /* CONFIG_PPC64 */
+
+#endif /* _ASM_POWERPC_VDSO_ARCH_DATA_H */
diff --git a/arch/powerpc/include/asm/vdso/getrandom.h b/arch/powerpc/include/asm/vdso/getrandom.h
index 80ce0709725eb89c1f3b69e0733038b458fbf24f..067a5396aac6e99c1a96f730459ec684bc7785d7 100644
--- a/arch/powerpc/include/asm/vdso/getrandom.h
+++ b/arch/powerpc/include/asm/vdso/getrandom.h
@@ -43,20 +43,21 @@ static __always_inline ssize_t getrandom_syscall(void *buffer, size_t len, unsig
(unsigned long)len, (unsigned long)flags);
}
-static __always_inline struct vdso_rng_data *__arch_get_vdso_rng_data(void)
+static __always_inline const struct vdso_rng_data *__arch_get_vdso_u_rng_data(void)
{
- struct vdso_arch_data *data;
+ struct vdso_rng_data *data;
asm (
" bcl 20, 31, .+4 ;"
"0: mflr %0 ;"
- " addis %0, %0, (_vdso_datapage - 0b)@ha ;"
- " addi %0, %0, (_vdso_datapage - 0b)@l ;"
+ " addis %0, %0, (vdso_u_rng_data - 0b)@ha ;"
+ " addi %0, %0, (vdso_u_rng_data - 0b)@l ;"
: "=r" (data) : : "lr"
);
- return &data->rng_data;
+ return data;
}
+#define __arch_get_vdso_u_rng_data __arch_get_vdso_u_rng_data
ssize_t __c_kernel_getrandom(void *buffer, size_t len, unsigned int flags, void *opaque_state,
size_t opaque_len);
diff --git a/arch/powerpc/include/asm/vdso/gettimeofday.h b/arch/powerpc/include/asm/vdso/gettimeofday.h
index c6390890a60c2fdcb608bf321b2945c3fb372f54..dc955f2e0cc51f44d46f488a292aa0dbee3dc16c 100644
--- a/arch/powerpc/include/asm/vdso/gettimeofday.h
+++ b/arch/powerpc/include/asm/vdso/gettimeofday.h
@@ -94,22 +94,12 @@ int clock_getres32_fallback(clockid_t _clkid, struct old_timespec32 *_ts)
#endif
static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
return get_tb();
}
-const struct vdso_data *__arch_get_vdso_data(void);
-
-#ifdef CONFIG_TIME_NS
-static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
-{
- return (void *)vd + (1U << CONFIG_PAGE_SHIFT);
-}
-#endif
-
-static inline bool vdso_clocksource_ok(const struct vdso_data *vd)
+static inline bool vdso_clocksource_ok(const struct vdso_time_data *vd)
{
return true;
}
@@ -135,21 +125,22 @@ static __always_inline u64 vdso_shift_ns(u64 ns, unsigned long shift)
#ifdef __powerpc64__
int __c_kernel_clock_gettime(clockid_t clock, struct __kernel_timespec *ts,
- const struct vdso_data *vd);
+ const struct vdso_time_data *vd);
int __c_kernel_clock_getres(clockid_t clock_id, struct __kernel_timespec *res,
- const struct vdso_data *vd);
+ const struct vdso_time_data *vd);
#else
int __c_kernel_clock_gettime(clockid_t clock, struct old_timespec32 *ts,
- const struct vdso_data *vd);
+ const struct vdso_time_data *vd);
int __c_kernel_clock_gettime64(clockid_t clock, struct __kernel_timespec *ts,
- const struct vdso_data *vd);
+ const struct vdso_time_data *vd);
int __c_kernel_clock_getres(clockid_t clock_id, struct old_timespec32 *res,
- const struct vdso_data *vd);
+ const struct vdso_time_data *vd);
#endif
int __c_kernel_gettimeofday(struct __kernel_old_timeval *tv, struct timezone *tz,
- const struct vdso_data *vd);
+ const struct vdso_time_data *vd);
__kernel_old_time_t __c_kernel_time(__kernel_old_time_t *time,
- const struct vdso_data *vd);
+ const struct vdso_time_data *vd);
+
#endif /* __ASSEMBLY__ */
#endif /* _ASM_POWERPC_VDSO_GETTIMEOFDAY_H */
diff --git a/arch/powerpc/include/asm/vdso/vsyscall.h b/arch/powerpc/include/asm/vdso/vsyscall.h
index 48560a11955956b8fbb59360334a81972723bd57..c2c9ae1b22e71a3f87e5a1a351699c7ab42b2f95 100644
--- a/arch/powerpc/include/asm/vdso/vsyscall.h
+++ b/arch/powerpc/include/asm/vdso/vsyscall.h
@@ -6,19 +6,6 @@
#include <asm/vdso_datapage.h>
-static __always_inline
-struct vdso_data *__arch_get_k_vdso_data(void)
-{
- return vdso_data->data;
-}
-#define __arch_get_k_vdso_data __arch_get_k_vdso_data
-
-static __always_inline
-struct vdso_rng_data *__arch_get_k_vdso_rng_data(void)
-{
- return &vdso_data->rng_data;
-}
-
/* The asm-generic header needs to be included after the definitions above */
#include <asm-generic/vdso/vsyscall.h>
diff --git a/arch/powerpc/include/asm/vdso_datapage.h b/arch/powerpc/include/asm/vdso_datapage.h
index a202f5b63479533a7f45a74df015feb59f3d7c87..95d45a50355d269454dd3e175a5b3844181536b5 100644
--- a/arch/powerpc/include/asm/vdso_datapage.h
+++ b/arch/powerpc/include/asm/vdso_datapage.h
@@ -11,56 +11,18 @@
#ifndef __ASSEMBLY__
-#include <linux/unistd.h>
-#include <linux/time.h>
#include <vdso/datapage.h>
-#define SYSCALL_MAP_SIZE ((NR_syscalls + 31) / 32)
-
-#ifdef CONFIG_PPC64
-
-struct vdso_arch_data {
- __u64 tb_ticks_per_sec; /* Timebase tics / sec */
- __u32 dcache_block_size; /* L1 d-cache block size */
- __u32 icache_block_size; /* L1 i-cache block size */
- __u32 dcache_log_block_size; /* L1 d-cache log block size */
- __u32 icache_log_block_size; /* L1 i-cache log block size */
- __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
- __u32 compat_syscall_map[SYSCALL_MAP_SIZE]; /* Map of compat syscalls */
-
- struct vdso_rng_data rng_data;
-
- struct vdso_data data[CS_BASES] __aligned(1 << CONFIG_PAGE_SHIFT);
-};
-
-#else /* CONFIG_PPC64 */
-
-struct vdso_arch_data {
- __u64 tb_ticks_per_sec; /* Timebase tics / sec */
- __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
- __u32 compat_syscall_map[0]; /* No compat syscalls on PPC32 */
- struct vdso_rng_data rng_data;
-
- struct vdso_data data[CS_BASES] __aligned(1 << CONFIG_PAGE_SHIFT);
-};
-
-#endif /* CONFIG_PPC64 */
-
-extern struct vdso_arch_data *vdso_data;
-
#else /* __ASSEMBLY__ */
-.macro get_datapage ptr offset=0
+.macro get_datapage ptr symbol
bcl 20, 31, .+4
999:
mflr \ptr
- addis \ptr, \ptr, (_vdso_datapage - 999b + \offset)@ha
- addi \ptr, \ptr, (_vdso_datapage - 999b + \offset)@l
+ addis \ptr, \ptr, (\symbol - 999b)@ha
+ addi \ptr, \ptr, (\symbol - 999b)@l
.endm
-#include <asm/asm-offsets.h>
-#include <asm/page.h>
-
#endif /* __ASSEMBLY__ */
#endif /* __KERNEL__ */
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index 7a390bd4f4af3c7408b3e3c5ef6d43b95b3b6463..b3048f6d3822c0c457f4aa2ccb5dc870495ba79b 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -334,7 +334,6 @@ int main(void)
#endif /* ! CONFIG_PPC64 */
/* datapage offsets for use by vdso */
- OFFSET(VDSO_DATA_OFFSET, vdso_arch_data, data);
OFFSET(CFG_TB_TICKS_PER_SEC, vdso_arch_data, tb_ticks_per_sec);
#ifdef CONFIG_PPC64
OFFSET(CFG_ICACHE_BLOCKSZ, vdso_arch_data, icache_block_size);
diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
index 0727332ad86fbcfcf8ca18b344ba04381e827c79..15784c5c95c77f1eccfa948a36ba69386a2c175b 100644
--- a/arch/powerpc/kernel/time.c
+++ b/arch/powerpc/kernel/time.c
@@ -950,7 +950,7 @@ void __init time_init(void)
sys_tz.tz_dsttime = 0;
}
- vdso_data->tb_ticks_per_sec = tb_ticks_per_sec;
+ vdso_k_arch_data->tb_ticks_per_sec = tb_ticks_per_sec;
#ifdef CONFIG_PPC64_PROC_SYSTEMCFG
systemcfg->tb_ticks_per_sec = tb_ticks_per_sec;
#endif
diff --git a/arch/powerpc/kernel/vdso.c b/arch/powerpc/kernel/vdso.c
index 43379365ce1b37cfba662ea58feca5e73dd5f700..219d67bcf747e79f48d09a50f5cb9624bcc0f7b1 100644
--- a/arch/powerpc/kernel/vdso.c
+++ b/arch/powerpc/kernel/vdso.c
@@ -17,7 +17,7 @@
#include <linux/elf.h>
#include <linux/security.h>
#include <linux/syscalls.h>
-#include <linux/time_namespace.h>
+#include <linux/vdso_datastore.h>
#include <vdso/datapage.h>
#include <asm/syscall.h>
@@ -32,6 +32,8 @@
#include <asm/vdso_datapage.h>
#include <asm/setup.h>
+static_assert(__VDSO_PAGES == VDSO_NR_PAGES);
+
/* The alignment of the vDSO */
#define VDSO_ALIGNMENT (1 << 16)
@@ -40,24 +42,6 @@ extern char vdso64_start, vdso64_end;
long sys_ni_syscall(void);
-/*
- * The vdso data page (aka. systemcfg for old ppc64 fans) is here.
- * Once the early boot kernel code no longer needs to muck around
- * with it, it will become dynamically allocated
- */
-static union {
- struct vdso_arch_data data;
- u8 page[2 * PAGE_SIZE];
-} vdso_data_store __page_aligned_data;
-struct vdso_arch_data *vdso_data = &vdso_data_store.data;
-
-enum vvar_pages {
- VVAR_BASE_PAGE_OFFSET,
- VVAR_TIME_PAGE_OFFSET,
- VVAR_TIMENS_PAGE_OFFSET,
- VVAR_NR_PAGES,
-};
-
static int vdso_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma,
unsigned long text_size)
{
@@ -96,14 +80,6 @@ static void vdso_close(const struct vm_special_mapping *sm, struct vm_area_struc
mm->context.vdso = NULL;
}
-static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
- struct vm_area_struct *vma, struct vm_fault *vmf);
-
-static struct vm_special_mapping vvar_spec __ro_after_init = {
- .name = "[vvar]",
- .fault = vvar_fault,
-};
-
static struct vm_special_mapping vdso32_spec __ro_after_init = {
.name = "[vdso]",
.mremap = vdso32_mremap,
@@ -116,73 +92,6 @@ static struct vm_special_mapping vdso64_spec __ro_after_init = {
.close = vdso_close,
};
-#ifdef CONFIG_TIME_NS
-struct vdso_data *arch_get_vdso_data(void *vvar_page)
-{
- return vvar_page;
-}
-
-/*
- * The vvar mapping contains data for a specific time namespace, so when a task
- * changes namespace we must unmap its vvar data for the old namespace.
- * Subsequent faults will map in data for the new namespace.
- *
- * For more details see timens_setup_vdso_data().
- */
-int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
-{
- struct mm_struct *mm = task->mm;
- VMA_ITERATOR(vmi, mm, 0);
- struct vm_area_struct *vma;
-
- mmap_read_lock(mm);
- for_each_vma(vmi, vma) {
- if (vma_is_special_mapping(vma, &vvar_spec))
- zap_vma_pages(vma);
- }
- mmap_read_unlock(mm);
-
- return 0;
-}
-#endif
-
-static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
- struct vm_area_struct *vma, struct vm_fault *vmf)
-{
- struct page *timens_page = find_timens_vvar_page(vma);
- unsigned long pfn;
-
- switch (vmf->pgoff) {
- case VVAR_BASE_PAGE_OFFSET:
- pfn = virt_to_pfn(vdso_data);
- break;
- case VVAR_TIME_PAGE_OFFSET:
- if (timens_page)
- pfn = page_to_pfn(timens_page);
- else
- pfn = virt_to_pfn(vdso_data->data);
- break;
-#ifdef CONFIG_TIME_NS
- case VVAR_TIMENS_PAGE_OFFSET:
- /*
- * If a task belongs to a time namespace then a namespace
- * specific VVAR is mapped with the VVAR_DATA_PAGE_OFFSET and
- * the real VVAR page is mapped with the VVAR_TIMENS_PAGE_OFFSET
- * offset.
- * See also the comment near timens_setup_vdso_data().
- */
- if (!timens_page)
- return VM_FAULT_SIGBUS;
- pfn = virt_to_pfn(vdso_data->data);
- break;
-#endif /* CONFIG_TIME_NS */
- default:
- return VM_FAULT_SIGBUS;
- }
-
- return vmf_insert_pfn(vma, vmf->address, pfn);
-}
-
/*
* This is called from binfmt_elf, we create the special vma for the
* vDSO and insert it into the mm struct tree
@@ -191,7 +100,7 @@ static int __arch_setup_additional_pages(struct linux_binprm *bprm, int uses_int
{
unsigned long vdso_size, vdso_base, mappings_size;
struct vm_special_mapping *vdso_spec;
- unsigned long vvar_size = VVAR_NR_PAGES * PAGE_SIZE;
+ unsigned long vvar_size = VDSO_NR_PAGES * PAGE_SIZE;
struct mm_struct *mm = current->mm;
struct vm_area_struct *vma;
@@ -217,9 +126,7 @@ static int __arch_setup_additional_pages(struct linux_binprm *bprm, int uses_int
/* Add required alignment. */
vdso_base = ALIGN(vdso_base, VDSO_ALIGNMENT);
- vma = _install_special_mapping(mm, vdso_base, vvar_size,
- VM_READ | VM_MAYREAD | VM_IO |
- VM_DONTDUMP | VM_PFNMAP, &vvar_spec);
+ vma = vdso_install_vvar_mapping(mm, vdso_base);
if (IS_ERR(vma))
return PTR_ERR(vma);
@@ -299,10 +206,10 @@ static void __init vdso_setup_syscall_map(void)
for (i = 0; i < NR_syscalls; i++) {
if (sys_call_table[i] != (void *)&sys_ni_syscall)
- vdso_data->syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
+ vdso_k_arch_data->syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
if (IS_ENABLED(CONFIG_COMPAT) &&
compat_sys_call_table[i] != (void *)&sys_ni_syscall)
- vdso_data->compat_syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
+ vdso_k_arch_data->compat_syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
}
}
@@ -352,10 +259,10 @@ static struct page ** __init vdso_setup_pages(void *start, void *end)
static int __init vdso_init(void)
{
#ifdef CONFIG_PPC64
- vdso_data->dcache_block_size = ppc64_caches.l1d.block_size;
- vdso_data->icache_block_size = ppc64_caches.l1i.block_size;
- vdso_data->dcache_log_block_size = ppc64_caches.l1d.log_block_size;
- vdso_data->icache_log_block_size = ppc64_caches.l1i.log_block_size;
+ vdso_k_arch_data->dcache_block_size = ppc64_caches.l1d.block_size;
+ vdso_k_arch_data->icache_block_size = ppc64_caches.l1i.block_size;
+ vdso_k_arch_data->dcache_log_block_size = ppc64_caches.l1d.log_block_size;
+ vdso_k_arch_data->icache_log_block_size = ppc64_caches.l1i.log_block_size;
#endif /* CONFIG_PPC64 */
vdso_setup_syscall_map();
diff --git a/arch/powerpc/kernel/vdso/cacheflush.S b/arch/powerpc/kernel/vdso/cacheflush.S
index 0085ae464dac9c32381625a6969a4e422ad34eb7..488d3ade11e64996b30f42777251df8499eda92c 100644
--- a/arch/powerpc/kernel/vdso/cacheflush.S
+++ b/arch/powerpc/kernel/vdso/cacheflush.S
@@ -30,7 +30,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_COHERENT_ICACHE)
#ifdef CONFIG_PPC64
mflr r12
.cfi_register lr,r12
- get_datapage r10
+ get_datapage r10 vdso_u_arch_data
mtlr r12
.cfi_restore lr
#endif
diff --git a/arch/powerpc/kernel/vdso/datapage.S b/arch/powerpc/kernel/vdso/datapage.S
index db8e167f01667eb95b3dc74f6771e610411bba90..d23b2e8e2a34ca9b142231eb3a492716a49b2248 100644
--- a/arch/powerpc/kernel/vdso/datapage.S
+++ b/arch/powerpc/kernel/vdso/datapage.S
@@ -28,7 +28,7 @@ V_FUNCTION_BEGIN(__kernel_get_syscall_map)
mflr r12
.cfi_register lr,r12
mr. r4,r3
- get_datapage r3
+ get_datapage r3 vdso_u_arch_data
mtlr r12
#ifdef __powerpc64__
addi r3,r3,CFG_SYSCALL_MAP64
@@ -52,7 +52,7 @@ V_FUNCTION_BEGIN(__kernel_get_tbfreq)
.cfi_startproc
mflr r12
.cfi_register lr,r12
- get_datapage r3
+ get_datapage r3 vdso_u_arch_data
#ifndef __powerpc64__
lwz r4,(CFG_TB_TICKS_PER_SEC + 4)(r3)
#endif
diff --git a/arch/powerpc/kernel/vdso/gettimeofday.S b/arch/powerpc/kernel/vdso/gettimeofday.S
index 5333848322ca6105018d501952e3bf42475f49df..79c967212444732da50805fd086c6f2a3c75b0cc 100644
--- a/arch/powerpc/kernel/vdso/gettimeofday.S
+++ b/arch/powerpc/kernel/vdso/gettimeofday.S
@@ -33,9 +33,9 @@
.cfi_rel_offset r2, PPC_MIN_STKFRM + STK_GOT
#endif
.ifeq \call_time
- get_datapage r5 VDSO_DATA_OFFSET
+ get_datapage r5 vdso_u_time_data
.else
- get_datapage r4 VDSO_DATA_OFFSET
+ get_datapage r4 vdso_u_time_data
.endif
bl CFUNC(DOTSYM(\funct))
PPC_LL r0, PPC_MIN_STKFRM + PPC_LR_STKOFF(r1)
diff --git a/arch/powerpc/kernel/vdso/vdso32.lds.S b/arch/powerpc/kernel/vdso/vdso32.lds.S
index 1a1b0b6d681a9977e4ef8042e52d8d33da61887e..72a1012b8a205c6357cecb4b53d2d8e1ff59b051 100644
--- a/arch/powerpc/kernel/vdso/vdso32.lds.S
+++ b/arch/powerpc/kernel/vdso/vdso32.lds.S
@@ -6,6 +6,7 @@
#include <asm/vdso.h>
#include <asm/page.h>
#include <asm-generic/vmlinux.lds.h>
+#include <vdso/datapage.h>
#ifdef __LITTLE_ENDIAN__
OUTPUT_FORMAT("elf32-powerpcle", "elf32-powerpcle", "elf32-powerpcle")
@@ -16,7 +17,8 @@ OUTPUT_ARCH(powerpc:common)
SECTIONS
{
- PROVIDE(_vdso_datapage = . - 3 * PAGE_SIZE);
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
diff --git a/arch/powerpc/kernel/vdso/vdso64.lds.S b/arch/powerpc/kernel/vdso/vdso64.lds.S
index e21b5506cad62b16e677be74fda7921ec917141a..32102a05eaa7e015e0f89e4a94a3c5e31da7d460 100644
--- a/arch/powerpc/kernel/vdso/vdso64.lds.S
+++ b/arch/powerpc/kernel/vdso/vdso64.lds.S
@@ -6,6 +6,7 @@
#include <asm/vdso.h>
#include <asm/page.h>
#include <asm-generic/vmlinux.lds.h>
+#include <vdso/datapage.h>
#ifdef __LITTLE_ENDIAN__
OUTPUT_FORMAT("elf64-powerpcle", "elf64-powerpcle", "elf64-powerpcle")
@@ -16,7 +17,8 @@ OUTPUT_ARCH(powerpc:common64)
SECTIONS
{
- PROVIDE(_vdso_datapage = . - 3 * PAGE_SIZE);
+ VDSO_VVAR_SYMS
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
diff --git a/arch/powerpc/kernel/vdso/vgettimeofday.c b/arch/powerpc/kernel/vdso/vgettimeofday.c
index 55a287c9a7366aa59ab4af1e760a8995f588a4d5..6f5167d81af5f3e6e755dbda4307769e45a28421 100644
--- a/arch/powerpc/kernel/vdso/vgettimeofday.c
+++ b/arch/powerpc/kernel/vdso/vgettimeofday.c
@@ -7,43 +7,43 @@
#ifdef __powerpc64__
int __c_kernel_clock_gettime(clockid_t clock, struct __kernel_timespec *ts,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
return __cvdso_clock_gettime_data(vd, clock, ts);
}
int __c_kernel_clock_getres(clockid_t clock_id, struct __kernel_timespec *res,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
return __cvdso_clock_getres_data(vd, clock_id, res);
}
#else
int __c_kernel_clock_gettime(clockid_t clock, struct old_timespec32 *ts,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
return __cvdso_clock_gettime32_data(vd, clock, ts);
}
int __c_kernel_clock_gettime64(clockid_t clock, struct __kernel_timespec *ts,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
return __cvdso_clock_gettime_data(vd, clock, ts);
}
int __c_kernel_clock_getres(clockid_t clock_id, struct old_timespec32 *res,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
return __cvdso_clock_getres_time32_data(vd, clock_id, res);
}
#endif
int __c_kernel_gettimeofday(struct __kernel_old_timeval *tv, struct timezone *tz,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
return __cvdso_gettimeofday_data(vd, tv, tz);
}
-__kernel_old_time_t __c_kernel_time(__kernel_old_time_t *time, const struct vdso_data *vd)
+__kernel_old_time_t __c_kernel_time(__kernel_old_time_t *time, const struct vdso_time_data *vd)
{
return __cvdso_time_data(vd, time);
}
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* Re: [PATCH v3 14/18] powerpc/vdso: Switch to generic storage implementation
2025-02-04 12:05 ` [PATCH v3 14/18] powerpc/vdso: " Thomas Weißschuh
@ 2025-02-05 8:50 ` Christophe Leroy
0 siblings, 0 replies; 29+ messages in thread
From: Christophe Leroy @ 2025-02-05 8:50 UTC (permalink / raw)
To: Thomas Weißschuh, James E.J. Bottomley, Helge Deller,
Andy Lutomirski, Thomas Gleixner, Vincenzo Frascino,
Anna-Maria Behnsen, Frederic Weisbecker, Andrew Morton,
Catalin Marinas, Will Deacon, Theodore Ts'o,
Jason A. Donenfeld, Paul Walmsley, Palmer Dabbelt, Albert Ou,
Huacai Chen, WANG Xuerui, Russell King, Heiko Carstens,
Vasily Gorbik, Alexander Gordeev, Christian Borntraeger,
Sven Schnelle, Thomas Bogendoerfer, Michael Ellerman,
Nicholas Piggin, Naveen N Rao, Madhavan Srinivasan, Ingo Molnar,
Borislav Petkov, Dave Hansen, x86, H. Peter Anvin, Arnd Bergmann,
Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, linux-csky
Le 04/02/2025 à 13:05, Thomas Weißschuh a écrit :
> The generic storage implementation provides the same features as the
> custom one. However it can be shared between architectures, making
> maintenance easier.
>
> Co-developed-by: Nam Cao <namcao@linutronix.de>
> Signed-off-by: Nam Cao <namcao@linutronix.de>
> Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
> ---
> arch/powerpc/Kconfig | 2 +
> arch/powerpc/include/asm/vdso.h | 1 +
> arch/powerpc/include/asm/vdso/arch_data.h | 37 +++++++++
> arch/powerpc/include/asm/vdso/getrandom.h | 11 +--
> arch/powerpc/include/asm/vdso/gettimeofday.h | 29 +++----
> arch/powerpc/include/asm/vdso/vsyscall.h | 13 ---
> arch/powerpc/include/asm/vdso_datapage.h | 44 +---------
> arch/powerpc/kernel/asm-offsets.c | 1 -
> arch/powerpc/kernel/time.c | 2 +-
> arch/powerpc/kernel/vdso.c | 115 +++------------------------
> arch/powerpc/kernel/vdso/cacheflush.S | 2 +-
> arch/powerpc/kernel/vdso/datapage.S | 4 +-
> arch/powerpc/kernel/vdso/gettimeofday.S | 4 +-
> arch/powerpc/kernel/vdso/vdso32.lds.S | 4 +-
> arch/powerpc/kernel/vdso/vdso64.lds.S | 4 +-
> arch/powerpc/kernel/vdso/vgettimeofday.c | 14 ++--
> 16 files changed, 89 insertions(+), 198 deletions(-)
>
> diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
> index 424f188e62d9886ee7f6d2531547f09a0606747d..8a2a6e403eb1e654254100cb7347477a3a82e3f7 100644
> --- a/arch/powerpc/Kconfig
> +++ b/arch/powerpc/Kconfig
> @@ -159,6 +159,7 @@ config PPC
> select ARCH_HAS_TICK_BROADCAST if GENERIC_CLOCKEVENTS_BROADCAST
> select ARCH_HAS_UACCESS_FLUSHCACHE
> select ARCH_HAS_UBSAN
> + select ARCH_HAS_VDSO_ARCH_DATA
> select ARCH_HAVE_NMI_SAFE_CMPXCHG
> select ARCH_HAVE_EXTRA_ELF_NOTES if SPU_BASE
> select ARCH_KEEP_MEMBLOCK
> @@ -209,6 +210,7 @@ config PPC
> select GENERIC_PTDUMP
> select GENERIC_SMP_IDLE_THREAD
> select GENERIC_TIME_VSYSCALL
> + select GENERIC_VDSO_DATA_STORE
> select GENERIC_VDSO_TIME_NS
> select HAS_IOPORT if PCI
> select HAVE_ARCH_AUDITSYSCALL
> diff --git a/arch/powerpc/include/asm/vdso.h b/arch/powerpc/include/asm/vdso.h
> index 8d972bc98b55fe916f23488ca9e2a5918046b9aa..1ca23fbfe087ae90b90c4286335f86d9f8121078 100644
> --- a/arch/powerpc/include/asm/vdso.h
> +++ b/arch/powerpc/include/asm/vdso.h
> @@ -3,6 +3,7 @@
> #define _ASM_POWERPC_VDSO_H
>
> #define VDSO_VERSION_STRING LINUX_2.6.15
> +#define __VDSO_PAGES 4
>
> #ifndef __ASSEMBLY__
>
> diff --git a/arch/powerpc/include/asm/vdso/arch_data.h b/arch/powerpc/include/asm/vdso/arch_data.h
> new file mode 100644
> index 0000000000000000000000000000000000000000..c240a6b875181ac4159f2e80b11f9bf214e22808
> --- /dev/null
> +++ b/arch/powerpc/include/asm/vdso/arch_data.h
> @@ -0,0 +1,37 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (C) 2002 Peter Bergner <bergner@vnet.ibm.com>, IBM
> + * Copyright (C) 2005 Benjamin Herrenschmidy <benh@kernel.crashing.org>,
> + * IBM Corp.
> + */
> +#ifndef _ASM_POWERPC_VDSO_ARCH_DATA_H
> +#define _ASM_POWERPC_VDSO_ARCH_DATA_H
> +
> +#include <linux/unistd.h>
> +#include <linux/types.h>
> +
> +#define SYSCALL_MAP_SIZE ((NR_syscalls + 31) / 32)
> +
> +#ifdef CONFIG_PPC64
> +
> +struct vdso_arch_data {
> + __u64 tb_ticks_per_sec; /* Timebase tics / sec */
> + __u32 dcache_block_size; /* L1 d-cache block size */
> + __u32 icache_block_size; /* L1 i-cache block size */
> + __u32 dcache_log_block_size; /* L1 d-cache log block size */
> + __u32 icache_log_block_size; /* L1 i-cache log block size */
> + __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
> + __u32 compat_syscall_map[SYSCALL_MAP_SIZE]; /* Map of compat syscalls */
> +};
> +
> +#else /* CONFIG_PPC64 */
> +
> +struct vdso_arch_data {
> + __u64 tb_ticks_per_sec; /* Timebase tics / sec */
> + __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
> + __u32 compat_syscall_map[0]; /* No compat syscalls on PPC32 */
> +};
> +
> +#endif /* CONFIG_PPC64 */
> +
> +#endif /* _ASM_POWERPC_VDSO_ARCH_DATA_H */
> diff --git a/arch/powerpc/include/asm/vdso/getrandom.h b/arch/powerpc/include/asm/vdso/getrandom.h
> index 80ce0709725eb89c1f3b69e0733038b458fbf24f..067a5396aac6e99c1a96f730459ec684bc7785d7 100644
> --- a/arch/powerpc/include/asm/vdso/getrandom.h
> +++ b/arch/powerpc/include/asm/vdso/getrandom.h
> @@ -43,20 +43,21 @@ static __always_inline ssize_t getrandom_syscall(void *buffer, size_t len, unsig
> (unsigned long)len, (unsigned long)flags);
> }
>
> -static __always_inline struct vdso_rng_data *__arch_get_vdso_rng_data(void)
> +static __always_inline const struct vdso_rng_data *__arch_get_vdso_u_rng_data(void)
> {
> - struct vdso_arch_data *data;
> + struct vdso_rng_data *data;
>
> asm (
> " bcl 20, 31, .+4 ;"
> "0: mflr %0 ;"
> - " addis %0, %0, (_vdso_datapage - 0b)@ha ;"
> - " addi %0, %0, (_vdso_datapage - 0b)@l ;"
> + " addis %0, %0, (vdso_u_rng_data - 0b)@ha ;"
> + " addi %0, %0, (vdso_u_rng_data - 0b)@l ;"
> : "=r" (data) : : "lr"
> );
>
> - return &data->rng_data;
> + return data;
> }
> +#define __arch_get_vdso_u_rng_data __arch_get_vdso_u_rng_data
>
> ssize_t __c_kernel_getrandom(void *buffer, size_t len, unsigned int flags, void *opaque_state,
> size_t opaque_len);
> diff --git a/arch/powerpc/include/asm/vdso/gettimeofday.h b/arch/powerpc/include/asm/vdso/gettimeofday.h
> index c6390890a60c2fdcb608bf321b2945c3fb372f54..dc955f2e0cc51f44d46f488a292aa0dbee3dc16c 100644
> --- a/arch/powerpc/include/asm/vdso/gettimeofday.h
> +++ b/arch/powerpc/include/asm/vdso/gettimeofday.h
> @@ -94,22 +94,12 @@ int clock_getres32_fallback(clockid_t _clkid, struct old_timespec32 *_ts)
> #endif
>
> static __always_inline u64 __arch_get_hw_counter(s32 clock_mode,
> - const struct vdso_data *vd)
> + const struct vdso_time_data *vd)
> {
> return get_tb();
> }
>
> -const struct vdso_data *__arch_get_vdso_data(void);
> -
> -#ifdef CONFIG_TIME_NS
> -static __always_inline
> -const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
> -{
> - return (void *)vd + (1U << CONFIG_PAGE_SHIFT);
> -}
> -#endif
> -
> -static inline bool vdso_clocksource_ok(const struct vdso_data *vd)
> +static inline bool vdso_clocksource_ok(const struct vdso_time_data *vd)
> {
> return true;
> }
> @@ -135,21 +125,22 @@ static __always_inline u64 vdso_shift_ns(u64 ns, unsigned long shift)
>
> #ifdef __powerpc64__
> int __c_kernel_clock_gettime(clockid_t clock, struct __kernel_timespec *ts,
> - const struct vdso_data *vd);
> + const struct vdso_time_data *vd);
> int __c_kernel_clock_getres(clockid_t clock_id, struct __kernel_timespec *res,
> - const struct vdso_data *vd);
> + const struct vdso_time_data *vd);
> #else
> int __c_kernel_clock_gettime(clockid_t clock, struct old_timespec32 *ts,
> - const struct vdso_data *vd);
> + const struct vdso_time_data *vd);
> int __c_kernel_clock_gettime64(clockid_t clock, struct __kernel_timespec *ts,
> - const struct vdso_data *vd);
> + const struct vdso_time_data *vd);
> int __c_kernel_clock_getres(clockid_t clock_id, struct old_timespec32 *res,
> - const struct vdso_data *vd);
> + const struct vdso_time_data *vd);
> #endif
> int __c_kernel_gettimeofday(struct __kernel_old_timeval *tv, struct timezone *tz,
> - const struct vdso_data *vd);
> + const struct vdso_time_data *vd);
> __kernel_old_time_t __c_kernel_time(__kernel_old_time_t *time,
> - const struct vdso_data *vd);
> + const struct vdso_time_data *vd);
> +
> #endif /* __ASSEMBLY__ */
>
> #endif /* _ASM_POWERPC_VDSO_GETTIMEOFDAY_H */
> diff --git a/arch/powerpc/include/asm/vdso/vsyscall.h b/arch/powerpc/include/asm/vdso/vsyscall.h
> index 48560a11955956b8fbb59360334a81972723bd57..c2c9ae1b22e71a3f87e5a1a351699c7ab42b2f95 100644
> --- a/arch/powerpc/include/asm/vdso/vsyscall.h
> +++ b/arch/powerpc/include/asm/vdso/vsyscall.h
> @@ -6,19 +6,6 @@
>
> #include <asm/vdso_datapage.h>
>
> -static __always_inline
> -struct vdso_data *__arch_get_k_vdso_data(void)
> -{
> - return vdso_data->data;
> -}
> -#define __arch_get_k_vdso_data __arch_get_k_vdso_data
> -
> -static __always_inline
> -struct vdso_rng_data *__arch_get_k_vdso_rng_data(void)
> -{
> - return &vdso_data->rng_data;
> -}
> -
> /* The asm-generic header needs to be included after the definitions above */
> #include <asm-generic/vdso/vsyscall.h>
>
> diff --git a/arch/powerpc/include/asm/vdso_datapage.h b/arch/powerpc/include/asm/vdso_datapage.h
> index a202f5b63479533a7f45a74df015feb59f3d7c87..95d45a50355d269454dd3e175a5b3844181536b5 100644
> --- a/arch/powerpc/include/asm/vdso_datapage.h
> +++ b/arch/powerpc/include/asm/vdso_datapage.h
> @@ -11,56 +11,18 @@
>
> #ifndef __ASSEMBLY__
>
> -#include <linux/unistd.h>
> -#include <linux/time.h>
> #include <vdso/datapage.h>
>
> -#define SYSCALL_MAP_SIZE ((NR_syscalls + 31) / 32)
> -
> -#ifdef CONFIG_PPC64
> -
> -struct vdso_arch_data {
> - __u64 tb_ticks_per_sec; /* Timebase tics / sec */
> - __u32 dcache_block_size; /* L1 d-cache block size */
> - __u32 icache_block_size; /* L1 i-cache block size */
> - __u32 dcache_log_block_size; /* L1 d-cache log block size */
> - __u32 icache_log_block_size; /* L1 i-cache log block size */
> - __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
> - __u32 compat_syscall_map[SYSCALL_MAP_SIZE]; /* Map of compat syscalls */
> -
> - struct vdso_rng_data rng_data;
> -
> - struct vdso_data data[CS_BASES] __aligned(1 << CONFIG_PAGE_SHIFT);
> -};
> -
> -#else /* CONFIG_PPC64 */
> -
> -struct vdso_arch_data {
> - __u64 tb_ticks_per_sec; /* Timebase tics / sec */
> - __u32 syscall_map[SYSCALL_MAP_SIZE]; /* Map of syscalls */
> - __u32 compat_syscall_map[0]; /* No compat syscalls on PPC32 */
> - struct vdso_rng_data rng_data;
> -
> - struct vdso_data data[CS_BASES] __aligned(1 << CONFIG_PAGE_SHIFT);
> -};
> -
> -#endif /* CONFIG_PPC64 */
> -
> -extern struct vdso_arch_data *vdso_data;
> -
> #else /* __ASSEMBLY__ */
>
> -.macro get_datapage ptr offset=0
> +.macro get_datapage ptr symbol
> bcl 20, 31, .+4
> 999:
> mflr \ptr
> - addis \ptr, \ptr, (_vdso_datapage - 999b + \offset)@ha
> - addi \ptr, \ptr, (_vdso_datapage - 999b + \offset)@l
> + addis \ptr, \ptr, (\symbol - 999b)@ha
> + addi \ptr, \ptr, (\symbol - 999b)@l
> .endm
>
> -#include <asm/asm-offsets.h>
> -#include <asm/page.h>
> -
> #endif /* __ASSEMBLY__ */
>
> #endif /* __KERNEL__ */
> diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
> index 7a390bd4f4af3c7408b3e3c5ef6d43b95b3b6463..b3048f6d3822c0c457f4aa2ccb5dc870495ba79b 100644
> --- a/arch/powerpc/kernel/asm-offsets.c
> +++ b/arch/powerpc/kernel/asm-offsets.c
> @@ -334,7 +334,6 @@ int main(void)
> #endif /* ! CONFIG_PPC64 */
>
> /* datapage offsets for use by vdso */
> - OFFSET(VDSO_DATA_OFFSET, vdso_arch_data, data);
> OFFSET(CFG_TB_TICKS_PER_SEC, vdso_arch_data, tb_ticks_per_sec);
> #ifdef CONFIG_PPC64
> OFFSET(CFG_ICACHE_BLOCKSZ, vdso_arch_data, icache_block_size);
> diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
> index 0727332ad86fbcfcf8ca18b344ba04381e827c79..15784c5c95c77f1eccfa948a36ba69386a2c175b 100644
> --- a/arch/powerpc/kernel/time.c
> +++ b/arch/powerpc/kernel/time.c
> @@ -950,7 +950,7 @@ void __init time_init(void)
> sys_tz.tz_dsttime = 0;
> }
>
> - vdso_data->tb_ticks_per_sec = tb_ticks_per_sec;
> + vdso_k_arch_data->tb_ticks_per_sec = tb_ticks_per_sec;
> #ifdef CONFIG_PPC64_PROC_SYSTEMCFG
> systemcfg->tb_ticks_per_sec = tb_ticks_per_sec;
> #endif
> diff --git a/arch/powerpc/kernel/vdso.c b/arch/powerpc/kernel/vdso.c
> index 43379365ce1b37cfba662ea58feca5e73dd5f700..219d67bcf747e79f48d09a50f5cb9624bcc0f7b1 100644
> --- a/arch/powerpc/kernel/vdso.c
> +++ b/arch/powerpc/kernel/vdso.c
> @@ -17,7 +17,7 @@
> #include <linux/elf.h>
> #include <linux/security.h>
> #include <linux/syscalls.h>
> -#include <linux/time_namespace.h>
> +#include <linux/vdso_datastore.h>
> #include <vdso/datapage.h>
>
> #include <asm/syscall.h>
> @@ -32,6 +32,8 @@
> #include <asm/vdso_datapage.h>
> #include <asm/setup.h>
>
> +static_assert(__VDSO_PAGES == VDSO_NR_PAGES);
> +
> /* The alignment of the vDSO */
> #define VDSO_ALIGNMENT (1 << 16)
>
> @@ -40,24 +42,6 @@ extern char vdso64_start, vdso64_end;
>
> long sys_ni_syscall(void);
>
> -/*
> - * The vdso data page (aka. systemcfg for old ppc64 fans) is here.
> - * Once the early boot kernel code no longer needs to muck around
> - * with it, it will become dynamically allocated
> - */
> -static union {
> - struct vdso_arch_data data;
> - u8 page[2 * PAGE_SIZE];
> -} vdso_data_store __page_aligned_data;
> -struct vdso_arch_data *vdso_data = &vdso_data_store.data;
> -
> -enum vvar_pages {
> - VVAR_BASE_PAGE_OFFSET,
> - VVAR_TIME_PAGE_OFFSET,
> - VVAR_TIMENS_PAGE_OFFSET,
> - VVAR_NR_PAGES,
> -};
> -
> static int vdso_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma,
> unsigned long text_size)
> {
> @@ -96,14 +80,6 @@ static void vdso_close(const struct vm_special_mapping *sm, struct vm_area_struc
> mm->context.vdso = NULL;
> }
>
> -static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
> - struct vm_area_struct *vma, struct vm_fault *vmf);
> -
> -static struct vm_special_mapping vvar_spec __ro_after_init = {
> - .name = "[vvar]",
> - .fault = vvar_fault,
> -};
> -
> static struct vm_special_mapping vdso32_spec __ro_after_init = {
> .name = "[vdso]",
> .mremap = vdso32_mremap,
> @@ -116,73 +92,6 @@ static struct vm_special_mapping vdso64_spec __ro_after_init = {
> .close = vdso_close,
> };
>
> -#ifdef CONFIG_TIME_NS
> -struct vdso_data *arch_get_vdso_data(void *vvar_page)
> -{
> - return vvar_page;
> -}
> -
> -/*
> - * The vvar mapping contains data for a specific time namespace, so when a task
> - * changes namespace we must unmap its vvar data for the old namespace.
> - * Subsequent faults will map in data for the new namespace.
> - *
> - * For more details see timens_setup_vdso_data().
> - */
> -int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
> -{
> - struct mm_struct *mm = task->mm;
> - VMA_ITERATOR(vmi, mm, 0);
> - struct vm_area_struct *vma;
> -
> - mmap_read_lock(mm);
> - for_each_vma(vmi, vma) {
> - if (vma_is_special_mapping(vma, &vvar_spec))
> - zap_vma_pages(vma);
> - }
> - mmap_read_unlock(mm);
> -
> - return 0;
> -}
> -#endif
> -
> -static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
> - struct vm_area_struct *vma, struct vm_fault *vmf)
> -{
> - struct page *timens_page = find_timens_vvar_page(vma);
> - unsigned long pfn;
> -
> - switch (vmf->pgoff) {
> - case VVAR_BASE_PAGE_OFFSET:
> - pfn = virt_to_pfn(vdso_data);
> - break;
> - case VVAR_TIME_PAGE_OFFSET:
> - if (timens_page)
> - pfn = page_to_pfn(timens_page);
> - else
> - pfn = virt_to_pfn(vdso_data->data);
> - break;
> -#ifdef CONFIG_TIME_NS
> - case VVAR_TIMENS_PAGE_OFFSET:
> - /*
> - * If a task belongs to a time namespace then a namespace
> - * specific VVAR is mapped with the VVAR_DATA_PAGE_OFFSET and
> - * the real VVAR page is mapped with the VVAR_TIMENS_PAGE_OFFSET
> - * offset.
> - * See also the comment near timens_setup_vdso_data().
> - */
> - if (!timens_page)
> - return VM_FAULT_SIGBUS;
> - pfn = virt_to_pfn(vdso_data->data);
> - break;
> -#endif /* CONFIG_TIME_NS */
> - default:
> - return VM_FAULT_SIGBUS;
> - }
> -
> - return vmf_insert_pfn(vma, vmf->address, pfn);
> -}
> -
> /*
> * This is called from binfmt_elf, we create the special vma for the
> * vDSO and insert it into the mm struct tree
> @@ -191,7 +100,7 @@ static int __arch_setup_additional_pages(struct linux_binprm *bprm, int uses_int
> {
> unsigned long vdso_size, vdso_base, mappings_size;
> struct vm_special_mapping *vdso_spec;
> - unsigned long vvar_size = VVAR_NR_PAGES * PAGE_SIZE;
> + unsigned long vvar_size = VDSO_NR_PAGES * PAGE_SIZE;
> struct mm_struct *mm = current->mm;
> struct vm_area_struct *vma;
>
> @@ -217,9 +126,7 @@ static int __arch_setup_additional_pages(struct linux_binprm *bprm, int uses_int
> /* Add required alignment. */
> vdso_base = ALIGN(vdso_base, VDSO_ALIGNMENT);
>
> - vma = _install_special_mapping(mm, vdso_base, vvar_size,
> - VM_READ | VM_MAYREAD | VM_IO |
> - VM_DONTDUMP | VM_PFNMAP, &vvar_spec);
> + vma = vdso_install_vvar_mapping(mm, vdso_base);
> if (IS_ERR(vma))
> return PTR_ERR(vma);
>
> @@ -299,10 +206,10 @@ static void __init vdso_setup_syscall_map(void)
>
> for (i = 0; i < NR_syscalls; i++) {
> if (sys_call_table[i] != (void *)&sys_ni_syscall)
> - vdso_data->syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
> + vdso_k_arch_data->syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
> if (IS_ENABLED(CONFIG_COMPAT) &&
> compat_sys_call_table[i] != (void *)&sys_ni_syscall)
> - vdso_data->compat_syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
> + vdso_k_arch_data->compat_syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f);
> }
> }
>
> @@ -352,10 +259,10 @@ static struct page ** __init vdso_setup_pages(void *start, void *end)
> static int __init vdso_init(void)
> {
> #ifdef CONFIG_PPC64
> - vdso_data->dcache_block_size = ppc64_caches.l1d.block_size;
> - vdso_data->icache_block_size = ppc64_caches.l1i.block_size;
> - vdso_data->dcache_log_block_size = ppc64_caches.l1d.log_block_size;
> - vdso_data->icache_log_block_size = ppc64_caches.l1i.log_block_size;
> + vdso_k_arch_data->dcache_block_size = ppc64_caches.l1d.block_size;
> + vdso_k_arch_data->icache_block_size = ppc64_caches.l1i.block_size;
> + vdso_k_arch_data->dcache_log_block_size = ppc64_caches.l1d.log_block_size;
> + vdso_k_arch_data->icache_log_block_size = ppc64_caches.l1i.log_block_size;
> #endif /* CONFIG_PPC64 */
>
> vdso_setup_syscall_map();
> diff --git a/arch/powerpc/kernel/vdso/cacheflush.S b/arch/powerpc/kernel/vdso/cacheflush.S
> index 0085ae464dac9c32381625a6969a4e422ad34eb7..488d3ade11e64996b30f42777251df8499eda92c 100644
> --- a/arch/powerpc/kernel/vdso/cacheflush.S
> +++ b/arch/powerpc/kernel/vdso/cacheflush.S
> @@ -30,7 +30,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_COHERENT_ICACHE)
> #ifdef CONFIG_PPC64
> mflr r12
> .cfi_register lr,r12
> - get_datapage r10
> + get_datapage r10 vdso_u_arch_data
> mtlr r12
> .cfi_restore lr
> #endif
> diff --git a/arch/powerpc/kernel/vdso/datapage.S b/arch/powerpc/kernel/vdso/datapage.S
> index db8e167f01667eb95b3dc74f6771e610411bba90..d23b2e8e2a34ca9b142231eb3a492716a49b2248 100644
> --- a/arch/powerpc/kernel/vdso/datapage.S
> +++ b/arch/powerpc/kernel/vdso/datapage.S
> @@ -28,7 +28,7 @@ V_FUNCTION_BEGIN(__kernel_get_syscall_map)
> mflr r12
> .cfi_register lr,r12
> mr. r4,r3
> - get_datapage r3
> + get_datapage r3 vdso_u_arch_data
> mtlr r12
> #ifdef __powerpc64__
> addi r3,r3,CFG_SYSCALL_MAP64
> @@ -52,7 +52,7 @@ V_FUNCTION_BEGIN(__kernel_get_tbfreq)
> .cfi_startproc
> mflr r12
> .cfi_register lr,r12
> - get_datapage r3
> + get_datapage r3 vdso_u_arch_data
> #ifndef __powerpc64__
> lwz r4,(CFG_TB_TICKS_PER_SEC + 4)(r3)
> #endif
> diff --git a/arch/powerpc/kernel/vdso/gettimeofday.S b/arch/powerpc/kernel/vdso/gettimeofday.S
> index 5333848322ca6105018d501952e3bf42475f49df..79c967212444732da50805fd086c6f2a3c75b0cc 100644
> --- a/arch/powerpc/kernel/vdso/gettimeofday.S
> +++ b/arch/powerpc/kernel/vdso/gettimeofday.S
> @@ -33,9 +33,9 @@
> .cfi_rel_offset r2, PPC_MIN_STKFRM + STK_GOT
> #endif
> .ifeq \call_time
> - get_datapage r5 VDSO_DATA_OFFSET
> + get_datapage r5 vdso_u_time_data
> .else
> - get_datapage r4 VDSO_DATA_OFFSET
> + get_datapage r4 vdso_u_time_data
> .endif
> bl CFUNC(DOTSYM(\funct))
> PPC_LL r0, PPC_MIN_STKFRM + PPC_LR_STKOFF(r1)
> diff --git a/arch/powerpc/kernel/vdso/vdso32.lds.S b/arch/powerpc/kernel/vdso/vdso32.lds.S
> index 1a1b0b6d681a9977e4ef8042e52d8d33da61887e..72a1012b8a205c6357cecb4b53d2d8e1ff59b051 100644
> --- a/arch/powerpc/kernel/vdso/vdso32.lds.S
> +++ b/arch/powerpc/kernel/vdso/vdso32.lds.S
> @@ -6,6 +6,7 @@
> #include <asm/vdso.h>
> #include <asm/page.h>
> #include <asm-generic/vmlinux.lds.h>
> +#include <vdso/datapage.h>
>
> #ifdef __LITTLE_ENDIAN__
> OUTPUT_FORMAT("elf32-powerpcle", "elf32-powerpcle", "elf32-powerpcle")
> @@ -16,7 +17,8 @@ OUTPUT_ARCH(powerpc:common)
>
> SECTIONS
> {
> - PROVIDE(_vdso_datapage = . - 3 * PAGE_SIZE);
> + VDSO_VVAR_SYMS
> +
> . = SIZEOF_HEADERS;
>
> .hash : { *(.hash) } :text
> diff --git a/arch/powerpc/kernel/vdso/vdso64.lds.S b/arch/powerpc/kernel/vdso/vdso64.lds.S
> index e21b5506cad62b16e677be74fda7921ec917141a..32102a05eaa7e015e0f89e4a94a3c5e31da7d460 100644
> --- a/arch/powerpc/kernel/vdso/vdso64.lds.S
> +++ b/arch/powerpc/kernel/vdso/vdso64.lds.S
> @@ -6,6 +6,7 @@
> #include <asm/vdso.h>
> #include <asm/page.h>
> #include <asm-generic/vmlinux.lds.h>
> +#include <vdso/datapage.h>
>
> #ifdef __LITTLE_ENDIAN__
> OUTPUT_FORMAT("elf64-powerpcle", "elf64-powerpcle", "elf64-powerpcle")
> @@ -16,7 +17,8 @@ OUTPUT_ARCH(powerpc:common64)
>
> SECTIONS
> {
> - PROVIDE(_vdso_datapage = . - 3 * PAGE_SIZE);
> + VDSO_VVAR_SYMS
> +
> . = SIZEOF_HEADERS;
>
> .hash : { *(.hash) } :text
> diff --git a/arch/powerpc/kernel/vdso/vgettimeofday.c b/arch/powerpc/kernel/vdso/vgettimeofday.c
> index 55a287c9a7366aa59ab4af1e760a8995f588a4d5..6f5167d81af5f3e6e755dbda4307769e45a28421 100644
> --- a/arch/powerpc/kernel/vdso/vgettimeofday.c
> +++ b/arch/powerpc/kernel/vdso/vgettimeofday.c
> @@ -7,43 +7,43 @@
>
> #ifdef __powerpc64__
> int __c_kernel_clock_gettime(clockid_t clock, struct __kernel_timespec *ts,
> - const struct vdso_data *vd)
> + const struct vdso_time_data *vd)
> {
> return __cvdso_clock_gettime_data(vd, clock, ts);
> }
>
> int __c_kernel_clock_getres(clockid_t clock_id, struct __kernel_timespec *res,
> - const struct vdso_data *vd)
> + const struct vdso_time_data *vd)
> {
> return __cvdso_clock_getres_data(vd, clock_id, res);
> }
> #else
> int __c_kernel_clock_gettime(clockid_t clock, struct old_timespec32 *ts,
> - const struct vdso_data *vd)
> + const struct vdso_time_data *vd)
> {
> return __cvdso_clock_gettime32_data(vd, clock, ts);
> }
>
> int __c_kernel_clock_gettime64(clockid_t clock, struct __kernel_timespec *ts,
> - const struct vdso_data *vd)
> + const struct vdso_time_data *vd)
> {
> return __cvdso_clock_gettime_data(vd, clock, ts);
> }
>
> int __c_kernel_clock_getres(clockid_t clock_id, struct old_timespec32 *res,
> - const struct vdso_data *vd)
> + const struct vdso_time_data *vd)
> {
> return __cvdso_clock_getres_time32_data(vd, clock_id, res);
> }
> #endif
>
> int __c_kernel_gettimeofday(struct __kernel_old_timeval *tv, struct timezone *tz,
> - const struct vdso_data *vd)
> + const struct vdso_time_data *vd)
> {
> return __cvdso_gettimeofday_data(vd, tv, tz);
> }
>
> -__kernel_old_time_t __c_kernel_time(__kernel_old_time_t *time, const struct vdso_data *vd)
> +__kernel_old_time_t __c_kernel_time(__kernel_old_time_t *time, const struct vdso_time_data *vd)
> {
> return __cvdso_time_data(vd, time);
> }
>
^ permalink raw reply [flat|nested] 29+ messages in thread
* [PATCH v3 15/18] x86/vdso: Switch to generic storage implementation
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (13 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 14/18] powerpc/vdso: " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 16/18] x86/vdso/vdso2c: Remove page handling Thomas Weißschuh
` (3 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The generic storage implementation provides the same features as the
custom one. However it can be shared between architectures, making
maintenance easier.
This switch also moves the random state data out of the time data page.
The currently used hardcoded __VDSO_RND_DATA_OFFSET does not take into
account changes to the time data page layout.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/x86/Kconfig | 1 +
arch/x86/entry/vdso/vdso-layout.lds.S | 14 ++--
arch/x86/entry/vdso/vma.c | 123 ++-----------------------------
arch/x86/include/asm/vdso/getrandom.h | 10 ---
arch/x86/include/asm/vdso/gettimeofday.h | 25 +------
arch/x86/include/asm/vdso/vsyscall.h | 24 +-----
6 files changed, 19 insertions(+), 178 deletions(-)
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 87198d957e2f11324743267b9921f6e18bf28e5c..e66ffd941d17d639a8ca3140a6384c5a800f9a2a 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -178,6 +178,7 @@ config X86
select GENERIC_SMP_IDLE_THREAD
select GENERIC_TIME_VSYSCALL
select GENERIC_GETTIMEOFDAY
+ select GENERIC_VDSO_DATA_STORE
select GENERIC_VDSO_TIME_NS
select GENERIC_VDSO_OVERFLOW_PROTECT
select GUP_GET_PXX_LOW_HIGH if X86_PAE
diff --git a/arch/x86/entry/vdso/vdso-layout.lds.S b/arch/x86/entry/vdso/vdso-layout.lds.S
index 918606ff92a988b14f5e64f984750ae04b3b6ede..e5cecdb0fedf000516bb789d6aca4c952c3035e5 100644
--- a/arch/x86/entry/vdso/vdso-layout.lds.S
+++ b/arch/x86/entry/vdso/vdso-layout.lds.S
@@ -1,6 +1,7 @@
/* SPDX-License-Identifier: GPL-2.0 */
#include <asm/vdso.h>
#include <asm/vdso/vsyscall.h>
+#include <vdso/datapage.h>
/*
* Linker script for vDSO. This is an ELF shared object prelinked to
@@ -17,17 +18,16 @@ SECTIONS
* segment.
*/
- vvar_start = . - __VVAR_PAGES * PAGE_SIZE;
- vvar_page = vvar_start;
+ VDSO_VVAR_SYMS
- vdso_rng_data = vvar_page + __VDSO_RND_DATA_OFFSET;
-
- timens_page = vvar_start + PAGE_SIZE;
-
- vclock_pages = VDSO_VCLOCK_PAGES_START(vvar_start);
+ vclock_pages = VDSO_VCLOCK_PAGES_START(vdso_u_data);
pvclock_page = vclock_pages + VDSO_PAGE_PVCLOCK_OFFSET * PAGE_SIZE;
hvclock_page = vclock_pages + VDSO_PAGE_HVCLOCK_OFFSET * PAGE_SIZE;
+ /* For compatibility with vdso2c */
+ vvar_page = vdso_u_data;
+ vvar_start = vdso_u_data;
+
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
diff --git a/arch/x86/entry/vdso/vma.c b/arch/x86/entry/vdso/vma.c
index aa62949335ecec3765d3b46eac7f7b83be5efdda..7245e95c573e1464d6eb744181d2306f17fc0329 100644
--- a/arch/x86/entry/vdso/vma.c
+++ b/arch/x86/entry/vdso/vma.c
@@ -14,7 +14,7 @@
#include <linux/elf.h>
#include <linux/cpu.h>
#include <linux/ptrace.h>
-#include <linux/time_namespace.h>
+#include <linux/vdso_datastore.h>
#include <asm/pvclock.h>
#include <asm/vgtod.h>
@@ -27,13 +27,7 @@
#include <asm/vdso/vsyscall.h>
#include <clocksource/hyperv_timer.h>
-struct vdso_data *arch_get_vdso_data(void *vvar_page)
-{
- return (struct vdso_data *)vvar_page;
-}
-
-static union vdso_data_store vdso_data_store __page_aligned_data;
-struct vdso_data *vdso_data = vdso_data_store.data;
+static_assert(VDSO_NR_PAGES + VDSO_NR_VCLOCK_PAGES == __VDSO_PAGES);
unsigned int vclocks_used __read_mostly;
@@ -54,7 +48,6 @@ int __init init_vdso_image(const struct vdso_image *image)
return 0;
}
-static const struct vm_special_mapping vvar_mapping;
struct linux_binprm;
static vm_fault_t vdso_fault(const struct vm_special_mapping *sm,
@@ -98,99 +91,6 @@ static int vdso_mremap(const struct vm_special_mapping *sm,
return 0;
}
-#ifdef CONFIG_TIME_NS
-/*
- * The vvar page layout depends on whether a task belongs to the root or
- * non-root time namespace. Whenever a task changes its namespace, the VVAR
- * page tables are cleared and then they will re-faulted with a
- * corresponding layout.
- * See also the comment near timens_setup_vdso_data() for details.
- */
-int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
-{
- struct mm_struct *mm = task->mm;
- struct vm_area_struct *vma;
- VMA_ITERATOR(vmi, mm, 0);
-
- mmap_read_lock(mm);
- for_each_vma(vmi, vma) {
- if (vma_is_special_mapping(vma, &vvar_mapping))
- zap_vma_pages(vma);
- }
- mmap_read_unlock(mm);
-
- return 0;
-}
-#endif
-
-static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
- struct vm_area_struct *vma, struct vm_fault *vmf)
-{
- const struct vdso_image *image = vma->vm_mm->context.vdso_image;
- unsigned long pfn;
- long sym_offset;
-
- if (!image)
- return VM_FAULT_SIGBUS;
-
- sym_offset = (long)(vmf->pgoff << PAGE_SHIFT) +
- image->sym_vvar_start;
-
- /*
- * Sanity check: a symbol offset of zero means that the page
- * does not exist for this vdso image, not that the page is at
- * offset zero relative to the text mapping. This should be
- * impossible here, because sym_offset should only be zero for
- * the page past the end of the vvar mapping.
- */
- if (sym_offset == 0)
- return VM_FAULT_SIGBUS;
-
- if (sym_offset == image->sym_vvar_page) {
- struct page *timens_page = find_timens_vvar_page(vma);
-
- pfn = __pa_symbol(vdso_data) >> PAGE_SHIFT;
-
- /*
- * If a task belongs to a time namespace then a namespace
- * specific VVAR is mapped with the sym_vvar_page offset and
- * the real VVAR page is mapped with the sym_timens_page
- * offset.
- * See also the comment near timens_setup_vdso_data().
- */
- if (timens_page) {
- unsigned long addr;
- vm_fault_t err;
-
- /*
- * Optimization: inside time namespace pre-fault
- * VVAR page too. As on timens page there are only
- * offsets for clocks on VVAR, it'll be faulted
- * shortly by VDSO code.
- */
- addr = vmf->address + (image->sym_timens_page - sym_offset);
- err = vmf_insert_pfn(vma, addr, pfn);
- if (unlikely(err & VM_FAULT_ERROR))
- return err;
-
- pfn = page_to_pfn(timens_page);
- }
-
- return vmf_insert_pfn(vma, vmf->address, pfn);
-
- } else if (sym_offset == image->sym_timens_page) {
- struct page *timens_page = find_timens_vvar_page(vma);
-
- if (!timens_page)
- return VM_FAULT_SIGBUS;
-
- pfn = __pa_symbol(vdso_data) >> PAGE_SHIFT;
- return vmf_insert_pfn(vma, vmf->address, pfn);
- }
-
- return VM_FAULT_SIGBUS;
-}
-
static vm_fault_t vvar_vclock_fault(const struct vm_special_mapping *sm,
struct vm_area_struct *vma, struct vm_fault *vmf)
{
@@ -212,7 +112,6 @@ static vm_fault_t vvar_vclock_fault(const struct vm_special_mapping *sm,
case VDSO_PAGE_HVCLOCK_OFFSET:
{
unsigned long pfn = hv_get_tsc_pfn();
-
if (pfn && vclock_was_used(VDSO_CLOCKMODE_HVCLOCK))
return vmf_insert_pfn(vma, vmf->address, pfn);
break;
@@ -228,10 +127,6 @@ static const struct vm_special_mapping vdso_mapping = {
.fault = vdso_fault,
.mremap = vdso_mremap,
};
-static const struct vm_special_mapping vvar_mapping = {
- .name = "[vvar]",
- .fault = vvar_fault,
-};
static const struct vm_special_mapping vvar_vclock_mapping = {
.name = "[vvar_vclock]",
.fault = vvar_vclock_fault,
@@ -253,13 +148,13 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
return -EINTR;
addr = get_unmapped_area(NULL, addr,
- image->size - image->sym_vvar_start, 0, 0);
+ image->size + __VDSO_PAGES * PAGE_SIZE, 0, 0);
if (IS_ERR_VALUE(addr)) {
ret = addr;
goto up_fail;
}
- text_start = addr - image->sym_vvar_start;
+ text_start = addr + __VDSO_PAGES * PAGE_SIZE;
/*
* MAYWRITE to allow gdb to COW and set breakpoints
@@ -276,13 +171,7 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
goto up_fail;
}
- vma = _install_special_mapping(mm,
- addr,
- (__VVAR_PAGES - VDSO_NR_VCLOCK_PAGES) * PAGE_SIZE,
- VM_READ|VM_MAYREAD|VM_IO|VM_DONTDUMP|
- VM_PFNMAP,
- &vvar_mapping);
-
+ vma = vdso_install_vvar_mapping(mm, addr);
if (IS_ERR(vma)) {
ret = PTR_ERR(vma);
do_munmap(mm, text_start, image->size, NULL);
@@ -327,7 +216,7 @@ int map_vdso_once(const struct vdso_image *image, unsigned long addr)
*/
for_each_vma(vmi, vma) {
if (vma_is_special_mapping(vma, &vdso_mapping) ||
- vma_is_special_mapping(vma, &vvar_mapping) ||
+ vma_is_special_mapping(vma, &vdso_vvar_mapping) ||
vma_is_special_mapping(vma, &vvar_vclock_mapping)) {
mmap_write_unlock(mm);
return -EEXIST;
diff --git a/arch/x86/include/asm/vdso/getrandom.h b/arch/x86/include/asm/vdso/getrandom.h
index 2bf9c0e970c3e7d2a2ddfcb1d007cb73da200494..af85539f6557b8e22f7c8fb142c5c83330219af1 100644
--- a/arch/x86/include/asm/vdso/getrandom.h
+++ b/arch/x86/include/asm/vdso/getrandom.h
@@ -27,16 +27,6 @@ static __always_inline ssize_t getrandom_syscall(void *buffer, size_t len, unsig
return ret;
}
-extern struct vdso_rng_data vdso_rng_data
- __attribute__((visibility("hidden")));
-
-static __always_inline const struct vdso_rng_data *__arch_get_vdso_rng_data(void)
-{
- if (IS_ENABLED(CONFIG_TIME_NS) && __arch_get_vdso_data()->clock_mode == VDSO_CLOCKMODE_TIMENS)
- return (void *)&vdso_rng_data + ((void *)&timens_page - (void *)__arch_get_vdso_data());
- return &vdso_rng_data;
-}
-
#endif /* !__ASSEMBLY__ */
#endif /* __ASM_VDSO_GETRANDOM_H */
diff --git a/arch/x86/include/asm/vdso/gettimeofday.h b/arch/x86/include/asm/vdso/gettimeofday.h
index 375a34b0f365792ec108381d7c7229f8351448f7..edec796832e08b73d6d58bda6408957048f4e80e 100644
--- a/arch/x86/include/asm/vdso/gettimeofday.h
+++ b/arch/x86/include/asm/vdso/gettimeofday.h
@@ -19,12 +19,6 @@
#include <asm/pvclock.h>
#include <clocksource/hyperv_timer.h>
-extern struct vdso_data vvar_page
- __attribute__((visibility("hidden")));
-
-extern struct vdso_data timens_page
- __attribute__((visibility("hidden")));
-
#define VDSO_HAS_TIME 1
#define VDSO_HAS_CLOCK_GETRES 1
@@ -59,14 +53,6 @@ extern struct ms_hyperv_tsc_page hvclock_page
__attribute__((visibility("hidden")));
#endif
-#ifdef CONFIG_TIME_NS
-static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
-{
- return &timens_page;
-}
-#endif
-
#ifndef BUILD_VDSO32
static __always_inline
@@ -250,7 +236,7 @@ static u64 vread_hvclock(void)
#endif
static inline u64 __arch_get_hw_counter(s32 clock_mode,
- const struct vdso_data *vd)
+ const struct vdso_time_data *vd)
{
if (likely(clock_mode == VDSO_CLOCKMODE_TSC))
return (u64)rdtsc_ordered() & S64_MAX;
@@ -275,12 +261,7 @@ static inline u64 __arch_get_hw_counter(s32 clock_mode,
return U64_MAX;
}
-static __always_inline const struct vdso_data *__arch_get_vdso_data(void)
-{
- return &vvar_page;
-}
-
-static inline bool arch_vdso_clocksource_ok(const struct vdso_data *vd)
+static inline bool arch_vdso_clocksource_ok(const struct vdso_time_data *vd)
{
return true;
}
@@ -319,7 +300,7 @@ static inline bool arch_vdso_cycles_ok(u64 cycles)
* declares everything with the MSB/Sign-bit set as invalid. Therefore the
* effective mask is S64_MAX.
*/
-static __always_inline u64 vdso_calc_ns(const struct vdso_data *vd, u64 cycles, u64 base)
+static __always_inline u64 vdso_calc_ns(const struct vdso_time_data *vd, u64 cycles, u64 base)
{
u64 delta = cycles - vd->cycle_last;
diff --git a/arch/x86/include/asm/vdso/vsyscall.h b/arch/x86/include/asm/vdso/vsyscall.h
index 88b31d4cdfaf331d2d597981d3f8ee0c5a339085..ebbf63420af03dd87349a390ed22655ebd6b5a54 100644
--- a/arch/x86/include/asm/vdso/vsyscall.h
+++ b/arch/x86/include/asm/vdso/vsyscall.h
@@ -2,11 +2,10 @@
#ifndef __ASM_VDSO_VSYSCALL_H
#define __ASM_VDSO_VSYSCALL_H
-#define __VDSO_RND_DATA_OFFSET 640
-#define __VVAR_PAGES 4
+#define __VDSO_PAGES 6
#define VDSO_NR_VCLOCK_PAGES 2
-#define VDSO_VCLOCK_PAGES_START(_b) ((_b) + (__VVAR_PAGES - VDSO_NR_VCLOCK_PAGES) * PAGE_SIZE)
+#define VDSO_VCLOCK_PAGES_START(_b) ((_b) + (__VDSO_PAGES - VDSO_NR_VCLOCK_PAGES) * PAGE_SIZE)
#define VDSO_PAGE_PVCLOCK_OFFSET 0
#define VDSO_PAGE_HVCLOCK_OFFSET 1
@@ -15,25 +14,6 @@
#include <vdso/datapage.h>
#include <asm/vgtod.h>
-extern struct vdso_data *vdso_data;
-
-/*
- * Update the vDSO data page to keep in sync with kernel timekeeping.
- */
-static __always_inline
-struct vdso_data *__x86_get_k_vdso_data(void)
-{
- return vdso_data;
-}
-#define __arch_get_k_vdso_data __x86_get_k_vdso_data
-
-static __always_inline
-struct vdso_rng_data *__x86_get_k_vdso_rng_data(void)
-{
- return (void *)vdso_data + __VDSO_RND_DATA_OFFSET;
-}
-#define __arch_get_k_vdso_rng_data __x86_get_k_vdso_rng_data
-
/* The asm-generic header needs to be included after the definitions above */
#include <asm-generic/vdso/vsyscall.h>
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 16/18] x86/vdso/vdso2c: Remove page handling
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (14 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 15/18] x86/vdso: " Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 17/18] vdso: Remove remnants of architecture-specific random state storage Thomas Weißschuh
` (2 subsequent siblings)
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
The values are not used anymore.
Also the sanity checks performed by vdso2c can never trigger as they
only validate invariants already enforced by the linker script.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
arch/x86/entry/vdso/vdso-layout.lds.S | 4 ----
arch/x86/entry/vdso/vdso2c.c | 21 ---------------------
arch/x86/entry/vdso/vdso2c.h | 20 --------------------
arch/x86/include/asm/vdso.h | 6 ------
4 files changed, 51 deletions(-)
diff --git a/arch/x86/entry/vdso/vdso-layout.lds.S b/arch/x86/entry/vdso/vdso-layout.lds.S
index e5cecdb0fedf000516bb789d6aca4c952c3035e5..ec1ac191a0578e55dc72ceff7f5b5eb104042881 100644
--- a/arch/x86/entry/vdso/vdso-layout.lds.S
+++ b/arch/x86/entry/vdso/vdso-layout.lds.S
@@ -24,10 +24,6 @@ SECTIONS
pvclock_page = vclock_pages + VDSO_PAGE_PVCLOCK_OFFSET * PAGE_SIZE;
hvclock_page = vclock_pages + VDSO_PAGE_HVCLOCK_OFFSET * PAGE_SIZE;
- /* For compatibility with vdso2c */
- vvar_page = vdso_u_data;
- vvar_start = vdso_u_data;
-
. = SIZEOF_HEADERS;
.hash : { *(.hash) } :text
diff --git a/arch/x86/entry/vdso/vdso2c.c b/arch/x86/entry/vdso/vdso2c.c
index 90d15f2a72055e37f2ef4292096dd780a051bc84..f84e8f8fa5fe6d1bc680895dfef28fe9251a7fe3 100644
--- a/arch/x86/entry/vdso/vdso2c.c
+++ b/arch/x86/entry/vdso/vdso2c.c
@@ -69,33 +69,12 @@
const char *outfilename;
-/* Symbols that we need in vdso2c. */
-enum {
- sym_vvar_start,
- sym_vvar_page,
- sym_pvclock_page,
- sym_hvclock_page,
- sym_timens_page,
-};
-
-const int special_pages[] = {
- sym_vvar_page,
- sym_pvclock_page,
- sym_hvclock_page,
- sym_timens_page,
-};
-
struct vdso_sym {
const char *name;
bool export;
};
struct vdso_sym required_syms[] = {
- [sym_vvar_start] = {"vvar_start", true},
- [sym_vvar_page] = {"vvar_page", true},
- [sym_pvclock_page] = {"pvclock_page", true},
- [sym_hvclock_page] = {"hvclock_page", true},
- [sym_timens_page] = {"timens_page", true},
{"VDSO32_NOTE_MASK", true},
{"__kernel_vsyscall", true},
{"__kernel_sigreturn", true},
diff --git a/arch/x86/entry/vdso/vdso2c.h b/arch/x86/entry/vdso/vdso2c.h
index 67b3e37576a64a29ecbdc6fd75e410fc52a82e58..78ed1c1f28b92b97973f57d3f65b9c4d4694f462 100644
--- a/arch/x86/entry/vdso/vdso2c.h
+++ b/arch/x86/entry/vdso/vdso2c.h
@@ -150,26 +150,6 @@ static void BITSFUNC(go)(void *raw_addr, size_t raw_len,
}
}
- /* Validate mapping addresses. */
- for (i = 0; i < sizeof(special_pages) / sizeof(special_pages[0]); i++) {
- INT_BITS symval = syms[special_pages[i]];
-
- if (!symval)
- continue; /* The mapping isn't used; ignore it. */
-
- if (symval % 4096)
- fail("%s must be a multiple of 4096\n",
- required_syms[i].name);
- if (symval + 4096 < syms[sym_vvar_start])
- fail("%s underruns vvar_start\n",
- required_syms[i].name);
- if (symval + 4096 > 0)
- fail("%s is on the wrong side of the vdso text\n",
- required_syms[i].name);
- }
- if (syms[sym_vvar_start] % 4096)
- fail("vvar_begin must be a multiple of 4096\n");
-
if (!image_name) {
fwrite(stripped_addr, stripped_len, 1, outfile);
return;
diff --git a/arch/x86/include/asm/vdso.h b/arch/x86/include/asm/vdso.h
index d7f6592b74a94f6fecba59ce8f23eb3990843db4..80be0da733df43de7520bef69ad120ce05d37994 100644
--- a/arch/x86/include/asm/vdso.h
+++ b/arch/x86/include/asm/vdso.h
@@ -18,12 +18,6 @@ struct vdso_image {
unsigned long extable_base, extable_len;
const void *extable;
- long sym_vvar_start; /* Negative offset to the vvar area */
-
- long sym_vvar_page;
- long sym_pvclock_page;
- long sym_hvclock_page;
- long sym_timens_page;
long sym_VDSO32_NOTE_MASK;
long sym___kernel_sigreturn;
long sym___kernel_rt_sigreturn;
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 17/18] vdso: Remove remnants of architecture-specific random state storage
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (15 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 16/18] x86/vdso/vdso2c: Remove page handling Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-04 12:05 ` [PATCH v3 18/18] vdso: Remove remnants of architecture-specific time storage Thomas Weißschuh
2025-02-06 9:31 ` [PATCH v3 00/18] vDSO: Introduce generic data storage David Woodhouse
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
All users of the random vDSO are using the generic storage
implementation. Remove the now unnecessary compatibility accessor
functions and symbols.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
include/asm-generic/vdso/vsyscall.h | 5 -----
include/vdso/datapage.h | 1 -
2 files changed, 6 deletions(-)
diff --git a/include/asm-generic/vdso/vsyscall.h b/include/asm-generic/vdso/vsyscall.h
index a5f973e4e2723ef1815c88a7846f7c477e3febd9..13e2ac3736ee9b4aea6800117997ee165a7e2b9d 100644
--- a/include/asm-generic/vdso/vsyscall.h
+++ b/include/asm-generic/vdso/vsyscall.h
@@ -32,11 +32,6 @@ static __always_inline struct vdso_data *__arch_get_k_vdso_data(void)
#define __arch_get_vdso_u_time_data __arch_get_vdso_data
-#ifndef __arch_get_vdso_u_rng_data
-#define __arch_get_vdso_u_rng_data() __arch_get_vdso_rng_data()
-#endif
-#define vdso_k_rng_data __arch_get_k_vdso_rng_data()
-
#endif /* CONFIG_GENERIC_VDSO_DATA_STORE */
#ifndef __arch_update_vsyscall
diff --git a/include/vdso/datapage.h b/include/vdso/datapage.h
index 46658b39c28250e977f5a3454224c3ed0fb4c81d..497907c3aa11fcae913f62ef7373bbe6a1026bd6 100644
--- a/include/vdso/datapage.h
+++ b/include/vdso/datapage.h
@@ -152,7 +152,6 @@ struct vdso_rng_data {
#ifndef CONFIG_GENERIC_VDSO_DATA_STORE
extern struct vdso_time_data _vdso_data[CS_BASES] __attribute__((visibility("hidden")));
extern struct vdso_time_data _timens_data[CS_BASES] __attribute__((visibility("hidden")));
-extern struct vdso_rng_data _vdso_rng_data __attribute__((visibility("hidden")));
#else
extern struct vdso_time_data vdso_u_time_data[CS_BASES] __attribute__((visibility("hidden")));
extern struct vdso_rng_data vdso_u_rng_data __attribute__((visibility("hidden")));
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* [PATCH v3 18/18] vdso: Remove remnants of architecture-specific time storage
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (16 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 17/18] vdso: Remove remnants of architecture-specific random state storage Thomas Weißschuh
@ 2025-02-04 12:05 ` Thomas Weißschuh
2025-02-06 9:31 ` [PATCH v3 00/18] vDSO: Introduce generic data storage David Woodhouse
18 siblings, 0 replies; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-04 12:05 UTC (permalink / raw)
To: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, Thomas Weißschuh, linux-csky
All users of the random vDSO are using the generic storage
implementation. Remove the now unnecessary compatibility accessor
functions and symbols.
Co-developed-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Nam Cao <namcao@linutronix.de>
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
---
include/asm-generic/vdso/vsyscall.h | 20 +++------------
include/linux/time_namespace.h | 3 ---
include/vdso/datapage.h | 19 +-------------
include/vdso/helpers.h | 8 +++---
kernel/time/namespace.c | 12 ++++-----
kernel/time/vsyscall.c | 19 +++++++-------
lib/vdso/datastore.c | 10 +++-----
lib/vdso/gettimeofday.c | 51 ++++++++++++++++++-------------------
8 files changed, 53 insertions(+), 89 deletions(-)
diff --git a/include/asm-generic/vdso/vsyscall.h b/include/asm-generic/vdso/vsyscall.h
index 13e2ac3736ee9b4aea6800117997ee165a7e2b9d..1fb3000f50364feeaaa9348d438b3ab8091bb265 100644
--- a/include/asm-generic/vdso/vsyscall.h
+++ b/include/asm-generic/vdso/vsyscall.h
@@ -20,31 +20,19 @@ static __always_inline const struct vdso_rng_data *__arch_get_vdso_u_rng_data(vo
}
#endif
-#else /* !CONFIG_GENERIC_VDSO_DATA_STORE */
-
-#ifndef __arch_get_k_vdso_data
-static __always_inline struct vdso_data *__arch_get_k_vdso_data(void)
-{
- return NULL;
-}
-#endif /* __arch_get_k_vdso_data */
-#define vdso_k_time_data __arch_get_k_vdso_data()
-
-#define __arch_get_vdso_u_time_data __arch_get_vdso_data
-
#endif /* CONFIG_GENERIC_VDSO_DATA_STORE */
#ifndef __arch_update_vsyscall
-static __always_inline void __arch_update_vsyscall(struct vdso_data *vdata)
+static __always_inline void __arch_update_vsyscall(struct vdso_time_data *vdata)
{
}
#endif /* __arch_update_vsyscall */
-#ifndef __arch_sync_vdso_data
-static __always_inline void __arch_sync_vdso_data(struct vdso_data *vdata)
+#ifndef __arch_sync_vdso_time_data
+static __always_inline void __arch_sync_vdso_time_data(struct vdso_time_data *vdata)
{
}
-#endif /* __arch_sync_vdso_data */
+#endif /* __arch_sync_vdso_time_data */
#endif /* !__ASSEMBLY__ */
diff --git a/include/linux/time_namespace.h b/include/linux/time_namespace.h
index 4b81db223f5450218dfaf553b24195be9ba97c08..0b8b32bf0655109d368c5dacc23ca5efcbb5aa80 100644
--- a/include/linux/time_namespace.h
+++ b/include/linux/time_namespace.h
@@ -8,7 +8,6 @@
#include <linux/ns_common.h>
#include <linux/err.h>
#include <linux/time64.h>
-#include <vdso/datapage.h>
struct user_namespace;
extern struct user_namespace init_user_ns;
@@ -166,6 +165,4 @@ static inline ktime_t timens_ktime_to_host(clockid_t clockid, ktime_t tim)
}
#endif
-struct vdso_data *arch_get_vdso_data(void *vvar_page);
-
#endif /* _LINUX_TIMENS_H */
diff --git a/include/vdso/datapage.h b/include/vdso/datapage.h
index 497907c3aa11fcae913f62ef7373bbe6a1026bd6..ed4fb4c06e3ee6423fe68ccb476565213f234863 100644
--- a/include/vdso/datapage.h
+++ b/include/vdso/datapage.h
@@ -128,8 +128,6 @@ struct vdso_time_data {
struct arch_vdso_time_data arch_data;
};
-#define vdso_data vdso_time_data
-
/**
* struct vdso_rng_data - vdso RNG state information
* @generation: counter representing the number of RNG reseeds
@@ -149,10 +147,7 @@ struct vdso_rng_data {
* With the hidden visibility, the compiler simply generates a PC-relative
* relocation, and this is what we need.
*/
-#ifndef CONFIG_GENERIC_VDSO_DATA_STORE
-extern struct vdso_time_data _vdso_data[CS_BASES] __attribute__((visibility("hidden")));
-extern struct vdso_time_data _timens_data[CS_BASES] __attribute__((visibility("hidden")));
-#else
+#ifdef CONFIG_GENERIC_VDSO_DATA_STORE
extern struct vdso_time_data vdso_u_time_data[CS_BASES] __attribute__((visibility("hidden")));
extern struct vdso_rng_data vdso_u_rng_data __attribute__((visibility("hidden")));
extern struct vdso_arch_data vdso_u_arch_data __attribute__((visibility("hidden")));
@@ -160,17 +155,6 @@ extern struct vdso_arch_data vdso_u_arch_data __attribute__((visibility("hidden"
extern struct vdso_time_data *vdso_k_time_data;
extern struct vdso_rng_data *vdso_k_rng_data;
extern struct vdso_arch_data *vdso_k_arch_data;
-#endif
-
-/**
- * union vdso_data_store - Generic vDSO data page
- */
-union vdso_data_store {
- struct vdso_time_data data[CS_BASES];
- u8 page[1U << CONFIG_PAGE_SHIFT];
-};
-
-#ifdef CONFIG_GENERIC_VDSO_DATA_STORE
#define VDSO_ARCH_DATA_SIZE ALIGN(sizeof(struct vdso_arch_data), PAGE_SIZE)
#define VDSO_ARCH_DATA_PAGES (VDSO_ARCH_DATA_SIZE >> PAGE_SHIFT)
@@ -189,7 +173,6 @@ enum vdso_pages {
/*
* The generic vDSO implementation requires that gettimeofday.h
* provides:
- * - __arch_get_vdso_data(): to get the vdso datapage.
* - __arch_get_hw_counter(): to get the hw counter based on the
* clock_mode.
* - gettimeofday_fallback(): fallback for gettimeofday.
diff --git a/include/vdso/helpers.h b/include/vdso/helpers.h
index 3ddb03bb05cbeefc110adf0e672c2cd68848a0ae..41c3087070c7ab21d7adec04e6cd30c4b32ea221 100644
--- a/include/vdso/helpers.h
+++ b/include/vdso/helpers.h
@@ -7,7 +7,7 @@
#include <asm/barrier.h>
#include <vdso/datapage.h>
-static __always_inline u32 vdso_read_begin(const struct vdso_data *vd)
+static __always_inline u32 vdso_read_begin(const struct vdso_time_data *vd)
{
u32 seq;
@@ -18,7 +18,7 @@ static __always_inline u32 vdso_read_begin(const struct vdso_data *vd)
return seq;
}
-static __always_inline u32 vdso_read_retry(const struct vdso_data *vd,
+static __always_inline u32 vdso_read_retry(const struct vdso_time_data *vd,
u32 start)
{
u32 seq;
@@ -28,7 +28,7 @@ static __always_inline u32 vdso_read_retry(const struct vdso_data *vd,
return seq != start;
}
-static __always_inline void vdso_write_begin(struct vdso_data *vd)
+static __always_inline void vdso_write_begin(struct vdso_time_data *vd)
{
/*
* WRITE_ONCE() is required otherwise the compiler can validly tear
@@ -40,7 +40,7 @@ static __always_inline void vdso_write_begin(struct vdso_data *vd)
smp_wmb();
}
-static __always_inline void vdso_write_end(struct vdso_data *vd)
+static __always_inline void vdso_write_end(struct vdso_time_data *vd)
{
smp_wmb();
/*
diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c
index 0775b9ec952af9639480db9480d06d12293d1ceb..12f55aa539adbc11cce4055f519dbeca8a73320c 100644
--- a/kernel/time/namespace.c
+++ b/kernel/time/namespace.c
@@ -165,18 +165,18 @@ static struct timens_offset offset_from_ts(struct timespec64 off)
* HVCLOCK
* VVAR
*
- * The check for vdso_data->clock_mode is in the unlikely path of
+ * The check for vdso_time_data->clock_mode is in the unlikely path of
* the seq begin magic. So for the non-timens case most of the time
* 'seq' is even, so the branch is not taken.
*
* If 'seq' is odd, i.e. a concurrent update is in progress, the extra check
- * for vdso_data->clock_mode is a non-issue. The task is spin waiting for the
+ * for vdso_time_data->clock_mode is a non-issue. The task is spin waiting for the
* update to finish and for 'seq' to become even anyway.
*
- * Timens page has vdso_data->clock_mode set to VDSO_CLOCKMODE_TIMENS which
+ * Timens page has vdso_time_data->clock_mode set to VDSO_CLOCKMODE_TIMENS which
* enforces the time namespace handling path.
*/
-static void timens_setup_vdso_data(struct vdso_data *vdata,
+static void timens_setup_vdso_data(struct vdso_time_data *vdata,
struct time_namespace *ns)
{
struct timens_offset *offset = vdata->offset;
@@ -219,7 +219,7 @@ static DEFINE_MUTEX(offset_lock);
static void timens_set_vvar_page(struct task_struct *task,
struct time_namespace *ns)
{
- struct vdso_data *vdata;
+ struct vdso_time_data *vdata;
unsigned int i;
if (ns == &init_time_ns)
@@ -235,7 +235,7 @@ static void timens_set_vvar_page(struct task_struct *task,
goto out;
ns->frozen_offsets = true;
- vdata = arch_get_vdso_data(page_address(ns->vvar_page));
+ vdata = page_address(ns->vvar_page);
for (i = 0; i < CS_BASES; i++)
timens_setup_vdso_data(&vdata[i], ns);
diff --git a/kernel/time/vsyscall.c b/kernel/time/vsyscall.c
index 09c1e39a6dd8e24aa161982c04078e78d52fa737..418192296ef7dd3c1772d50f129e7838883cf00c 100644
--- a/kernel/time/vsyscall.c
+++ b/kernel/time/vsyscall.c
@@ -15,8 +15,7 @@
#include "timekeeping_internal.h"
-static inline void update_vdso_data(struct vdso_data *vdata,
- struct timekeeper *tk)
+static inline void update_vdso_time_data(struct vdso_time_data *vdata, struct timekeeper *tk)
{
struct vdso_timestamp *vdso_ts;
u64 nsec, sec;
@@ -77,7 +76,7 @@ static inline void update_vdso_data(struct vdso_data *vdata,
void update_vsyscall(struct timekeeper *tk)
{
- struct vdso_data *vdata = vdso_k_time_data;
+ struct vdso_time_data *vdata = vdso_k_time_data;
struct vdso_timestamp *vdso_ts;
s32 clock_mode;
u64 nsec;
@@ -117,23 +116,23 @@ void update_vsyscall(struct timekeeper *tk)
* update of the high resolution parts.
*/
if (clock_mode != VDSO_CLOCKMODE_NONE)
- update_vdso_data(vdata, tk);
+ update_vdso_time_data(vdata, tk);
__arch_update_vsyscall(vdata);
vdso_write_end(vdata);
- __arch_sync_vdso_data(vdata);
+ __arch_sync_vdso_time_data(vdata);
}
void update_vsyscall_tz(void)
{
- struct vdso_data *vdata = vdso_k_time_data;
+ struct vdso_time_data *vdata = vdso_k_time_data;
vdata[CS_HRES_COARSE].tz_minuteswest = sys_tz.tz_minuteswest;
vdata[CS_HRES_COARSE].tz_dsttime = sys_tz.tz_dsttime;
- __arch_sync_vdso_data(vdata);
+ __arch_sync_vdso_time_data(vdata);
}
/**
@@ -150,7 +149,7 @@ void update_vsyscall_tz(void)
*/
unsigned long vdso_update_begin(void)
{
- struct vdso_data *vdata = vdso_k_time_data;
+ struct vdso_time_data *vdata = vdso_k_time_data;
unsigned long flags = timekeeper_lock_irqsave();
vdso_write_begin(vdata);
@@ -167,9 +166,9 @@ unsigned long vdso_update_begin(void)
*/
void vdso_update_end(unsigned long flags)
{
- struct vdso_data *vdata = vdso_k_time_data;
+ struct vdso_time_data *vdata = vdso_k_time_data;
vdso_write_end(vdata);
- __arch_sync_vdso_data(vdata);
+ __arch_sync_vdso_time_data(vdata);
timekeeper_unlock_irqrestore(flags);
}
diff --git a/lib/vdso/datastore.c b/lib/vdso/datastore.c
index 0959d62d78586ef458eb5f7c1e152f6b5d4cbf85..e227fbbcb79694f9a40606ac864f52cf1fdbfcf4 100644
--- a/lib/vdso/datastore.c
+++ b/lib/vdso/datastore.c
@@ -12,7 +12,10 @@
* The vDSO data page.
*/
#ifdef CONFIG_HAVE_GENERIC_VDSO
-static union vdso_data_store vdso_time_data_store __page_aligned_data;
+static union {
+ struct vdso_time_data data[CS_BASES];
+ u8 page[PAGE_SIZE];
+} vdso_time_data_store __page_aligned_data;
struct vdso_time_data *vdso_k_time_data = vdso_time_data_store.data;
static_assert(sizeof(vdso_time_data_store) == PAGE_SIZE);
#endif /* CONFIG_HAVE_GENERIC_VDSO */
@@ -123,9 +126,4 @@ int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
return 0;
}
-
-struct vdso_time_data *arch_get_vdso_data(void *vvar_page)
-{
- return (struct vdso_time_data *)vvar_page;
-}
#endif
diff --git a/lib/vdso/gettimeofday.c b/lib/vdso/gettimeofday.c
index 20c5b8752fcc81d9a044169b9a3ae534d840ef91..299f027116ee0e50a69c5a8a17218004e4af0ea1 100644
--- a/lib/vdso/gettimeofday.c
+++ b/lib/vdso/gettimeofday.c
@@ -17,12 +17,12 @@
#endif
#ifdef CONFIG_GENERIC_VDSO_OVERFLOW_PROTECT
-static __always_inline bool vdso_delta_ok(const struct vdso_data *vd, u64 delta)
+static __always_inline bool vdso_delta_ok(const struct vdso_time_data *vd, u64 delta)
{
return delta < vd->max_cycles;
}
#else
-static __always_inline bool vdso_delta_ok(const struct vdso_data *vd, u64 delta)
+static __always_inline bool vdso_delta_ok(const struct vdso_time_data *vd, u64 delta)
{
return true;
}
@@ -39,7 +39,7 @@ static __always_inline u64 vdso_shift_ns(u64 ns, u32 shift)
* Default implementation which works for all sane clocksources. That
* obviously excludes x86/TSC.
*/
-static __always_inline u64 vdso_calc_ns(const struct vdso_data *vd, u64 cycles, u64 base)
+static __always_inline u64 vdso_calc_ns(const struct vdso_time_data *vd, u64 cycles, u64 base)
{
u64 delta = (cycles - vd->cycle_last) & VDSO_DELTA_MASK(vd);
@@ -58,7 +58,7 @@ static inline bool __arch_vdso_hres_capable(void)
#endif
#ifndef vdso_clocksource_ok
-static inline bool vdso_clocksource_ok(const struct vdso_data *vd)
+static inline bool vdso_clocksource_ok(const struct vdso_time_data *vd)
{
return vd->clock_mode != VDSO_CLOCKMODE_NONE;
}
@@ -79,21 +79,20 @@ const struct vdso_time_data *__arch_get_vdso_u_timens_data(const struct vdso_tim
{
return (void *)vd + PAGE_SIZE;
}
-#define __arch_get_timens_vdso_data(vd) __arch_get_vdso_u_timens_data(vd)
#endif /* CONFIG_GENERIC_VDSO_DATA_STORE */
-static __always_inline int do_hres_timens(const struct vdso_data *vdns, clockid_t clk,
+static __always_inline int do_hres_timens(const struct vdso_time_data *vdns, clockid_t clk,
struct __kernel_timespec *ts)
{
const struct timens_offset *offs = &vdns->offset[clk];
const struct vdso_timestamp *vdso_ts;
- const struct vdso_data *vd;
+ const struct vdso_time_data *vd;
u64 cycles, ns;
u32 seq;
s64 sec;
vd = vdns - (clk == CLOCK_MONOTONIC_RAW ? CS_RAW : CS_HRES_COARSE);
- vd = __arch_get_timens_vdso_data(vd);
+ vd = __arch_get_vdso_u_timens_data(vd);
if (clk != CLOCK_MONOTONIC_RAW)
vd = &vd[CS_HRES_COARSE];
else
@@ -128,19 +127,19 @@ static __always_inline int do_hres_timens(const struct vdso_data *vdns, clockid_
}
#else
static __always_inline
-const struct vdso_data *__arch_get_timens_vdso_data(const struct vdso_data *vd)
+const struct vdso_time_data *__arch_get_vdso_u_timens_data(const struct vdso_time_data *vd)
{
return NULL;
}
-static __always_inline int do_hres_timens(const struct vdso_data *vdns, clockid_t clk,
+static __always_inline int do_hres_timens(const struct vdso_time_data *vdns, clockid_t clk,
struct __kernel_timespec *ts)
{
return -EINVAL;
}
#endif
-static __always_inline int do_hres(const struct vdso_data *vd, clockid_t clk,
+static __always_inline int do_hres(const struct vdso_time_data *vd, clockid_t clk,
struct __kernel_timespec *ts)
{
const struct vdso_timestamp *vdso_ts = &vd->basetime[clk];
@@ -192,10 +191,10 @@ static __always_inline int do_hres(const struct vdso_data *vd, clockid_t clk,
}
#ifdef CONFIG_TIME_NS
-static __always_inline int do_coarse_timens(const struct vdso_data *vdns, clockid_t clk,
+static __always_inline int do_coarse_timens(const struct vdso_time_data *vdns, clockid_t clk,
struct __kernel_timespec *ts)
{
- const struct vdso_data *vd = __arch_get_timens_vdso_data(vdns);
+ const struct vdso_time_data *vd = __arch_get_vdso_u_timens_data(vdns);
const struct vdso_timestamp *vdso_ts = &vd->basetime[clk];
const struct timens_offset *offs = &vdns->offset[clk];
u64 nsec;
@@ -221,14 +220,14 @@ static __always_inline int do_coarse_timens(const struct vdso_data *vdns, clocki
return 0;
}
#else
-static __always_inline int do_coarse_timens(const struct vdso_data *vdns, clockid_t clk,
+static __always_inline int do_coarse_timens(const struct vdso_time_data *vdns, clockid_t clk,
struct __kernel_timespec *ts)
{
return -1;
}
#endif
-static __always_inline int do_coarse(const struct vdso_data *vd, clockid_t clk,
+static __always_inline int do_coarse(const struct vdso_time_data *vd, clockid_t clk,
struct __kernel_timespec *ts)
{
const struct vdso_timestamp *vdso_ts = &vd->basetime[clk];
@@ -255,7 +254,7 @@ static __always_inline int do_coarse(const struct vdso_data *vd, clockid_t clk,
}
static __always_inline int
-__cvdso_clock_gettime_common(const struct vdso_data *vd, clockid_t clock,
+__cvdso_clock_gettime_common(const struct vdso_time_data *vd, clockid_t clock,
struct __kernel_timespec *ts)
{
u32 msk;
@@ -282,7 +281,7 @@ __cvdso_clock_gettime_common(const struct vdso_data *vd, clockid_t clock,
}
static __maybe_unused int
-__cvdso_clock_gettime_data(const struct vdso_data *vd, clockid_t clock,
+__cvdso_clock_gettime_data(const struct vdso_time_data *vd, clockid_t clock,
struct __kernel_timespec *ts)
{
int ret = __cvdso_clock_gettime_common(vd, clock, ts);
@@ -300,7 +299,7 @@ __cvdso_clock_gettime(clockid_t clock, struct __kernel_timespec *ts)
#ifdef BUILD_VDSO32
static __maybe_unused int
-__cvdso_clock_gettime32_data(const struct vdso_data *vd, clockid_t clock,
+__cvdso_clock_gettime32_data(const struct vdso_time_data *vd, clockid_t clock,
struct old_timespec32 *res)
{
struct __kernel_timespec ts;
@@ -326,7 +325,7 @@ __cvdso_clock_gettime32(clockid_t clock, struct old_timespec32 *res)
#endif /* BUILD_VDSO32 */
static __maybe_unused int
-__cvdso_gettimeofday_data(const struct vdso_data *vd,
+__cvdso_gettimeofday_data(const struct vdso_time_data *vd,
struct __kernel_old_timeval *tv, struct timezone *tz)
{
@@ -343,7 +342,7 @@ __cvdso_gettimeofday_data(const struct vdso_data *vd,
if (unlikely(tz != NULL)) {
if (IS_ENABLED(CONFIG_TIME_NS) &&
vd->clock_mode == VDSO_CLOCKMODE_TIMENS)
- vd = __arch_get_timens_vdso_data(vd);
+ vd = __arch_get_vdso_u_timens_data(vd);
tz->tz_minuteswest = vd[CS_HRES_COARSE].tz_minuteswest;
tz->tz_dsttime = vd[CS_HRES_COARSE].tz_dsttime;
@@ -360,13 +359,13 @@ __cvdso_gettimeofday(struct __kernel_old_timeval *tv, struct timezone *tz)
#ifdef VDSO_HAS_TIME
static __maybe_unused __kernel_old_time_t
-__cvdso_time_data(const struct vdso_data *vd, __kernel_old_time_t *time)
+__cvdso_time_data(const struct vdso_time_data *vd, __kernel_old_time_t *time)
{
__kernel_old_time_t t;
if (IS_ENABLED(CONFIG_TIME_NS) &&
vd->clock_mode == VDSO_CLOCKMODE_TIMENS)
- vd = __arch_get_timens_vdso_data(vd);
+ vd = __arch_get_vdso_u_timens_data(vd);
t = READ_ONCE(vd[CS_HRES_COARSE].basetime[CLOCK_REALTIME].sec);
@@ -384,7 +383,7 @@ static __maybe_unused __kernel_old_time_t __cvdso_time(__kernel_old_time_t *time
#ifdef VDSO_HAS_CLOCK_GETRES
static __maybe_unused
-int __cvdso_clock_getres_common(const struct vdso_data *vd, clockid_t clock,
+int __cvdso_clock_getres_common(const struct vdso_time_data *vd, clockid_t clock,
struct __kernel_timespec *res)
{
u32 msk;
@@ -396,7 +395,7 @@ int __cvdso_clock_getres_common(const struct vdso_data *vd, clockid_t clock,
if (IS_ENABLED(CONFIG_TIME_NS) &&
vd->clock_mode == VDSO_CLOCKMODE_TIMENS)
- vd = __arch_get_timens_vdso_data(vd);
+ vd = __arch_get_vdso_u_timens_data(vd);
/*
* Convert the clockid to a bitmask and use it to check which
@@ -425,7 +424,7 @@ int __cvdso_clock_getres_common(const struct vdso_data *vd, clockid_t clock,
}
static __maybe_unused
-int __cvdso_clock_getres_data(const struct vdso_data *vd, clockid_t clock,
+int __cvdso_clock_getres_data(const struct vdso_time_data *vd, clockid_t clock,
struct __kernel_timespec *res)
{
int ret = __cvdso_clock_getres_common(vd, clock, res);
@@ -443,7 +442,7 @@ int __cvdso_clock_getres(clockid_t clock, struct __kernel_timespec *res)
#ifdef BUILD_VDSO32
static __maybe_unused int
-__cvdso_clock_getres_time32_data(const struct vdso_data *vd, clockid_t clock,
+__cvdso_clock_getres_time32_data(const struct vdso_time_data *vd, clockid_t clock,
struct old_timespec32 *res)
{
struct __kernel_timespec ts;
--
2.48.1
^ permalink raw reply related [flat|nested] 29+ messages in thread* Re: [PATCH v3 00/18] vDSO: Introduce generic data storage
2025-02-04 12:05 [PATCH v3 00/18] vDSO: Introduce generic data storage Thomas Weißschuh
` (17 preceding siblings ...)
2025-02-04 12:05 ` [PATCH v3 18/18] vdso: Remove remnants of architecture-specific time storage Thomas Weißschuh
@ 2025-02-06 9:31 ` David Woodhouse
2025-02-06 10:59 ` Thomas Weißschuh
2025-02-14 11:34 ` Thomas Gleixner
18 siblings, 2 replies; 29+ messages in thread
From: David Woodhouse @ 2025-02-06 9:31 UTC (permalink / raw)
To: Thomas Weißschuh, James E.J. Bottomley, Helge Deller,
Andy Lutomirski, Thomas Gleixner, Vincenzo Frascino,
Anna-Maria Behnsen, Frederic Weisbecker, Andrew Morton,
Catalin Marinas, Will Deacon, Theodore Ts'o,
Jason A. Donenfeld, Paul Walmsley, Palmer Dabbelt, Albert Ou,
Huacai Chen, WANG Xuerui, Russell King, Heiko Carstens,
Vasily Gorbik, Alexander Gordeev, Christian Borntraeger,
Sven Schnelle, Thomas Bogendoerfer, Michael Ellerman,
Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, linux-csky, Ridoux, Julien, Luu, Ryan, kvm
[-- Attachment #1: Type: text/plain, Size: 1724 bytes --]
On Tue, 2025-02-04 at 13:05 +0100, Thomas Weißschuh wrote:
> Currently each architecture defines the setup of the vDSO data page on
> its own, mostly through copy-and-paste from some other architecture.
> Extend the existing generic vDSO implementation to also provide generic
> data storage.
> This removes duplicated code and paves the way for further changes to
> the generic vDSO implementation without having to go through a lot of
> per-architecture changes.
>
> Based on v6.14-rc1 and intended to be merged through the tip tree.
Thanks for working on this. Is there a plan to expose the time data
directly to userspace in a form which is usable *other* than by
function calls which get the value of the clock at a given moment?
For populating the vmclock device¹ we need to know the actual
relationship between the hardware counter (TSC, arch timer, etc.) and
real time in order to propagate that to the guest.
I see two options for doing this:
1. Via userspace, exposing the vdso time data (and a notification when
it changes?) and letting the userspace VMM populate the vmclock.
This is complex for x86 because of TSC scaling; in fact userspace
doesn't currently know the precise scaling from host to guest TSC
so we'd have to be able to extract that from KVM.
2. In kernel, asking KVM to populate the vmclock structure much like
it does other pvclocks shared with the guest. KVM/x86 already uses
pvclock_gtod_register_notifier() to hook changes; should we expand
on that? The problem with that notifier is that it seems to be
called far more frequently than I'd expect.
¹ https://gitlab.com/qemu-project/qemu/-/commit/3634039b93cc5
[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5069 bytes --]
^ permalink raw reply [flat|nested] 29+ messages in thread* Re: [PATCH v3 00/18] vDSO: Introduce generic data storage
2025-02-06 9:31 ` [PATCH v3 00/18] vDSO: Introduce generic data storage David Woodhouse
@ 2025-02-06 10:59 ` Thomas Weißschuh
2025-02-07 10:15 ` David Woodhouse
2025-02-14 11:34 ` Thomas Gleixner
1 sibling, 1 reply; 29+ messages in thread
From: Thomas Weißschuh @ 2025-02-06 10:59 UTC (permalink / raw)
To: David Woodhouse
Cc: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren, linux-parisc,
linux-kernel, linux-arm-kernel, linux-riscv, loongarch,
linux-s390, linux-mips, linuxppc-dev, linux-arch, Nam Cao,
linux-csky, Ridoux, Julien, Luu, Ryan, kvm
On Thu, Feb 06, 2025 at 09:31:42AM +0000, David Woodhouse wrote:
> On Tue, 2025-02-04 at 13:05 +0100, Thomas Weißschuh wrote:
> > Currently each architecture defines the setup of the vDSO data page on
> > its own, mostly through copy-and-paste from some other architecture.
> > Extend the existing generic vDSO implementation to also provide generic
> > data storage.
> > This removes duplicated code and paves the way for further changes to
> > the generic vDSO implementation without having to go through a lot of
> > per-architecture changes.
> >
> > Based on v6.14-rc1 and intended to be merged through the tip tree.
Note: The real answer will need to come from the timekeeping
maintainers, my personal two cents below.
> Thanks for working on this. Is there a plan to expose the time data
> directly to userspace in a form which is usable *other* than by
> function calls which get the value of the clock at a given moment?
There are no current plans that I am aware of.
> For populating the vmclock device¹ we need to know the actual
> relationship between the hardware counter (TSC, arch timer, etc.) and
> real time in order to propagate that to the guest.
>
> I see two options for doing this:
>
> 1. Via userspace, exposing the vdso time data (and a notification when
> it changes?) and letting the userspace VMM populate the vmclock.
> This is complex for x86 because of TSC scaling; in fact userspace
> doesn't currently know the precise scaling from host to guest TSC
> so we'd have to be able to extract that from KVM.
Exposing the raw vdso time data is problematic as it precludes any
evolution to its datastructures, like the one we are currently doing.
An additional, trimmed down and stable data structure could be used.
But I don't think it makes sense. The vDSO is all about a stable
highlevel function interface on top of an unstable data interface.
However the vmclock needs the lowlevel data to populate its own
datastructure, wrapping raw data access in function calls is unnecessary.
If no functions are involved then the vDSO is not needed. The data can
be maintained separately in any other place in the kernel and accessed
or mapped by userspace from there.
Also the vDSO does not have an active notification mechanism, this would
probably be implemented through a filedescriptor, but then the data
can also be mapped through exactly that fd.
> 2. In kernel, asking KVM to populate the vmclock structure much like
> it does other pvclocks shared with the guest. KVM/x86 already uses
> pvclock_gtod_register_notifier() to hook changes; should we expand
> on that? The problem with that notifier is that it seems to be
> called far more frequently than I'd expect.
This sounds better, especially as any custom ABI from the host kernel to
the VMM would look a lot like the vmclock structure anyways.
Timekeeper updates are indeed very frequent, but what are the concrete
issues? That frequency is fine for regular vDSO data page updates,
updating the vmclock data page should be very similar.
The timekeeper core can pass context to the notifier callbacks, maybe
this can be used to skip some expensive steps where possible.
> ¹ https://gitlab.com/qemu-project/qemu/-/commit/3634039b93cc5
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v3 00/18] vDSO: Introduce generic data storage
2025-02-06 10:59 ` Thomas Weißschuh
@ 2025-02-07 10:15 ` David Woodhouse
0 siblings, 0 replies; 29+ messages in thread
From: David Woodhouse @ 2025-02-07 10:15 UTC (permalink / raw)
To: Thomas Weißschuh
Cc: James E.J. Bottomley, Helge Deller, Andy Lutomirski,
Thomas Gleixner, Vincenzo Frascino, Anna-Maria Behnsen,
Frederic Weisbecker, Andrew Morton, Catalin Marinas, Will Deacon,
Theodore Ts'o, Jason A. Donenfeld, Paul Walmsley,
Palmer Dabbelt, Albert Ou, Huacai Chen, WANG Xuerui, Russell King,
Heiko Carstens, Vasily Gorbik, Alexander Gordeev,
Christian Borntraeger, Sven Schnelle, Thomas Bogendoerfer,
Michael Ellerman, Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren, linux-parisc,
linux-kernel, linux-arm-kernel, linux-riscv, loongarch,
linux-s390, linux-mips, linuxppc-dev, linux-arch, Nam Cao,
linux-csky, Ridoux, Julien, Luu, Ryan, kvm
[-- Attachment #1: Type: text/plain, Size: 4065 bytes --]
On Thu, 2025-02-06 at 11:59 +0100, Thomas Weißschuh wrote:
> On Thu, Feb 06, 2025 at 09:31:42AM +0000, David Woodhouse wrote:
> > On Tue, 2025-02-04 at 13:05 +0100, Thomas Weißschuh wrote:
> > > Currently each architecture defines the setup of the vDSO data page on
> > > its own, mostly through copy-and-paste from some other architecture.
> > > Extend the existing generic vDSO implementation to also provide generic
> > > data storage.
> > > This removes duplicated code and paves the way for further changes to
> > > the generic vDSO implementation without having to go through a lot of
> > > per-architecture changes.
> > >
> > > Based on v6.14-rc1 and intended to be merged through the tip tree.
>
> Note: The real answer will need to come from the timekeeping
> maintainers, my personal two cents below.
>
> > Thanks for working on this. Is there a plan to expose the time data
> > directly to userspace in a form which is usable *other* than by
> > function calls which get the value of the clock at a given moment?
>
> There are no current plans that I am aware of.
>
> > For populating the vmclock device¹ we need to know the actual
> > relationship between the hardware counter (TSC, arch timer, etc.) and
> > real time in order to propagate that to the guest.
> >
> > I see two options for doing this:
> >
> > 1. Via userspace, exposing the vdso time data (and a notification when
> > it changes?) and letting the userspace VMM populate the vmclock.
> > This is complex for x86 because of TSC scaling; in fact userspace
> > doesn't currently know the precise scaling from host to guest TSC
> > so we'd have to be able to extract that from KVM.
>
> Exposing the raw vdso time data is problematic as it precludes any
> evolution to its datastructures, like the one we are currently doing.
>
> An additional, trimmed down and stable data structure could be used.
> But I don't think it makes sense. The vDSO is all about a stable
> highlevel function interface on top of an unstable data interface.
> However the vmclock needs the lowlevel data to populate its own
> datastructure, wrapping raw data access in function calls is unnecessary.
> If no functions are involved then the vDSO is not needed. The data can
> be maintained separately in any other place in the kernel and accessed
> or mapped by userspace from there.
> Also the vDSO does not have an active notification mechanism, this would
> probably be implemented through a filedescriptor, but then the data
> can also be mapped through exactly that fd.
>
> > 2. In kernel, asking KVM to populate the vmclock structure much like
> > it does other pvclocks shared with the guest. KVM/x86 already uses
> > pvclock_gtod_register_notifier() to hook changes; should we expand
> > on that? The problem with that notifier is that it seems to be
> > called far more frequently than I'd expect.
>
> This sounds better, especially as any custom ABI from the host kernel to
> the VMM would look a lot like the vmclock structure anyways.
>
> Timekeeper updates are indeed very frequent, but what are the concrete
> issues? That frequency is fine for regular vDSO data page updates,
> updating the vmclock data page should be very similar.
> The timekeeper core can pass context to the notifier callbacks, maybe
> this can be used to skip some expensive steps where possible.
In the context of a hypervisor with lots of guests running, that's a
lot of pointless steal time. But it isn't just that; ISTR the result
was also *inaccurate*.
I need to go back and reproduce the testing, but I think it was
constantly adjusting the apparent rate even with no changed inputs from
NTP. Where the number of clock counts per jiffy wasn't an integer, the
notification would be constantly changing, for example to report 333333
counts per jiffy for most of the time, and occasionally 333334 counts
for a single jiffy before flipping back again. Or something like that.
[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5069 bytes --]
^ permalink raw reply [flat|nested] 29+ messages in thread
* Re: [PATCH v3 00/18] vDSO: Introduce generic data storage
2025-02-06 9:31 ` [PATCH v3 00/18] vDSO: Introduce generic data storage David Woodhouse
2025-02-06 10:59 ` Thomas Weißschuh
@ 2025-02-14 11:34 ` Thomas Gleixner
2025-02-14 12:04 ` David Woodhouse
1 sibling, 1 reply; 29+ messages in thread
From: Thomas Gleixner @ 2025-02-14 11:34 UTC (permalink / raw)
To: David Woodhouse, Thomas Weißschuh, James E.J. Bottomley,
Helge Deller, Andy Lutomirski, Vincenzo Frascino,
Anna-Maria Behnsen, Frederic Weisbecker, Andrew Morton,
Catalin Marinas, Will Deacon, Theodore Ts'o,
Jason A. Donenfeld, Paul Walmsley, Palmer Dabbelt, Albert Ou,
Huacai Chen, WANG Xuerui, Russell King, Heiko Carstens,
Vasily Gorbik, Alexander Gordeev, Christian Borntraeger,
Sven Schnelle, Thomas Bogendoerfer, Michael Ellerman,
Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, linux-csky, Ridoux, Julien, Luu, Ryan, kvm
David!
On Thu, Feb 06 2025 at 09:31, David Woodhouse wrote:
> Thanks for working on this. Is there a plan to expose the time data
> directly to userspace in a form which is usable *other* than by
> function calls which get the value of the clock at a given moment?
>
> For populating the vmclock device¹ we need to know the actual
> relationship between the hardware counter (TSC, arch timer, etc.) and
> real time in order to propagate that to the guest.
>
> I see two options for doing this:
>
> 1. Via userspace, exposing the vdso time data (and a notification when
> it changes?) and letting the userspace VMM populate the vmclock.
> This is complex for x86 because of TSC scaling; in fact userspace
> doesn't currently know the precise scaling from host to guest TSC
> so we'd have to be able to extract that from KVM.
Exposing the raw data is not going to happen as we would create an ABI
preventing any modifications to the internals. VDSO data is considered a
fully internal (think kernel) representation and the accessor functions
create an ABI around it. So if at all you can add a accessor function
which exposes data to user space so that the internal data
representation can still be modified as necessary.
> 2. In kernel, asking KVM to populate the vmclock structure much like
> it does other pvclocks shared with the guest. KVM/x86 already uses
> pvclock_gtod_register_notifier() to hook changes; should we expand
> on that? The problem with that notifier is that it seems to be
> called far more frequently than I'd expect.
It's called once per tick to expose the continous updates to the
conversion factors and related internal data.
Thanks,
tglx
^ permalink raw reply [flat|nested] 29+ messages in thread* Re: [PATCH v3 00/18] vDSO: Introduce generic data storage
2025-02-14 11:34 ` Thomas Gleixner
@ 2025-02-14 12:04 ` David Woodhouse
0 siblings, 0 replies; 29+ messages in thread
From: David Woodhouse @ 2025-02-14 12:04 UTC (permalink / raw)
To: Thomas Gleixner, Thomas Weißschuh, James E.J. Bottomley,
Helge Deller, Andy Lutomirski, Vincenzo Frascino,
Anna-Maria Behnsen, Frederic Weisbecker, Andrew Morton,
Catalin Marinas, Will Deacon, Theodore Ts'o,
Jason A. Donenfeld, Paul Walmsley, Palmer Dabbelt, Albert Ou,
Huacai Chen, WANG Xuerui, Russell King, Heiko Carstens,
Vasily Gorbik, Alexander Gordeev, Christian Borntraeger,
Sven Schnelle, Thomas Bogendoerfer, Michael Ellerman,
Nicholas Piggin, Christophe Leroy, Naveen N Rao,
Madhavan Srinivasan, Ingo Molnar, Borislav Petkov, Dave Hansen,
x86, H. Peter Anvin, Arnd Bergmann, Guo Ren
Cc: linux-parisc, linux-kernel, linux-arm-kernel, linux-riscv,
loongarch, linux-s390, linux-mips, linuxppc-dev, linux-arch,
Nam Cao, linux-csky, Ridoux, Julien, Luu, Ryan, kvm
[-- Attachment #1: Type: text/plain, Size: 2581 bytes --]
On Fri, 2025-02-14 at 12:34 +0100, Thomas Gleixner wrote:
> > 2. In kernel, asking KVM to populate the vmclock structure much like
> > it does other pvclocks shared with the guest. KVM/x86 already uses
> > pvclock_gtod_register_notifier() to hook changes; should we expand
> > on that? The problem with that notifier is that it seems to be
> > called far more frequently than I'd expect.
>
> It's called once per tick to expose the continous updates to the
> conversion factors and related internal data.
My recollection (a vague one) is that it's called, and reports
"changes", even when there *are* no changes to underlying conversion
factors. Something along the lines of "N ticks at 333 counts per tick,
then one tick at 334 counts per tick to catch up" because it can't
express the division factor completely without that discontinuity?
The actual 'error' caused by the apparent fluctuation in rate is
probably entirely negligible, but I am slightly concerned about the
steal time, if the hypervisor then spends stolen CPU time relaying all
those "changes" to the guest, and then the guest has to spend time
feeding the "changes" into its own timekeeping.
I'd like to strive for a mode where we only adjust what we tell guests,
when adjtimex actually changes the real timing factors.
In fact if we have a userspace tool like chrony feeding adjtimex based
on external NTP/PPS/whatever, that tool could probably calibrate a
stable host TSC directly against the external real time. And in that
mode maybe we don't even need to feed the guest from the kernel's
CLOCK_REALTIME; that would be just another conversion step to introduce
noise.
We might end up with the direct setup for dedicated hosting
environments, but I do also want to support the general-purpose QEMU-
based setup where we expose the host's CLOCK_REALTIME as efficiently as
possible.
How about this: A KVM feature to provide/populate the VMCLOCK, since
only KVM knows the precise TSC scaling (and can immediately flip the
VMCLOCK to report invalid state if the TSC becomes unreliable).
It can *either* be fed the precise TSC/realtime relationship from
userspace (maybe in a vmclock structure that *userspace* populates, so
all the kernel has to do is scale/offset to account for the guest TSC
being different from the host TSC).
Or it can be in 'automatic' mode, where it derives from the host's
timekeeping. Which at the moment would have "too many" updates for my
liking, but we can worry about that later if necessary.
[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5069 bytes --]
^ permalink raw reply [flat|nested] 29+ messages in thread