[PATCH v10 0/6] ARM: VDSO

[PATCH v10 0/6] ARM: VDSO

[Nathan Lynch]

Please consider this a call for review and test results.  I believe
this code is stable and ready for inclusion, but patches 3-6 lack
explicit acks/reviews so far.

Russell, I plan to add at least the sigpage changes (patches 1 and 2)
to your patch system shortly after this posting, as they have
Reviewed-by tags and are useful independent of the VDSO.

Provide fast userspace implementations of gettimeofday and
clock_gettime on systems that implement the generic timers extension
defined in ARMv7.  This follows the example of arm64 in conception but
significantly differs in some aspects of the implementation (C vs
assembly, mainly).

Clocks supported:
- CLOCK_REALTIME
- CLOCK_MONOTONIC
- CLOCK_REALTIME_COARSE
- CLOCK_MONOTONIC_COARSE

This also provides clock_getres (as arm64 does).

getcpu support is planned but not included at this time.

For applications to transparently benefit from this change,
ARM-specific support code needs to be added to glibc.  I hope to have
that code added to glibc for the 2.21 release.

The VDSO symbols are available for lookup via dlsym even with an
unpatched glibc.

Note that while the high-precision realtime and monotonic clock
support depends on the generic timers extension, support for
clock_getres and coarse clocks is independent of the timer
implementation and is provided unconditionally.  High-resolution clock
support requires changes to the arch timer code, posted here:

http://lists.infradead.org/pipermail/linux-arm-kernel/2014-September/288452.html

The VDSO will function correctly without those changes, but
gettimeofday and clock_gettime with CLOCK_REALTIME/CLOCK_MONOTONIC
will not be accelerated.

Tested on OMAP5 and i.MX6, verifying that results obtained with the
vdso are consistent with those obtained from the kernel.  On OMAP5 I
observe a 3- to 4-fold speedup for gettimeofday / CLOCK_REALTIME, with
even better (if less interesting) speedups for the coarse clock ids
and clock_getres.

I've been testing and benchmarking this with some custom test code
which I have hosted here:

https://github.com/nlynch-mentor/vdsotest

Unpatched FSF GDB may complain "warning: Could not load shared library
symbols for linux-vdso.so.1."  This is not an ARM-specific issue.
Current Fedora and Debian patch their GDB packages to prevent this
warning.

Changes since v9:
- Update to 3.17-rc5.
- Fixed minor checkpatch warnings (line length etc).
- Removed useless VDSO_LBASE constant.
- Added 'notrace' attributes to VDSO-internal functions.
- Removed unused debug code from vdsomunge host program.

Changes since v8:
- Update to 3.17-rc1.
- Split out arch timer changes into separate series.
- Ensure that VDSO will not attempt to read the counter if access is
  not enabled; this check can be removed after the arch timer changes
  are merged.  See update_vsyscall and vdso_can_use_arch_timer in
  patch #5.

Changes since v7:
- Update to next-20140801.
- In arch_setup_additional_pages, fix call to get_unmapped_area - use
  bytes (not pages) for length argument.
- As x86 does, separate data and text into two VMAs, [vvar] and [vdso]
  respectively.  These have different permissions; [vdso] will allow
  debuggers to set breakpoints, but [vvar] is read-only and cannot be
  modified even via ptrace.
- Use _install_special_mapping for signal page, vvar, and vdso
  mappings.
- Add -DDISABLE_BRANCH_PROFILING.
- Add --no-undefined -Bsymbolic to link options to cause linker to
  error out on unresolved references, making checkundef script
  unnecessary.
- Specify max-page-size, common-page-size in linker options so the
  true alignment is reflected in program header; otherwise gdb gets
  confused.
- Fix incremental build vs. generated auxvec.h.
- Use appropriate unwind directives in __get_datapage.
- Added vdso_install target and help text.  Install build-id symlinks
  as x86 does.
- Adjust update_vsyscall for changes in struct timekeeper.

Changes since v6:
- Update to 3.16-rc1.
- Remove -lgcc from link step - need to support GCC installations
  without libgcc.
- Force -O2 compilation to prevent GCC from emitting calls to libgcc
  math routines.
- Use custom post-processing to clear the EF_ARM_ABI_FLOAT_SOFT flag
  if set in the ELF header to produce a shared object which is
  architecturally allowed to be used by both soft- and hard-float
  code.
- Consolidate common arch timer code instead of duplicating it.
- Prevent the VDSO from attempting CP15 access on memory-only
  architected timer implementations by renaming the clocksource.

Changes since v5:
- Update to 3.15-rc1.
- Place vdso at a randomized offset above the stack along with the
  sigpage.
- Properly export asm/auxvec.h.
- Split patch into series for ease of review.

Changes since v4:
- Map data page at the beginning of the VMA to prevent orphan
  sections at the end of output invalidating the calculated offset.
- Move checkundef into cmd_vdsold to avoid spurious rebuilds.
- Change vdso_init message to pr_debug.
- Add -fno-stack-protector to cflags.

Changes since v3:
- Update to 3.14-rc6.
- Record vdso base in mm context before installing mapping (for the
  sake of perf_mmap_event).
- Use a more seqcount-like API for critical sections.  Using seqcount
  API directly, however, would leak kernel pointers to userspace when
  lockdep is enabled.
- Trap instead of looping forever in division-by-zero stubs.

Changes since v2:
- Update to 3.14-rc4.
- Make vDSO configurable, depending on AEABI and MMU.
- Defer shifting of nanosecond component of timespec: fixes observed
  1ns inconsistencies for CLOCK_REALTIME, CLOCK_MONOTONIC (see
  45a7905fc48f for arm64 equivalent).
- Force reload of seq_count when spinning: without a memory clobber
  after the load of vdata->seq_count, GCC can generate code like this:
    2f8:   e59c9020        ldr     r9, [ip, #32]
    2fc:   e3190001        tst     r9, #1
    300:   1a000033        bne     3d4 <do_realtime+0x104>
    304:   f57ff05b        dmb     ish
    308:   e59c3034        ldr     r3, [ip, #52]   ; 0x34
    ...
    3d4:   eafffffe        b       3d4 <do_realtime+0x104>
- Build vdso.so with -lgcc: calls to __lshrdi3, __divsi3 sometimes
  emitted (especially with -Os).  Override certain libgcc functions to
  prevent undefined symbols.
- Do not clear PG_reserved on vdso pages.
- Remove unnecessary get_page calls.
- Simplify ELF signature check during init.
- Use volatile for asm syscall fallbacks.
- Check whether vdso_pagelist is initialized in arm_install_vdso.
- Record clocksource mask in data page.
- Reduce code duplication in do_realtime, do_monotonic.
- Reduce calculations performed in critical sections.
- Simplify coarse clock handling.
- Move datapage load to its own assembly routine.
- Tune vdso_data layout and tweak field names.
- Check vdso shared object for undefined symbols during build.

Changes since v1:
- update to 3.14-rc1
- ensure cache coherency for data page
- Document the kernel-to-userspace protocol for vdso data page updates,
  and note that the timekeeping core prevents concurrent updates.
- update wall-to-monotonic fields unconditionally
- move vdso_start, vdso_end declarations to vdso.h
- correctly build and run when CONFIG_ARM_ARCH_TIMER=n
- rearrange linker script to avoid overlapping sections when CONFIG_DEBUGINFO=n
- remove use_syscall checks from coarse clock paths
- crib BUG_INSTR (0xe7f001f2) from asm/bug.h for text fill

Nathan Lynch (6):
  ARM: use _install_special_mapping for sigpage
  ARM: place sigpage at a random offset above stack
  ARM: miscellaneous vdso infrastructure, preparation
  ARM: add vdso user-space code
  ARM: vdso initialization, mapping, and synchronization
  ARM: add CONFIG_VDSO Kconfig and Makefile bits

 arch/arm/Makefile                    |   8 +
 arch/arm/include/asm/Kbuild          |   1 -
 arch/arm/include/asm/auxvec.h        |   1 +
 arch/arm/include/asm/elf.h           |   9 +
 arch/arm/include/asm/mmu.h           |   3 +
 arch/arm/include/asm/vdso.h          |  32 ++++
 arch/arm/include/asm/vdso_datapage.h |  60 +++++++
 arch/arm/include/uapi/asm/Kbuild     |   1 +
 arch/arm/include/uapi/asm/auxvec.h   |   7 +
 arch/arm/kernel/Makefile             |   1 +
 arch/arm/kernel/asm-offsets.c        |   5 +
 arch/arm/kernel/process.c            |  72 +++++++-
 arch/arm/kernel/vdso.c               | 211 ++++++++++++++++++++++
 arch/arm/mm/Kconfig                  |  15 ++
 arch/arm/vdso/.gitignore             |   1 +
 arch/arm/vdso/Makefile               |  74 ++++++++
 arch/arm/vdso/datapage.S             |  15 ++
 arch/arm/vdso/vdso.S                 |  35 ++++
 arch/arm/vdso/vdso.lds.S             |  88 ++++++++++
 arch/arm/vdso/vdsomunge.c            | 201 +++++++++++++++++++++
 arch/arm/vdso/vgettimeofday.c        | 328 +++++++++++++++++++++++++++++++++++
 21 files changed, 1158 insertions(+), 10 deletions(-)
 create mode 100644 arch/arm/include/asm/auxvec.h
 create mode 100644 arch/arm/include/asm/vdso.h
 create mode 100644 arch/arm/include/asm/vdso_datapage.h
 create mode 100644 arch/arm/include/uapi/asm/auxvec.h
 create mode 100644 arch/arm/kernel/vdso.c
 create mode 100644 arch/arm/vdso/.gitignore
 create mode 100644 arch/arm/vdso/Makefile
 create mode 100644 arch/arm/vdso/datapage.S
 create mode 100644 arch/arm/vdso/vdso.S
 create mode 100644 arch/arm/vdso/vdso.lds.S
 create mode 100644 arch/arm/vdso/vdsomunge.c
 create mode 100644 arch/arm/vdso/vgettimeofday.c

-- 
1.9.3

[PATCH v10 1/6] ARM: use _install_special_mapping for sigpage

[Nathan Lynch]

_install_special_mapping allows the VMA to be identifed in
/proc/pid/maps without the use of arch_vma_name, providing a
slight net reduction in object size:

  text    data     bss     dec     hex filename
  2996      96     144    3236     ca4 arch/arm/kernel/process.o (before)
  2956     104     144    3204     c84 arch/arm/kernel/process.o (after)

Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
---
 arch/arm/kernel/process.c | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index 81ef686a91ca..46fbbb3701a0 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -472,19 +472,23 @@ int in_gate_area_no_mm(unsigned long addr)
 
 const char *arch_vma_name(struct vm_area_struct *vma)
 {
-	return is_gate_vma(vma) ? "[vectors]" :
-		(vma->vm_mm && vma->vm_start == vma->vm_mm->context.sigpage) ?
-		 "[sigpage]" : NULL;
+	return is_gate_vma(vma) ? "[vectors]" : NULL;
 }
 
 static struct page *signal_page;
 extern struct page *get_signal_page(void);
 
+static const struct vm_special_mapping sigpage_mapping = {
+	.name = "[sigpage]",
+	.pages = &signal_page,
+};
+
 int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 {
 	struct mm_struct *mm = current->mm;
+	struct vm_area_struct *vma;
 	unsigned long addr;
-	int ret;
+	int ret = 0;
 
 	if (!signal_page)
 		signal_page = get_signal_page();
@@ -498,12 +502,16 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 		goto up_fail;
 	}
 
-	ret = install_special_mapping(mm, addr, PAGE_SIZE,
+	vma = _install_special_mapping(mm, addr, PAGE_SIZE,
 		VM_READ | VM_EXEC | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC,
-		&signal_page);
+		&sigpage_mapping);
+
+	if (IS_ERR(vma)) {
+		ret = PTR_ERR(vma);
+		goto up_fail;
+	}
 
-	if (ret == 0)
-		mm->context.sigpage = addr;
+	mm->context.sigpage = addr;
 
  up_fail:
 	up_write(&mm->mmap_sem);
-- 
1.9.3

[PATCH v10 2/6] ARM: place sigpage at a random offset above stack

[Nathan Lynch]

The sigpage is currently placed alongside shared libraries etc in the
address space.  Similar to what x86_64 does for its VDSO, place the
sigpage at a randomized offset above the stack so that learning the
base address of the sigpage doesn't help expose where shared libraries
are loaded in the address space (and vice versa).

Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
---
 arch/arm/kernel/process.c | 37 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 36 insertions(+), 1 deletion(-)

diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index 46fbbb3701a0..edd2ac3d7c44 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -475,6 +475,39 @@ const char *arch_vma_name(struct vm_area_struct *vma)
 	return is_gate_vma(vma) ? "[vectors]" : NULL;
 }
 
+/* If possible, provide a placement hint at a random offset from the
+ * stack for the signal page.
+ */
+static unsigned long sigpage_addr(const struct mm_struct *mm,
+				  unsigned int npages)
+{
+	unsigned long offset;
+	unsigned long first;
+	unsigned long last;
+	unsigned long addr;
+	unsigned int slots;
+
+	first = PAGE_ALIGN(mm->start_stack);
+
+	last = TASK_SIZE - (npages << PAGE_SHIFT);
+
+	/* No room after stack? */
+	if (first > last)
+		return 0;
+
+	/* Just enough room? */
+	if (first == last)
+		return first;
+
+	slots = ((last - first) >> PAGE_SHIFT) + 1;
+
+	offset = get_random_int() % slots;
+
+	addr = first + (offset << PAGE_SHIFT);
+
+	return addr;
+}
+
 static struct page *signal_page;
 extern struct page *get_signal_page(void);
 
@@ -488,6 +521,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct *vma;
 	unsigned long addr;
+	unsigned long hint;
 	int ret = 0;
 
 	if (!signal_page)
@@ -496,7 +530,8 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 		return -ENOMEM;
 
 	down_write(&mm->mmap_sem);
-	addr = get_unmapped_area(NULL, 0, PAGE_SIZE, 0, 0);
+	hint = sigpage_addr(mm, 1);
+	addr = get_unmapped_area(NULL, hint, PAGE_SIZE, 0, 0);
 	if (IS_ERR_VALUE(addr)) {
 		ret = addr;
 		goto up_fail;
-- 
1.9.3

[PATCH v10 3/6] ARM: miscellaneous vdso infrastructure, preparation

[Nathan Lynch]

Define the layout of the data structure shared between kernel and
userspace.

Track the vdso address in the mm_context; needed for communicating
AT_SYSINFO_EHDR to the ELF loader.

Add declarations for arm_install_vdso; implementation is in a
following patch.

Define AT_SYSINFO_EHDR, and, if CONFIG_VDSO=y, report the vdso shared
object address via the ELF auxiliary vector.

Note - this adds the AT_SYSINFO_EHDR in a new user-visible header
asm/auxvec.h; this is consistent with other architectures.

Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
---
 arch/arm/include/asm/Kbuild          |  1 -
 arch/arm/include/asm/auxvec.h        |  1 +
 arch/arm/include/asm/elf.h           |  9 ++++++
 arch/arm/include/asm/mmu.h           |  3 ++
 arch/arm/include/asm/vdso.h          | 32 +++++++++++++++++++
 arch/arm/include/asm/vdso_datapage.h | 60 ++++++++++++++++++++++++++++++++++++
 arch/arm/include/uapi/asm/Kbuild     |  1 +
 arch/arm/include/uapi/asm/auxvec.h   |  7 +++++
 8 files changed, 113 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm/include/asm/auxvec.h
 create mode 100644 arch/arm/include/asm/vdso.h
 create mode 100644 arch/arm/include/asm/vdso_datapage.h
 create mode 100644 arch/arm/include/uapi/asm/auxvec.h

diff --git a/arch/arm/include/asm/Kbuild b/arch/arm/include/asm/Kbuild
index 70cd84eb7fda..9fb902425606 100644
--- a/arch/arm/include/asm/Kbuild
+++ b/arch/arm/include/asm/Kbuild
@@ -1,6 +1,5 @@
 
 
-generic-y += auxvec.h
 generic-y += bitsperlong.h
 generic-y += cputime.h
 generic-y += current.h
diff --git a/arch/arm/include/asm/auxvec.h b/arch/arm/include/asm/auxvec.h
new file mode 100644
index 000000000000..fbd388c46299
--- /dev/null
+++ b/arch/arm/include/asm/auxvec.h
@@ -0,0 +1 @@
+#include <uapi/asm/auxvec.h>
diff --git a/arch/arm/include/asm/elf.h b/arch/arm/include/asm/elf.h
index afb9cafd3786..ac3f17fb4c8d 100644
--- a/arch/arm/include/asm/elf.h
+++ b/arch/arm/include/asm/elf.h
@@ -1,7 +1,9 @@
 #ifndef __ASMARM_ELF_H
 #define __ASMARM_ELF_H
 
+#include <asm/auxvec.h>
 #include <asm/hwcap.h>
+#include <asm/vdso_datapage.h>
 
 /*
  * ELF register definitions..
@@ -130,6 +132,13 @@ extern unsigned long arch_randomize_brk(struct mm_struct *mm);
 #define arch_randomize_brk arch_randomize_brk
 
 #ifdef CONFIG_MMU
+#ifdef CONFIG_VDSO
+#define ARCH_DLINFO						\
+do {								\
+	NEW_AUX_ENT(AT_SYSINFO_EHDR,				\
+		    (elf_addr_t)current->mm->context.vdso);	\
+} while (0)
+#endif
 #define ARCH_HAS_SETUP_ADDITIONAL_PAGES 1
 struct linux_binprm;
 int arch_setup_additional_pages(struct linux_binprm *, int);
diff --git a/arch/arm/include/asm/mmu.h b/arch/arm/include/asm/mmu.h
index 64fd15159b7d..a5b47421059d 100644
--- a/arch/arm/include/asm/mmu.h
+++ b/arch/arm/include/asm/mmu.h
@@ -11,6 +11,9 @@ typedef struct {
 #endif
 	unsigned int	vmalloc_seq;
 	unsigned long	sigpage;
+#ifdef CONFIG_VDSO
+	unsigned long	vdso;
+#endif
 } mm_context_t;
 
 #ifdef CONFIG_CPU_HAS_ASID
diff --git a/arch/arm/include/asm/vdso.h b/arch/arm/include/asm/vdso.h
new file mode 100644
index 000000000000..d0295f1dd1a3
--- /dev/null
+++ b/arch/arm/include/asm/vdso.h
@@ -0,0 +1,32 @@
+#ifndef __ASM_VDSO_H
+#define __ASM_VDSO_H
+
+#ifdef __KERNEL__
+
+#ifndef __ASSEMBLY__
+
+struct mm_struct;
+
+#ifdef CONFIG_VDSO
+
+void arm_install_vdso(struct mm_struct *mm, unsigned long addr);
+
+extern char vdso_start, vdso_end;
+
+extern unsigned int vdso_total_pages;
+
+#else /* CONFIG_VDSO */
+
+static inline void arm_install_vdso(struct mm_struct *mm, unsigned long addr)
+{
+}
+
+#define vdso_total_pages 0
+
+#endif /* CONFIG_VDSO */
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* __KERNEL__ */
+
+#endif /* __ASM_VDSO_H */
diff --git a/arch/arm/include/asm/vdso_datapage.h b/arch/arm/include/asm/vdso_datapage.h
new file mode 100644
index 000000000000..f08bdb73d3f4
--- /dev/null
+++ b/arch/arm/include/asm/vdso_datapage.h
@@ -0,0 +1,60 @@
+/*
+ * Adapted from arm64 version.
+ *
+ * Copyright (C) 2012 ARM Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __ASM_VDSO_DATAPAGE_H
+#define __ASM_VDSO_DATAPAGE_H
+
+#ifdef __KERNEL__
+
+#ifndef __ASSEMBLY__
+
+#include <asm/page.h>
+
+/* Try to be cache-friendly on systems that don't implement the
+ * generic timer: fit the unconditionally updated fields in the first
+ * 32 bytes.
+ */
+struct vdso_data {
+	u32 seq_count;		/* sequence count - odd during updates */
+	u16 use_syscall;	/* whether to fall back to syscalls */
+	u16 cs_shift;		/* clocksource shift */
+	u32 xtime_coarse_sec;	/* coarse time */
+	u32 xtime_coarse_nsec;
+
+	u32 wtm_clock_sec;	/* wall to monotonic offset */
+	u32 wtm_clock_nsec;
+	u32 xtime_clock_sec;	/* CLOCK_REALTIME - seconds */
+	u32 cs_mult;		/* clocksource multiplier */
+
+	u64 cs_cycle_last;	/* last cycle value */
+	u64 cs_mask;		/* clocksource mask */
+
+	u64 xtime_clock_snsec;	/* CLOCK_REALTIME sub-ns base */
+	u32 tz_minuteswest;	/* timezone info for gettimeofday(2) */
+	u32 tz_dsttime;
+};
+
+union vdso_data_store {
+	struct vdso_data data;
+	u8 page[PAGE_SIZE];
+};
+
+#endif /* !__ASSEMBLY__ */
+
+#endif /* __KERNEL__ */
+
+#endif /* __ASM_VDSO_DATAPAGE_H */
diff --git a/arch/arm/include/uapi/asm/Kbuild b/arch/arm/include/uapi/asm/Kbuild
index 70a1c9da30ca..a1c05f93d920 100644
--- a/arch/arm/include/uapi/asm/Kbuild
+++ b/arch/arm/include/uapi/asm/Kbuild
@@ -1,6 +1,7 @@
 # UAPI Header export list
 include include/uapi/asm-generic/Kbuild.asm
 
+header-y += auxvec.h
 header-y += byteorder.h
 header-y += fcntl.h
 header-y += hwcap.h
diff --git a/arch/arm/include/uapi/asm/auxvec.h b/arch/arm/include/uapi/asm/auxvec.h
new file mode 100644
index 000000000000..f56936b97ec2
--- /dev/null
+++ b/arch/arm/include/uapi/asm/auxvec.h
@@ -0,0 +1,7 @@
+#ifndef __ASM_AUXVEC_H
+#define __ASM_AUXVEC_H
+
+/* vDSO location */
+#define AT_SYSINFO_EHDR	33
+
+#endif
-- 
1.9.3

[PATCH v10 4/6] ARM: add vdso user-space code

[Nathan Lynch]

Place vdso-related user-space code in arch/arm/kernel/vdso/.

It is almost completely written in C with some assembly helpers to
load the data page address, sample the counter, and fall back to
system calls when necessary.

If CONFIG_ARM_ARCH_TIMER is not enabled, the vdso cannot service
high-resolution clocks and falls back to syscalls.  Low-resolution
clocks e.g. CLOCK_REALTIME_COARSE can be serviced regardless.

Of particular note is that a post-processing step ("vdsomunge") is
necessary to produce a shared object which is architecturally allowed
to be used by both soft- and hard-float EABI programs.

The 2012 edition of the ARM ABI defines Tag_ABI_VFP_args = 3 "Code is
compatible with both the base and VFP variants; the user did not
permit non-variadic functions to pass FP parameters/results."
Unfortunately current toolchains do not support this tag, which is
ideally what we would use.

The best available option is to ensure that both EF_ARM_ABI_FLOAT_SOFT
and EF_ARM_ABI_FLOAT_HARD are unset in the ELF header's e_flags,
indicating that the shared object is "old" and should be accepted for
backward compatibility's sake.  While binutils < 2.24 appear to
produce a vdso.so with both flags clear, 2.24 always sets
EF_ARM_ABI_FLOAT_SOFT, with no way to inhibit this behavior.  So we
have to fix things up with a custom post-processing step.

In fact, the VDSO code in glibc does much less validation (including
checking these flags) than the code for handling conventional
file-backed shared libraries, so this is a bit moot unless glibc's
VDSO code becomes more strict.

Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
---
 arch/arm/kernel/asm-offsets.c |   5 +
 arch/arm/vdso/.gitignore      |   1 +
 arch/arm/vdso/Makefile        |  74 ++++++++++
 arch/arm/vdso/datapage.S      |  15 ++
 arch/arm/vdso/vdso.S          |  35 +++++
 arch/arm/vdso/vdso.lds.S      |  88 ++++++++++++
 arch/arm/vdso/vdsomunge.c     | 201 ++++++++++++++++++++++++++
 arch/arm/vdso/vgettimeofday.c | 328 ++++++++++++++++++++++++++++++++++++++++++
 8 files changed, 747 insertions(+)
 create mode 100644 arch/arm/vdso/.gitignore
 create mode 100644 arch/arm/vdso/Makefile
 create mode 100644 arch/arm/vdso/datapage.S
 create mode 100644 arch/arm/vdso/vdso.S
 create mode 100644 arch/arm/vdso/vdso.lds.S
 create mode 100644 arch/arm/vdso/vdsomunge.c
 create mode 100644 arch/arm/vdso/vgettimeofday.c

diff --git a/arch/arm/kernel/asm-offsets.c b/arch/arm/kernel/asm-offsets.c
index 713e807621d2..72b4c3873dc8 100644
--- a/arch/arm/kernel/asm-offsets.c
+++ b/arch/arm/kernel/asm-offsets.c
@@ -24,6 +24,7 @@
 #include <asm/memory.h>
 #include <asm/procinfo.h>
 #include <asm/suspend.h>
+#include <asm/vdso_datapage.h>
 #include <asm/hardware/cache-l2x0.h>
 #include <linux/kbuild.h>
 
@@ -200,5 +201,9 @@ int main(void)
 #endif
   DEFINE(KVM_VTTBR,		offsetof(struct kvm, arch.vttbr));
 #endif
+  BLANK();
+#ifdef CONFIG_VDSO
+  DEFINE(VDSO_DATA_SIZE,	sizeof(union vdso_data_store));
+#endif
   return 0; 
 }
diff --git a/arch/arm/vdso/.gitignore b/arch/arm/vdso/.gitignore
new file mode 100644
index 000000000000..f8b69d84238e
--- /dev/null
+++ b/arch/arm/vdso/.gitignore
@@ -0,0 +1 @@
+vdso.lds
diff --git a/arch/arm/vdso/Makefile b/arch/arm/vdso/Makefile
new file mode 100644
index 000000000000..bab0a8be7924
--- /dev/null
+++ b/arch/arm/vdso/Makefile
@@ -0,0 +1,74 @@
+hostprogs-y := vdsomunge
+
+obj-vdso := vgettimeofday.o datapage.o
+
+# Build rules
+targets := $(obj-vdso) vdso.so vdso.so.dbg vdso.so.raw vdso.lds
+obj-vdso := $(addprefix $(obj)/, $(obj-vdso))
+
+ccflags-y := -shared -fPIC -fno-common -fno-builtin -fno-stack-protector
+ccflags-y += -nostdlib -Wl,-soname=linux-vdso.so.1 -DDISABLE_BRANCH_PROFILING
+ccflags-y += -Wl,--no-undefined $(call cc-ldoption, -Wl$(comma)--hash-style=sysv)
+
+obj-y += vdso.o
+extra-y += vdso.lds
+CPPFLAGS_vdso.lds += -P -C -U$(ARCH)
+
+CFLAGS_REMOVE_vdso.o = -pg
+
+# Force -O2 to avoid libgcc dependencies
+CFLAGS_REMOVE_vgettimeofday.o = -pg -Os
+CFLAGS_vgettimeofday.o = -O2
+
+# Disable gcov profiling for VDSO code
+GCOV_PROFILE := n
+
+# Force dependency
+$(obj)/vdso.o : $(obj)/vdso.so
+
+# Link rule for the .so file
+$(obj)/vdso.so.raw: $(src)/vdso.lds $(obj-vdso) FORCE
+	$(call if_changed,vdsold)
+
+$(obj)/vdso.so.dbg: $(obj)/vdso.so.raw $(obj)/vdsomunge FORCE
+	$(call if_changed,vdsomunge)
+
+# Strip rule for the .so file
+$(obj)/%.so: OBJCOPYFLAGS := -S
+$(obj)/%.so: $(obj)/%.so.dbg FORCE
+	$(call if_changed,objcopy)
+
+# Actual build commands
+quiet_cmd_vdsold = VDSO    $@
+      cmd_vdsold = $(CC) $(c_flags) -Wl,-T $(filter %.lds,$^) $(filter %.o,$^) \
+                   $(call cc-ldoption, -Wl$(comma)--build-id) \
+                   -Wl,-Bsymbolic -Wl,-z,max-page-size=4096 \
+                   -Wl,-z,common-page-size=4096 -o $@
+
+quiet_cmd_vdsomunge = MUNGE   $@
+      cmd_vdsomunge = $(objtree)/$(obj)/vdsomunge $< $@
+
+#
+# Install the unstripped copy of vdso.so.dbg.  If our toolchain
+# supports build-id, install .build-id links as well.
+#
+# Cribbed from arch/x86/vdso/Makefile.
+#
+quiet_cmd_vdso_install = INSTALL $<
+define cmd_vdso_install
+	cp $< "$(MODLIB)/vdso/vdso.so"; \
+	if readelf -n $< | grep -q 'Build ID'; then \
+	  buildid=`readelf -n $< |grep 'Build ID' |sed -e 's/^.*Build ID: \(.*\)$$/\1/'`; \
+	  first=`echo $$buildid | cut -b-2`; \
+	  last=`echo $$buildid | cut -b3-`; \
+	  mkdir -p "$(MODLIB)/vdso/.build-id/$$first"; \
+	  ln -sf "../../vdso.so" "$(MODLIB)/vdso/.build-id/$$first/$$last.debug"; \
+	fi
+endef
+
+$(MODLIB)/vdso: FORCE
+	@mkdir -p $(MODLIB)/vdso
+
+PHONY += vdso_install
+vdso_install: $(obj)/vdso.so.dbg $(MODLIB)/vdso FORCE
+	$(call cmd,vdso_install)
diff --git a/arch/arm/vdso/datapage.S b/arch/arm/vdso/datapage.S
new file mode 100644
index 000000000000..a2e60367931b
--- /dev/null
+++ b/arch/arm/vdso/datapage.S
@@ -0,0 +1,15 @@
+#include <linux/linkage.h>
+#include <asm/asm-offsets.h>
+
+	.align 2
+.L_vdso_data_ptr:
+	.long	_start - . - VDSO_DATA_SIZE
+
+ENTRY(__get_datapage)
+	.fnstart
+	adr	r0, .L_vdso_data_ptr
+	ldr	r1, [r0]
+	add	r0, r0, r1
+	bx	lr
+	.fnend
+ENDPROC(__get_datapage)
diff --git a/arch/arm/vdso/vdso.S b/arch/arm/vdso/vdso.S
new file mode 100644
index 000000000000..b2b97e3e7bab
--- /dev/null
+++ b/arch/arm/vdso/vdso.S
@@ -0,0 +1,35 @@
+/*
+ * Adapted from arm64 version.
+ *
+ * Copyright (C) 2012 ARM Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * Author: Will Deacon <will.deacon@arm.com>
+ */
+
+#include <linux/init.h>
+#include <linux/linkage.h>
+#include <linux/const.h>
+#include <asm/page.h>
+
+	__PAGE_ALIGNED_DATA
+
+	.globl vdso_start, vdso_end
+	.balign PAGE_SIZE
+vdso_start:
+	.incbin "arch/arm/vdso/vdso.so"
+	.balign PAGE_SIZE
+vdso_end:
+
+	.previous
diff --git a/arch/arm/vdso/vdso.lds.S b/arch/arm/vdso/vdso.lds.S
new file mode 100644
index 000000000000..d8a9583c0450
--- /dev/null
+++ b/arch/arm/vdso/vdso.lds.S
@@ -0,0 +1,88 @@
+/*
+ * Adapted from arm64 version.
+ *
+ * GNU linker script for the VDSO library.
+ *
+ * Copyright (C) 2012 ARM Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * Author: Will Deacon <will.deacon@arm.com>
+ * Heavily based on the vDSO linker scripts for other archs.
+ */
+
+#include <linux/const.h>
+#include <asm/page.h>
+#include <asm/vdso.h>
+
+OUTPUT_FORMAT("elf32-littlearm", "elf32-bigarm", "elf32-littlearm")
+OUTPUT_ARCH(arm)
+
+SECTIONS
+{
+	PROVIDE(_start = .);
+
+	. = SIZEOF_HEADERS;
+
+	.hash		: { *(.hash) }			:text
+	.gnu.hash	: { *(.gnu.hash) }
+	.dynsym		: { *(.dynsym) }
+	.dynstr		: { *(.dynstr) }
+	.gnu.version	: { *(.gnu.version) }
+	.gnu.version_d	: { *(.gnu.version_d) }
+	.gnu.version_r	: { *(.gnu.version_r) }
+
+	.note		: { *(.note.*) }		:text	:note
+
+
+	.eh_frame_hdr	: { *(.eh_frame_hdr) }		:text	:eh_frame_hdr
+	.eh_frame	: { KEEP (*(.eh_frame)) }	:text
+
+	.dynamic	: { *(.dynamic) }		:text	:dynamic
+
+	.rodata		: { *(.rodata*) }		:text
+
+	.text		: { *(.text*) }			:text	=0xe7f001f2
+
+	.got		: { *(.got) }
+	.rel.plt	: { *(.rel.plt) }
+
+	/DISCARD/	: {
+		*(.note.GNU-stack)
+		*(.data .data.* .gnu.linkonce.d.* .sdata*)
+		*(.bss .sbss .dynbss .dynsbss)
+	}
+}
+
+/*
+ * We must supply the ELF program headers explicitly to get just one
+ * PT_LOAD segment, and set the flags explicitly to make segments read-only.
+ */
+PHDRS
+{
+	text		PT_LOAD		FLAGS(5) FILEHDR PHDRS; /* PF_R|PF_X */
+	dynamic		PT_DYNAMIC	FLAGS(4);		/* PF_R */
+	note		PT_NOTE		FLAGS(4);		/* PF_R */
+	eh_frame_hdr	PT_GNU_EH_FRAME;
+}
+
+VERSION
+{
+	LINUX_2.6 {
+	global:
+		__vdso_clock_getres;
+		__vdso_clock_gettime;
+		__vdso_gettimeofday;
+	local: *;
+	};
+}
diff --git a/arch/arm/vdso/vdsomunge.c b/arch/arm/vdso/vdsomunge.c
new file mode 100644
index 000000000000..55e3e4226e4d
--- /dev/null
+++ b/arch/arm/vdso/vdsomunge.c
@@ -0,0 +1,201 @@
+/*
+ * Copyright 2014 Mentor Graphics Corporation.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2 of the
+ * License.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ *
+ * vdsomunge - Host program which produces a shared object
+ * architecturally specified to be usable by both soft- and hard-float
+ * programs.
+ *
+ * The Procedure Call Standard for the ARM Architecture (ARM IHI
+ * 0042E) says:
+ *
+ *	6.4.1 VFP and Base Standard Compatibility
+ *
+ *	Code compiled for the VFP calling standard is compatible with
+ *	the base standard (and vice-versa) if no floating-point or
+ *	containerized vector arguments or results are used.
+ *
+ * And ELF for the ARM Architecture (ARM IHI 0044E) (Table 4-2) says:
+ *
+ *	If both EF_ARM_ABI_FLOAT_XXXX bits are clear, conformance to the
+ *	base procedure-call standard is implied.
+ *
+ * The VDSO is built with -msoft-float, as with the rest of the ARM
+ * kernel, and uses no floating point arguments or results.  The build
+ * process will produce a shared object that may or may not have the
+ * EF_ARM_ABI_FLOAT_SOFT flag set (it seems to depend on the binutils
+ * version; binutils starting with 2.24 appears to set it).  The
+ * EF_ARM_ABI_FLOAT_HARD flag should definitely not be set, and this
+ * program will error out if it is.
+ *
+ * If the soft-float flag is set, this program clears it.  That's all
+ * it does.
+ */
+
+#define _GNU_SOURCE
+
+#include <byteswap.h>
+#include <elf.h>
+#include <errno.h>
+#include <error.h>
+#include <fcntl.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define HOST_ORDER ELFDATA2LSB
+#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+#define HOST_ORDER ELFDATA2MSB
+#endif
+
+/* Some of the ELF constants we'd like to use were added to <elf.h>
+ * relatively recently.
+ */
+#ifndef EF_ARM_EABI_VER5
+#define EF_ARM_EABI_VER5 0x05000000
+#endif
+
+#ifndef EF_ARM_ABI_FLOAT_SOFT
+#define EF_ARM_ABI_FLOAT_SOFT 0x200
+#endif
+
+#ifndef EF_ARM_ABI_FLOAT_HARD
+#define EF_ARM_ABI_FLOAT_HARD 0x400
+#endif
+
+static const char *outfile;
+
+static void cleanup(void)
+{
+	if (error_message_count > 0 && outfile != NULL)
+		unlink(outfile);
+}
+
+static Elf32_Word read_elf_word(Elf32_Word word, bool swap)
+{
+	return swap ? bswap_32(word) : word;
+}
+
+static Elf32_Half read_elf_half(Elf32_Half half, bool swap)
+{
+	return swap ? bswap_16(half) : half;
+}
+
+static void write_elf_word(Elf32_Word val, Elf32_Word *dst, bool swap)
+{
+	*dst = swap ? bswap_32(val) : val;
+}
+
+int main(int argc, char **argv)
+{
+	const Elf32_Ehdr *inhdr;
+	bool clear_soft_float;
+	const char *infile;
+	Elf32_Word e_flags;
+	const void *inbuf;
+	struct stat stat;
+	void *outbuf;
+	bool swap;
+	int outfd;
+	int infd;
+
+	atexit(cleanup);
+
+	if (argc != 3)
+		error(EXIT_FAILURE, 0, "Usage: %s [infile] [outfile]", argv[0]);
+
+	infile = argv[1];
+	outfile = argv[2];
+
+	infd = open(infile, O_RDONLY);
+	if (infd < 0)
+		error(EXIT_FAILURE, errno, "Cannot open %s", infile);
+
+	if (fstat(infd, &stat) != 0)
+		error(EXIT_FAILURE, errno, "Failed stat for %s", infile);
+
+	inbuf = mmap(NULL, stat.st_size, PROT_READ, MAP_PRIVATE, infd, 0);
+	if (inbuf == MAP_FAILED)
+		error(EXIT_FAILURE, errno, "Failed to map %s", infile);
+
+	close(infd);
+
+	inhdr = inbuf;
+
+	if (memcmp(&inhdr->e_ident, ELFMAG, SELFMAG) != 0)
+		error(EXIT_FAILURE, 0, "Not an ELF file");
+
+	if (inhdr->e_ident[EI_CLASS] != ELFCLASS32)
+		error(EXIT_FAILURE, 0, "Unsupported ELF class");
+
+	swap = inhdr->e_ident[EI_DATA] != HOST_ORDER;
+
+	if (read_elf_half(inhdr->e_type, swap) != ET_DYN)
+		error(EXIT_FAILURE, 0, "Not a shared object");
+
+	if (read_elf_half(inhdr->e_machine, swap) != EM_ARM) {
+		error(EXIT_FAILURE, 0, "Unsupported architecture %#x",
+		      inhdr->e_machine);
+	}
+
+	e_flags = read_elf_word(inhdr->e_flags, swap);
+
+	if (EF_ARM_EABI_VERSION(e_flags) != EF_ARM_EABI_VER5) {
+		error(EXIT_FAILURE, 0, "Unsupported EABI version %#x",
+		      EF_ARM_EABI_VERSION(e_flags));
+	}
+
+	if (e_flags & EF_ARM_ABI_FLOAT_HARD)
+		error(EXIT_FAILURE, 0,
+		      "Unexpected hard-float flag set in e_flags");
+
+	clear_soft_float = !!(e_flags & EF_ARM_ABI_FLOAT_SOFT);
+
+	outfd = open(outfile, O_RDWR | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR);
+	if (outfd < 0)
+		error(EXIT_FAILURE, errno, "Cannot open %s", outfile);
+
+	if (ftruncate(outfd, stat.st_size) != 0)
+		error(EXIT_FAILURE, errno, "Cannot truncate %s", outfile);
+
+	outbuf = mmap(NULL, stat.st_size, PROT_READ | PROT_WRITE, MAP_SHARED,
+		      outfd, 0);
+	if (outbuf == MAP_FAILED)
+		error(EXIT_FAILURE, errno, "Failed to map %s", outfile);
+
+	close(outfd);
+
+	memcpy(outbuf, inbuf, stat.st_size);
+
+	if (clear_soft_float) {
+		Elf32_Ehdr *outhdr;
+
+		outhdr = outbuf;
+		e_flags &= ~EF_ARM_ABI_FLOAT_SOFT;
+		write_elf_word(e_flags, &outhdr->e_flags, swap);
+	}
+
+	if (msync(outbuf, stat.st_size, MS_SYNC) != 0)
+		error(EXIT_FAILURE, errno, "Failed to sync %s", outfile);
+
+	return EXIT_SUCCESS;
+}
diff --git a/arch/arm/vdso/vgettimeofday.c b/arch/arm/vdso/vgettimeofday.c
new file mode 100644
index 000000000000..9a0e8d093ac0
--- /dev/null
+++ b/arch/arm/vdso/vgettimeofday.c
@@ -0,0 +1,328 @@
+/*
+ * Copyright 2014 Mentor Graphics Corporation.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2 of the
+ * License.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/compiler.h>
+#include <linux/hrtimer.h>
+#include <linux/time.h>
+#include <asm/arch_timer.h>
+#include <asm/barrier.h>
+#include <asm/bug.h>
+#include <asm/page.h>
+#include <asm/unistd.h>
+#include <asm/vdso_datapage.h>
+
+#ifndef CONFIG_AEABI
+#error This code depends on AEABI system call conventions
+#endif
+
+extern struct vdso_data *__get_datapage(void);
+
+static notrace u32 __vdso_read_begin(const struct vdso_data *vdata)
+{
+	u32 seq;
+repeat:
+	seq = ACCESS_ONCE(vdata->seq_count);
+	if (seq & 1) {
+		cpu_relax();
+		goto repeat;
+	}
+	return seq;
+}
+
+static notrace u32 vdso_read_begin(const struct vdso_data *vdata)
+{
+	u32 seq;
+
+	seq = __vdso_read_begin(vdata);
+
+	smp_rmb(); /* Pairs with smp_wmb in vdso_write_end */
+	return seq;
+}
+
+static notrace int vdso_read_retry(const struct vdso_data *vdata, u32 start)
+{
+	smp_rmb(); /* Pairs with smp_wmb in vdso_write_begin */
+	return vdata->seq_count != start;
+}
+
+static notrace long clock_gettime_fallback(clockid_t _clkid,
+					   struct timespec *_ts)
+{
+	register struct timespec *ts asm("r1") = _ts;
+	register clockid_t clkid asm("r0") = _clkid;
+	register long ret asm ("r0");
+	register long nr asm("r7") = __NR_clock_gettime;
+
+	asm volatile(
+	"	swi #0\n"
+	: "=r" (ret)
+	: "r" (clkid), "r" (ts), "r" (nr)
+	: "memory");
+
+	return ret;
+}
+
+static notrace int do_realtime_coarse(struct timespec *ts,
+				      struct vdso_data *vdata)
+{
+	u32 seq;
+
+	do {
+		seq = vdso_read_begin(vdata);
+
+		ts->tv_sec = vdata->xtime_coarse_sec;
+		ts->tv_nsec = vdata->xtime_coarse_nsec;
+
+	} while (vdso_read_retry(vdata, seq));
+
+	return 0;
+}
+
+static notrace int do_monotonic_coarse(struct timespec *ts,
+				       struct vdso_data *vdata)
+{
+	struct timespec tomono;
+	u32 seq;
+
+	do {
+		seq = vdso_read_begin(vdata);
+
+		ts->tv_sec = vdata->xtime_coarse_sec;
+		ts->tv_nsec = vdata->xtime_coarse_nsec;
+
+		tomono.tv_sec = vdata->wtm_clock_sec;
+		tomono.tv_nsec = vdata->wtm_clock_nsec;
+
+	} while (vdso_read_retry(vdata, seq));
+
+	ts->tv_sec += tomono.tv_sec;
+	timespec_add_ns(ts, tomono.tv_nsec);
+
+	return 0;
+}
+
+#ifdef CONFIG_ARM_ARCH_TIMER
+
+static notrace u64 get_ns(struct vdso_data *vdata)
+{
+	u64 cycle_delta;
+	u64 cycle_now;
+	u64 nsec;
+
+	cycle_now = arch_counter_get_cntvct();
+
+	cycle_delta = (cycle_now - vdata->cs_cycle_last) & vdata->cs_mask;
+
+	nsec = (cycle_delta * vdata->cs_mult) + vdata->xtime_clock_snsec;
+	nsec >>= vdata->cs_shift;
+
+	return nsec;
+}
+
+static notrace int do_realtime(struct timespec *ts, struct vdso_data *vdata)
+{
+	u64 nsecs;
+	u32 seq;
+
+	do {
+		seq = vdso_read_begin(vdata);
+
+		if (vdata->use_syscall)
+			return -1;
+
+		ts->tv_sec = vdata->xtime_clock_sec;
+		nsecs = get_ns(vdata);
+
+	} while (vdso_read_retry(vdata, seq));
+
+	ts->tv_nsec = 0;
+	timespec_add_ns(ts, nsecs);
+
+	return 0;
+}
+
+static notrace int do_monotonic(struct timespec *ts, struct vdso_data *vdata)
+{
+	struct timespec tomono;
+	u64 nsecs;
+	u32 seq;
+
+	do {
+		seq = vdso_read_begin(vdata);
+
+		if (vdata->use_syscall)
+			return -1;
+
+		ts->tv_sec = vdata->xtime_clock_sec;
+		nsecs = get_ns(vdata);
+
+		tomono.tv_sec = vdata->wtm_clock_sec;
+		tomono.tv_nsec = vdata->wtm_clock_nsec;
+
+	} while (vdso_read_retry(vdata, seq));
+
+	ts->tv_sec += tomono.tv_sec;
+	ts->tv_nsec = 0;
+	timespec_add_ns(ts, nsecs + tomono.tv_nsec);
+
+	return 0;
+}
+
+#else /* CONFIG_ARM_ARCH_TIMER */
+
+static notrace int do_realtime(struct timespec *ts, struct vdso_data *vdata)
+{
+	return -1;
+}
+
+static notrace int do_monotonic(struct timespec *ts, struct vdso_data *vdata)
+{
+	return -1;
+}
+
+#endif /* CONFIG_ARM_ARCH_TIMER */
+
+notrace int __vdso_clock_gettime(clockid_t clkid, struct timespec *ts)
+{
+	struct vdso_data *vdata;
+	int ret = -1;
+
+	vdata = __get_datapage();
+
+	switch (clkid) {
+	case CLOCK_REALTIME_COARSE:
+		ret = do_realtime_coarse(ts, vdata);
+		break;
+	case CLOCK_MONOTONIC_COARSE:
+		ret = do_monotonic_coarse(ts, vdata);
+		break;
+	case CLOCK_REALTIME:
+		ret = do_realtime(ts, vdata);
+		break;
+	case CLOCK_MONOTONIC:
+		ret = do_monotonic(ts, vdata);
+		break;
+	default:
+		break;
+	}
+
+	if (ret)
+		ret = clock_gettime_fallback(clkid, ts);
+
+	return ret;
+}
+
+static notrace long clock_getres_fallback(clockid_t _clkid,
+					  struct timespec *_ts)
+{
+	register struct timespec *ts asm("r1") = _ts;
+	register clockid_t clkid asm("r0") = _clkid;
+	register long ret asm ("r0");
+	register long nr asm("r7") = __NR_clock_getres;
+
+	asm volatile(
+	"	swi #0\n"
+	: "=r" (ret)
+	: "r" (clkid), "r" (ts), "r" (nr)
+	: "memory");
+
+	return ret;
+}
+
+notrace int __vdso_clock_getres(clockid_t clkid, struct timespec *ts)
+{
+	int ret;
+
+	switch (clkid) {
+	case CLOCK_REALTIME:
+	case CLOCK_MONOTONIC:
+		if (ts) {
+			ts->tv_sec = 0;
+			ts->tv_nsec = MONOTONIC_RES_NSEC;
+		}
+		ret = 0;
+		break;
+	case CLOCK_REALTIME_COARSE:
+	case CLOCK_MONOTONIC_COARSE:
+		if (ts) {
+			ts->tv_sec = 0;
+			ts->tv_nsec = LOW_RES_NSEC;
+		}
+		ret = 0;
+		break;
+	default:
+		ret = clock_getres_fallback(clkid, ts);
+		break;
+	}
+
+	return ret;
+}
+
+static notrace long gettimeofday_fallback(struct timeval *_tv,
+					  struct timezone *_tz)
+{
+	register struct timezone *tz asm("r1") = _tz;
+	register struct timeval *tv asm("r0") = _tv;
+	register long ret asm ("r0");
+	register long nr asm("r7") = __NR_gettimeofday;
+
+	asm volatile(
+	"	swi #0\n"
+	: "=r" (ret)
+	: "r" (tv), "r" (tz), "r" (nr)
+	: "memory");
+
+	return ret;
+}
+
+notrace int __vdso_gettimeofday(struct timeval *tv, struct timezone *tz)
+{
+	struct timespec ts;
+	struct vdso_data *vdata;
+	int ret;
+
+	vdata = __get_datapage();
+
+	ret = do_realtime(&ts, vdata);
+	if (ret)
+		return gettimeofday_fallback(tv, tz);
+
+	if (tv) {
+		tv->tv_sec = ts.tv_sec;
+		tv->tv_usec = ts.tv_nsec / 1000;
+	}
+	if (tz) {
+		tz->tz_minuteswest = vdata->tz_minuteswest;
+		tz->tz_dsttime = vdata->tz_dsttime;
+	}
+
+	return ret;
+}
+
+/* Avoid unresolved references emitted by GCC */
+
+void __aeabi_unwind_cpp_pr0(void)
+{
+}
+
+void __aeabi_unwind_cpp_pr1(void)
+{
+}
+
+void __aeabi_unwind_cpp_pr2(void)
+{
+}
-- 
1.9.3

[PATCH v10 5/6] ARM: vdso initialization, mapping, and synchronization

[Nathan Lynch]

Initialize the vdso page list at boot, install the vdso mapping at
exec time, and update the data page during timer ticks.  This code is
not built if CONFIG_VDSO is not enabled.

Account for the vdso length when randomizing the offset from the
stack.  The vdso is placed immediately following the sigpage with a
separate install_special_mapping call in arm_install_vdso.

Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
---
 arch/arm/kernel/process.c |  17 +++-
 arch/arm/kernel/vdso.c    | 211 ++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 225 insertions(+), 3 deletions(-)
 create mode 100644 arch/arm/kernel/vdso.c

diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index edd2ac3d7c44..7926c2829c05 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -41,6 +41,7 @@
 #include <asm/system_misc.h>
 #include <asm/mach/time.h>
 #include <asm/tls.h>
+#include <asm/vdso.h>
 
 #ifdef CONFIG_CC_STACKPROTECTOR
 #include <linux/stackprotector.h>
@@ -476,7 +477,7 @@ const char *arch_vma_name(struct vm_area_struct *vma)
 }
 
 /* If possible, provide a placement hint at a random offset from the
- * stack for the signal page.
+ * stack for the sigpage and vdso pages.
  */
 static unsigned long sigpage_addr(const struct mm_struct *mm,
 				  unsigned int npages)
@@ -520,6 +521,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 {
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct *vma;
+	unsigned long npages;
 	unsigned long addr;
 	unsigned long hint;
 	int ret = 0;
@@ -529,9 +531,12 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 	if (!signal_page)
 		return -ENOMEM;
 
+	npages = 1; /* for sigpage */
+	npages += vdso_total_pages;
+
 	down_write(&mm->mmap_sem);
-	hint = sigpage_addr(mm, 1);
-	addr = get_unmapped_area(NULL, hint, PAGE_SIZE, 0, 0);
+	hint = sigpage_addr(mm, npages);
+	addr = get_unmapped_area(NULL, hint, npages << PAGE_SHIFT, 0, 0);
 	if (IS_ERR_VALUE(addr)) {
 		ret = addr;
 		goto up_fail;
@@ -548,6 +553,12 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 
 	mm->context.sigpage = addr;
 
+	/* Unlike the sigpage, failure to install the vdso is unlikely
+	 * to be fatal to the process, so no error check needed
+	 * here.
+	 */
+	arm_install_vdso(mm, addr + PAGE_SIZE);
+
  up_fail:
 	up_write(&mm->mmap_sem);
 	return ret;
diff --git a/arch/arm/kernel/vdso.c b/arch/arm/kernel/vdso.c
new file mode 100644
index 000000000000..0f8d9a5f9a24
--- /dev/null
+++ b/arch/arm/kernel/vdso.c
@@ -0,0 +1,211 @@
+/*
+ * Adapted from arm64 version.
+ *
+ * Copyright (C) 2012 ARM Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/err.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/slab.h>
+#include <linux/timekeeper_internal.h>
+#include <linux/vmalloc.h>
+
+#include <asm/barrier.h>
+#include <asm/cacheflush.h>
+#include <asm/page.h>
+#include <asm/vdso.h>
+#include <asm/vdso_datapage.h>
+
+#include <asm/arch_timer.h>
+
+static struct page **vdso_text_pagelist;
+
+/* Total number of pages needed for the data and text portions of the VDSO. */
+unsigned int vdso_total_pages __read_mostly;
+
+/*
+ * The vDSO data page.
+ */
+static union vdso_data_store vdso_data_store __page_aligned_data;
+static struct vdso_data *vdso_data = &vdso_data_store.data;
+
+static struct page *vdso_data_page;
+static struct vm_special_mapping vdso_data_mapping = {
+	.name = "[vvar]",
+	.pages = &vdso_data_page,
+};
+
+static struct vm_special_mapping vdso_text_mapping = {
+	.name = "[vdso]",
+};
+
+static int __init vdso_init(void)
+{
+	unsigned int text_pages;
+	int i;
+
+	if (memcmp(&vdso_start, "\177ELF", 4)) {
+		pr_err("vDSO is not a valid ELF object!\n");
+		return -ENOEXEC;
+	}
+
+	text_pages = (&vdso_end - &vdso_start) >> PAGE_SHIFT;
+	pr_debug("vdso: %i text pages at base %p\n", text_pages, &vdso_start);
+
+	/* Allocate the vDSO text pagelist */
+	vdso_text_pagelist = kcalloc(text_pages, sizeof(struct page *),
+				     GFP_KERNEL);
+	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;
+
+		page = virt_to_page(&vdso_start + i * PAGE_SIZE);
+		vdso_text_pagelist[i] = page;
+	}
+
+	vdso_text_mapping.pages = vdso_text_pagelist;
+
+	vdso_total_pages = 1; /* for the data/vvar page */
+	vdso_total_pages += text_pages;
+
+	return 0;
+}
+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 IS_ERR(vma) ? PTR_ERR(vma) : 0;
+}
+
+/* assumes mmap_sem is write-locked */
+void arm_install_vdso(struct mm_struct *mm, unsigned long addr)
+{
+	struct vm_area_struct *vma;
+	unsigned long len;
+
+	mm->context.vdso = 0;
+
+	if (vdso_text_pagelist == NULL)
+		return;
+
+	if (install_vvar(mm, addr))
+		return;
+
+	/* Account for vvar page. */
+	addr += PAGE_SIZE;
+	len = (vdso_total_pages - 1) << PAGE_SHIFT;
+
+	vma = _install_special_mapping(mm, addr, len,
+		VM_READ | VM_EXEC | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC,
+		&vdso_text_mapping);
+
+	if (!IS_ERR(vma))
+		mm->context.vdso = addr;
+}
+
+static void vdso_write_begin(struct vdso_data *vdata)
+{
+	++vdso_data->seq_count;
+	smp_wmb(); /* Pairs with smp_rmb in vdso_read_retry */
+}
+
+static void vdso_write_end(struct vdso_data *vdata)
+{
+	smp_wmb(); /* Pairs with smp_rmb in vdso_read_begin */
+	++vdso_data->seq_count;
+}
+
+static bool vdso_can_use_arch_timer(const struct timekeeper *tk)
+{
+#ifdef CONFIG_ARM_ARCH_TIMER
+	u32 cntkctl;
+
+	if (strcmp(tk->tkr.clock->name, "arch_sys_counter") != 0)
+		return false;
+
+	cntkctl = arch_timer_get_cntkctl();
+
+	return cntkctl & ARCH_TIMER_USR_VCT_ACCESS_EN;
+#else
+	return false;
+#endif
+}
+
+/**
+ * update_vsyscall - update the vdso data page
+ *
+ * Increment the sequence counter, making it odd, indicating to
+ * userspace that an update is in progress.  Update the fields used
+ * for coarse clocks and, if the architected system timer is in use,
+ * the fields used for high precision clocks.  Increment the sequence
+ * counter again, making it even, indicating to userspace that the
+ * update is finished.
+ *
+ * Userspace is expected to sample seq_count before reading any other
+ * fields from the data page.  If seq_count is odd, userspace is
+ * expected to wait until it becomes even.  After copying data from
+ * the page, userspace must sample seq_count again; if it has changed
+ * from its previous value, userspace must retry the whole sequence.
+ *
+ * Calls to update_vsyscall are serialized by the timekeeping core.
+ */
+void update_vsyscall(struct timekeeper *tk)
+{
+	struct timespec xtime_coarse;
+	struct timespec64 *wtm = &tk->wall_to_monotonic;
+	bool use_syscall = !vdso_can_use_arch_timer(tk);
+
+	vdso_write_begin(vdso_data);
+
+	xtime_coarse = __current_kernel_time();
+	vdso_data->use_syscall			= use_syscall;
+	vdso_data->xtime_coarse_sec		= xtime_coarse.tv_sec;
+	vdso_data->xtime_coarse_nsec		= xtime_coarse.tv_nsec;
+	vdso_data->wtm_clock_sec		= wtm->tv_sec;
+	vdso_data->wtm_clock_nsec		= wtm->tv_nsec;
+
+	if (!use_syscall) {
+		vdso_data->cs_cycle_last	= tk->tkr.cycle_last;
+		vdso_data->xtime_clock_sec	= tk->xtime_sec;
+		vdso_data->xtime_clock_snsec	= tk->tkr.xtime_nsec;
+		vdso_data->cs_mult		= tk->tkr.mult;
+		vdso_data->cs_shift		= tk->tkr.shift;
+		vdso_data->cs_mask		= tk->tkr.mask;
+	}
+
+	vdso_write_end(vdso_data);
+
+	flush_dcache_page(virt_to_page(vdso_data));
+}
+
+void update_vsyscall_tz(void)
+{
+	vdso_data->tz_minuteswest	= sys_tz.tz_minuteswest;
+	vdso_data->tz_dsttime		= sys_tz.tz_dsttime;
+	flush_dcache_page(virt_to_page(vdso_data));
+}
-- 
1.9.3

[PATCH v10 6/6] ARM: add CONFIG_VDSO Kconfig and Makefile bits

[Nathan Lynch]

Allow users to enable the vdso in Kconfig; include the vdso in the
build if CONFIG_VDSO is enabled.  Add 'vdso_install' target.

Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
---
 arch/arm/Makefile        |  8 ++++++++
 arch/arm/kernel/Makefile |  1 +
 arch/arm/mm/Kconfig      | 15 +++++++++++++++
 3 files changed, 24 insertions(+)

diff --git a/arch/arm/Makefile b/arch/arm/Makefile
index 0ce9d0f71f2a..109c011ef4ed 100644
--- a/arch/arm/Makefile
+++ b/arch/arm/Makefile
@@ -264,6 +264,7 @@ core-$(CONFIG_FPE_FASTFPE)	+= $(FASTFPE_OBJ)
 core-$(CONFIG_VFP)		+= arch/arm/vfp/
 core-$(CONFIG_XEN)		+= arch/arm/xen/
 core-$(CONFIG_KVM_ARM_HOST) 	+= arch/arm/kvm/
+core-$(CONFIG_VDSO)		+= arch/arm/vdso/
 
 # If we have a machine-specific directory, then include it in the build.
 core-y				+= arch/arm/kernel/ arch/arm/mm/ arch/arm/common/
@@ -316,6 +317,12 @@ PHONY += dtbs dtbs_install
 dtbs dtbs_install: prepare scripts
 	$(Q)$(MAKE) $(build)=$(boot)/dts MACHINE=$(MACHINE) $@
 
+PHONY += vdso_install
+vdso_install:
+ifeq ($(CONFIG_VDSO),y)
+	$(Q)$(MAKE) $(build)=arch/arm/vdso $@
+endif
+
 # We use MRPROPER_FILES and CLEAN_FILES now
 archclean:
 	$(Q)$(MAKE) $(clean)=$(boot)
@@ -340,4 +347,5 @@ define archhelp
   echo  '                  Install using (your) ~/bin/$(INSTALLKERNEL) or'
   echo  '                  (distribution) /sbin/$(INSTALLKERNEL) or'
   echo  '                  install to $$(INSTALL_PATH) and run lilo'
+  echo  '  vdso_install  - Install unstripped vdso.so to $$(INSTALL_MOD_PATH)/vdso'
 endef
diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
index 38ddd9f83d0e..3e2b80658344 100644
--- a/arch/arm/kernel/Makefile
+++ b/arch/arm/kernel/Makefile
@@ -86,6 +86,7 @@ obj-$(CONFIG_PERF_EVENTS)	+= perf_regs.o
 obj-$(CONFIG_HW_PERF_EVENTS)	+= perf_event.o perf_event_cpu.o
 AFLAGS_iwmmxt.o			:= -Wa,-mcpu=iwmmxt
 obj-$(CONFIG_ARM_CPU_TOPOLOGY)  += topology.o
+obj-$(CONFIG_VDSO)		+= vdso.o
 
 ifneq ($(CONFIG_ARCH_EBSA110),y)
   obj-y		+= io.o
diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index ae69809a9e47..c256323040d6 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -824,6 +824,21 @@ config KUSER_HELPERS
 	  Say N here only if you are absolutely certain that you do not
 	  need these helpers; otherwise, the safe option is to say Y.
 
+config VDSO
+	bool "Enable vDSO for acceleration of some system calls"
+	depends on AEABI && MMU
+	default y if ARM_ARCH_TIMER
+	select GENERIC_TIME_VSYSCALL
+	help
+	  Place in the process address space an ELF shared object
+	  providing fast implementations of several system calls,
+	  including gettimeofday and clock_gettime.  Systems that
+	  implement the ARM architected timer will receive maximum
+	  benefit.
+
+	  You must have glibc 2.21 or later for programs to seamlessly
+	  take advantage of this.
+
 config DMA_CACHE_RWFO
 	bool "Enable read/write for ownership DMA cache maintenance"
 	depends on CPU_V6K && SMP
-- 
1.9.3

[PATCH v10 4/6] ARM: add vdso user-space code

[Venkappa Mala]

Nathan Lynch <nathan_lynch <at> mentor.com> writes:

> +#ifdef CONFIG_ARM_ARCH_TIMER
> +
> +static notrace u64 get_ns(struct vdso_data *vdata)
> +{
> +	u64 cycle_delta;
> +	u64 cycle_now;
> +	u64 nsec;
> +
> +	cycle_now = arch_counter_get_cntvct();

Hello Nathan,

Regarding ARM:vDSO, I have enabled your patch set on Cortex-A7/ARMv7, 
the vDSO is up and running using generic arch timer using physical 
counter but not with VCT.
I anticipate, the issue could be due to VCT cycles.

let me explain the problem first,
   The current validation system at the following state,
    
   commit 65b5732d241b8b39e07653794eefffd0d8028cbb(branch: linux-3.19.y)
   clocksource: arch_timer: Allow the device tree to specify 
uninitialized timer registers

    Some 32-bit (ARMv7) systems are architected like this:

    * The firmware doesn't know and doesn't care about hypervisor mode
      and we don't want to add the complexity of hypervisor there.

    * The firmware isn't involved in SMP bringup or resume.

    * The ARCH timer come up with an uninitialized offset (CNTVOFF)
      between the virtual and physical counters.  Each core gets a
      different random offset.

    * The device boots in "Secure SVC" mode .

    On systems like the above, it doesn't make sense to use the virtual
    counter.  There's nobody managing the offset and each time a core
    goes down and comes back up it will get reinitialized to some other
    random value.
                
    Based on the above, physical timer like this really only makes sense
    for ARMv7 systems. 

    To support vDSO on the above configuration,
        Will arch_counter_get_cntvct() work?
         OR
        Should we use arch_counter_get_cntpct()?

Perhaps, either we need to revise the vDSO frame work to choose VCT or 
PCT dynamically Otherwise CNTVOFF reset to be zero but CNTVOFF could not 
be accessed in SVC mode (PL1).
  
BR,
Venkappa

[PATCH v10 4/6] ARM: add vdso user-space code

[Catalin Marinas]

On Fri, Feb 13, 2015 at 04:36:25AM +0000, Venkappa Mala wrote:
> Nathan Lynch <nathan_lynch <at> mentor.com> writes:
> 
> > +#ifdef CONFIG_ARM_ARCH_TIMER
> > +
> > +static notrace u64 get_ns(struct vdso_data *vdata)
> > +{
> > +	u64 cycle_delta;
> > +	u64 cycle_now;
> > +	u64 nsec;
> > +
> > +	cycle_now = arch_counter_get_cntvct();
> 
> Regarding ARM:vDSO, I have enabled your patch set on Cortex-A7/ARMv7, 
> the vDSO is up and running using generic arch timer using physical 
> counter but not with VCT.
> I anticipate, the issue could be due to VCT cycles.
[...]
> Perhaps, either we need to revise the vDSO frame work to choose VCT or 
> PCT dynamically Otherwise CNTVOFF reset to be zero but CNTVOFF could not 
> be accessed in SVC mode (PL1).

I assume on your platform, the DT also contains
"arm,cpu-registers-not-fw-configured". In such case, the VDSO
optimisation for gettimeofday should just be disabled (as it would be if
the generic timers are not present). Really, just fix the firmware.

-- 
Catalin

[PATCH v10 4/6] ARM: add vdso user-space code

[Venkappa Mala]

Catalin Marinas <catalin.marinas <at> arm.com> writes:

> 
> On Fri, Feb 13, 2015 at 04:36:25AM +0000, Venkappa Mala wrote:
> > Nathan Lynch <nathan_lynch <at> mentor.com> writes:
> > 
> > > +#ifdef CONFIG_ARM_ARCH_TIMER
> > > +
> > > +static notrace u64 get_ns(struct vdso_data *vdata)
> > > +{
> > > +	u64 cycle_delta;
> > > +	u64 cycle_now;
> > > +	u64 nsec;
> > > +
> > > +	cycle_now = arch_counter_get_cntvct();
> > 
> > Regarding ARM:vDSO, I have enabled your patch set on Cortex-
A7/ARMv7, 
> > the vDSO is up and running using generic arch timer using physical 
> > counter but not with VCT.
> > I anticipate, the issue could be due to VCT cycles.
> [...]
> > Perhaps, either we need to revise the vDSO frame work to choose VCT 
or 
> > PCT dynamically Otherwise CNTVOFF reset to be zero but CNTVOFF could 
not 
> > be accessed in SVC mode (PL1).
> 
> I assume on your platform, the DT also contains
> "arm,cpu-registers-not-fw-configured". In such case, the VDSO
> optimisation for gettimeofday should just be disabled (as it would be 
if
> the generic timers are not present). Really, just fix the firmware.
> 
Thanks Catalin for your promote reply.
Yes, DTS contained arm,cpu-registers-not-fw-configured so that it uses 
generic physical timer instead of virtual timer.
Now, the vDSO gettimeofday optimization is also possible since the 
generic timer is presented and enabled.
However, vDSO gettimeofday implementation has been using virtual timer 
(VCT  user access) but the current implementation
will not be sufficient to handle when the firmware is not taken care 
about reset CNTVOFF to zero.
Let?s assume,neither firmware taken care nor booted with HYP mode, then 
Can we use physical timer (PCT user access needs to be enabled) for vDSO 
in the safe manner?

[PATCH v10 4/6] ARM: add vdso user-space code

[Catalin Marinas]

On Fri, Feb 13, 2015 at 12:01:40PM +0000, Venkappa Mala wrote:
> Catalin Marinas <catalin.marinas <at> arm.com> writes:
> > On Fri, Feb 13, 2015 at 04:36:25AM +0000, Venkappa Mala wrote:
> > > Nathan Lynch <nathan_lynch <at> mentor.com> writes:
> > > > +#ifdef CONFIG_ARM_ARCH_TIMER
> > > > +
> > > > +static notrace u64 get_ns(struct vdso_data *vdata)
> > > > +{
> > > > +	u64 cycle_delta;
> > > > +	u64 cycle_now;
> > > > +	u64 nsec;
> > > > +
> > > > +	cycle_now = arch_counter_get_cntvct();
> > > 
> > > Regarding ARM:vDSO, I have enabled your patch set on Cortex-A7/ARMv7, 
> > > the vDSO is up and running using generic arch timer using physical 
> > > counter but not with VCT.
> > > I anticipate, the issue could be due to VCT cycles.
> > [...]
> > > Perhaps, either we need to revise the vDSO frame work to choose VCT or 
> > > PCT dynamically Otherwise CNTVOFF reset to be zero but CNTVOFF could not 
> > > be accessed in SVC mode (PL1).
> > 
> > I assume on your platform, the DT also contains
> > "arm,cpu-registers-not-fw-configured". In such case, the VDSO
> > optimisation for gettimeofday should just be disabled (as it would be if
> > the generic timers are not present). Really, just fix the firmware.
> 
> Thanks Catalin for your promote reply. Yes, DTS contained
> arm,cpu-registers-not-fw-configured so that it uses generic physical
> timer instead of virtual timer. Now, the vDSO gettimeofday
> optimization is also possible since the generic timer is presented and
> enabled. However, vDSO gettimeofday implementation has been using
> virtual timer (VCT  user access) but the current implementation will
> not be sufficient to handle when the firmware is not taken care about
> reset CNTVOFF to zero. Let?s assume,neither firmware taken care nor
> booted with HYP mode, then Can we use physical timer (PCT user access
> needs to be enabled) for vDSO in the safe manner?

The short answer: no.

The problem is that user space is very "innovative" in making use of
features in a different way than what the kernel people intended and
it will be claimed "user ABI" afterwards. Take this glibc revert for
example, luckily we noticed it early:

https://sourceware.org/git/?p=glibc.git;a=commitdiff;h=33ef2f0c763b51e1df7896d7d39d585824558c75

CNTVCT is not intended to be used outside of the VDSO but, apparently,
the VDSO->kernel interface is seen as user ABI by some.

CNTPCT cannot always be used in user space (VDSO and not even the
kernel), especially when the kernel boots at EL1. Choosing CNTPCT vs
CNTVCT in VDSO dynamically (or at boot time) is technically doable but
there is a high risk that user space ends up using them directly when it
detects a gettimeofday VDSO and such code will break when CNTPCT is no
longer accessible.

-- 
Catalin

[PATCH v10 4/6] ARM: add vdso user-space code

[Venkappa Mala]

Catalin Marinas <catalin.marinas <at> arm.com> writes:

> 
> On Fri, Feb 13, 2015 at 12:01:40PM +0000, Venkappa Mala wrote:
> > Catalin Marinas <catalin.marinas <at> arm.com> writes:
> > > On Fri, Feb 13, 2015 at 04:36:25AM +0000, Venkappa Mala wrote:
> > > > Nathan Lynch <nathan_lynch <at> mentor.com> writes:
> > > > > +#ifdef CONFIG_ARM_ARCH_TIMER
> > > > > +
> > > > > +static notrace u64 get_ns(struct vdso_data *vdata)
> > > > > +{
> > > > > +	u64 cycle_delta;
> > > > > +	u64 cycle_now;
> > > > > +	u64 nsec;
> > > > > +
> > > > > +	cycle_now = arch_counter_get_cntvct();
> > > > 
> > > > Regarding ARM:vDSO, I have enabled your patch set on Cortex-
A7/ARMv7, 
> > > > the vDSO is up and running using generic arch timer using 
physical 
> > > > counter but not with VCT.
> > > > I anticipate, the issue could be due to VCT cycles.
> > > [...]
> > > > Perhaps, either we need to revise the vDSO frame work to choose 
VCT or 
> > > > PCT dynamically Otherwise CNTVOFF reset to be zero but CNTVOFF 
could not 
> > > > be accessed in SVC mode (PL1).
> > > 
> > > I assume on your platform, the DT also contains
> > > "arm,cpu-registers-not-fw-configured". In such case, the VDSO
> > > optimisation for gettimeofday should just be disabled (as it would 
be if
> > > the generic timers are not present). Really, just fix the 
firmware.
> > 
> > Thanks Catalin for your promote reply. Yes, DTS contained
> > arm,cpu-registers-not-fw-configured so that it uses generic physical
> > timer instead of virtual timer. Now, the vDSO gettimeofday
> > optimization is also possible since the generic timer is presented 
and
> > enabled. However, vDSO gettimeofday implementation has been using
> > virtual timer (VCT  user access) but the current implementation will
> > not be sufficient to handle when the firmware is not taken care 
about
> > reset CNTVOFF to zero. Let?s assume,neither firmware taken care nor
> > booted with HYP mode, then Can we use physical timer (PCT user 
access
> > needs to be enabled) for vDSO in the safe manner?
> 
> The short answer: no.
> 
> The problem is that user space is very "innovative" in making use of
> features in a different way than what the kernel people intended and
> it will be claimed "user ABI" afterwards. Take this glibc revert for
> example, luckily we noticed it early:
> 
> https://sourceware.org/git/?
p=glibc.git;a=commitdiff;h=33ef2f0c763b51e1df7896d7d39d585824558c75
> 
> CNTVCT is not intended to be used outside of the VDSO but, apparently,
> the VDSO->kernel interface is seen as user ABI by some.
> 
> CNTPCT cannot always be used in user space (VDSO and not even the
> kernel), especially when the kernel boots at EL1. Choosing CNTPCT vs
> CNTVCT in VDSO dynamically (or at boot time) is technically doable but
> there is a high risk that user space ends up using them directly when 
it
> detects a gettimeofday VDSO and such code will break when CNTPCT is no
> longer accessible.
> 

Ok Catalin, I have understood about your concern.
>From ARMv7 TRM,
An implementation of the Generic Timer always includes a virtual 
counter, that indicates virtual time:
[1]  In a processor implementation that does not include the 
Virtualization Extensions, virtual time is identical to physical time, 
and the virtual counter contains the same value as the physical counter.

[2]  If In a processor implementation that includes the Virtualization 
Extensions, then it is already taken care in the kernel with the 
following if boot loader boots with HYP mode,
       commit 0af0b189abf73d232af782df2f999235cd2fed7f 
	ARM: hyp: initialize CNTVOFF to zero
	
Currently, we have been noticing an issue with [1] with virtual offset 
when we enable VCT since it is not an initialize to zero.
Perhaps, the best way to handle this one in firmware itself rather than 
bringing too much complexity.
Are you recommending this one?

If that is true, then we are going to force everyone to take care 
CNTVOFF initialization in the firmware if they want to utilize vDSO on 
ARMv7 where HYP mode is disabled.

BR,
Venkappa

[PATCH v10 4/6] ARM: add vdso user-space code

[Catalin Marinas]

On Fri, Feb 13, 2015 at 02:56:41PM +0000, Venkappa Mala wrote:
> Catalin Marinas <catalin.marinas <at> arm.com> writes:
> > On Fri, Feb 13, 2015 at 12:01:40PM +0000, Venkappa Mala wrote:
> > > Thanks Catalin for your promote reply. Yes, DTS contained
> > > arm,cpu-registers-not-fw-configured so that it uses generic physical
> > > timer instead of virtual timer. Now, the vDSO gettimeofday
> > > optimization is also possible since the generic timer is presented and
> > > enabled. However, vDSO gettimeofday implementation has been using
> > > virtual timer (VCT  user access) but the current implementation will
> > > not be sufficient to handle when the firmware is not taken care about
> > > reset CNTVOFF to zero. Let?s assume,neither firmware taken care nor
> > > booted with HYP mode, then Can we use physical timer (PCT user access
> > > needs to be enabled) for vDSO in the safe manner?
> > 
> > The short answer: no.
> > 
> > The problem is that user space is very "innovative" in making use of
> > features in a different way than what the kernel people intended and
> > it will be claimed "user ABI" afterwards. Take this glibc revert for
> > example, luckily we noticed it early:
> > 
> > https://sourceware.org/git/?p=glibc.git;a=commitdiff;h=33ef2f0c763b51e1df7896d7d39d585824558c75
> > 
> > CNTVCT is not intended to be used outside of the VDSO but, apparently,
> > the VDSO->kernel interface is seen as user ABI by some.
> > 
> > CNTPCT cannot always be used in user space (VDSO and not even the
> > kernel), especially when the kernel boots at EL1. Choosing CNTPCT vs
> > CNTVCT in VDSO dynamically (or at boot time) is technically doable but
> > there is a high risk that user space ends up using them directly when it
> > detects a gettimeofday VDSO and such code will break when CNTPCT is no
> > longer accessible.
> 
> Ok Catalin, I have understood about your concern.
> From ARMv7 TRM,
> An implementation of the Generic Timer always includes a virtual 
> counter, that indicates virtual time:
> [1]  In a processor implementation that does not include the 
> Virtualization Extensions, virtual time is identical to physical time, 
> and the virtual counter contains the same value as the physical counter.
> 
> [2]  If In a processor implementation that includes the Virtualization 
> Extensions, then it is already taken care in the kernel with the 
> following if boot loader boots with HYP mode,
>        commit 0af0b189abf73d232af782df2f999235cd2fed7f 
> 	ARM: hyp: initialize CNTVOFF to zero
> 	
> Currently, we have been noticing an issue with [1] with virtual offset 
> when we enable VCT since it is not an initialize to zero.

I highly doubt you are having a problem with [1] since you mentioned
Cortex-A7 which falls under [2]. If you have a _processor_
implementation with generic timers but no Hyp, as per the ARMv7 ARM the
CNTVCT == CNTPCT, otherwise it is a CPU bug.

What you actually have is [2] but without code to initialise the Hyp
mode. If your kernel is running non-secure, sane Hyp registers
initialisation is mandatory even when the kernel does not boot in Hyp.

> Perhaps, the best way to handle this one in firmware itself rather than 
> bringing too much complexity.
> Are you recommending this one?

Yes, definitely. But note that it's not only (secondary) boot but coming
out of a lower power mode which normally involves firmware anyway.

> If that is true, then we are going to force everyone to take care 
> CNTVOFF initialization in the firmware if they want to utilize vDSO on 
> ARMv7 where HYP mode is disabled.

Even for the kernel, this should be the case unless
"arm,cpu-registers-not-fw-configured" is passed in the DT (this was
really only meant as workaround for broken firmware; I think
Documentation/arm/Booting needs to be updated with the state of the
generic timers on entry).

IIRC there were some discussions about Linux booting in secure mode on
ARMv7 with no firmware to initialise the Hyp registers. If that's the
case for you, we could restart that discussion and try to initialise
CNTVOFF in the kernel running in _secure_ mode (possibly with some CPU
notifiers in the generic timer code).

-- 
Catalin

[PATCH v10 4/6] ARM: add vdso user-space code

[Nathan Lynch]

On 02/12/2015 10:36 PM, Venkappa Mala wrote:
> Nathan Lynch <nathan_lynch <at> mentor.com> writes:
> 
>> +#ifdef CONFIG_ARM_ARCH_TIMER
>> +
>> +static notrace u64 get_ns(struct vdso_data *vdata)
>> +{
>> +	u64 cycle_delta;
>> +	u64 cycle_now;
>> +	u64 nsec;
>> +
>> +	cycle_now = arch_counter_get_cntvct();
> 
> Hello Nathan,
> 
> Regarding ARM:vDSO, I have enabled your patch set on Cortex-A7/ARMv7, 
> the vDSO is up and running using generic arch timer using physical 
> counter but not with VCT.

Hi,

Sorry for not seeing this earlier but it doesn't look like my address
was in the recipients list, and I just happened to catch up on list
traffic today.

It looks like Catalin already covered everything, but just to sum up --
the patches need to be updated to avoid enabling the VDSO gettimeofday
on systems where CNTVCT is not usable, for whatever reason.  I do not
intend to enable user space / VDSO access to CNTPCT to accommodate
systems that don't initialize the virtual counter/offset correctly.

[PATCH v10 4/6] ARM: add vdso user-space code

[Venkappa Mala]

Catalin Marinas <catalin.marinas <at> arm.com> writes:


> I highly doubt you are having a problem with [1] since you mentioned
> Cortex-A7 which falls under [2]. If you have a _processor_
> implementation with generic timers but no Hyp, as per the ARMv7 ARM 
the
> CNTVCT == CNTPCT, otherwise it is a CPU bug.
> 
> What you actually have is [2] but without code to initialise the Hyp
> mode. If your kernel is running non-secure, sane Hyp registers
> initialisation is mandatory even when the kernel does not boot in Hyp.
yes, cortexA7 is falling under [2]. i.e. my mistake. Thanks for the 
correction.

> Even for the kernel, this should be the case unless
> "arm,cpu-registers-not-fw-configured" is passed in the DT (this was
> really only meant as workaround for broken firmware; I think
> Documentation/arm/Booting needs to be updated with the state of the
> generic timers on entry).
It will be good and it helps a lot if it updates.
 
> IIRC there were some discussions about Linux booting in secure mode on
> ARMv7 with no firmware to initialise the Hyp registers. If that's the
> case for you, we could restart that discussion and try to initialise
> CNTVOFF in the kernel running in _secure_ mode (possibly with some CPU
> notifiers in the generic timer code).
My test environment is like CNTVOFF initialization is not taken care in 
the firmware, HYP mode is disabled and kernel is booted with Secure SVC 
PL1 mode. As you have mentioned about restarting the thread, perhaps, it 
will be good to have a discussion in the following  thread rather than 
restarting the thread.
arm: Handle starting up in secure mode
https://lkml.org/lkml/2014/9/16/552

May I know your opinion about this one?

BR,
Venkappa

[PATCH v10 4/6] ARM: add vdso user-space code

[Venkappa Mala]

Nathan Lynch <Nathan_Lynch <at> mentor.com> writes:

> Hi,
> 
> Sorry for not seeing this earlier but it doesn't look like my address
> was in the recipients list, and I just happened to catch up on list
> traffic today.
> 
> It looks like Catalin already covered everything, but just to sum up -
-
> the patches need to be updated to avoid enabling the VDSO gettimeofday
> on systems where CNTVCT is not usable, for whatever reason.  I do not
> intend to enable user space / VDSO access to CNTPCT to accommodate
> systems that don't initialize the virtual counter/offset correctly.
> 

Hello Nathan,

Now, it's clear about current vDSO patch set and it is validated with 
virtual timer/VCT by using workaround with the below results on
ARMv7-A: CortexA7: Quad core, 1.2GHz,

About 1 iteration (gettimeofday),
WITHOUT VDSO   : 30-35 ?sec 
WITH VDSO      : 4-7 ?sec

About 1000 iteration (gettimeofday),
WITHOUT VDSO   : 27-30 msec 
WITH VDSO      : 13-15 msec

As you have mentioned about avoid enabling the VDSO gettimeofday
on systems where CNTVCT is not usable, for whatever reason, please take 
care about this one in the updated patch list so that the things will be 
very clean.
Meanwhile, i will discuss further on "firmware is broken about CNTVOFF 
initialization zero + HYP disabled" in the following thread as mentioned 
in the previous reply to Catalin,
arm: Handle starting up in secure mode
https://lkml.org/lkml/2014/9/16/552

Thanks for the patch list,

BR,
Venkappa

[PATCH v10 4/6] ARM: add vdso user-space code

[Christopher Covington]

Hi Venkappa,

On 02/18/2015 07:12 AM, Venkappa Mala wrote:
> Catalin Marinas <catalin.marinas <at> arm.com> writes:
> 
> 
>> I highly doubt you are having a problem with [1] since you mentioned
>> Cortex-A7 which falls under [2]. If you have a _processor_
>> implementation with generic timers but no Hyp, as per the ARMv7 ARM 
> the
>> CNTVCT == CNTPCT, otherwise it is a CPU bug.
>>
>> What you actually have is [2] but without code to initialise the Hyp
>> mode. If your kernel is running non-secure, sane Hyp registers
>> initialisation is mandatory even when the kernel does not boot in Hyp.
> yes, cortexA7 is falling under [2]. i.e. my mistake. Thanks for the 
> correction.
> 
>> Even for the kernel, this should be the case unless
>> "arm,cpu-registers-not-fw-configured" is passed in the DT (this was
>> really only meant as workaround for broken firmware; I think
>> Documentation/arm/Booting needs to be updated with the state of the
>> generic timers on entry).
> It will be good and it helps a lot if it updates.
>  
>> IIRC there were some discussions about Linux booting in secure mode on
>> ARMv7 with no firmware to initialise the Hyp registers. If that's the
>> case for you, we could restart that discussion and try to initialise
>> CNTVOFF in the kernel running in _secure_ mode (possibly with some CPU
>> notifiers in the generic timer code).
> My test environment is like CNTVOFF initialization is not taken care in 
> the firmware, HYP mode is disabled and kernel is booted with Secure SVC 
> PL1 mode. As you have mentioned about restarting the thread, perhaps, it 
> will be good to have a discussion in the following  thread rather than 
> restarting the thread.
> arm: Handle starting up in secure mode
> https://lkml.org/lkml/2014/9/16/552
> 
> May I know your opinion about this one?

Catalin pointed out that accesses to NSACR may trap. I have not yet found a
reason why trying to write to CNTFRQ wouldn't work. Peter pointed out that
VBAR can't be used on systems without the security extensions, so something
has to be changed there as well.

In the long term I may be able to post an updated version, but I have no idea
how long that might take. In the short term feel free to CC me on any related
work you do, which I might be able to review, test, or both.

Chris

-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH v10 4/6] ARM: add vdso user-space code

[Catalin Marinas]

On Fri, Feb 20, 2015 at 04:52:16PM -0500, Christopher Covington wrote:
> On 02/18/2015 07:12 AM, Venkappa Mala wrote:
> > Catalin Marinas <catalin.marinas <at> arm.com> writes:
> >> IIRC there were some discussions about Linux booting in secure mode on
> >> ARMv7 with no firmware to initialise the Hyp registers. If that's the
> >> case for you, we could restart that discussion and try to initialise
> >> CNTVOFF in the kernel running in _secure_ mode (possibly with some CPU
> >> notifiers in the generic timer code).
> > 
> > My test environment is like CNTVOFF initialization is not taken care in 
> > the firmware, HYP mode is disabled and kernel is booted with Secure SVC 
> > PL1 mode. As you have mentioned about restarting the thread, perhaps, it 
> > will be good to have a discussion in the following  thread rather than 
> > restarting the thread.
> > arm: Handle starting up in secure mode
> > https://lkml.org/lkml/2014/9/16/552
> > 
> > May I know your opinion about this one?
> 
> Catalin pointed out that accesses to NSACR may trap. I have not yet found a
> reason why trying to write to CNTFRQ wouldn't work.

Writing CNTFRQ from non-secure mode would trap as well. For whatever
reason, the architecture doesn't allow you to easily detect whether the
code runs in secure or non-secure mode (unless it is EL2/Hyp which is
always non-secure). The detection mechanism would have to involve
trapping of an undefined instruction but it doesn't need to be done
early during boot.

Otherwise, I think in the arch timer code we could do something like:

	if (!is_hyp_mode_available() && !arch_timer_user_virtual &&
	    arch_timer_ppi[PHYS_SECURE_PPI]) {
		/*
		 * Assume the kernel is running in secure mode. Program
		 * the CNVOFF and CNTFRQ registers.
		 */
		...
	}

-- 
Catalin

[PATCH v10 4/6] ARM: add vdso user-space code

[Christopher Covington]

Hi Catalin,

On 02/24/2015 05:17 AM, Catalin Marinas wrote:

[snip]

> Writing CNTFRQ from non-secure mode would trap as well.

Are you saying it could A) cause an exception targeting the current EL/mode
but not others, or B) cause an exception targeting the current or another EL/mode?

[snip]

> The detection mechanism would have to involve trapping of an undefined
> instruction but it doesn't need to be done early during boot.

I put detection during early boot because I wanted to potentially change
ELs/modes during early boot.

Thanks,
Chris

-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH v10 4/6] ARM: add vdso user-space code

[Catalin Marinas]

On Wed, Feb 25, 2015 at 11:56:36AM -0500, Christopher Covington wrote:
> On 02/24/2015 05:17 AM, Catalin Marinas wrote:
> > Writing CNTFRQ from non-secure mode would trap as well.
> 
> Are you saying it could A) cause an exception targeting the current EL/mode
> but not others, or B) cause an exception targeting the current or another EL/mode?

IIRC, it only traps at the current mode (if non-secure EL1 or EL2).

> > The detection mechanism would have to involve trapping of an undefined
> > instruction but it doesn't need to be done early during boot.
> 
> I put detection during early boot because I wanted to potentially change
> ELs/modes during early boot.

Yes, but it is simpler to install an undef hook later in the boot
process.

-- 
Catalin