Hi Alistair,
Às 02:54 de 21/10/21, Alistair Francis escreveu:
From: Alistair Francis <redacted>
This commit adds two futex syscall wrappers that are exposed to
userspace.
Neither the kernel or glibc currently expose a futex wrapper, so
userspace is left performing raw syscalls. This has mostly been becuase
because
the overloading of one of the arguments makes it impossible to provide a
single type safe function.
Until recently the single syscall has worked fine. With the introduction
of a 64-bit time_t futex call on 32-bit architectures, this has become
more complex. The logic of handling the two possible futex syscalls is
complex and often implemented incorrectly.
This patch adds two futux syscall functions that correctly handle the
time_t complexity for userspace.
This idea is based on previous discussions: https://lkml.org/lkml/2021/9/21/143
I would use lore
https://lore.kernel.org/lkml/CAK8P3a3x_EyCiPDpMK54y=Rtm-Wb08ym2TNiuAZgXhYrThcWTw@mail.gmail.com/
Signed-off-by: Alistair Francis <redacted>
Thanks for working on that :)
quoted hunk
---
include/uapi/linux/futex_syscall.h | 81 ++++++++++++++++++++++++++++++
1 file changed, 81 insertions(+)
create mode 100644 include/uapi/linux/futex_syscall.h
diff --git a/include/uapi/linux/futex_syscall.h b/include/uapi/linux/futex_syscall.h
new file mode 100644
index 0000000000000..f84a0c68baf78
--- /dev/null
+++ b/include/uapi/linux/futex_syscall.h
@@ -0,0 +1,81 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef _UAPI_LINUX_FUTEX_SYSCALL_H
+#define _UAPI_LINUX_FUTEX_SYSCALL_H
+
+#include <asm/unistd.h>
+#include <errno.h>
+#include <linux/types.h>
+#include <linux/time_types.h>
+#include <sys/syscall.h>
+
+/**
+ * futex_syscall_timeout() - __NR_futex/__NR_futex_time64 syscall wrapper
+ * @uaddr: address of first futex
+ * @op: futex op code
+ * @val: typically expected value of uaddr, but varies by op
+ * @timeout: an absolute struct timespec
+ * @uaddr2: address of second futex for some ops
+ * @val3: varies by op
+ */
+static inline int
+__kernel_futex_syscall_timeout(volatile u_int32_t *uaddr, int op, u_int32_t val,
+ struct timespec *timeout, volatile u_int32_t *uaddr2, int val3)
I tried to write an example[0] that uses this header, but I can't
compile given that u_int32_t isn't defined. Maybe change to uint32_t and
include <stdint.h>?
Also, I got some invalid use of undefined type 'struct timespec', and
#include <time.h> solved.
[0] https://paste.debian.net/1216834/
+{
+#if defined(__NR_futex_time64)
+ if (sizeof(*timeout) != sizeof(struct __kernel_old_timespec)) {
+ int ret = syscall(__NR_futex_time64, uaddr, op, val, timeout, uaddr2, val3);
+
+ if (ret == 0 || errno != ENOSYS)
+ return ret;
+ }
+#endif
+
+#if defined(__NR_futex)
+ if (sizeof(*timeout) == sizeof(struct __kernel_old_timespec))
+ return syscall(__NR_futex, uaddr, op, val, timeout, uaddr2, val3);
+
+ if (timeout && timeout->tv_sec == (long)timeout->tv_sec) {
+ struct __kernel_old_timespec ts32;
+
+ ts32.tv_sec = (__kernel_long_t) timeout->tv_sec;> + ts32.tv_nsec = (__kernel_long_t) timeout->tv_nsec;
+
+ return syscall(__NR_futex, uaddr, op, val, &ts32, uaddr2, val3);
+ } else if (!timeout) {
+ return syscall(__NR_futex, uaddr, op, val, NULL, uaddr2, val3);
+ }
+#endif
If I read this part right, you will always use ts32 for __NR_futex. I
know that it can be misleading, but __NR_futex uses ts64 in 64-bit
archs, so they shouldn't be converted to ts32 in those cases.
Just to make it clear, there's no __NR_futex_time64 at 64-bit archs.
+
+ errno = ENOSYS;
+ return -1;
+}
+
+/**
+ * futex_syscall_nr_requeue() - __NR_futex/__NR_futex_time64 syscall wrapper
+ * @uaddr: address of first futex
+ * @op: futex op code
+ * @val: typically expected value of uaddr, but varies by op
+ * @nr_requeue: an op specific meaning
+ * @uaddr2: address of second futex for some ops
+ * @val3: varies by op
+ */
+static inline int
+__kernel_futex_syscall_nr_requeue(volatile u_int32_t *uaddr, int op, u_int32_t val,
+ u_int32_t nr_requeue, volatile u_int32_t *uaddr2, int val3)
I would always assume that op is FUTEX_CMP_REQUEUE, given that
FUTEX_REQUEUE is racy. From `man futex`:
The FUTEX_CMP_REQUEUE operation was added as a replacement for the
earlier FUTEX_REQUEUE. The difference is that the check of the value at
uaddr can be used to ensure that requeueing happens only under certain
conditions, which allows race conditions to be avoided in certain use cases.
And then we can drop `int op` from the args and give defined
descriptions for the args.
+{
+#if defined(__NR_futex_time64)
+ int ret = syscall(__NR_futex_time64, uaddr, op, val, nr_requeue, uaddr2, val3);
+
+ if (ret == 0 || errno != ENOSYS)
+ return ret;
+#endif
+
+#if defined(__NR_futex)
+ return syscall(__NR_futex, uaddr, op, val, nr_requeue, uaddr2, val3);
+#endif
+
+ errno = ENOSYS;
+ return -1;
+}
+
+#endif /* _UAPI_LINUX_FUTEX_SYSCALL_H */
Sorry if this question was already asked but I didn't find it in the
thread: Should we go with wrappers for the most common op? Like:
__kernel_futex_wait(volatile uint32_t *uaddr, uint32_t val, struct
timespec *timeout)
__kernel_futex_wake(volatile uint32_t *uaddr, uint32_t nr_wake)
Thanks!
André
On Mon, Oct 25, 2021 at 6:33 PM André Almeida [off-list ref] wrote:
Às 02:54 de 21/10/21, Alistair Francis escreveu:
quoted
From: Alistair Francis <redacted>
quoted
+#if defined(__NR_futex)
+ if (sizeof(*timeout) == sizeof(struct __kernel_old_timespec))
+ return syscall(__NR_futex, uaddr, op, val, timeout, uaddr2, val3);
+
+ if (timeout && timeout->tv_sec == (long)timeout->tv_sec) {
+ struct __kernel_old_timespec ts32;
+
+ ts32.tv_sec = (__kernel_long_t) timeout->tv_sec;> + ts32.tv_nsec = (__kernel_long_t) timeout->tv_nsec;
+
+ return syscall(__NR_futex, uaddr, op, val, &ts32, uaddr2, val3);
+ } else if (!timeout) {
+ return syscall(__NR_futex, uaddr, op, val, NULL, uaddr2, val3);
+ }
+#endif
If I read this part right, you will always use ts32 for __NR_futex. I
know that it can be misleading, but __NR_futex uses ts64 in 64-bit
archs, so they shouldn't be converted to ts32 in those cases.
__kernel_old_timespec is the correct type for sys_futex() on all
architectures.
Maybe name the local variable 'ts' or 'ts_old' instead of 'ts32' then?
quoted
+{
+#if defined(__NR_futex_time64)
+ int ret = syscall(__NR_futex_time64, uaddr, op, val, nr_requeue, uaddr2, val3);
+
+ if (ret == 0 || errno != ENOSYS)
+ return ret;
+#endif
+
+#if defined(__NR_futex)
+ return syscall(__NR_futex, uaddr, op, val, nr_requeue, uaddr2, val3);
+#endif
+
+ errno = ENOSYS;
+ return -1;
+}
+
+#endif /* _UAPI_LINUX_FUTEX_SYSCALL_H */
Sorry if this question was already asked but I didn't find it in the
thread: Should we go with wrappers for the most common op? Like:
__kernel_futex_wait(volatile uint32_t *uaddr, uint32_t val, struct
timespec *timeout)
__kernel_futex_wake(volatile uint32_t *uaddr, uint32_t nr_wake)
I had suggested having just a single function definition here, but having one
per argument type seems reasonable as well. Having one definition
per futex_op value would also be possible, but in that case I suppose
we need all 13 of them, not just two.
Arnd
On Tue, Oct 26, 2021 at 3:33 AM Arnd Bergmann [off-list ref] wrote:
On Mon, Oct 25, 2021 at 6:33 PM André Almeida [off-list ref] wrote:
quoted
Às 02:54 de 21/10/21, Alistair Francis escreveu:
quoted
From: Alistair Francis <redacted>
quoted
quoted
+#if defined(__NR_futex)
+ if (sizeof(*timeout) == sizeof(struct __kernel_old_timespec))
+ return syscall(__NR_futex, uaddr, op, val, timeout, uaddr2, val3);
+
+ if (timeout && timeout->tv_sec == (long)timeout->tv_sec) {
+ struct __kernel_old_timespec ts32;
+
+ ts32.tv_sec = (__kernel_long_t) timeout->tv_sec;> + ts32.tv_nsec = (__kernel_long_t) timeout->tv_nsec;
+
+ return syscall(__NR_futex, uaddr, op, val, &ts32, uaddr2, val3);
+ } else if (!timeout) {
+ return syscall(__NR_futex, uaddr, op, val, NULL, uaddr2, val3);
+ }
+#endif
If I read this part right, you will always use ts32 for __NR_futex. I
know that it can be misleading, but __NR_futex uses ts64 in 64-bit
archs, so they shouldn't be converted to ts32 in those cases.
__kernel_old_timespec is the correct type for sys_futex() on all
architectures.
Maybe name the local variable 'ts' or 'ts_old' instead of 'ts32' then?
quoted
quoted
+{
+#if defined(__NR_futex_time64)
+ int ret = syscall(__NR_futex_time64, uaddr, op, val, nr_requeue, uaddr2, val3);
+
+ if (ret == 0 || errno != ENOSYS)
+ return ret;
+#endif
+
+#if defined(__NR_futex)
+ return syscall(__NR_futex, uaddr, op, val, nr_requeue, uaddr2, val3);
+#endif
+
+ errno = ENOSYS;
+ return -1;
+}
+
+#endif /* _UAPI_LINUX_FUTEX_SYSCALL_H */
Sorry if this question was already asked but I didn't find it in the
thread: Should we go with wrappers for the most common op? Like:
__kernel_futex_wait(volatile uint32_t *uaddr, uint32_t val, struct
timespec *timeout)
__kernel_futex_wake(volatile uint32_t *uaddr, uint32_t nr_wake)
I had suggested having just a single function definition here, but having one
per argument type seems reasonable as well. Having one definition
per futex_op value would also be possible, but in that case I suppose
we need all 13 of them, not just two.
Sorry I have taken so long to look at this again. I have addressed
your other comments.
I don't have a strong preference here. I do like that by having a
generic __kernel_futex_syscall_timeout() it should be easier to
convert existing uses of SYSCALL() to the new function (the requeue is
it's own thing anyway).
Alistair
Arnd
On Tue, Oct 26, 2021 at 2:34 AM André Almeida [off-list ref] wrote:
Hi Alistair,
Às 02:54 de 21/10/21, Alistair Francis escreveu:
quoted
From: Alistair Francis <redacted>
This commit adds two futex syscall wrappers that are exposed to
userspace.
Neither the kernel or glibc currently expose a futex wrapper, so
userspace is left performing raw syscalls. This has mostly been becuase
because
quoted
the overloading of one of the arguments makes it impossible to provide a
single type safe function.
Until recently the single syscall has worked fine. With the introduction
of a 64-bit time_t futex call on 32-bit architectures, this has become
more complex. The logic of handling the two possible futex syscalls is
complex and often implemented incorrectly.
This patch adds two futux syscall functions that correctly handle the
time_t complexity for userspace.
This idea is based on previous discussions: https://lkml.org/lkml/2021/9/21/143
I would use lore
https://lore.kernel.org/lkml/CAK8P3a3x_EyCiPDpMK54y=Rtm-Wb08ym2TNiuAZgXhYrThcWTw@mail.gmail.com/
quoted
Signed-off-by: Alistair Francis <redacted>
Thanks for working on that :)
quoted
---
include/uapi/linux/futex_syscall.h | 81 ++++++++++++++++++++++++++++++
1 file changed, 81 insertions(+)
create mode 100644 include/uapi/linux/futex_syscall.h
diff --git a/include/uapi/linux/futex_syscall.h b/include/uapi/linux/futex_syscall.h
new file mode 100644
index 0000000000000..f84a0c68baf78
--- /dev/null
+++ b/include/uapi/linux/futex_syscall.h
@@ -0,0 +1,81 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef _UAPI_LINUX_FUTEX_SYSCALL_H
+#define _UAPI_LINUX_FUTEX_SYSCALL_H
+
+#include <asm/unistd.h>
+#include <errno.h>
+#include <linux/types.h>
+#include <linux/time_types.h>
+#include <sys/syscall.h>
+
+/**
+ * futex_syscall_timeout() - __NR_futex/__NR_futex_time64 syscall wrapper
+ * @uaddr: address of first futex
+ * @op: futex op code
+ * @val: typically expected value of uaddr, but varies by op
+ * @timeout: an absolute struct timespec
+ * @uaddr2: address of second futex for some ops
+ * @val3: varies by op
+ */
+static inline int
+__kernel_futex_syscall_timeout(volatile u_int32_t *uaddr, int op, u_int32_t val,
+ struct timespec *timeout, volatile u_int32_t *uaddr2, int val3)
I tried to write an example[0] that uses this header, but I can't
compile given that u_int32_t isn't defined. Maybe change to uint32_t and
include <stdint.h>?
Also, I got some invalid use of undefined type 'struct timespec', and
#include <time.h> solved.
[0] https://paste.debian.net/1216834/
quoted
+{
+#if defined(__NR_futex_time64)
+ if (sizeof(*timeout) != sizeof(struct __kernel_old_timespec)) {
+ int ret = syscall(__NR_futex_time64, uaddr, op, val, timeout, uaddr2, val3);
+
+ if (ret == 0 || errno != ENOSYS)
+ return ret;
+ }
+#endif
+
+#if defined(__NR_futex)
+ if (sizeof(*timeout) == sizeof(struct __kernel_old_timespec))
+ return syscall(__NR_futex, uaddr, op, val, timeout, uaddr2, val3);
+
+ if (timeout && timeout->tv_sec == (long)timeout->tv_sec) {
+ struct __kernel_old_timespec ts32;
+
+ ts32.tv_sec = (__kernel_long_t) timeout->tv_sec;> + ts32.tv_nsec = (__kernel_long_t) timeout->tv_nsec;
+
+ return syscall(__NR_futex, uaddr, op, val, &ts32, uaddr2, val3);
+ } else if (!timeout) {
+ return syscall(__NR_futex, uaddr, op, val, NULL, uaddr2, val3);
+ }
+#endif
If I read this part right, you will always use ts32 for __NR_futex. I
know that it can be misleading, but __NR_futex uses ts64 in 64-bit
archs, so they shouldn't be converted to ts32 in those cases.
Just to make it clear, there's no __NR_futex_time64 at 64-bit archs.
quoted
+
+ errno = ENOSYS;
+ return -1;
+}
+
+/**
+ * futex_syscall_nr_requeue() - __NR_futex/__NR_futex_time64 syscall wrapper
+ * @uaddr: address of first futex
+ * @op: futex op code
+ * @val: typically expected value of uaddr, but varies by op
+ * @nr_requeue: an op specific meaning
+ * @uaddr2: address of second futex for some ops
+ * @val3: varies by op
+ */
+static inline int
+__kernel_futex_syscall_nr_requeue(volatile u_int32_t *uaddr, int op, u_int32_t val,
+ u_int32_t nr_requeue, volatile u_int32_t *uaddr2, int val3)
I would always assume that op is FUTEX_CMP_REQUEUE, given that
FUTEX_REQUEUE is racy. From `man futex`:
There are other ops that this could be though. From just the kernel
futex self tests it could be FUTEX_WAKE_OP, FUTEX_WAIT_REQUEUE_PI or
FUTEX_CMP_REQUEUE_PI
Alistair
The FUTEX_CMP_REQUEUE operation was added as a replacement for the
earlier FUTEX_REQUEUE. The difference is that the check of the value at
uaddr can be used to ensure that requeueing happens only under certain
conditions, which allows race conditions to be avoided in certain use cases.
And then we can drop `int op` from the args and give defined
descriptions for the args.