nptl: Reinstate pthread_timedjoin_np as a cancellation point (BZ#24215)

Patch ce7eb0e90315 ("nptl: Cleanup cancellation macros") changed the
join sequence for internal common __pthread_timedjoin_ex to use the
new macro lll_wait_tid.  The idea was this macro would issue the
cancellable futex operation depending whether the timeout is used or
not.  However if a timeout is used, __lll_timedwait_tid is called and
it is not a cancellable entrypoint.

This patch fixes it by simplifying the code in various ways:

  - Instead of adding the cancellation handling on __lll_timedwait_tid,
    it moves the generic implementation to pthread_join_common.c (called
    now timedwait_tid with some fixes to use the correct type for pid).

  - The llvm_wait_tid macro is removed, along with its replication on
    x86_64, i686, and sparc arch-specific lowlevellock.h.

  - sparc32 __lll_timedwait_tid is also removed, since the code is similar
    to generic one.

  - x86_64 and i386 provides arch-specific __lll_timedwait_tid which is
    also removed since they are similar in functionality to generic C code
    and there is no indication it is better than compiler generated code.

New tests, tst-join8 and tst-join9, are provided to check if
pthread_timedjoin_np acts as a cancellation point.

Checked on x86_64-linux-gnu, i686-linux-gnu, sparcv9-linux-gnu, and
aarch64-linux-gnu.

	[BZ #24215]
	* nptl/Makefile (lpthread-routines): Remove lll_timedwait_tid.
	(tests): Add tst-join8 tst-join9.
	* nptl/lll_timedwait_tid.c: Remove file.
	* sysdeps/sparc/sparc32/lll_timedwait_tid.c: Likewise.
	* sysdeps/unix/sysv/linux/i386/lll_timedwait_tid.c: Likewise.
	* sysdeps/sysv/linux/x86_64/lll_timedwait_tid.c: Likewise.
	* nptl/pthread_join_common.c (timedwait_tid): New function.
	(__pthread_timedjoin_ex): Act as cancellation entrypoint is block
	is set.
	* nptl/tst-join5.c (thread_join): New function.
	(tf1, tf2, do_test): Use libsupport and add pthread_timedjoin_np
	check.
	* nptl/tst-join8.c: New file.
	* nptl/tst-join9.c: Likewise.
	* sysdeps/nptl/lowlevellock-futex.h (lll_futex_wait_cancel,
	lll_futex_timed_wait_cancel): Add generic macros.
	* sysdeps/nptl/lowlevellock.h (__lll_timedwait_tid, lll_wait_tid):
	Remove definitions.
	* sysdeps/unix/sysv/linux/i386/lowlevellock.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/lowlevellock.h: Likewise.
	* sysdeps/unix/sysv/linux/x86_64/lowlevellock.h: Likewise.
	* sysdeps/sparc/sparc32/lowlevellock.c (__lll_timedwait_tid):
	Remove function.
	* sysdeps/unix/sysv/linux/i386/lowlevellock.S (__lll_timedwait_tid):
	Likewise.
	* sysdeps/unix/sysv/linux/x86_64/lowlevellock.S: Likewise.
	* sysdeps/unix/sysv/linux/lowlevellock-futex.h
	(lll_futex_timed_wait_cancel): New macro.

8 files changed, 7 insertions, 226 deletions

diff --git a/sysdeps/unix/sysv/linux/i386/lll_timedwait_tid.c b/sysdeps/unix/sysv/linux/i386/lll_timedwait_tid.c
deleted file mode 100644
index 43900c6294..0000000000
--- a/sysdeps/unix/sysv/linux/i386/lll_timedwait_tid.c
+++ /dev/null
@@ -1 +0,0 @@
-/* __lll_timedwait_tid is in lowlevellock.S.  */
diff --git a/sysdeps/unix/sysv/linux/i386/lowlevellock.S b/sysdeps/unix/sysv/linux/i386/lowlevellock.S
index b5f5b5b11c..83191a3899 100644
--- a/sysdeps/unix/sysv/linux/i386/lowlevellock.S
+++ b/sysdeps/unix/sysv/linux/i386/lowlevellock.S
@@ -365,70 +365,4 @@ __lll_unlock_wake:
 	ret
 	cfi_endproc
 	.size	__lll_unlock_wake,.-__lll_unlock_wake
-
-	.globl	__lll_timedwait_tid
-	.type	__lll_timedwait_tid,@function
-	.hidden	__lll_timedwait_tid
-	.align	16
-__lll_timedwait_tid:
-	pushl	%edi
-	pushl	%esi
-	pushl	%ebx
-	pushl	%ebp
-
-	movl	%eax, %ebp
-	movl	%edx, %edi
-	subl	$8, %esp
-
-	/* Get current time.  */
-2:	movl	%esp, %ebx
-	xorl	%ecx, %ecx
-	movl	$__NR_gettimeofday, %eax
-	ENTER_KERNEL
-
-	/* Compute relative timeout.  */
-	movl	4(%esp), %eax
-	movl	$1000, %edx
-	mul	%edx		/* Milli seconds to nano seconds.  */
-	movl	(%edi), %ecx
-	movl	4(%edi), %edx
-	subl	(%esp), %ecx
-	subl	%eax, %edx
-	jns	5f
-	addl	$1000000000, %edx
-	subl	$1, %ecx
-5:	testl	%ecx, %ecx
-	js	6f		/* Time is already up.  */
-
-	movl	%ecx, (%esp)	/* Store relative timeout.  */
-	movl	%edx, 4(%esp)
-
-	movl	(%ebp), %edx
-	testl	%edx, %edx
-	jz	4f
-
-	movl	%esp, %esi
-	/* XXX The kernel so far uses global futex for the wakeup at
-	   all times.  */
-	xorl	%ecx, %ecx	/* movl $FUTEX_WAIT, %ecx */
-	movl	%ebp, %ebx
-	movl	$SYS_futex, %eax
-	ENTER_KERNEL
-
-	cmpl	$0, (%ebx)
-	jne	1f
-4:	xorl	%eax, %eax
-
-3:	addl	$8, %esp
-	popl	%ebp
-	popl	%ebx
-	popl	%esi
-	popl	%edi
-	ret
-
-1:	cmpl	$-ETIMEDOUT, %eax
-	jne	2b
-6:	movl	$ETIMEDOUT, %eax
-	jmp	3b
-	.size	__lll_timedwait_tid,.-__lll_timedwait_tid
 #endif
diff --git a/sysdeps/unix/sysv/linux/i386/lowlevellock.h b/sysdeps/unix/sysv/linux/i386/lowlevellock.h
index aa8f4bfdf6..94dccc4ce7 100644
--- a/sysdeps/unix/sysv/linux/i386/lowlevellock.h
+++ b/sysdeps/unix/sysv/linux/i386/lowlevellock.h
@@ -219,31 +219,6 @@ extern int __lll_timedlock_elision (int *futex, short *adapt_count,
 #define lll_islocked(futex) \
   (futex != LLL_LOCK_INITIALIZER)
 
-extern int __lll_timedwait_tid (int *tid, const struct timespec *abstime)
-     __attribute__ ((regparm (2))) attribute_hidden;
-
-/* The kernel notifies a process which uses CLONE_CHILD_CLEARTID via futex
-   wake-up when the clone terminates.  The memory location contains the
-   thread ID while the clone is running and is reset to zero by the kernel
-   afterwards.  The kernel up to version 3.16.3 does not use the private futex
-   operations for futex wake-up when the clone terminates.
-   If ABSTIME is not NULL, is used a timeout for futex call.  If the timeout
-   occurs then return ETIMEOUT, if ABSTIME is invalid, return EINVAL.
-   The futex operation are issues with cancellable versions.  */
-#define lll_wait_tid(tid, abstime)					\
-  ({									\
-    int __res = 0;							\
-    __typeof (tid) __tid;						\
-    if (abstime != NULL)						\
-      __res = __lll_timedwait_tid (&(tid), (abstime));			\
-    else								\
-      /* We need acquire MO here so that we synchronize with the 	\
-	 kernel's store to 0 when the clone terminates. (see above)  */	\
-      while ((__tid = atomic_load_acquire (&(tid))) != 0)		\
-        lll_futex_wait_cancel (&(tid), __tid, LLL_SHARED);		\
-    __res;								\
-  })
-
 extern int __lll_lock_elision (int *futex, short *adapt_count, int private)
   attribute_hidden;
 
diff --git a/sysdeps/unix/sysv/linux/lowlevellock-futex.h b/sysdeps/unix/sysv/linux/lowlevellock-futex.h
index 8ca3417aaa..6f060b1739 100644
--- a/sysdeps/unix/sysv/linux/lowlevellock-futex.h
+++ b/sysdeps/unix/sysv/linux/lowlevellock-futex.h
@@ -135,6 +135,13 @@
     __err;							       \
   })
 
+#define lll_futex_timed_wait_cancel(futexp, val, timeout, private)	   \
+  ({									   \
+    int __oldtype = CANCEL_ASYNC ();				       	   \
+    long int __err = lll_futex_timed_wait (futexp, val, timeout, private); \
+    CANCEL_RESET (__oldtype);						   \
+    __err;								   \
+  })
 
 #endif  /* !__ASSEMBLER__  */
 
diff --git a/sysdeps/unix/sysv/linux/sparc/lowlevellock.h b/sysdeps/unix/sysv/linux/sparc/lowlevellock.h
index 5aed572d43..c778f4259b 100644
--- a/sysdeps/unix/sysv/linux/sparc/lowlevellock.h
+++ b/sysdeps/unix/sysv/linux/sparc/lowlevellock.h
@@ -108,29 +108,4 @@ __lll_timedlock (int *futex, const struct timespec *abstime, int private)
 #define LLL_LOCK_INITIALIZER		(0)
 #define LLL_LOCK_INITIALIZER_LOCKED	(1)
 
-extern int __lll_timedwait_tid (int *, const struct timespec *)
-     attribute_hidden;
-
-/* The kernel notifies a process which uses CLONE_CHILD_CLEARTID via futex
-   wake-up when the clone terminates.  The memory location contains the
-   thread ID while the clone is running and is reset to zero by the kernel
-   afterwards.  The kernel up to version 3.16.3 does not use the private futex
-   operations for futex wake-up when the clone terminates.
-   If ABSTIME is not NULL, is used a timeout for futex call.  If the timeout
-   occurs then return ETIMEOUT, if ABSTIME is invalid, return EINVAL.
-   The futex operation are issues with cancellable versions.  */
-#define lll_wait_tid(tid, abstime)					\
-  ({									\
-    int __res = 0;							\
-    __typeof (tid) __tid;						\
-    if (abstime != NULL)						\
-      __res = __lll_timedwait_tid (&(tid), (abstime));			\
-    else								\
-      /* We need acquire MO here so that we synchronize with the 	\
-	 kernel's store to 0 when the clone terminates. (see above)  */	\
-      while ((__tid = atomic_load_acquire (&(tid))) != 0)		\
-        lll_futex_wait_cancel (&(tid), __tid, LLL_SHARED);		\
-    __res;								\
-  })
-
 #endif	/* lowlevellock.h */
diff --git a/sysdeps/unix/sysv/linux/x86_64/lll_timedwait_tid.c b/sysdeps/unix/sysv/linux/x86_64/lll_timedwait_tid.c
deleted file mode 100644
index 43900c6294..0000000000
--- a/sysdeps/unix/sysv/linux/x86_64/lll_timedwait_tid.c
+++ /dev/null
@@ -1 +0,0 @@
-/* __lll_timedwait_tid is in lowlevellock.S.  */
diff --git a/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S b/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S
index f9abb4cb1f..166dbcfd3b 100644
--- a/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S
+++ b/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S
@@ -345,87 +345,4 @@ __lll_unlock_wake:
 	retq
 	cfi_endproc
 	.size	__lll_unlock_wake,.-__lll_unlock_wake
-
-	.globl	__lll_timedwait_tid
-	.type	__lll_timedwait_tid,@function
-	.hidden	__lll_timedwait_tid
-	.align	16
-__lll_timedwait_tid:
-	cfi_startproc
-	pushq	%r12
-	cfi_adjust_cfa_offset(8)
-	pushq	%r13
-	cfi_adjust_cfa_offset(8)
-	cfi_offset(%r12, -16)
-	cfi_offset(%r13, -24)
-
-	movq	%rdi, %r12
-	movq	%rsi, %r13
-
-	/* Align stack to 16 bytes when calling __gettimeofday.  */
-	subq	$24, %rsp
-	cfi_adjust_cfa_offset(24)
-
-	/* Get current time.  */
-2:	movq	%rsp, %rdi
-	xorl	%esi, %esi
-	/* This call works because we directly jump to a system call entry
-	   which preserves all the registers.  */
-	call	JUMPTARGET(__gettimeofday)
-
-	/* Compute relative timeout.  */
-	movq	8(%rsp), %rax
-	movl	$1000, %edi
-	mul	%rdi		/* Milli seconds to nano seconds.  */
-	movq	(%r13), %rdi
-	movq	8(%r13), %rsi
-	subq	(%rsp), %rdi
-	subq	%rax, %rsi
-	jns	5f
-	addq	$1000000000, %rsi
-	decq	%rdi
-5:	testq	%rdi, %rdi
-	js	6f		/* Time is already up.  */
-
-	movq	%rdi, (%rsp)	/* Store relative timeout.  */
-	movq	%rsi, 8(%rsp)
-
-	movl	(%r12), %edx
-	testl	%edx, %edx
-	jz	4f
-
-	movq	%rsp, %r10
-	/* XXX The kernel so far uses global futex for the wakeup at
-	   all times.  */
-#if FUTEX_WAIT == 0
-	xorl	%esi, %esi
-#else
-	movl	$FUTEX_WAIT, %esi
-#endif
-	movq	%r12, %rdi
-	movl	$SYS_futex, %eax
-	syscall
-
-	cmpl	$0, (%rdi)
-	jne	1f
-4:	xorl	%eax, %eax
-
-8:	addq	$24, %rsp
-	cfi_adjust_cfa_offset(-24)
-	popq	%r13
-	cfi_adjust_cfa_offset(-8)
-	cfi_restore(%r13)
-	popq	%r12
-	cfi_adjust_cfa_offset(-8)
-	cfi_restore(%r12)
-	retq
-
-	cfi_adjust_cfa_offset(32)
-1:	cmpq	$-ETIMEDOUT, %rax
-	jne	2b
-
-6:	movl	$ETIMEDOUT, %eax
-	jmp	8b
-	cfi_endproc
-	.size	__lll_timedwait_tid,.-__lll_timedwait_tid
 #endif
diff --git a/sysdeps/unix/sysv/linux/x86_64/lowlevellock.h b/sysdeps/unix/sysv/linux/x86_64/lowlevellock.h
index 47d4eabd61..8cbc1caa5b 100644
--- a/sysdeps/unix/sysv/linux/x86_64/lowlevellock.h
+++ b/sysdeps/unix/sysv/linux/x86_64/lowlevellock.h
@@ -222,31 +222,6 @@ extern int __lll_timedlock_elision (int *futex, short *adapt_count,
 #define lll_islocked(futex) \
   (futex != LLL_LOCK_INITIALIZER)
 
-extern int __lll_timedwait_tid (int *, const struct timespec *)
-     attribute_hidden;
-
-/* The kernel notifies a process which uses CLONE_CHILD_CLEARTID via futex
-   wake-up when the clone terminates.  The memory location contains the
-   thread ID while the clone is running and is reset to zero by the kernel
-   afterwards.  The kernel up to version 3.16.3 does not use the private futex
-   operations for futex wake-up when the clone terminates.
-   If ABSTIME is not NULL, is used a timeout for futex call.  If the timeout
-   occurs then return ETIMEOUT, if ABSTIME is invalid, return EINVAL.
-   The futex operation are issues with cancellable versions.  */
-#define lll_wait_tid(tid, abstime)					\
-  ({									\
-    int __res = 0;							\
-    __typeof (tid) __tid;						\
-    if (abstime != NULL)						\
-      __res = __lll_timedwait_tid (&(tid), (abstime));			\
-    else								\
-      /* We need acquire MO here so that we synchronize with the 	\
-	 kernel's store to 0 when the clone terminates. (see above)  */	\
-      while ((__tid = atomic_load_acquire (&(tid))) != 0)		\
-        lll_futex_wait_cancel (&(tid), __tid, LLL_SHARED);		\
-    __res;								\
-  })
-
 extern int __lll_lock_elision (int *futex, short *adapt_count, int private)
   attribute_hidden;