TLS (GNU/C11 thread-local storage) support for static-linked programs

the design for TLS in dynamic-linked programs is mostly complete too,
but I have not yet implemented it. cost is nonzero but still low for
programs which do not use TLS and/or do not use threads (a few hundred
bytes of new code, plus dependency on memcpy). i believe it can be
made smaller at some point by merging __init_tls and __init_security
into __libc_start_main and avoiding duplicate auxv-parsing code.

at the same time, I've also slightly changed the logic pthread_create
uses to allocate guard pages to ensure that guard pages are not
counted towards commit charge.

2 files changed, 33 insertions, 14 deletions

diff --git a/src/thread/pthread_create.c b/src/thread/pthread_create.c
index d8b8f0fc..f53fc1ba 100644
--- a/src/thread/pthread_create.c
+++ b/src/thread/pthread_create.c
@@ -84,6 +84,8 @@ static void init_file_lock(FILE *f)
 	if (f && f->lock<0) f->lock = 0;
 }
 
+void *__copy_tls(unsigned char *, size_t);
+
 int pthread_create(pthread_t *restrict res, const pthread_attr_t *restrict attr, void *(*entry)(void *), void *restrict arg)
 {
 	int ret;
@@ -92,6 +94,8 @@ int pthread_create(pthread_t *restrict res, const pthread_attr_t *restrict attr,
 	struct pthread *self = pthread_self(), *new;
 	unsigned char *map, *stack, *tsd;
 	unsigned flags = 0x7d8f00;
+	size_t tls_cnt = libc.tls_cnt;
+	size_t tls_size = libc.tls_size;
 
 	if (!self) return ENOSYS;
 	if (!libc.threaded) {
@@ -109,15 +113,24 @@ int pthread_create(pthread_t *restrict res, const pthread_attr_t *restrict attr,
 	} else {
 		if (attr) {
 			guard = ROUND(attr->_a_guardsize + DEFAULT_GUARD_SIZE);
-			size = guard + ROUND(attr->_a_stacksize + DEFAULT_STACK_SIZE);
+			size = guard + ROUND(attr->_a_stacksize
+				+ DEFAULT_STACK_SIZE + tls_size);
 		}
 		size += __pthread_tsd_size;
-		map = mmap(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANON, -1, 0);
-		if (map == MAP_FAILED) return EAGAIN;
-		if (guard) mprotect(map, guard, PROT_NONE);
+		if (guard) {
+			map = mmap(0, size, PROT_NONE, MAP_PRIVATE|MAP_ANON, -1, 0);
+			if (map == MAP_FAILED) return EAGAIN;
+			if (mprotect(map+guard, size-guard, PROT_READ|PROT_WRITE)) {
+				munmap(map, size);
+				return EAGAIN;
+			}
+		} else {
+			map = mmap(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANON, -1, 0);
+			if (map == MAP_FAILED) return EAGAIN;
+		}
 		tsd = map + size - __pthread_tsd_size;
 	}
-	new = (void *)(tsd - sizeof *new - PAGE_SIZE%sizeof *new);
+	new = __copy_tls(tsd - tls_size, tls_cnt);
 	new->map_base = map;
 	new->map_size = size;
 	new->pid = self->pid;
diff --git a/src/thread/pthread_self.c b/src/thread/pthread_self.c
index a8cc80b5..a93d8129 100644
--- a/src/thread/pthread_self.c
+++ b/src/thread/pthread_self.c
@@ -1,6 +1,6 @@
 #include "pthread_impl.h"
 
-static struct pthread main_thread;
+static struct pthread *main_thread = &(struct pthread){0};
 
 /* pthread_key_create.c overrides this */
 static const void *dummy[1] = { 0 };
@@ -10,14 +10,14 @@ static int init_main_thread()
 {
 	__syscall(SYS_rt_sigprocmask, SIG_UNBLOCK,
 		SIGPT_SET, 0, __SYSCALL_SSLEN);
-	if (__set_thread_area(&main_thread) < 0) return -1;
-	main_thread.canceldisable = libc.canceldisable;
-	main_thread.tsd = (void **)__pthread_tsd_main;
-	main_thread.errno_ptr = __errno_location();
-	main_thread.self = &main_thread;
-	main_thread.tid = main_thread.pid =
-		__syscall(SYS_set_tid_address, &main_thread.tid);
-	libc.main_thread = &main_thread;
+	if (__set_thread_area(main_thread) < 0) return -1;
+	main_thread->canceldisable = libc.canceldisable;
+	main_thread->tsd = (void **)__pthread_tsd_main;
+	main_thread->errno_ptr = __errno_location();
+	main_thread->self = main_thread;
+	main_thread->tid = main_thread->pid =
+		__syscall(SYS_set_tid_address, &main_thread->tid);
+	libc.main_thread = main_thread;
 	return 0;
 }
 
@@ -35,3 +35,9 @@ pthread_t __pthread_self_def()
 
 weak_alias(__pthread_self_def, pthread_self);
 weak_alias(__pthread_self_def, __pthread_self_init);
+
+void *__install_initial_tls(void *p)
+{
+	main_thread = p;
+	return __pthread_self_def();
+}