Skip to contentSkip to conversation

syslets: core code

PATCH In review Ingo Molnar 4 participants 4 messages Feb 15, 2007 – Feb 17, 2007

lore
Share & tools
lore
#0 lore raw

PATCH 5/14 syslets: core code

From: Ingo Molnar <[email protected]>

the core syslet / async system calls infrastructure code.

Is built only if CONFIG_ASYNC_SUPPORT is enabled.

---
 kernel/Makefile |    1 
 kernel/async.c  |  897 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 898 insertions(+)
kernel/Makefile +1 -0
Index: linux/kernel/Makefile
===================================================================
@@ -10,6 +10,7 @@ obj-y     = sched.o fork.o exec_domain.o
 	    kthread.o wait.o kfifo.o sys_ni.o posix-cpu-timers.o mutex.o \
 	    hrtimer.o rwsem.o latency.o nsproxy.o srcu.o
 
+obj-$(CONFIG_ASYNC_SUPPORT) += async.o
 obj-$(CONFIG_STACKTRACE) += stacktrace.o
 obj-y += time/
 obj-$(CONFIG_DEBUG_MUTEXES) += mutex-debug.o
kernel/async.c new file +897 -0

897 lines not on this page · raw message

#1 lore raw

kernel/async.c line 65

Quoted 13 lines
(quoted diff) +2 -0
+static void
+__mark_async_thread_ready(struct async_thread *at, struct  

async_head *ah)

kernel/async.c new file +4 -0
+{
+	list_del(&at->entry);
+	list_add_tail(&at->entry, &ah->ready_async_threads);
 
+__mark_async_thread_busy(struct async_thread *at, struct  

async_head *ah)

kernel/async.c new file +3 -0
+{
+	list_del(&at->entry);
+	list_add_tail(&at->entry, &ah->busy_async_threads);

list_move_tail()?

- z

#2 lore raw
In reply to Ingo Molnar

kernel/async.c line 418

Quoted 9 lines · Ingo Molnar

From: Ingo Molnar <[email protected]>

kernel/async.c new file +8 -0
+static struct syslet_uatom __user *
+exec_atom(struct async_head *ah, struct task_struct *t,
+	  struct syslet_uatom __user *uatom)
+{
+	struct syslet_uatom __user *last_uatom;
+	struct syslet_atom atom;
+	long ret;
+

- run_next:
+ do {

kernel/async.c line 425

Quoted 6 lines
kernel/async.c new file +6 -0
+	if (unlikely(copy_uatom(&atom, uatom)))
+		return ERR_PTR(-EFAULT);
+
+	last_uatom = uatom;
+	ret = __exec_atom(t, &atom);
+	if (unlikely(signal_pending(t) || need_resched()))

+ break;
- goto stop;

kernel/async.c line 428

Quoted 2 lines
kernel/async.c new file +2 -0
+
+	uatom = next_uatom(&atom, uatom, ret);

+ } while(uatom);
- if (uatom)
- goto run_next;
- stop:

kernel/async.c line 439

Quoted 8 lines
kernel/async.c new file +8 -0
+	/*
+	 * We do completion only in async context:
+	 */
+	if (t->at && complete_uatom(ah, t, &atom, last_uatom))
+		return ERR_PTR(-EFAULT);
+
+	return last_uatom;
+}

Goto's are cool granted :), but IMO the "do {} while()" is more natural here.

kernel/async.c line 707

Quoted 9 lines
kernel/async.c new file +9 -0
+asmlinkage long
+sys_async_register(struct async_head_user __user *ahu, unsigned int len)
+{
+	struct task_struct *t = current;
+
+	/*
+	 * This 'len' check enables future extension of
+	 * the async_head ABI:
+	 */

I'm not sure I understand why this isn't a '>' check. If you want to extend the ABI, you'll add members to async_head_user. So you just need to check that you understand all the information that userspace passes to you, or did I missed something?

kernel/async.c line 718

Quoted 11 lines
kernel/async.c new file +11 -0
+	if (len != sizeof(struct async_head_user))
+		return -EINVAL;
+	/*
+	 * Already registered?
+	 */
+	if (t->ah)
+		return -EEXIST;
+
+	return async_head_init(t, ahu);
+}
+

Regards,
Frederik

#3 lore raw
In reply to Ingo Molnar

kernel/async.c line 150

Quoted 13 lines · Ingo Molnar
kernel/async.c new file +13 -0
+	spin_lock(&ah->lock);
+
+	new_async_thread = pick_ready_cachemiss_thread(ah);
+	if (!new_async_thread)
+		goto out_unlock;
+
+	async_ready = t->async_ready;
+	WARN_ON(!async_ready);
+	t->async_ready = NULL;
+
+	new_task = new_async_thread->task;
+
+	move_user_context(new_task, t);

Since I lost the email where you were describing the todo list, I'll hook into this one.
I haven't seen you mentioning it, but at least TID and signal handling need to be swapped to. So it'd be better to have a generic move_user_context() that does system independent work, and an arch_move_user_context() that takes care of system dependent stuff (that is called by move_user_context()).

- Davide