Re: [3/4] ntdll: Implement the timer queue thread.

[Rob Shearman]

Hi Dan,

2008/7/23 Dan Hipschman <dsh(a)linux.ucla.edu>:
> This isn't the most efficient implementation, but it works, and it should
> not be difficult to tweak.  Hence, I'd rather get this version in and add
> the optimizations one at a time, in little patches.  "Get it working first..."

Yes, I think you're right. I don't think it will be too hard to change
the architecture of this to that of my suggestions.

However, I have one minor nit.
> +static void WINAPI timer_queue_thread_proc(LPVOID p)
> +{
> +    struct timer_queue *q = p;
> +    ULONG timeout_ms;
> +    BOOL done;
> +
> +    timeout_ms = INFINITE;
> +    while (!q->quit)
> +    {
> +        LARGE_INTEGER timeout;
> +        DWORD ret = NtWaitForSingleObject(q->event, FALSE,
> +                                          get_nt_timeout(&timeout, timeout_ms));

This should be NTSTATUS.
> +
> +        if (ret == STATUS_TIMEOUT)
> +            queue_timer_expire_next(q);
> +
> +        RtlEnterCriticalSection(&q->cs);
> +        timeout_ms = timer_queue_update(q);
> +        RtlLeaveCriticalSection(&q->cs);
> +    }

-- 
Rob Shearman

[3/4] ntdll: Implement the timer queue thread.

[Alexandre Julliard]

"Rob Shearman" <robertshearman(a)gmail.com> writes:
> 2008/7/23 Dan Hipschman <dsh(a)linux.ucla.edu>:
> > This isn't the most efficient implementation, but it works, and it should
> > not be difficult to tweak.  Hence, I'd rather get this version in and add
> > the optimizations one at a time, in little patches.  "Get it working first..."
>
> Yes, I think you're right. I don't think it will be too hard to change
> the architecture of this to that of my suggestions.

Actually I think the code will be much simpler with a sorted timer list,
so I'd suggest to start with that.

-- 
Alexandre Julliard
julliard(a)winehq.org