lkml.org 
[lkml]   [2003]   [Jul]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] select fix
Valdis.Kletnieks@vt.edu wrote:

>On Tue, 29 Jul 2003 10:36:30 PDT, Andrew Morton said:
>
>
>
>>>- if (tty->driver->chars_in_buffer(tty) < WAKEUP_CHARS)
>>>+ if (!tty->stopped && tty->driver->chars_in_buffer(tty) < WAKEUP_CHARS)
>>> mask |= POLLOUT | POLLWRNORM;
>>>
>>>
>>Manfred sent a patch through esterday which addresses it this way:
>>
>>- if (tty->driver->chars_in_buffer(tty) < WAKEUP_CHARS)
>>+ if (tty->driver->chars_in_buffer(tty) < WAKEUP_CHARS &&
>>+ tty->driver->write_room(tty) > 0)
>>
>>Any preferences?
>>
>>
>
>Would including all 3 conditions make sense? Not sure if it should be A&B&C, or
>A&(B|C) though, but it certainly smells like the write_room() and tty->stopped
>checks are covering 2 different corner cases....
>
>
No. select() and write() must agree when -EAGAIN happens.
write() will fail if write_room() returns 0. Additionally, we want to
delay wakeups a bit, to reduce context switches.
The problem is that the console driver implements stopping by returning
0 from ->write_room() - therefore "less than WAKEUP_CHARS in buffer" is
not equivalent to "write will not return -EAGAIN", and thus user space
loops. My patch fixes that by checking ->write_room() in normal_poll.

Perhaps the Right Thing (tm) is
> if (tty->driver->write_room(tty) > WAKEUP_CHARS)
> mask |= POLLOUT | POLLWRNORM;

but I simply to not understand the tty layer at all, thus I proposed the
minimal patch.

--
Manfred

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:47    [W:0.023 / U:1.608 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site