Resent-Date: Sun, 29 Nov 1998 05:48:50 +0100 (MET)
From: micha@schmitzm.hip.berkeley.edu (Michael Schmitz)
Subject: Re: 2.1.130 modules patch
In-Reply-To: <199811281302.NAA24714@pcitpdp01.cern.ch> from Jes Sorensen at "Nov 28, 98 01:02:32 pm"
To: Jes.Sorensen@cern.ch (Jes Sorensen)
Date: Sat, 28 Nov 1998 20:41:53 -0800 (PST)
Cc: linux-m68k@lists.linux-m68k.org
Reply-To: mschmitz@lbl.gov
Resent-From: linux-m68k@phil.uni-sb.de

Hi,

> For those of you who got this strange idea about using modules ;-)

For those of us that have these funny ideas about serial devices ...

--- linux-2.1.130/drivers/char/m68kserial.c.org	Sat Nov 28 17:59:45 1998
+++ linux-2.1.130/drivers/char/m68kserial.c	Sat Nov 28 21:31:35 1998
@@ -982,7 +982,6 @@
 		timeout = jiffies+HZ;
 		while (!(info->sw->trans_empty(info))) {
 			current->state = TASK_INTERRUPTIBLE;
-			current->timeout = jiffies + info->timeout;
 			schedule_timeout(info->timeout);
 			if (signal_pending(current))
 				break;
@@ -1002,7 +1001,6 @@
 	if (info->blocked_open) {
 		if (info->close_delay) {
 			current->state = TASK_INTERRUPTIBLE;
-			current->timeout = jiffies + info->close_delay;
 			schedule_timeout(info->close_delay);
 		}
 		wake_up_interruptible(&info->open_wait);

current->timeout is no more (I presume schedule_timeout() is sufficient
here, couldn't actually test it yet. But it compiles, at least).

The rest (Mac stuff) follows later. Is Geert still doing the framebuffer 
device / console patches? 

Alan: using the io_request spinlock indeed fixed all race conditions we had
in the 5380 driver, at least on my Mac. Thanks for dropping that hint ...

	Michael