lkml.org 
[lkml]   [2012]   [Jul]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: long boot delays caused by 070ad7e7 floppy change
On Tue, 3 Jul 2012, Linus Torvalds wrote:

> > It looks like the code meant for -1 to mean "regular delay"
> > (current_reqD), and -2 to mean MAXTIMEOUT. For some reason the "-1"
> > logic has gotten screwed.
>
> So it looks like current_reqD still exists, and sets the drive to
> "current_drive". Which *should* be 0, as far as I can tell. In fact,
> since "current_drive" is printed out by DPRINTK(), we *know* it's 0
> (it's the "0" in the "floppy0:" message).
>
> But how the heck does it pick that bogus 20 second delay then? The -1
> should have been caught by
>
> if (drive == current_reqD)
> drive = current_drive;
>
> and then we should have used the regular delay for drive 0.
>
> Jiri, any ideas?

I am not seeing how this could be happening at the moment. I am also a bit
puzzled by Andi's statement that reverting the conversion to the single wq
is not fixing it.

Have we by any chance changed anything in wq processing during early
bootup lately?

Calvin, Andi, could you please dump the value of UDP->timeout in
__reschedule_timeout(), please?

diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c
index cce7df3..9850120 100644
--- a/drivers/block/floppy.c
+++ b/drivers/block/floppy.c
@@ -675,8 +675,11 @@ static void __reschedule_timeout(int drive, const char *message)
if (drive < 0 || drive >= N_DRIVE) {
delay = 20UL * HZ;
drive = 0;
- } else
+ } else {
+ DPRINT("__reschedule_timeout: setting timeout to %ld\n",
+ UDP->timeout);
delay = UDP->timeout;
+ }

queue_delayed_work(floppy_wq, &fd_timeout, delay);
if (UDP->flags & FD_DEBUG)
--
Jiri Kosina
SUSE Labs


\
 
 \ /
  Last update: 2012-07-03 20:41    [W:0.053 / U:0.308 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site