Re: long boot delays caused by 070ad7e7 floppy change

[Andi Kleen <andi@firstfloor.org>]

On Tue, Jul 03, 2012 at 07:57:15PM +0200, Jiri Kosina wrote:
> On Tue, 3 Jul 2012, Linus Torvalds wrote:
> 
> > > Ok, here's the dmesg output that I get:
> > >
> > > [    0.714316] floppy0: lock_fdc: -1 (out of 8)
> > 
> > Ok, thanks. That -1 definitely explains the long delay. It's using
> > "drive=-1" because it has no actual drive number, just the
> > *controller* number.
> > 
> > And the backtrace shows what's going on too: it's the
> > 
> >   user_reset_fdc(-1, FD_RESET_ALWAYS, false);
> > 
> > that passes in -1 as the floppy controller number.
> > 
> > Ugh.
> > 
> > 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.
> 
> I am still trying to understand the exact situation.
> 
> Andi, do I understand you correctly that even if you revert 070ad7e793dc 
> on top of current Linus' tree, you are still seeing the 20 second delay 
> inside the initcall?

Correct. Here's the debug output with just reverted (without Linus
patch):

[    2.489559] calling  floppy_init+0x0/0xe46 @ 1
[    2.504363] floppy0: lock_fdc: -1 (out of 8)
[    2.504410] floppy0: reschedule timeout lock fdc
[   22.498784] floppy0: reschedule timeout do wakeup
[   22.498811] floppy0: no floppy controllers found
[   22.498944] initcall floppy_init+0x0/0xe46 returned -19 after
19577158 usecs