Re: loop-AES: timer softirq blocked after USB disconnect?

Matthias Kaehlcke <[email protected]> Thu, 23 Oct 2008 15:59:57 +0200
Newsgroups gmane.linux.cryptography
Organization Electronic Engineering Solutions
Message-ID <[email protected]>
Hi Markus,

markus reichelt wrote:
> * Matthias Kaehlcke <[email protected]> wrote:
> 
>> i totally agree that unplugging the device without unmounting it
>> previously isn't how things should be done and can have side
>> effects.  but is it really obvious that the timer softirq isn't
>> executed during ~250ms because an USB device is uncleanly
>> unplugged?
> 
> I've had my fair share of USB fun myself, yeah.
> 
> The problem is difficult to track, you are using an old kernel. If
> it's possible to use a recent kernel, I'd try that first.

i tried it with 2.6.26.3 (have some prob with USB and 2.6.27) but with 
the same result :(

adding more traces i see that the softirq is raised by the timer ISR but 
not executed

> Also, the USB setup: Is the stick connected directly or via hub? What
> about scheduler? PREEMPT? I'm just taking a wild guess here.

it's connected directly. PREEMPT is enabled, i'll try what happens when 
disabling it

> The USB mailing list at [email protected] might be a better
> place to ask, people there have a deeper knowledge about things that
> could cause your problem. Worth a try.

thanks for the pointer, i'll give it a try

thanks for you response!
mkaehlcke.vcf (text/x-vcard, 336 B)
begin:vcard
fn:Matthias Kaehlcke
n:Kaehlcke;Matthias
org:Electronic Engineering Solutions;Dept. de Software
adr:;;C/ Sant Ferran, 10-16 bajos;Barcelona;;08031;Spain
email;internet:[email protected]
title:Ingeniero de Firmware
tel;work:93 407 30 80
tel;fax:93 433 56 71
x-mozilla-html:FALSE
url:http://www.e2s.net
version:2.1
end:vcard