linux-kernel - Re: [ANNOUNCE] 3.6.11-rt26

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <1379352.tj3b5LpAGs@dabox>
Date:	Thu, 07 Feb 2013 18:11:07 +0100
From:	Tim Sander <tim@...eglstein.org>
To:	Thomas Gleixner <tglx@...utronix.de>
Cc:	LKML <linux-kernel@...r.kernel.org>,
	linux-rt-users <linux-rt-users@...r.kernel.org>
Subject: Re: [ANNOUNCE] 3.6.11-rt26 - interrupts again

Dear Thomas and the preempt-rt community

> I'm pleased to announce the 3.6.11-rt26 release.
Thanks for the release! Unfortunatly it gives me some problems:

I tried to start it on the pcm043 arm1136 but an attached kernel driver of my 
own creates an interrupt storm as i have seen with the serial imx driver as 
reported yesterday on the linux-rt-users list 
(http://marc.info/?l=linux-rt-users&m=136016739407045&w=2). In fact i have 
tried earlier versions of the 3.6-rt kernel series (e.g. 3.6.9-rt21) with the 
same error but i didn't know why the system freeze so i just gave up on this. 
But now i just realized that this error looked exactly like the serial error 
reported!

Some more background: The driver (code attached) reads data from a fifo within 
a fpga attached to the data bus of the processor and distributes the 
characters read to the 4 devices created by the driver. The driver works for 
preempt rt kernels 3.0,3.2,3.4 but not 3.6. The error occurs directly when 
loading the module. Commenting out the request_irq (line 386) the interrupt 
storm disappears but of course the driver stops working. As i have access to 
the interrupt lines on the fpga i have verified the proper function of the 
interrupt line in question.
(It also works with older kernel versions). When loading the module the 
interrupt line is high (i.e. not active) but i still get an interrupt storm? 
So even the line enabling the interrupt within the fpga is not called (line 
144) and obviously it doesn't make a difference commenting out this line.
All these errors seem to indicate that the error lies within the kernel 
infrastructure used by this module and not the module itself. I don't 
understand that only this module shows this strange behaviour and not the 
other modules. Its also strange that the interrupt handler gets called even if 
the interrupt line is high at this moment?! It also seems as if the system is 
not completly unresponsive as i get some printk output from other modules
(albeit very slowly). Serial line and network is completly unresponsive though 
after loading the module.

The kernel itself is not a vanilla rt but with some minor local platform 
adaptions (but definetly not concerning the interrupt subsystem). The kernel 
version is 3.6.11-rt26 but i changed it to 2.6.42 to make the hw debugger 
happy. 

Best regards
Tim
View attachment "msp.c" of type "text/x-csrc" (13990 bytes)

View attachment "kernelconfig" of type "text/plain" (52288 bytes)