Hello!
Ang Way Chuang wrote:
Hi all,
I am resending this email because the last email bounced back due to the large attachment.
Now the attachment has been trimmed down.
CRC-32 checking during ULE decapsulation always failed on x86_64 system due to the
size of a variable used to store CRC? Discovered this bug on Fedora Core 6 with
kernel-2.6.18-1.2849. The i686 counterpart has no such problem. Attached is
the output of dmesg showing the problem. Look for the last few lines for indication.
This patch has been tested on 64-bit system as well as 32-bit system.
Guess this bug should affect 2.6.19 as well since dvb_net.c hasn't
changed. Other kernel versions may be affected too. I am not sure why there is extra 0xff... for
Because it is initialized with the expression ~0. If it were 0xFFFFFFFF, things would work fine.
64 bit unsigned long (I know it is 64 bit wide in 64-bit system) :-D. If this patch is relevant,
should I send a similar patch to Adrian Bunk for his 2.6.16.x tree? I haven't tested 2.6.16.x on my
64 bit system. Thanks in advance
In my opinion, this patch should be included into the Linux-DVB kernel tree ASAP. I think Mauro would do that.
Did you also send your patch to A. Morton (the Linux 2.6 kernel maintainer)?
Cheers,
Christian.
--
Christian Praehauser <cpraehaus@xxxxxxxxxxxxxx>
Department of Computer Sciences, University of Salzburg
_______________________________________________
linux-dvb mailing list
linux-dvb@xxxxxxxxxxx
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
