[01:35] <fabbione> BenC: i just pushed the e1000 fixes from .16 into my tree
[01:35] <fabbione> they fix the driver hang i was experiencing here every 30 secs
[02:13] <fabbione> BenC: ping?
[04:04] <zul> heylo
[04:04] <zul> BenC: the ipw3945 driver im working is partially working it just cant find the ieee80211 header files.
[04:59] <BenC> fabbione: ok, I'll pull again
[05:00] <fabbione> BenC: great thanks
[05:00] <fabbione> BenC: i also have some issues about ioctl32
[05:00] <fabbione> i am testing a patch right now, but i need you to review it
[05:00] <fabbione> i think you are familiar with that problem since you use ioctl32 compat stuff in firewire too
[05:06] <BenC> yeah, ioctl32 is pretty sticky stuff
[05:07] <fabbione> BenC: well the point is that the entire redhat-cluster-suite seems to require that kind of love
[05:07] <fabbione> i did try to stub them
[05:07] <fabbione> some of them seems to work
[05:07] <fabbione> other don't
[05:07] <BenC> guessing some of them need real conversions
[05:09] <fabbione> example:
[05:09] <fabbione> #define SIOCCLUSTER_ISACTIVE          _IO( 'x', 0x0b)
[05:10] <fabbione> HANDLE_IOCTL(SIOCCLUSTER_ISACTIVE, do_cluster_ioctl)
[05:10] <fabbione> static int do_cluster_ioctl(unsigned int fd, unsigned int cmd, unsigned long arg)
[05:10] <fabbione> {
[05:10] <fabbione>         return sys_ioctl(fd, cmd, (unsigned long)compat_ptr(arg));
[05:10] <fabbione> }
[05:10] <fabbione> this one seems to work just fine
[05:10] <fabbione> #define SIOCCLUSTER_SET_NODENAME      _IOW('x', 0x0b1, char *)
[05:11] <fabbione> HANDLE_IOCTL(SIOCCLUSTER_SET_NODENAME, do_cluster_ioctl)
[05:11] <fabbione> this one gives me:
[05:11] <fabbione> ioctl32(cman_tool:5880): Unknown cmd fd(3) cmd(800478b1){00} arg(ff8f37d8) on socket:[11451]    
[05:11] <fabbione> but the cmd is there..
[05:11] <fabbione> so i am a bit puzzled
[05:13] <fabbione> BenC: http://people.ubuntu.com/~fabbione/cluster-ioctl32.diff this is what i have now
[05:13] <BenC> are you sure oxb1 is right?
[05:13] <BenC> 0x0b1 I mean
[05:14] <BenC> well, I guess 0x0b1 was already in the code
[05:14] <fabbione> i include directly cluster/cnxcman-socket.h
[05:14] <fabbione> so i yes
[05:14] <fabbione> so i think they are correct :)
[05:14] <fabbione> given they work on all other arches :)
[05:14] <fabbione> it's only sparc64 suffering of this problem
[05:15] <fabbione> afair not even hppa was complaining
[05:15] <fabbione> but i might be wrong
[05:15] <fabbione> i need to go out for about 30 minutes
[05:15] <BenC> I think I know why
[05:15] <BenC> it's the char *
[05:15] <fabbione> if you can give it a look or a fix, i will test it when i am back
[05:16] <fabbione> ok i will read when i am back :)
[05:16] <fabbione> thanks Ben
[05:17] <BenC> you  may want to handle this differently
[05:17] <BenC> is this operating on a device file with file_ops?
[05:18] <fabbione> BenC: code here is cluster/* and fs/gfs*
[05:19] <fabbione> i am not sure how i want to operate...
[05:19] <fabbione> personally i don't even care... it needs to work :)
[05:19] <BenC> yeah, it is
[05:19] <BenC> look in cluster/cman/cnxman.c
[05:19] <BenC> and then follow what was done in drivers/ieee1394/dv1394.c
[05:19] <fabbione> i really need to go
[05:19] <fabbione> i will look at it later
[05:19] <BenC> the compat_ioctl entry 
[05:19] <BenC> ok
[05:19] <BenC> I think I did the same for one of the acpi drivers
[05:19] <fabbione> note that these are socket operation
[05:19] <fabbione> +s
[05:20] <fabbione> there are also files_ops
[05:20] <BenC> yeah, the file ops should handle it I believe
[05:44] <fabbione> re
[05:45] <fabbione> so BenC you suggest to remove that patch and go for what's in dv1394.c?
[05:45] <BenC> yeah
[05:47] <fabbione> i can try..
[05:48] <fabbione> but see. there is no file_operations there
[05:50] <fabbione> static struct proto_ops cl_proto_ops = {
[05:51] <fabbione>         .ioctl       = cl_ioctl,
[05:51] <fabbione> }
[05:51] <fabbione> cl_ioctl is the func that knows what userland is calling
[05:51] <fabbione> userland is in redhat-cluster-suite source cman/cman_tools/join.c
[05:51] <fabbione>     error = ioctl(cluster_sock, SIOCCLUSTER_ISACTIVE);
[05:52] <fabbione> it's like SIOCCLUSTER_ISACTIVE is not exported to 32bit userland
[05:52] <fabbione> but proto_ops doesn't have .compat_ioctl
[05:52] <fabbione> (from include/...)
[06:06] <BenC> hmm
[06:07] <BenC> I'd have to take a closer look then
[06:42] <fabbione> BenC: yeah.. i can't get in it
[06:42] <fabbione> no more than what i showed to you
[06:46] <zul_> wq
[06:46] <zul_> damn it
[06:47] <zul_> hey fabbione how is it goinng?
[06:47] <fabbione> okish
[06:47] <zul_> thats good..
[06:47] <fabbione> i just powered down the SAN
[06:47] <fabbione> i was getting almost stupid from the noise
[06:47] <zul_> thats why you wear earphones when you work..
[06:48] <zul_> i went home for lunch..yippe skippe
[06:57] <fabbione> it doesn't help
[06:58] <fabbione> not when a 19" x 6ft racks is completely powered on 3 feet behind you
[06:59] <zul_> meh...you need your own data center
[06:59] <fabbione> did you see the last evolution in my office?
[06:59] <fabbione> the desktop extender?
[06:59] <zul_> nope i havent
[06:59] <fabbione> http://people.ubuntu.com/~fabbione/office/second_layer2.jpg
[06:59] <fabbione> http://people.ubuntu.com/~fabbione/office/second_layer.jpg
[07:00] <zul_> lol...my wife would kill me
[07:01] <fabbione> my wife did look at it for 10 seconds in religious silence...
[07:01] <fabbione> then i looked her and told her: "Why i am so sure you don't like it?"
[07:01] <fabbione> she answered: "If it wasn't for the fact that all this junk bring us food, i would have never married you"
[07:02] <fabbione> so to say, she didn't like it
[07:02] <zul_> lol
[07:03] <fabbione> anyway.. food time
[07:03] <fabbione> i am hungry
[10:19] <zul_> ok i dont like the launchpad changing of descriptions