August 2003 - Darwin-kernel

Apple3Com3C90x kernel extension and OS X 10.2
by Robert J. Accettura 13 Aug '03

13 Aug '03

Looking to get Apple3Com3C90x/ <http://www.opendarwin.org/cgi-bin/cvsweb.cgi/src/Apple3Com3C90x/> built for OS X 10.2.... does anyone know how to do so? Would be great to get working. Regards, Robert J. Accettura robert(a)accettura.com [demime 0.98b removed an attachment of type application/x-pkcs7-signature which had a name of smime.p7s] _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: MALLOC failing..
by Brian Tabone 13 Aug '03

13 Aug '03

Andrew, There is a limit to how much memory you can allocate in the kernel space, if I remember correctly. On top of this, kernel memory is wired, so allocating large chunks has a very negative affect on system performance. I forget specifcally how to set the limit for kernel allocations unfortunately. -Brian On Wednesday, August 13, 2003, at 11:36 AM, Andrew Gallatin wrote: I've got a kext which does some BSD things already, so I'm using the BSD MALLOC(). I've found that MALLOC (ret, void *, len, M_DEVBUF, M_WAITOK); will fail when I attempt to allocate 135167 bytes, and there are ~700MB free (according to top). Is there anything I can do to determine why this allocation is failing? Or, better, to make it succeed? Since there seems to be plenty of ram, I'm guessing that the problem must be a lack of space in the kernel map, or a limit on the number of allocations allowed in some zone. This is Darwin Kernel Version 6.6: Thu May 1 21:48:54 PDT 2003 on a dual g4 with 1280MB of ram. FWIW, it also fails on 6.3. Thanks, Drew _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored. _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

MALLOC failing..
by Andrew Gallatin 13 Aug '03

13 Aug '03

I've got a kext which does some BSD things already, so I'm using the BSD MALLOC(). I've found that MALLOC (ret, void *, len, M_DEVBUF, M_WAITOK); will fail when I attempt to allocate 135167 bytes, and there are ~700MB free (according to top). Is there anything I can do to determine why this allocation is failing? Or, better, to make it succeed? Since there seems to be plenty of ram, I'm guessing that the problem must be a lack of space in the kernel map, or a limit on the number of allocations allowed in some zone. This is Darwin Kernel Version 6.6: Thu May 1 21:48:54 PDT 2003 on a dual g4 with 1280MB of ram. FWIW, it also fails on 6.3. Thanks, Drew _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Is snprintf available?
by Paul R Potts 11 Aug '03

11 Aug '03

Is snprintf available in the MacOS X kernel? It appears to be defined in the header sys/systm.h but including this file in my IOKit driver seems to conflict with including IOKit/IOLib.h; I get a bunch of parse errors in libkern.h. If this function is available, is there a usable public header file I can include for it? Thanks, Paul "Since flak jackets or infrared goggles aren't available to the destitute Liberian fighters, they opt for evening gowns and frilly blouses." --http://slate.msn.com/id/2086490/ _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: How to send an all ones broadcast
by Joshua Graessley 08 Aug '03

08 Aug '03

On Friday, August 8, 2003, at 13:26, Robert G Palmer Jr wrote: Pardon my post if this is not the appropriate list - I looked at darwin-userlevel and it seemed too general. If this isn't the appropriate place, please let me know. I'm working on my cocoa application, which does some network communication. I am using sockets with the sin_addr portion of the sockaddr_in set to INADDR_BROADCAST. I have run into a problem and need some help. I need to send a general network broadcast (IP = 255.255.255.255), but under BSD implementations of sockets, if I specify the general broadcast address, the sockets layer automatically masks it to the interface's subnet, thus creating a subnet broadcast (e.g. if the subnet mask is 255.255.255.0 and my IP is 192.168.20.73, then the broadcast would be 192.168.20.255). I MUST have a full general broadcast. Steven's book says that BSD v3.0 includes a new flag IP_ONESBROADCAST (pg 471) that will do this, but I can't find any reference to this in the OS X headers or documentation (or man pages). Does anyone know how to get OS X to send a general broadcast instead of a subnet broadcast? If you can use a link-local multicast instead of fighting the system to send an all ones broadcast, you'll probably end up much happier. The multicast APIs let you specify which interface a packet should be sent on. You'll have no way of doing this with the broadcast packet, so you can only broadcast on one interface. I think there is a way to send the all ones multicast, but I don't recall what it is. I think it does require root privileges. -josh _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

How to send an all ones broadcast
by Robert G Palmer Jr 08 Aug '03

08 Aug '03

Pardon my post if this is not the appropriate list - I looked at darwin-userlevel and it seemed too general. If this isn't the appropriate place, please let me know. I'm working on my cocoa application, which does some network communication. I am using sockets with the sin_addr portion of the sockaddr_in set to INADDR_BROADCAST. I have run into a problem and need some help. I need to send a general network broadcast (IP = 255.255.255.255), but under BSD implementations of sockets, if I specify the general broadcast address, the sockets layer automatically masks it to the interface's subnet, thus creating a subnet broadcast (e.g. if the subnet mask is 255.255.255.0 and my IP is 192.168.20.73, then the broadcast would be 192.168.20.255). I MUST have a full general broadcast. Steven's book says that BSD v3.0 includes a new flag IP_ONESBROADCAST (pg 471) that will do this, but I can't find any reference to this in the OS X headers or documentation (or man pages). Does anyone know how to get OS X to send a general broadcast instead of a subnet broadcast? Robert Palmer, Jr ----------------------------- Robert G. Palmer, Jr. Lead Engineer, 360 Video Surveillance Internet Pictures Corp. robert.palmer(a)ipix.com (865) 220-6574 _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

TCP Bandwidth Management on OSX: Can the same TCP traffic be passed through TWO ipfw divert rules?
by R Blake 08 Aug '03

08 Aug '03

hi all, I recently discovered a GPL'd util named "Throttled" which bring ipfw-based bandwidth management to OSX, in lieu of "dummynet" (ne1 know WHY dummynet is missing from OSX in the 1st place?). For reference, "Throttled" can be found at: http://www.intrarts.com/throttled.html Throttled "does its thing" by running port-specific tcp traffic through an ipfw divert rule; e.g. - in order to limit upload (outbound) traffic from a port:80 Apache server to $MAXSPEED KB/s, you first initialize throttled (here, to ITS default port:17777) /usr/local/sbin/throttled -s $MAXSPEED -d 17777 and then add an ipfw rule, e.g.: /sbin/ipfw add divert 17777 tcp from $IP 80 to any out xmit $INTERFACE where $IP == machine's PUBLIC IP address, or "any", and $INTERFACE == machine's WAN ethernet interface (e.g., "en0") this works quite nicely, and is straightforward/clear to me, ..... but, NOT when I consider integrating into my NATd env/config. I'm interested in integrating Throttled into my OSX ipfw+natd+dhcp environment. specifically, my NATd runs on $INTERFACE, as is common, at a dynamically assigned public IP (requiring "any" for $IP ...) my gateway LAN address is IP = 10.0.0.1, and my internal web server is IP = 10.0.0.2 NATd is command-line configured with: /usr/sbin/natd -interface $INTERFACE -dynamic -port 8668 -unregistered_only -use_sockets -redirect_port tcp 10.0.0.2:80 80 and the natd ipfw rule is: /sbin/ipfw add 1000 divert natd ip from any to any via $INTERFACE i'm a bit confused as to whether the two divert rules (throttled and natd) will conflict, and/or whether there is an order/priority issue .... bottom line: can throttled & natd CO-EXIST? and if so, what's the proper configuration? i've contacted the Throttled author re: this question, and altho he wasn't certain of the answer, he speculated that divert sockets DO NOT GET REINSERTED into the ipfw rule list after they are processed, thus preventing passing the TCP traffic through BOTH divert rules ... but that there are changeable kernel directives to change this behavior ... any wisdom/insight is much appreciated! blakers _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Q: Darwin panicing too soon?
by Jim Magee 07 Aug '03

07 Aug '03

On Thursday, August 7, 2003, at 5:50 AM, Stiphane Sudre wrote: Le jeudi, 7 ao{ 2003, ` 04:54 Europe/Paris, Jim Magee a icrit : Well, free space is a bit tough to track - especially since the kernel carves its own address space up into a few large submaps. One of those is the zones submap (where all the zones virtual memory allocations reside). You can see what allocations have been made in there by using the zprint(1) utility. But I'm pretty sure that only the kernel debugger will let you see what the maximum size of the submap is - to allow you to compute the remaining free space. Since there's no man page (and a quick search in google pointed to no related zprint stuff), would someone know what the option of this tools mean: usage: zprint [-w] [-s] [-c] [-h] [name] -h is not working the way I'm trying to use it This tool will be very useful to check that I'm not leaking mbuf. But the source is available! ;-) It's part of the system_cmds project. Here's the manpage data (the manpage is installed as part of Panther): ZPRINT(1) ZPRINT(1) NAME zprint - show information about kernel zones SYNOPSIS zprint [-w] [-s] [-c] [-h] [name] DESCRIPTION zprint(1) displays data about Mach zones. By default, zprint will print out information about all Mach zones. If the optional name is specified, zprint will print information about each zone for which name is a substring of the zone's name. zprint interprets the following options: -w For each zone, zprint calculates how much space is allocated but not currently in use, the space wasted by the zone. -s zprint sorts the zones, showing the zones wasting the most memory first. -c (Default) zprint prints zone info in columns. Long zone names are truncated with '$', and spaces are replaced with '.', to allow for sorting by column. Pageable and collectible zones are shown with 'P' and 'C' on the far right. Zones with prepos- terously large maximum sizes are shown with '----' in the max size and max num elts fields. -h (Default) Shows headings for the columns printed with the -c option. It may be useful to override this option when sorting by column. Any option (including default options) can be overridden by specifying the option in upper-case; for example, -C overrides the (default) option -c. --Jim _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Q: Darwin panicing too soon?
by Stéphane Sudre 07 Aug '03

07 Aug '03

Le jeudi, 7 ao{ 2003, ` 04:54 Europe/Paris, Jim Magee a icrit : Well, free space is a bit tough to track - especially since the kernel carves its own address space up into a few large submaps. One of those is the zones submap (where all the zones virtual memory allocations reside). You can see what allocations have been made in there by using the zprint(1) utility. But I'm pretty sure that only the kernel debugger will let you see what the maximum size of the submap is - to allow you to compute the remaining free space. Since there's no man page (and a quick search in google pointed to no related zprint stuff), would someone know what the option of this tools mean: usage: zprint [-w] [-s] [-c] [-h] [name] -h is not working the way I'm trying to use it This tool will be very useful to check that I'm not leaking mbuf. _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Q: Darwin panicing too soon?
by Jim Magee 07 Aug '03

07 Aug '03

On Wednesday, August 6, 2003, at 1:24 PM, Stiphane Sudre wrote: Note that it panics if retval is *different* from KERN_RESOURCE_SHORTAGE". So in case of KERN_RESOURCE_SHORTAGE, whatever comes below will be executed. Correct. The issue is that most kernel code expects zalloc() to wait until memory is available. They only expect a NULL return in the cases that they specified that they are unwilling to block for memory allocation. However, one of the "different from KERN_RESOURCE_SHORTAGE" errors actually occurs as a result of the kernel's virtual address space being exhausted (rather than just being short of free page frames at the moment). It is extremely unlikely that this "exhausted virtual memory space" situation will rectify itself by having the current thread wait. So we panic. This is somewhat analogous to malloc(). It will normally wait for memory to become available (including waiting for the kernel to select and page out other application pages, etc...). But if the virtual address space of your application is completely exhausted, there's not much that can be done. The good news is that the Darwin 10.7 (Panther) PowerPC[1] kernel supports a significantly larger kernel virtual address space. So the odds of hitting that panic go WAY down (most existing machines can't handle enough physical memory to trigger it). There are also additional hand and soft resource limits added to keep applications from inducing run-away kernel allocations. Side question: Is there a way or a tool to monitor kernel memory (free) space? Well, free space is a bit tough to track - especially since the kernel carves its own address space up into a few large submaps. One of those is the zones submap (where all the zones virtual memory allocations reside). You can see what allocations have been made in there by using the zprint(1) utility. But I'm pretty sure that only the kernel debugger will let you see what the maximum size of the submap is - to allow you to compute the remaining free space. --Jim _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0