June 2002 - Darwin-kernel - lists.apple.com

Re: VFS Stack Development
by Justin C. Walker 06 Jun '02

06 Jun '02

On Wednesday, June 5, 2002, at 07:41 , Zack Angelo wrote: Hi all, I'm looking for information/code on developing VFS stack components. I've checked all over ADC and the mailing-list archives and the only thing I could find is the 3 or 4 pages in the Kernel developer doc. If anyone could point me to any information I'd greatly appreciate it. Thanks in advance. There's precious little information on this, in the form of documentation. You can check source at Regards, Justin -- /~\ The ASCII Justin C. Walker, Curmudgeon-at-Large \ / Ribbon Campaign X Against HTML / \ Email _______________________________________________ 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?: Kernel Memory Mapped to User Space
by Andrew Gallatin 06 Jun '02

06 Jun '02

Godfrey van der Linden writes: > As Jim said it is usually better to map user memory into the kernel > and have the driver wire and unwire it. You don't really have to > worry about the app aborting without your being aware of it as the > act of wiring means that the memory stays valid until the DRIVER is > ready to get rid of it. Unless you're wiring things for short bounded amounts of time, then you *do* need to worry about it. If the app attempts to free the wired memory, it will deadlock because the VM system is waiting for the driver to unwire the memory. But the driver has no idea the VM system wants it to unwire the memory. Take a look at vm_map_delete() in System/xnu/osfmk/vm/vm_map.c The comments explain things clearly: <...> /* * All our DMA I/O operations in IOKit are currently done by * wiring through the map entries of the task requesting the I/O. * Because of this, we must always wait for kernel wirings * to go away on the entries before deleting them. * * Any caller who wants to actually remove a kernel wiring * should explicitly set the VM_MAP_REMOVE_KUNWIRE flag to * properly remove one wiring instead of blasting through * them all. */ flags |= VM_MAP_REMOVE_WAIT_FOR_KWIRE; <...> Note that VM_MAP_REMOVE_WAIT_FOR_KWIRE is set unconditionally. Later on when trying to remove an entry whose wired count is nonzero: if (flags & VM_MAP_REMOVE_WAIT_FOR_KWIRE) { s = entry->vme_start; entry->needs_wakeup = TRUE; vm_map_entry_wait(map, interruptible); At this point the interruptible flag is false (if it came through vm_deallocate, which it almost certainly did), so the process is unkillable. As far as I can tell, OS-X gives a driver no way to know that an app has shot its foot off by attempting to free memory you've wired for it. The only way to recover from this is to have the driver unwire the memory through operator intervention. This is a serious problem for our software -- GM driver for Myrinet (low-latency, high-bandwidth cluster interconnect, http://www.myri.com) where user processes "register" (wire) large parts of their address space for unbounded amounts of time for the purpose of zero-copy OS-bypass networking. The problem only happens when users violate our API which says you must "deregister" (unwire) memory before freeing it. But its still a nasty situation & nearly impossible for a hapless user to debug. (They'll just complain their app "works fine on Linux" & potentially not use OS-X.). I really wish I could get somebody at Apple interested in fixing this by providing some way for a driver to detect this situation. (Setting a flag you can get to off the proc struct like Solaris does would be nice) Thanks, -- Drew Gallatin (direct email gallatin AT myri DOT com) Member of Technical Staff Myricom Inc. 1575 Yauger Rd., Suite #34 Mount Vernon, OH. 43050 (740) 397-8629 _______________________________________________ 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?: Kernel Memory Mapped to User Space
by Godfrey van der Linden 06 Jun '02

06 Jun '02

G'day Scott, At 18:08 -0700 02-6-5, Scott Taggart wrote: Hi, I have seen various posts, responses, etc. on this subject, but quite frankly, I got lost. I have the need to take a slug of kernel memory in my driver and make it available to my user app so that my app can access it directly (I'll take care of arbitration and protection with my driver). Can someone give me the simplest set of calls I need to do this? I am willing to research the calls but just need the general set of calls required and order they should be called along with the appropriate caveats. The memory I need to map to user land is allocated by my driver at startup and is wired. As Jim said it is usually better to map user memory into the kernel and have the driver wire and unwire it. You don't really have to worry about the app aborting without your being aware of it as the act of wiring means that the memory stays valid until the DRIVER is ready to get rid of it. By far the easiest way of doing this sort of thing is to pass a buffer in with an open style routine. Wrap it in a memory descriptor and register it with your driver. Your driver can then prepare() (i.e. wire) it and map() it to get a virtual address to access the memory. Now is also a good time to grab the physical addresses for any future DMA you may want to do. Godfrey _______________________________________________ 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?: Kernel Memory Mapped to User Space
by Jim Magee 06 Jun '02

06 Jun '02

On Wednesday, June 5, 2002, at 09:08 PM, Scott Taggart wrote: I have seen various posts, responses, etc. on this subject, but quite frankly, I got lost. I have the need to take a slug of kernel memory in my driver and make it available to my user app so that my app can access it directly Does it have to go this way? Can you take a slug of user memory and make it available in the driver? That's the easiest and most straightforward (create a IOMemoryDescriptor from the user task and address range, and map it into the kernel). (I'll take care of arbitration and protection with my driver). ...he says unwittingly before really knowing what he is signing up for.... ;-) Oh, you mean inside that hunk of memory. That's the easy part. It's the managing of access to the memory as a whole (from outside pressures like copy-on-write, etc...) that's the tricky part. And it's also why you need to stick with the standard approaches. --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

How To?: Kernel Memory Mapped to User Space
by Scott Taggart 06 Jun '02

06 Jun '02

Hi, I have seen various posts, responses, etc. on this subject, but quite frankly, I got lost. I have the need to take a slug of kernel memory in my driver and make it available to my user app so that my app can access it directly (I'll take care of arbitration and protection with my driver). Can someone give me the simplest set of calls I need to do this? I am willing to research the calls but just need the general set of calls required and order they should be called along with the appropriate caveats. The memory I need to map to user land is allocated by my driver at startup and is wired. Thanks, Scott _______________________________________________ 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: Threading in IOKit (was: Protection Questions)
by Sean McBride 06 Jun '02

06 Jun '02

Godfrey van der Linden (gvdl(a)apple.com) wrote: >Also you have to watch out for various scheduling priorities. > >Broadly we have > >1> Primary Interrupts >2> Real Time Threads (user land only) >3> Timer threads (Thread Call threads, kernel only) >4> Kernel threads (IOWorkLoop aka 'secondary interrupts') >5> Window Manager Threads >6> The rest of user land threads. > >Sometimes the priorites seems a bit screwed up Indeed. It seems the only way a driver can have as high priority as a userland app is to use primary interrupts, which are strongly discouraged. _______________________________________________ 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

VFS Stack Development
by Zack Angelo 05 Jun '02

05 Jun '02

Hi all, I'm looking for information/code on developing VFS stack components. I've checked all over ADC and the mailing-list archives and the only thing I could find is the 3 or 4 pages in the Kernel developer doc. If anyone could point me to any information I'd greatly appreciate it. Thanks in advance. -Zack _______________________________________________ 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: IORegistry crash
by Francis bouchard 05 Jun '02

05 Jun '02

Hi, Thanks all for your help. But I sill have the problem. But more info now. I've managed to removed a lot of code from my effect driver. I also made sure that it was working fine without the framebuffer. Here is what happen: I have three static library used by this project. It seems to have problem "big time" to link with those libraries. Now if I play (adding and removing them from the project) gets me in trouble. I can kextload my dirver, but the ioregistry will stop working and I won't be able to kextunload... If I stub my kext and don'T link with those libraries it works fine(for what it is supposed to do whithout the libs!) Any Idea ? Can I merge the two libs in one with ld ? Francis _______________________________________________ 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: Undefined "atexit" in kext driver
by Jed Davis 04 Jun '02

04 Jun '02

Scott Taggart <taggart(a)taggarts.org> writes: > Last time I tracked this down, it came down to me having a static > defined in a function!!! > > f() > { > static something x; > } > > I moved x out of the function and the ref to atexit() went away. I > wrote it off to compiler weirdness and went on. This time, I can't go > through the dozens of source files doing a binary search with #ifdefs, > etc. That makes a certain amount of sense, actually: with C++ running in a hosted environment (i.e. a normal user program), a static (or global?) object will have its destructor called when the program exits; the compiler, assuming a full C library, implements this with atexit(). These semantics of course make no sense for a KEXT, but apparently the compiler doesn't know that. --Jed -- #!/usr/bin/perl -- ## "But life wasn't yes-no, on-off. Life was shades of gray, sub f{(($n,$d,@_)=@_)?(substr(## and rainbows not in the order of the spectrum." " ExhortJavelinBus",$n&&$d/$n,1),$n?f($d## -- L. E. Modesitt, Jr., _Adiamante_ %$n,$n,@_):&f):("\n")}print f 1461,10324,55001,444162,1208,1341,5660480,79715997 _______________________________________________ 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: Undefined "atexit" in kext driver
by Scott Taggart 04 Jun '02

04 Jun '02

Last time I tracked this down, it came down to me having a static defined in a function!!! f() { static something x; } I moved x out of the function and the ref to atexit() went away. I wrote it off to compiler weirdness and went on. This time, I can't go through the dozens of source files doing a binary search with #ifdefs, etc. Scott At 02:10 PM 6/4/2002, Justin C. Walker wrote: On Monday, June 3, 2002, at 08:46 , Scott Taggart wrote: Hi, I get an undefined "atexit" symbol when I load my kext. I have seen this before but it was very hard to track down (I have dozens of files and none explicitly reference atexit()). I presume this is my code referencing some library function that is referencing atexit. I searched the archives and could find no mention of this. Any ideas on how to find where this is referenced? Have you tried 'nm' on your various objects? Note that you can't use the system's normal "C Library" in kernel extensions (don't know if you know this; just thought I'd point it out, since the appearance of 'atexit' is pretty weird). Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------*-- _______________________________________________ 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