November 2006 - Darwin-kernel

Re: [Solved] Re: Kernel Panic Cluedo
by Michael Smith 23 Nov '06

23 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Thunderbird 1.5.0.8 (Macintosh/20061025) On Nov 22, 2006, at 6:04 PM, Michael Smith wrote: Stephane wrote: It turned out to be a copyin of a bigger size than required. = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Stephane wrote: When you do some bad things in your Kernel Extension (such as writing where you should not, reading a bit too far in buffers, etc...), should a Kernel Panic occur in your Kernel Extension or can it occur later? It depends on what you damage. If you read/write completely out of bounds, your code will fail. But if you destroy a pointer belonging to someone else, there is no way for the hardware/kernel to know a) that that's what you're doing, or b) to blame you when some other code attempts to use it. It could be cool if there was a Kernel simulator with memory protection... Protection of what? Read/write memory in the kernel map is freely writable by any code within the kernel; this is what being a "monolithic" kernel is all about. Given how much vilifiction Darwin has received for being a "microkernel" because we "all know" how inefficient they are, I see this rapidly becoming amusing. This email sent to site_archiver(a)lists.apple.com

1 0

[Solved] Re: Kernel Panic Cluedo
by Stephane 23 Nov '06

23 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 22, 2006, at 6:04 PM, Michael Smith wrote: It turned out to be a copyin of a bigger size than required. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Stephane wrote: When you do some bad things in your Kernel Extension (such as writing where you should not, reading a bit too far in buffers, etc...), should a Kernel Panic occur in your Kernel Extension or can it occur later? It depends on what you damage. If you read/write completely out of bounds, your code will fail. But if you destroy a pointer belonging to someone else, there is no way for the hardware/kernel to know a) that that's what you're doing, or b) to blame you when some other code attempts to use it. It could be cool if there was a Kernel simulator with memory protection... This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kernel Panic Cluedo
by Michael Smith 22 Nov '06

22 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Thunderbird 1.5.0.8 (Macintosh/20061025) = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Stephane wrote: When you do some bad things in your Kernel Extension (such as writing where you should not, reading a bit too far in buffers, etc...), should a Kernel Panic occur in your Kernel Extension or can it occur later? It depends on what you damage. If you read/write completely out of bounds, your code will fail. But if you destroy a pointer belonging to someone else, there is no way for the hardware/kernel to know a) that that's what you're doing, or b) to blame you when some other code attempts to use it. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kernel Panic Cluedo
by Shawn Erickson 22 Nov '06

22 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 22, 2006, at 2:54 AM, Stephane wrote: -Shawn _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… - can you kernel panic outside your own kernel extension if you're doing bad things with memory (till now I've always KPed in my code)? <http://developer.apple.com/documentation/DeviceDrivers/Conceptual/ IOKitFundamentals/Features/chapter_2_section_8.html> This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kernel Panic Cluedo
by Joseph Oreste Bruni 22 Nov '06

22 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 22, 2006, at 3:54 AM, Stephane wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com The kernel is sort of one big program in a single address space. If some function "A" corrupted a memory structure of another function "B", most likely "B" would cause the kernel to crash. Indeed, if memory structure "B" suddenly started pointing into structure "C" because of "A's" mistake, it would be "C" that crashed, etc. Memory protection exists between the kernel and user processes (and between user processes), but not between various regions of the kernel itself. If you ever did programming on OS 9 and earlier you probably experienced lots of random system crashes caused by wayward applications, extensions, etc. since the whole world was one giant address space. - can you kernel panic outside your own kernel extension if you're doing bad things with memory (till now I've always KPed in my code)? smime.p7s

1 0

Re: Kernel Panic Cluedo
by Andrew Gallatin 22 Nov '06

22 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Stephane writes: > > - can you kernel panic outside your own kernel extension if you're > doing bad things with memory (till now I've always KPed in my code)? Sure; it is all one not-so-big address space shared by the kernel and all extensions. Consider what would happen if you malloced some memory, and that memory was located right next to memory used by another driver. You then access that memory out of bounds, changing the memory owned by another driver. That could be bad, especially of the other drive is using the memory to store pointers. Drew _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Kernel Panic Cluedo
by Stephane 22 Nov '06

22 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com or So 2 questions: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… When you do some bad things in your Kernel Extension (such as writing where you should not, reading a bit too far in buffers, etc...), should a Kernel Panic occur in your Kernel Extension or can it occur later? Since yesterday, I'm "enjoying" random Kernel Panics. I know that I changed something in my work-in-progress kernel extensions but every panic.log I look at is pointing to KP in: com.apple.ATIRadeon9700 (Data access DAR=0x0000000000000000 PC=0x00000000002FB7C8) com.apple.driver.AppleFWOHCI (Data access DAR=0x0000000000000000 PC=0x000000000070D4C0) I know that GPU drivers are not 100% perfect (for instance there are still bugs in the current NVidia drivers) and that since yesterday I have an external Firewire drive (actually a Mac running in Firewire Target mode) plugged to my machine. I tested my code in user land for different scenarios and I didn't experience any crash. - are there known issues with these (both PCI) com.apple. drivers (Mac OS X 10.4.8)? - can you kernel panic outside your own kernel extension if you're doing bad things with memory (till now I've always KPed in my code)? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Re: Call function on all processors
by Paul Litvak 21 Nov '06

21 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=SNi/jPGAUJUAggyJA2eRJ5pqW1TiU/9KlqpyCvZmaNvtKq5aeBLiGh9nmjFrtsrSY75TAE8qTb5qfj60h0taz5fDNcqE8emScepaK7jAXZG0jD9Wckvsc38q+05h6pBOmz9HvXjsnMyyzlo03kSDuklwyupk6apHw4uxE1Iurik= Paul Litvak wrote: > How would you call a function on all cpus from a kernel extension? I'm curious; why would you want to do this? The only reasons I can think of have to do with changing per-CPU (vs. per context) state, and the OS owns all of that state. I am working with the Intel virtualization technology and want to invoke the VMXON instruction on all CPUs. - Paul _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… On 11/17/06, Michael Smith <drivers(a)mu.org> wrote: This email sent to site_archiver(a)lists.apple.com

1 0

Re: Call function on all processors
by Michael Smith 17 Nov '06

17 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Thunderbird 1.5.0.8 (Macintosh/20061025) I'm curious; why would you want to do this? For one, I'm guessing this is the mechanism that CHUD tools uses to get the performance counters out of all the CPUs. But I could be wrong. = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Paul Ripke wrote: How would you call a function on all cpus from a kernel extension? The only reasons I can think of have to do with changing per-CPU (vs. per context) state, and the OS owns all of that state. I suppose any other per-CPU state would be fair game - SpeedStep state, thermal state... I can see why some of that might be interesting. Most of it is exposed via other interfaces, or not practical to obtain without racing. The very act of trying to read the SpeedStep state can cause it to change, for example... This email sent to site_archiver(a)lists.apple.com

1 0

Re: Call function on all processors
by Paul Ripke 17 Nov '06

17 Nov '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Mail-followup-to: darwin-kernel(a)lists.apple.com, Paul Litvak <paullitvak(a)gmail.com>, Michael Smith <drivers(a)mu.org> User-agent: Mutt/1.4.2.1i On Fri, Nov 17, 2006 at 03:40:54AM -0800, Michael Smith wrote: > Paul Litvak wrote: > >How would you call a function on all cpus from a kernel extension? > > I'm curious; why would you want to do this? > > The only reasons I can think of have to do with changing per-CPU (vs. > per context) state, and the OS owns all of that state. For one, I'm guessing this is the mechanism that CHUD tools uses to get the performance counters out of all the CPUs. But I could be wrong. I suppose any other per-CPU state would be fair game - SpeedStep state, thermal state... I can see why some of that might be interesting. -- stix _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0