September 2015 - Darwin-kernel

Re: Kernel crash and Restart (warm boot)
by Prokash Sinha 25 Sep '15

25 Sep '15

Thanks Shantonu, Yeah, that was my reasoning too ( i.e. if I ask for break, it should wait for it to be hooked up). For example, on windows the behavior is to dump the crashed memory. Come to think of it, I think I am asking too much. Not all kernel exception could case a worm reboot, so that I know I’ve a log etc. And that is why I was asking if there is any configuration, that can coerce it do just that. -Prokash On Sep 25, 2015, at 1:36 PM, Shantonu Sen <ssen(a)apple.com> wrote: > Sounds like it’s doing exactly what you asked, which was to wait for the debugger. > > What if you insert a “ud2” instruction? > > Shantonu > >> On Sep 25, 2015, at 11:47 AM, Prokash Sinha <prokash(a)garlic.com> wrote: >> >> Folks, >> >> I’m looking for a way to configure a VM instance off of 10.8.2 plus iso, so that any exception from the kernel cause a system panic, and warm reboots. >> >> What I found, just by testing it that if I try to insert a int3 into a kernel extension , without any debugger attached, the systems GUI hangs. Basically frozen. >> >> Whereas if I do a null de-ref, then I get the crash, and warm reboot. >> >> Is there any configuration that I need ? >> >> Thanks, >> P >> >> >> _______________________________________________ >> 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: >> https://lists.apple.com/mailman/options/darwin-kernel/ssen%40apple.com >> >> This email sent to ssen(a)apple.com > _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Kernel crash and Restart (warm boot)
by Prokash Sinha 25 Sep '15

25 Sep '15

Folks, I’m looking for a way to configure a VM instance off of 10.8.2 plus iso, so that any exception from the kernel cause a system panic, and warm reboots. What I found, just by testing it that if I try to insert a int3 into a kernel extension , without any debugger attached, the systems GUI hangs. Basically frozen. Whereas if I do a null de-ref, then I get the crash, and warm reboot. Is there any configuration that I need ? Thanks, P _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdp_match_name
by Sam Manzer 18 Sep '15

18 Sep '15

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdp_match_name
by Phil Dennis-Jordan 18 Sep '15

18 Sep '15

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdp_match_name
by Stephen Checkoway 18 Sep '15

18 Sep '15

Agreed. -- Stephen Checkoway > On Sep 18, 2015, at 00:58, Stan Sieler <sieler(a)me.com> wrote: > > And that would make sense…if both arguments were potentially unbounded. > But when one’s a constant string, you *know* it has a null at a decent spot! > >> On Sep 17, 2015, at 8:48 PM, Stephen Checkoway <s(a)pahtak.org> wrote: >> >> >>> On Sep 17, 2015, at 5:23 PM, Stan Sieler <sieler(a)me.com> wrote: >>> >>> I’m asking partially to learn if my understanding of when to use >>> strncmp is wrong or not. >> >> strncmp and strncpy are for working with fixed-width fields which may not be NUL-terminated. It's extremely rare that these functions should be used. >> >> The code in question appears to have >> >> char kdpname[80]; >> >> so using strncmp doesn't really matter here. My guess is that it was done because some code paths in PE_parse_boot_argn (those for arguments that aren't strings) don't NUL-terminate. >> >> >> -- >> Stephen Checkoway > > > _______________________________________________ > 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: > https://lists.apple.com/mailman/options/darwin-kernel/s%40pahtak.org > > This email sent to s(a)pahtak.org _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdp_match_name
by Stan Sieler 18 Sep '15

18 Sep '15

And that would make sense…if both arguments were potentially unbounded. But when one’s a constant string, you *know* it has a null at a decent spot! > On Sep 17, 2015, at 8:48 PM, Stephen Checkoway <s(a)pahtak.org> wrote: > > >> On Sep 17, 2015, at 5:23 PM, Stan Sieler <sieler(a)me.com> wrote: >> >> I’m asking partially to learn if my understanding of when to use >> strncmp is wrong or not. > > strncmp and strncpy are for working with fixed-width fields which may not be NUL-terminated. It's extremely rare that these functions should be used. > > The code in question appears to have > > char kdpname[80]; > > so using strncmp doesn't really matter here. My guess is that it was done because some code paths in PE_parse_boot_argn (those for arguments that aren't strings) don't NUL-terminate. > > > -- > Stephen Checkoway > > > > > _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdp_match_name
by Stephen Checkoway 18 Sep '15

18 Sep '15

> On Sep 17, 2015, at 5:23 PM, Stan Sieler <sieler(a)me.com> wrote: > > I’m asking partially to learn if my understanding of when to use > strncmp is wrong or not. strncmp and strncpy are for working with fixed-width fields which may not be NUL-terminated. It's extremely rare that these functions should be used. The code in question appears to have char kdpname[80]; so using strncmp doesn't really matter here. My guess is that it was done because some code paths in PE_parse_boot_argn (those for arguments that aren't strings) don't NUL-terminate. -- Stephen Checkoway _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdp_match_name
by Stan Sieler 17 Sep '15

17 Sep '15

Not a answer for you, but, re: > […] strncmp(kdpname, "serial", sizeof(kdpname)) != 0) That struck me as odd. Wouldn’t ’6’ be much better than sizeof(kdpname)? (Or, wouldn’t using ’strcmp’ be much better here?) After all, if ’kdpname’ happens to be a pointer (not an array), then the size of it could be 4 or 8 … and 4 is too short (allowing “serious” to match :) (Thus, it would be a bug.) Alternatively, if kdpname is an array of 7 or more bytes, why bother with the ’n’ of strncmp, since we know the comparison will stop after loading the 7th byte, worst case (thus, no ’n’ needed to limit the loop). If kdpname is an array of 6 or less bytes, then 'strcmp (kdpname, “serial”)’ would give the compiler a chance to say “hey, that will never match”, but the ’strncmp’ wouldn’t provide that opportunity. This looks like a great candidate for a simple “strcmp”. I’m asking partially to learn if my understanding of when to use strncmp is wrong or not. (I’m not interested in the performance difference here … yeah, the strncmp is slower, but by such a small amount that the savings earned (across all future runs) by recoding is more than lost by the cost of the time to read this sentence :) thanks, Stan _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Specifying vector number when registering and ISR. Is it possible ?
by Prokash Sinha 16 Sep '15

16 Sep '15

Hi All, I’m trying to find a way, if any, to craft a driver ( without any hardware) to install an ISR at some IDT vector. It is going to run under VM ( always ), and want to be able to use say some high vector number like 0x7e ( usually the ones that are expected to be used by device interrupts). May be this is not possible, if the underlying mechanism is MSI or MSIe and vector assignment is purely under kernel control. I would appreciate, any thoughts and answer(s) on it. Regards, prokash _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdp_match_name
by Brian Bechtel 16 Sep '15

16 Sep '15

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0