May 2005 - Darwin-kernel - lists.apple.com

64-bit problem with VM statistics
by Tony Scaminaci 31 May '05

31 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.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: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com vm_size_t mac_pagesize; unsigned long long FreeMem; // Free real (physical) memory in bytes . . . FreeMem = ((unsigned long long) vm_stat.free_count) * mac_pagesize; // Calculate total free memory in bytes FreeMem = (unsigned long long) (vm_stat.free_count * mac_pagesize);

1 0

Re: encryption in kernel
by Jake Luck 27 May '05

27 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Does anyone know of an encryption library and/or source that can be used in a kernel extension? xnu/bsd/crypto/ jake _______________________________________________ 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

Re: kpi socket filter: Safe to block in callbacks?
by Brian Bergstrand 27 May '05

27 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I'm thinking worse case scenario here. It would be possible for us to issue our own callback to be serviced on another thread, but I'm afraid of lag issues with that model. This brings up a question: why are there no try variants for R/W and mutex locks in the new lock KPI (lck_*)? On May 26, 2005, at 8:36 PM, Vincent Lubet wrote: > It is safe up to the point it takes a long time to resolve the > contention because some events are notified on the input thread and > you don't want to hold up all the incoming network traffic. > > Vincent > > On May 25, 2005, at 3:05 PM, Josh Graessley wrote: > > >> >> That should be safe. >> >> -josh >> >> On May 25, 2005, at 2:38 PM, Brian Bergstrand wrote: >> >> >> >>> -----BEGIN PGP SIGNED MESSAGE----- >>> Hash: SHA1 >>> >>> Is it safe to block on a R/W lock inside a socket filter callback? >>> I'm specifically interested in sf_notify_func(). I doubt lock >>> contention will be an issue for us at the point the notify callback >>> happens, but it's always possible. > Brian Bergstrand <http://www.bergstrand.org/brian/> PGP Key ID: 0xB6C7B6A2 -----BEGIN PGP SIGNATURE----- Version: PGP Desktop 9.0.1 (Build 2130) iQA/AwUBQpaYO3nR2Fu2x7aiEQIHQACeNtwDFQa5Z0SKSYDy8/3SuhRaposAni8a A0/z2ta9SnKxiJ64wb4cbSed =cyv+ -----END PGP SIGNATURE----- _______________________________________________ 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

Re: kpi socket filter: Safe to block in callbacks?
by Vincent Lubet 27 May '05

27 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Vincent On May 25, 2005, at 3:05 PM, Josh Graessley wrote: That should be safe. -josh On May 25, 2005, at 2:38 PM, Brian Bergstrand wrote: -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Is it safe to block on a R/W lock inside a socket filter callback? I'm specifically interested in sf_notify_func(). I doubt lock contention will be an issue for us at the point the notify callback happens, but it's always possible. Thanks. Brian Bergstrand <http://www.bergstrand.org/brian/> PGP Key ID: 0xB6C7B6A2 -----BEGIN PGP SIGNATURE----- Version: PGP Desktop 9.0.1 (Build 2130) This email sent to jgraessley(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: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… It is safe up to the point it takes a long time to resolve the contention because some events are notified on the input thread and you don't want to hold up all the incoming network traffic. iQA/AwUBQpTwTXnR2Fu2x7aiEQLNgwCglmmXfGDO2+bs5Qs0n+GB5lN1ORUAn2zN 6Z4+Pzx1v65Y7efnWDgw8ShM =nVmj -----END PGP SIGNATURE----- _______________________________________________ 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/jgraessley% 40apple.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: http://lists.apple.com/mailman/options/darwin-kernel/vlubet% 40apple.com This email sent to vlubet(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: encryption in kernel
by Josh Graessley 26 May '05

26 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -josh On May 27, 2005, at 3:08 PM, Jake Luck wrote: Does anyone know of an encryption library and/or source that can be used in a kernel extension? xnu/bsd/crypto/ _______________________________________________ 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 crypto code from those headers is not guarteed to not change. I believe it isn't exported in the 10.4.x kernel and later. We may solidify the API to those various things in the future and make them available to kexts. If there is something specific that you'd like access to, it's best to file a bug. jake smime.p7s

1 0

Re: kauth and creds in Darwin 8
by Godfrey van der Linden 26 May '05

26 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com ml_get_interrupts_enabled():- Use preemption_enabled() from Mach KPIs. klogwakeup is internal API use IOLog to achieve the same effect. I personally don't know anything about the kauth APIs. Godfrey On 05/26/2005, at 9:08 , Brian Bergstrand wrote: -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I'm in the same boat as Sam. There's some exports missing (1 critical to our product) and it sounds like linking against the kernel proper is going to be a support nightmare. I've filed radars about the missing symbols, but Apple seems to move slower than dirt in integrating these changes and I'm really afraid of a "behaves correctly". Here's the missing symbols: ml_get_interrupts_enabled() -- this is the critical one -- we use it as a switch to determine whether to call OSMalloc or OSMalloc_noblock thread_policy_get() The weird thing about these two is that their _set counterparts are exported. Why would set be exported but not get? host_info() klogwakeup() The last three I can probably disable or workaround, but I don't know of a way to work around ml_get_interrupts_enabled() missing. I thought the new preemption_enabled() would work, but it's not exported either! Radar: 4130443 On May 26, 2005, at 6:35 AM, Andrew Gallatin wrote: Sam Vaughan writes: On 26/05/2005, at 2:48 PM, Sam Vaughan wrote: What does a dependency on the kernel itself look like in a kext's OSBundleLibraries dict? This seems to do the trick: <key>com.apple.kernel</key> <string>8.1.0</string> ... but I assume it'll then refuse to load when I install 10.4.2. Yep. It will break every time your customers upgrade. :-( Drew Brian Bergstrand <http://www.bergstrand.org/brian/> PGP Key ID: 0xB6C7B6A2 -----BEGIN PGP SIGNATURE----- Version: PGP Desktop 9.0.1 (Build 2130) iQA/AwUBQpX0dHnR2Fu2x7aiEQL2PACgt/9YFGfImOmlMIBR+zDVe7QJPfoAoIdn jueR9najqAIVpGx4gAO0q2Hw =5pof -----END PGP SIGNATURE----- _______________________________________________ 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/gvdl%40apple.com This email sent to gvdl(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: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… thread_policy_get() - "There is no need for developer to call thread_policy_get()" host_info() - Use sysctlbyname() from Libkern KPIs to get exported system info. "In general, we haven't removing any KPIs. Tiger does publish the first KPI set and new KPIs can be added to cover missing functionality" This email sent to site_archiver(a)lists.apple.com

1 0

Re: kauth and creds in Darwin 8
by Brian Bergstrand 26 May '05

26 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On May 26, 2005, at 11:08 AM, Brian Bergstrand wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > I'm in the same boat as Sam. There's some exports missing (1 critical > to our product) and it sounds like linking against the kernel proper > is going to be a support nightmare. I've filed radars about the > missing symbols, but Apple seems to move slower than dirt in > integrating these changes and I'm really afraid of a "behaves > correctly". > > Here's the missing symbols: > > ml_get_interrupts_enabled() -- this is the critical one -- we use it > as a switch to determine whether to call OSMalloc or OSMalloc_noblock > FFR, preemption_enabled() is exported by the Mach KPI, but is not in the headers. My fault. Brian Bergstrand <http://www.bergstrand.org/brian/> PGP Key ID: 0xB6C7B6A2 -----BEGIN PGP SIGNATURE----- Version: PGP Desktop 9.0.1 (Build 2130) iQA/AwUBQpYssXnR2Fu2x7aiEQI7rACgkNnJ4t6+s3vIqOQR270v9twBkdYAoKLu XU6EvyQjro7J+/Ffngo43Ia2 =axnG -----END PGP SIGNATURE----- _______________________________________________ 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

Re: kauth and creds in Darwin 8
by Brian Bergstrand 26 May '05

26 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I'm in the same boat as Sam. There's some exports missing (1 critical to our product) and it sounds like linking against the kernel proper is going to be a support nightmare. I've filed radars about the missing symbols, but Apple seems to move slower than dirt in integrating these changes and I'm really afraid of a "behaves correctly". Here's the missing symbols: ml_get_interrupts_enabled() -- this is the critical one -- we use it as a switch to determine whether to call OSMalloc or OSMalloc_noblock thread_policy_get() The weird thing about these two is that their _set counterparts are exported. Why would set be exported but not get? host_info() klogwakeup() The last three I can probably disable or workaround, but I don't know of a way to work around ml_get_interrupts_enabled() missing. I thought the new preemption_enabled() would work, but it's not exported either! Radar: 4130443 On May 26, 2005, at 6:35 AM, Andrew Gallatin wrote: > > Sam Vaughan writes: > >> On 26/05/2005, at 2:48 PM, Sam Vaughan wrote: >> >> >>> What does a dependency on the kernel itself look like in a kext's >>> OSBundleLibraries dict? >>> >> >> This seems to do the trick: >> >> <key>com.apple.kernel</key> >> <string>8.1.0</string> >> >> ... but I assume it'll then refuse to load when I install 10.4.2. >> > > Yep. It will break every time your customers upgrade. :-( > > Drew Brian Bergstrand <http://www.bergstrand.org/brian/> PGP Key ID: 0xB6C7B6A2 -----BEGIN PGP SIGNATURE----- Version: PGP Desktop 9.0.1 (Build 2130) iQA/AwUBQpX0dHnR2Fu2x7aiEQL2PACgt/9YFGfImOmlMIBR+zDVe7QJPfoAoIdn jueR9najqAIVpGx4gAO0q2Hw =5pof -----END PGP SIGNATURE----- _______________________________________________ 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

Re: posting message from kernel extension
by Justin Walker 26 May '05

26 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On May 25, 2005, at 21:41, Herbert wrote: Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large Institute for General Semantics -------- It's not whether you win or lose... It's whether *I* win or lose. -------- _______________________________________________ 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.… Does anyone know how to do this? I've found some docs on receiving them, but nothing on posting them as of yet. I'm sure it exists somewhere... Can you be more clear about what you want to do? "posting message" can mean many things. This email sent to site_archiver(a)lists.apple.com

1 0

Re: async io (aio_read, aio_write) in 64-bit mode
by Steve Olson 26 May '05

26 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Wednesday, May 25, 2005, at 06:20PM, Steve Olson <steveo1951(a)mac.com> wrote: >Does anyone know of any issues with using the aio routines (aio_read, aio_write, aio_error, aio_return) with -arch ppc64 on Tiger? > >I'm seeing some strange behavior that suggests what's in my write buffer is not what actually gets written to disk. When i read the buffer back, all I get is zeroes (all bits off). > >This works great for me in 32-bit mode on Tiger, but compiling in 64-bit mode seems to be problematic. Of course, it could be me, but I've been wrestling with this for several days, and thought maybe 2 or more heads are better than one (or none!!) > >Regards, >Steve > I believe I found the problem. I used fcntl(fd, F_NOCACHE, 1) on the file being accessed. While this worked very well in Panther, Tiger has a problem with it. I need writes to hit the disk without caching, so this is a rather large problem for me. Will enter a bug report. Regards, Steve _______________________________________________ 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