April 2008 - Darwin-dev - lists.apple.com

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Steve Sisak 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I want to be able to use this from user space. Are there any functions that are supported from user space and capable of all that the OSAtomic* functions are? CompareAndSwap() from <DriverServices.h> is available back to Mac OS 8.0. ------- These just implement a singly lined list with atomic enqueue/dequeue. You enqueue (OTLIFOEnqueue()) to head and (OTLIFODequeue()) from tail. If tail becomes empty: if (head) is not empty: Take the lock retry the deque from case it's been filled while waiting If you get nil: Steal the contents of head (OTLIFOStealList()) If not nil: Reverse the list you get (OTReverseList()) Set fHead of tail to the reversed list Deque the first element as the result Release the lock This has the desirable attributes of: 1) enqueuing to head never blocks 2) dequeing from tail only blocks when tail is empty. Implementing this is a good interview question. :-) Cheers, -Steve -- _________________________________________________________________________ Steve Sisak, CTO steve.sisak(a)ioxperts.com IOXperts, Inc. +1 617 876-2572 _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… At 11:11 AM -0400 4/29/08, Army Research Lab wrote: I'm not sure how far back the OS* versions are declared by that name, but there's some version of CompareAndSwap32 in every Mac OS version since OpenTransport (Unix STREAMS) was introduced -- Mac OS 7.5, I think. OTCompareAndSwap32() from <OpenTransport.h> is available back to the beginning of time. As I mentioned previously, given CompareAndSwap32(), and CompareAndSwapPtr (correct size for a pointer) you can implement everything else. BTW, if you ever need really fast atomic queues from user space, look at OTLifo in <OpenTransport.h>. To implement a fast FIFO, you use two OTLifo's (call them head and tail) and a lock. (This is part of the reason why OpenTransport was so fast -- it didn't block much and used very clever memory allocation to avoid copying data much) The above can all be implemented in C given an implementation of CompareAndSwapPtr(). (I could probably send you some code if you need it) This email sent to site_archiver(a)lists.apple.com

1 0

Re: userland supported APIs (was: Re: Synchronization primitives)
by Dave Zarzycki 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Apr 29, 2008, at 8:58 AM, Army Research Lab wrote: On 4/29/08 11:20 AM, "Dave Zarzycki" <zarzycki(a)apple.com> wrote: On Apr 29, 2008, at 8:11 AM, Army Research Lab wrote: I want to be able to use this from user space. Are there any functions that are supported from user space and capable of all that the OSAtomic* functions are? Please forgive us for the poorly named /usr/include/libkern/*.h davez _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Actually it is of quite a bit of use! My concern about the OSAtomic* functions is that they are all defined in libkern/OSAtomic.h. If I understand Terry Lambert's earlier reply (sent to the list on Thursday, April 24, 2008, with the subject 'Re: Where is atomicqueue?'), I should only expect to be able to use this within the kernel (Note that that earlier conversation was about atomicqueue, and not about the other functions, so I'm interpreting what he said, and may be interpreting it incorrectly. Terry, please forgive me [and correct me!] if I'm wrong). We, Apple, have made a small confusing mess when it comes to OSAtomic.h. Both userland and kernel-land have a header by that name. Unfortunately, not all of the API within have the same names, with the same arguments. The userland header is located here: /usr/include/ libkern/OSAtomic.h and is supported API. Does that mean that anything I find in /usr/include is supported in userland? I mean, what Terry said made a lot of sense to mean, that since it is in libkern, we shouldn't use it. So, is there anything we should stay out of in /usr/include when in userland? To the best of my knowledge, there are no kernel ONLY headers in /usr/ include. Kernel only headers are supposed to live in /System/Library/ Frameworks/Kernel.framework/Headers. This email sent to site_archiver(a)lists.apple.com

1 0

userland supported APIs (was: Re: Synchronization primitives)
by Army Research Lab 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AciqEe5YLJ9VghYFEd23XAAdT0T19A== Thread-topic: userland supported APIs (was: Re: Synchronization primitives) User-agent: Microsoft-Entourage/11.4.0.080122 On 4/29/08 11:20 AM, "Dave Zarzycki" <zarzycki(a)apple.com> wrote: > On Apr 29, 2008, at 8:11 AM, Army Research Lab wrote: > >> Actually it is of quite a bit of use! My concern about the OSAtomic* >> functions is that they are all defined in libkern/OSAtomic.h. If I >> understand Terry Lambert's earlier reply (sent to the list on >> Thursday, >> April 24, 2008, with the subject 'Re: Where is atomicqueue?'), I >> should only >> expect to be able to use this within the kernel (Note that that >> earlier >> conversation was about atomicqueue, and not about the other >> functions, so >> I'm interpreting what he said, and may be interpreting it incorrectly. >> Terry, please forgive me [and correct me!] if I'm wrong). >> >> I want to be able to use this from user space. Are there any >> functions that >> are supported from user space and capable of all that the OSAtomic* >> functions are? > > We, Apple, have made a small confusing mess when it comes to > OSAtomic.h. Both userland and kernel-land have a header by that name. > Unfortunately, not all of the API within have the same names, with the > same arguments. The userland header is located here: /usr/include/ > libkern/OSAtomic.h and is supported API. Does that mean that anything I find in /usr/include is supported in userland? I mean, what Terry said made a lot of sense to mean, that since it is in libkern, we shouldn't use it. So, is there anything we should stay out of in /usr/include when in userland? Thanks, Cem Karan _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Dave Zarzycki 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Apr 29, 2008, at 8:43 AM, Army Research Lab wrote: davez _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… The problem is that although I can find it in the library, I can't always find it in the headers. Please review /usr/include/libkern/OSAtomic.h. We have supported spin lock API within that file. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Army Research Lab 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AciqD9XnFKy5IxYDEd23XAAdT0T19A== Thread-topic: Synchronization primitives (was:Re: Where is atomicqueue?) User-agent: Microsoft-Entourage/11.4.0.080122 On 4/29/08 11:22 AM, "Shawn Erickson" <shawnce(a)gmail.com> wrote: > On Tue, Apr 29, 2008 at 8:18 AM, Shawn Erickson <shawnce(a)gmail.com> wrote: >> On Tue, Apr 29, 2008 at 8:11 AM, Army Research Lab <ckaran(a)arl.army.mil> >> wrote: >> >>> I want to be able to use this from user space. Are there any functions >>> that >>> are supported from user space and capable of all that the OSAtomic* >>> functions are? >> >> If you are interested in support 10.4 and later then yes. Review... >> >> >> <http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/atom >> ic.3.html> >> >> <http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/spin >> lock.3.html> > > As a note about how to explore things (man pages state libc is the > library that provides these)... > > Note this is on a Leopard (10.5) system. > > [0:588] > nm -arch i386 /usr/lib/libc.dylib | grep OS <<SNIP>> > > [0:590] > nm -arch ppc /usr/lib/libc.dylib | grep OS <<SNIP>> The problem is that although I can find it in the library, I can't always find it in the headers. E.g. spinlock (and please forgive the line breaks; I'm at work, and they force us to use Entourage, and I have yet to figure out how to fix its line-breaking habits): anvil:/usr/lib ckaran$ find . -type f -exec nm -Afg -arch i386 {} \; 2>/dev/null | grep spin ./libauto.dylib: U __spin_lock ./libauto.dylib: U __spin_unlock ./libgmalloc.B.dylib: U __spin_lock_try ./libgmalloc.B.dylib: U __spin_unlock ./libMallocDebug.a:debug_malloc.o: U __spin_lock_try ./libMallocDebug.a:debug_malloc.o: U __spin_unlock ./libMallocDebug.A.dylib: U __spin_lock_try ./libMallocDebug.A.dylib: U __spin_unlock ./libnetsnmphelpers.5.2.1.dylib: 0000515d T _netsnmp_get_watched_spinlock_handler ./libnetsnmphelpers.5.2.1.dylib: 000054a7 T _netsnmp_register_watched_spinlock ./libnetsnmphelpers.5.2.1.dylib: 0000518f T _netsnmp_watched_spinlock_handler ./libSystem.B.dylib: 9010d48c T __spin_lock ./libSystem.B.dylib: 9010d484 T __spin_lock_try ./libSystem.B.dylib: a0002004 D __spin_tries ./libSystem.B.dylib: 9010d494 T __spin_unlock ./libSystem.B.dylib: 9010d48c T _spin_lock ./libSystem.B.dylib: 9010d494 T _spin_unlock ./libSystem.B_debug.dylib: 000fe210 T __spin_lock ./libSystem.B_debug.dylib: 000fe208 T __spin_lock_try ./libSystem.B_debug.dylib: 00184050 D __spin_tries ./libSystem.B_debug.dylib: 000fe218 T __spin_unlock ./libSystem.B_debug.dylib: 000fe210 T _spin_lock ./libSystem.B_debug.dylib: 000fe218 T _spin_unlock ./libSystem.B_profile.dylib: 00112f1c T __spin_lock ./libSystem.B_profile.dylib: 00112f14 T __spin_lock_try ./libSystem.B_profile.dylib: 0017c004 D __spin_tries ./libSystem.B_profile.dylib: 00112f24 T __spin_unlock ./libSystem.B_profile.dylib: 00112f1c T _spin_lock ./libSystem.B_profile.dylib: 00112f24 T _spin_unlock which means there is a spin_lock defined. anvil:/usr/include ckaran$ man spin_lock No manual entry for spin_lock anvil:/usr/include ckaran$ apropos spin_lock spin_lock: nothing appropriate anvil:/usr/include ckaran$ find . -type f -exec grep -H spin_lock {} \; anvil:/usr/include ckaran$ Which results in nothing. So, as Terry said earlier, it might be in there, but it may not be supported, which isn't a problem, as long as we know. Its knowing if it *IS* supported or not that is the trick. Thanks, Cem Karan _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Shawn Erickson 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=1Yujcns0qKody4F6wR13UTZgpLki1IUS+AGYjz94ueE=; b=hH9MyBw2R5V/PV/2PSnzO7cETPORbm3J3jN8gyKjmbtk6UFi+qaro5bW6izSzJCvcxISbFpOoIlsQc1UJ0qfnsDhZQKysPvE3TwLj7w/c3JF6NzXrZFn7fkdi9lWfU9AhlYitKPUN02VbhPMwRO8nqLTvoJlcv8i4sHJ7FogOOU= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=GUlpqkhaSuIOd1OJiBwPCRa/G2QAqlJ3qXjo6yLGKhN0T4fzZURo5QmgGZcEvY82RmGyEWBCdMeePE3SQMCBmR55tY53oZc2hylBzBofYjj/zMqqEwCZXmwOB8lk2YlZwJs4S/CbdzFhMXuDzgbNQpKmj/5CyUR/RFFmdjaxpX4= On Tue, Apr 29, 2008 at 8:18 AM, Shawn Erickson <shawnce(a)gmail.com> wrote: > On Tue, Apr 29, 2008 at 8:11 AM, Army Research Lab <ckaran(a)arl.army.mil> wrote: > > > I want to be able to use this from user space. Are there any functions that > > are supported from user space and capable of all that the OSAtomic* > > functions are? > > If you are interested in support 10.4 and later then yes. Review... > > <http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/ato…> > <http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/spi…> As a note about how to explore things (man pages state libc is the library that provides these)... Note this is on a Leopard (10.5) system. [0:588] > nm -arch i386 /usr/lib/libc.dylib | grep OS 0000af00 T _OSAtomicAdd32 0000af00 T _OSAtomicAdd32Barrier 0000af14 T _OSAtomicAdd64 0000af14 T _OSAtomicAdd64Barrier 0000ae2c T _OSAtomicAnd32 0000ae2c T _OSAtomicAnd32Barrier 0000ae74 T _OSAtomicAnd32Orig 0000ae74 T _OSAtomicAnd32OrigBarrier 0000aebc T _OSAtomicCompareAndSwap32 0000aebc T _OSAtomicCompareAndSwap32Barrier 0000aed8 T _OSAtomicCompareAndSwap64 0000aed8 T _OSAtomicCompareAndSwap64Barrier 0000aebc T _OSAtomicCompareAndSwapInt 0000aebc T _OSAtomicCompareAndSwapIntBarrier 0000aebc T _OSAtomicCompareAndSwapLong 0000aebc T _OSAtomicCompareAndSwapLongBarrier 0000aebc T _OSAtomicCompareAndSwapPtr 0000aebc T _OSAtomicCompareAndSwapPtrBarrier 0000afcc T _OSAtomicDequeue 0000afa8 T _OSAtomicEnqueue 0000ae44 T _OSAtomicOr32 0000ae44 T _OSAtomicOr32Barrier 0000ae8c T _OSAtomicOr32Orig 0000ae8c T _OSAtomicOr32OrigBarrier 0000af60 T _OSAtomicTestAndClear 0000af60 T _OSAtomicTestAndClearBarrier 0000af3c T _OSAtomicTestAndSet 0000af3c T _OSAtomicTestAndSetBarrier 0000ae5c T _OSAtomicXor32 0000ae5c T _OSAtomicXor32Barrier 0000aea4 T _OSAtomicXor32Orig 0000aea4 T _OSAtomicXor32OrigBarrier 0000afa0 T _OSMemoryBarrier 0000af8c T _OSSpinLockLock 0000af84 T _OSSpinLockTry 0000af94 T _OSSpinLockUnlock 000b9f4c T _OSSystemInfo [0:590] > nm -arch ppc /usr/lib/libc.dylib | grep OS 0000bc94 T _OSAtomicAdd32 0000be4c T _OSAtomicAdd32Barrier 0000bcc4 T _OSAtomicAnd32 0000bea4 T _OSAtomicAnd32Barrier 0000bd4c T _OSAtomicAnd32Orig 0000bf2c T _OSAtomicAnd32OrigBarrier 0000bdac T _OSAtomicCompareAndSwap32 0000bf8c T _OSAtomicCompareAndSwap32Barrier 0000bdb0 T _OSAtomicCompareAndSwapInt 0000bf90 T _OSAtomicCompareAndSwapIntBarrier 0000bdb8 T _OSAtomicCompareAndSwapLong 0000bf98 T _OSAtomicCompareAndSwapLongBarrier 0000bdb4 T _OSAtomicCompareAndSwapPtr 0000bf94 T _OSAtomicCompareAndSwapPtrBarrier 0000c040 T _OSAtomicDequeue 0000c03c T _OSAtomicEnqueue 0000bc98 T _OSAtomicOr32 0000be78 T _OSAtomicOr32Barrier 0000bd1c T _OSAtomicOr32Orig 0000befc T _OSAtomicOr32OrigBarrier 0000be04 T _OSAtomicTestAndClear 0000bfe4 T _OSAtomicTestAndClearBarrier 0000bdbc T _OSAtomicTestAndSet 0000bf9c T _OSAtomicTestAndSetBarrier 0000bcf0 T _OSAtomicXor32 0000bed0 T _OSAtomicXor32Barrier 0000bd7c T _OSAtomicXor32Orig 0000bf5c T _OSAtomicXor32OrigBarrier 0000c038 T _OSMemoryBarrier 0000c030 T _OSSpinLockLock 0000c02c T _OSSpinLockTry 0000c034 T _OSSpinLockUnlock 000dda38 T _OSSystemInfo -Shawn _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Dave Zarzycki 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Apr 29, 2008, at 8:11 AM, Army Research Lab wrote: davez _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Actually it is of quite a bit of use! My concern about the OSAtomic* functions is that they are all defined in libkern/OSAtomic.h. If I understand Terry Lambert's earlier reply (sent to the list on Thursday, April 24, 2008, with the subject 'Re: Where is atomicqueue?'), I should only expect to be able to use this within the kernel (Note that that earlier conversation was about atomicqueue, and not about the other functions, so I'm interpreting what he said, and may be interpreting it incorrectly. Terry, please forgive me [and correct me!] if I'm wrong). I want to be able to use this from user space. Are there any functions that are supported from user space and capable of all that the OSAtomic* functions are? We, Apple, have made a small confusing mess when it comes to OSAtomic.h. Both userland and kernel-land have a header by that name. Unfortunately, not all of the API within have the same names, with the same arguments. The userland header is located here: /usr/include/ libkern/OSAtomic.h and is supported API. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Shawn Erickson 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=gTRO9ngaAj+G2N+0RtDlqar9CSxajnhAvwhmJmqMb1M=; b=kAemwDCv737m1jpzXKZx6RSuFbKj+9HNlv8xXgtZkEoXsZha59eERqAW3RHziFm2TTCB20Fa73jEthr9hO5Bd5vFe/fA5QXjLnVboL7zplLIn4GJ64G63yrMtaioPw7VfhJs2Vpjh5sTZROlSapG8TWsoDXkXJ4a3ru72AtoLyw= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=HIUDX8G8WTTlngEGs8F7y9ETmyVjmH59T1GE7aAlw7bkf04SfwYVqRPoGjlkeLRlHVw2WQPAIGz5wA8KkCrK/j+rd1+dC/1cWy10or1LiHFgTrH3msGVRbNmeQ2hVH9A2G0KWE0Om7yYHfBP263Y9Dtw4rmx7hfoX6bIQW/COAM= On Tue, Apr 29, 2008 at 8:11 AM, Army Research Lab <ckaran(a)arl.army.mil> wrote: > I want to be able to use this from user space. Are there any functions that > are supported from user space and capable of all that the OSAtomic* > functions are? If you are interested in support 10.4 and later then yes. Review... <http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/ato…> <http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/spi…> -Shawn _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Army Research Lab 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AciqC1scmX7+0hX+Ed23XAAdT0T19A== Thread-topic: Synchronization primitives (was:Re: Where is atomicqueue?) User-agent: Microsoft-Entourage/11.4.0.080122 On 4/29/08 10:53 AM, "Steve Sisak" <sgs-lists(a)codewell.com> wrote: > At 7:25 AM -0400 4/29/08, Army Research Lab wrote: >> I understand the problems with them; however, the case I plan on supporting >> generally locks for just long enough to write a value to a shared location, >> with the values being 32 or 64 bit ints. That was why I got interested in >> the various atomic functions. I know that they have a much higher >> likelihood of being optimized than using pthread lock/write/unlock. > > If this is all you're interested in, you can implement an atomic > update of a variable with one of the CompareAndSwap variants. (Note > that the 64-bit versions are only available to 64-bit code, and the > values must be on natural alignment, etc.) > > int32_t *shared_value = <whatever> > int32_t new_value; > int32_t old_value; > > do > { > old_value = *shared_value; > new_value = <insert your calculation>; > } > while (!OSAtomicCompareAndSwap32Barrier(old_value, new_value, shared_value); > > At this point new_value has been successfully written and old_value > contains the > > This is actually the basis for most of the OSAtomic*32 operations. > > For 64-bit (or non-32-bit) values on a 32-bit system, you would have > to protect access using a semaphore, spinlock, etc. > >>> I would recommend you restrict yourself to supported interfaces. >> >> I don't plan on doing otherwise. If I can't find a supported interface for >> it, then I'll have to find another way. > > You can implement a simple spinlock-like protection mechanism for > multiple values with compare and swap as well (I'll use the OSAtomic* > convenience functions and leave how to implement them using compare > and swap as an exercise), > > ------------ > > #define VALUE1_MASK (1 << 0) > #define VALUE2_MASK (1 << 1) > > int32_t *shared_locks = <whatever> > > int32_t old_locks; > > do > { > old_locks = OSAtomicOr32Barrier(VALUE1_MASK, shared_locks); > > // old_locks is the value of *shared_locks before > // you changed it -- if your bit wasn't set, then > // YOU changed it and now own the lock -- otherwise > // someone else set it and you don't > } > while (old_locks & VALUE1_MASK); > > // do your stuff > > old_locks = OSAtomicAnd32Barrier(~VALUE1_MASK, shared_locks); > > ------------ > > to implement non-blocking protection you would do: > > old_locks = OSAtomicOr32Barrier(VALUE1_MASK, shared_locks); > > if (!(old_locks & VALUE1_MASK)) > { > // you got the lock -- do your stuff > > old_locks = OSAtomicAnd32Barrier(~VALUE1_MASK, shared_locks); > } > > ------------ > > Read the relevant tecnnotes to determine if you need the barrier > versions or if you can get away with the non-barrier version. > > When in doubt, use the barrier version, which forces all writes > started before it to complete before it. I think you can use > non-barrier for the unlock as long as you use barrier for the lock -- > there's probably someone here who can correct me if I'm wrong -- but > the code above is safe. > > You'd use a bit to protect each structure you wanted to update > atomically and more words as needed. > > Hopefully this of use. > > -Steve > > *Disclaimer, this code was typed in Eudora, not a compiler. YMMV. Actually it is of quite a bit of use! My concern about the OSAtomic* functions is that they are all defined in libkern/OSAtomic.h. If I understand Terry Lambert's earlier reply (sent to the list on Thursday, April 24, 2008, with the subject 'Re: Where is atomicqueue?'), I should only expect to be able to use this within the kernel (Note that that earlier conversation was about atomicqueue, and not about the other functions, so I'm interpreting what he said, and may be interpreting it incorrectly. Terry, please forgive me [and correct me!] if I'm wrong). I want to be able to use this from user space. Are there any functions that are supported from user space and capable of all that the OSAtomic* functions are? Thanks, Cem Karan Quoting Terry's earlier post: """ Short answer: (1) It's protected by #ifdef __ppc__, so it's PPC only (2) It's in a header <libkern/OSAtomic.h>, so you should only _expect_ to be able to use it in the kernel; the fact that there are some inline functions there notwithstanding (3) The atual implementaiton is as an external symbol in the Libkern and System6.0 exports for PPC only So to get at it you will need to be linked against the kernel (you will need to be either kernel code or a KEXT), you will need to be running PPC-only, and you will need to declare an explicit dependency for your KEXT on an exported symbol set or on the entire kernel. -- Terry """ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Synchronization primitives (was:Re: Where is atomicqueue?)
by Steve Sisak 29 Apr '08

29 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I understand the problems with them; however, the case I plan on supporting generally locks for just long enough to write a value to a shared location, with the values being 32 or 64 bit ints. That was why I got interested in the various atomic functions. I know that they have a much higher likelihood of being optimized than using pthread lock/write/unlock. int32_t *shared_value = <whatever> int32_t new_value; int32_t old_value; do { old_value = *shared_value; new_value = <insert your calculation>; } while (!OSAtomicCompareAndSwap32Barrier(old_value, new_value, shared_value); This is actually the basis for most of the OSAtomic*32 operations. > I would recommend you restrict yourself to supported interfaces. I don't plan on doing otherwise. If I can't find a supported interface for it, then I'll have to find another way. ------------ #define VALUE1_MASK (1 << 0) #define VALUE2_MASK (1 << 1) int32_t *shared_locks = <whatever> int32_t old_locks; do { old_locks = OSAtomicOr32Barrier(VALUE1_MASK, shared_locks); // old_locks is the value of *shared_locks before // you changed it -- if your bit wasn't set, then // YOU changed it and now own the lock -- otherwise // someone else set it and you don't } while (old_locks & VALUE1_MASK); // do your stuff old_locks = OSAtomicAnd32Barrier(~VALUE1_MASK, shared_locks); ------------ to implement non-blocking protection you would do: old_locks = OSAtomicOr32Barrier(VALUE1_MASK, shared_locks); if (!(old_locks & VALUE1_MASK)) { // you got the lock -- do your stuff old_locks = OSAtomicAnd32Barrier(~VALUE1_MASK, shared_locks); } ------------ Hopefully this of use. -Steve *Disclaimer, this code was typed in Eudora, not a compiler. YMMV. -- _________________________________________________________________________ Steve Sisak, CTO steve.sisak(a)ioxperts.com IOXperts, Inc. +1 617 876-2572 _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… At 7:25 AM -0400 4/29/08, Army Research Lab wrote: If this is all you're interested in, you can implement an atomic update of a variable with one of the CompareAndSwap variants. (Note that the 64-bit versions are only available to 64-bit code, and the values must be on natural alignment, etc.) At this point new_value has been successfully written and old_value contains the For 64-bit (or non-32-bit) values on a 32-bit system, you would have to protect access using a semaphore, spinlock, etc. You can implement a simple spinlock-like protection mechanism for multiple values with compare and swap as well (I'll use the OSAtomic* convenience functions and leave how to implement them using compare and swap as an exercise), Read the relevant tecnnotes to determine if you need the barrier versions or if you can get away with the non-barrier version. When in doubt, use the barrier version, which forces all writes started before it to complete before it. I think you can use non-barrier for the unlock as long as you use barrier for the lock -- there's probably someone here who can correct me if I'm wrong -- but the code above is safe. You'd use a bit to protect each structure you wanted to update atomically and more words as needed. This email sent to site_archiver(a)lists.apple.com

1 0