ia64_cmpxchg on United Linux 1.0
Hi -
The macro used for test_and_set (ia64_cmpxchg) asm/atomic.h orasm/spinlock.h on United Linux 1.0 is under #ifdef _KERNEL_ and therefore can not be used in user applications. In RHAT 2.1 it was not under #ifdef _KERNEL_ . Is there another macro for test_and_set that user applications can use?
thanks,
Umut
You might want to look at the atomic operations package that's included
here: http://www.hpl.hp.com/research/linux/qprof/download.php4 .
This isn't exactly a mature package, but I'm trying to get it there. IA64 is relatively well-supported. And AFAIK, it's the only such package that handles memory barriers explicitly and correctly (at least by my definition of "correctly" :-) ).
Suggestions/bug reports/patches are welcome.
Hans
Quoted 30 lines
-----Original Message-----
From: Umut Aymakoglu [mailto:[email protected]]
Sent: Tuesday, September 30, 2003 3:39 PM
To: [email protected]
Subject:ia64_cmpxchgon United Linux 1.0Hi -
The macro used fortest_and_set(ia64_cmpxchg)asm/atomic.horasm/spinlock.hon United Linux 1.0 is under #ifdef _KERNEL_ and therefore
can not be used in user applications. In RHAT 2.1 it was not under #ifdef
_KERNEL_ . Is there another macro fortest_and_setthat user applications
can use?thanks,
Umut-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Quoted 2 lines · Umut Aymakoglu
Is there another macro for
test_and_setthat user applications
can use?
The psABI defines a set of synchronization primitives in section 7.3. The psABI is available from the Intel IPF web site.
http://developer.intel.com/design/itanium/arch_spec.htm
Gcc implements these primitives in the ia64intrin.h file. This header file should be available with any system that has gcc installed on it. It is found in the gcc install directory, probably under /usr/lib.
These are very basic primitives that expand to single IPF instructions. They won't be very useful unless you know what you are doing. You might be better off using libraries implemented on top of these primitives.
These primitives aren't used much. HJ found a couple of simple type errors in them when he looked at them earlier this year. The patches for these problems will be in gcc-3.4 which hasn't been released yet. They should be trivial to backpatch into older gcc releases if you need them.
Signature · 1 line
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com