mem* intrinsics are not 64 bit safe

[llvmbot]

Bugzilla Link	709
Resolution	FIXED
Resolved on	Feb 22, 2010 12:51
Version	1.0
OS	All
Reporter	LLVM Bugzilla Contributor

Extended Description

The definition of the memcpy and friends do match the definition needed and used
on 64 bit hosts. the length field should be a ulong.

However, to stay consistent with the fixed type based on name, we need 2
intrinsics for each mem* intrinsic. A 32 bit version and a 64 bit version.

on alpha llvm.memcpy is void %llvm.memcpy(sbyte*, sbyte*, ulong, uint), whereas
SimplifyLibCalls was inserting llvm.memcpy(sbyte*, sbyte*, uint, uint) which is
not correct (but matches LangRef).

[llvmbot]

assigned to @lattner

[lattner]

Just to be perfectly clear, this bug is about adding an llvm.memcpy.i32 and llvm.memcpy.i64 intrinsic,
which llvm.memcpy gets autoupgraded to (similarly for memset/memmove). We have the infrastructure to
do this now, we just have to do it.

-Chris

[lattner]

Fixed. Testcase here: test/Regression/Bytecode/memcpy.ll

Patches here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20060227/032253.html
...
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20060227/032258.html

-Chris