[Valgrind-developers] [valgrind] memcheck: Allow unaligned loads of 128bit vectors on ppc64[le].

[Mark W. <ma...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=8d12697b157bb50ad98467b565b3a7a39097bf31

commit 8d12697b157bb50ad98467b565b3a7a39097bf31
Author: Mark Wielaard <ma...@kl...>
Date:   Sun Dec 9 14:26:39 2018 +0100

    memcheck: Allow unaligned loads of 128bit vectors on ppc64[le].
    
    On powerpc partial unaligned loads of vectors from partially invalid
    addresses are OK and could be generated by our translation of lxvd2x.
    
    Adjust partial_load memcheck tests to allow partial loads of 16 byte
    vectors on powerpc64.
    
    Part of resolving bug #386945.

Diff:
---
 memcheck/mc_main.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/memcheck/mc_main.c b/memcheck/mc_main.c
index 737f79d..101916b 100644
--- a/memcheck/mc_main.c
+++ b/memcheck/mc_main.c
@@ -1354,6 +1354,9 @@ void mc_LOADV_128_or_256_slow ( /*OUT*/ULong* res,
    tl_assert(szB == 16); // s390 doesn't have > 128 bit SIMD
    /* OK if all loaded bytes are from the same page. */
    Bool alignedOK = ((a & 0xfff) <= 0x1000 - szB);
+#  elif defined(VGA_ppc64be) || defined(VGA_ppc64le)
+   /* lxvd2x might generate an unaligned 128 bit vector load.  */
+   Bool alignedOK = (szB == 16);
 #  else
    /* OK if the address is aligned by the load size. */
    Bool alignedOK = (0 == (a & (szB - 1)));