[asan] [mips] changed ShadowOffset32 for systems having 16kb PageSize

[sdkie]

Summary

Earlier ShadowOffset32 address was not aligned to 16kb PageSize
This patch is for compiler-rt project.
The link to corresponding asan Instrumentation part of LLVM - http://reviews.llvm.org/D6080

[kcc]

LGTM

[sdkie]

Thanks.
Similar change need to be done in llvm code, I will submit its patch tomorrow.

I don't have commit access, can you please commit it for me.

[kcc]

I can commit both changes at once, when you have both.
Or ask petarj to commit.

[kcc]

Ah, wait a moment.
Please also update the comment above these lines:
// Default Linux/MIPS mapping:

You can get the mapping printed by asan-ified program with ASAN_OPTIONS=verbosity=1

[sdkie]

Change "Default Linux/MIPS mapping" as commented by kcc

[sdkie]

increased the patch context using -U999999

[kcc]

r221288.

lib/asan/asan_mapping.h

 // || `[0x00000000, 0x1fffffff]` || LowMem     ||
 //
 // Default Linux/MIPS mapping:
-// || `[0x2aaa8000, 0xffffffff]` || HighMem    ||
-// || `[0x0fffd000, 0x2aaa7fff]` || HighShadow ||
-// || `[0x0bffd000, 0x0fffcfff]` || ShadowGap  ||
-// || `[0x0aaa8000, 0x0bffcfff]` || LowShadow  ||
-// || `[0x00000000, 0x0aaa7fff]` || LowMem     ||
+// || `[0x2aaa0000, 0xffffffff]` || HighMem    ||
+// || `[0x0fff4000, 0x2aa9ffff]` || HighShadow ||
+// || `[0x0bff4000, 0x0fff3fff]` || ShadowGap  ||
+// || `[0x0aaa0000, 0x0bff3fff]` || LowShadow  ||
+// || `[0x00000000, 0x0aa9ffff]` || LowMem     ||
 //
 // Shadow mapping on FreeBSD/x86-64 with SHADOW_OFFSET == 0x400000000000:
 // || `[0x500000000000, 0x7fffffffffff]` || HighMem    ||

 static const u64 kDefaultShadowOffset64 = 1ULL << 44;
 static const u64 kDefaultShort64bitShadowOffset = 0x7FFF8000;  // < 2G.
 static const u64 kAArch64_ShadowOffset64 = 1ULL << 36;
-static const u64 kMIPS32_ShadowOffset32 = 0x0aaa8000;
+static const u64 kMIPS32_ShadowOffset32 = 0x0aaa0000;
 #if defined(__powerpc64__) && defined(__BIG_ENDIAN__)
 static const u64 kPPC64_ShadowOffset64 = 1ULL << 41;
 #elif defined(__powerpc64__) && defined(__LITTLE_ENDIAN__)