[scudo] Add support for LoongArch hardware CRC32 checksumming

[xen0n]

One has to probe for platform capability prior to use with HWCAP, according to LoongArch documentation.

[llvmbot]

@llvm/pr-subscribers-compiler-rt-sanitizer

Author: WÁNG Xuěruì (xen0n)

Changes

The support is available for processors implementing the LoongArch64 subset, but one still has to probe for platform capability prior to use, according to LoongArch documentation.

---

Full diff: https://github.com/llvm/llvm-project/pull/83113.diff

3 Files Affected:

• (modified) compiler-rt/lib/scudo/standalone/checksum.cpp (+10)
• (modified) compiler-rt/lib/scudo/standalone/checksum.h (+4)
• (modified) compiler-rt/lib/scudo/standalone/crc32_hw.cpp (+7)

diff --git a/compiler-rt/lib/scudo/standalone/checksum.cpp b/compiler-rt/lib/scudo/standalone/checksum.cpp
index 2c277391a2ec16..09fa5417653a7a 100644
--- a/compiler-rt/lib/scudo/standalone/checksum.cpp
+++ b/compiler-rt/lib/scudo/standalone/checksum.cpp
@@ -19,6 +19,8 @@
 #else
 #include <sys/auxv.h>
 #endif
+#elif defined(__loongarch_lp64)
+#include <sys/auxv.h>
 #endif
 
 namespace scudo {
@@ -75,6 +77,14 @@ bool hasHardwareCRC32() {
   return !!(getauxval(AT_HWCAP) & HWCAP_CRC32);
 #endif // SCUDO_FUCHSIA
 }
+#elif defined(__loongarch_lp64)
+// Query HWCAP for platform capability, according to *Software Development and
+// Build Convention for LoongArch Architectures* v0.1, Section 9.1.
+//
+// Link: https://github.com/loongson/la-softdev-convention/blob/v0.1/la-softdev-convention.adoc#kernel-development
+bool hasHardwareCRC32() {
+  return !!(getauxval(AT_HWCAP) & HWCAP_LOONGARCH_CRC32);
+}
 #else
 // No hardware CRC32 implemented in Scudo for other architectures.
 bool hasHardwareCRC32() { return false; }
diff --git a/compiler-rt/lib/scudo/standalone/checksum.h b/compiler-rt/lib/scudo/standalone/checksum.h
index f8eda81fd91288..b5f97fe5111f04 100644
--- a/compiler-rt/lib/scudo/standalone/checksum.h
+++ b/compiler-rt/lib/scudo/standalone/checksum.h
@@ -30,6 +30,10 @@
 #include <arm_acle.h>
 #define CRC32_INTRINSIC FIRST_32_SECOND_64(__crc32cw, __crc32cd)
 #endif
+#ifdef __loongarch_lp64
+#include <larchintrin.h>
+#define CRC32_INTRINSIC FIRST_32_SECOND_64(__crcc_w_w_w, __crcc_w_d_w)
+#endif
 
 namespace scudo {
 
diff --git a/compiler-rt/lib/scudo/standalone/crc32_hw.cpp b/compiler-rt/lib/scudo/standalone/crc32_hw.cpp
index 73f2ae000c63d6..3f8c423fa0f49e 100644
--- a/compiler-rt/lib/scudo/standalone/crc32_hw.cpp
+++ b/compiler-rt/lib/scudo/standalone/crc32_hw.cpp
@@ -17,4 +17,11 @@ u32 computeHardwareCRC32(u32 Crc, uptr Data) {
 #endif // defined(__CRC32__) || defined(__SSE4_2__) ||
        // defined(__ARM_FEATURE_CRC32)
 
+#if defined(__loongarch_lp64)
+u32 computeHardwareCRC32(u32 Crc, uptr Data) {
+  // The LoongArch CRC intrinsics have the two input arguments swapped.
+  return static_cast<u32>(CRC32_INTRINSIC(Data, Crc));
+}
+#endif // defined(__loongarch_lp64)
+
 } // namespace scudo

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[xen0n]

cc @SixWeining @heiher @xry111

[ChiaHungDuan]

Thanks for adding this

[SixWeining]

Thanks for the patch.

[SixWeining]

When I click the merge button, I get this prompt "This commit will be authored by [email protected]". Do you mind to change your email settings before this getting landed?

[SixWeining]

FYI: https://docs.github.com/en/account-and-profile/setting-up-and-managing-your-personal-account-on-github/managing-email-preferences/setting-your-commit-email-address