diff --git a/lld/COFF/DLL.cpp b/lld/COFF/DLL.cpp
--- a/lld/COFF/DLL.cpp
+++ b/lld/COFF/DLL.cpp
@@ -201,18 +201,18 @@
     0x52,                               // push    rdx
     0x41, 0x50,                         // push    r8
     0x41, 0x51,                         // push    r9
-    0x48, 0x83, 0xEC, 0x48,             // sub     rsp, 48h
-    0x66, 0x0F, 0x7F, 0x04, 0x24,       // movdqa  xmmword ptr [rsp], xmm0
-    0x66, 0x0F, 0x7F, 0x4C, 0x24, 0x10, // movdqa  xmmword ptr [rsp+10h], xmm1
-    0x66, 0x0F, 0x7F, 0x54, 0x24, 0x20, // movdqa  xmmword ptr [rsp+20h], xmm2
-    0x66, 0x0F, 0x7F, 0x5C, 0x24, 0x30, // movdqa  xmmword ptr [rsp+30h], xmm3
+    0x48, 0x83, 0xEC, 0x68,             // sub     rsp, 68h
+    0x66, 0x0F, 0x7F, 0x44, 0x24, 0x20, // movdqa  xmmword ptr [rsp+20h], xmm0
+    0x66, 0x0F, 0x7F, 0x4C, 0x24, 0x30, // movdqa  xmmword ptr [rsp+30h], xmm1
+    0x66, 0x0F, 0x7F, 0x54, 0x24, 0x40, // movdqa  xmmword ptr [rsp+40h], xmm2
+    0x66, 0x0F, 0x7F, 0x5C, 0x24, 0x50, // movdqa  xmmword ptr [rsp+50h], xmm3
     0x48, 0x8B, 0xD0,                   // mov     rdx, rax
     0x48, 0x8D, 0x0D, 0, 0, 0, 0,       // lea     rcx, [___DELAY_IMPORT_...]
     0xE8, 0, 0, 0, 0,                   // call    __delayLoadHelper2
-    0x66, 0x0F, 0x6F, 0x04, 0x24,       // movdqa  xmm0, xmmword ptr [rsp]
-    0x66, 0x0F, 0x6F, 0x4C, 0x24, 0x10, // movdqa  xmm1, xmmword ptr [rsp+10h]
-    0x66, 0x0F, 0x6F, 0x54, 0x24, 0x20, // movdqa  xmm2, xmmword ptr [rsp+20h]
-    0x66, 0x0F, 0x6F, 0x5C, 0x24, 0x30, // movdqa  xmm3, xmmword ptr [rsp+30h]
+    0x66, 0x0F, 0x6F, 0x44, 0x24, 0x20, // movdqa  xmm0, xmmword ptr [rsp+20h]
+    0x66, 0x0F, 0x6F, 0x4C, 0x24, 0x30, // movdqa  xmm1, xmmword ptr [rsp+30h]
+    0x66, 0x0F, 0x6F, 0x54, 0x24, 0x40, // movdqa  xmm2, xmmword ptr [rsp+40h]
+    0x66, 0x0F, 0x6F, 0x5C, 0x24, 0x50, // movdqa  xmm3, xmmword ptr [rsp+50h]
     0x48, 0x83, 0xC4, 0x48,             // add     rsp, 48h
     0x41, 0x59,                         // pop     r9
     0x41, 0x58,                         // pop     r8
@@ -316,8 +316,8 @@
 
   void writeTo(uint8_t *buf) const override {
     memcpy(buf, tailMergeX64, sizeof(tailMergeX64));
-    write32le(buf + 39, desc->getRVA() - rva - 43);
-    write32le(buf + 44, helper->getRVA() - rva - 48);
+    write32le(buf + 40, desc->getRVA() - rva - 44);
+    write32le(buf + 45, helper->getRVA() - rva - 49);
   }
 
   Chunk *desc = nullptr;