diff options
Diffstat (limited to 'sysdeps/x86_64/dl-trampoline.S')
| -rw-r--r-- | sysdeps/x86_64/dl-trampoline.S | 267 |
1 files changed, 219 insertions, 48 deletions
diff --git a/sysdeps/x86_64/dl-trampoline.S b/sysdeps/x86_64/dl-trampoline.S index d8d9bc12a4..49d239f075 100644 --- a/sysdeps/x86_64/dl-trampoline.S +++ b/sysdeps/x86_64/dl-trampoline.S @@ -17,7 +17,9 @@ Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA. */ +#include <config.h> #include <sysdep.h> +#include <link-defines.h> .text .globl _dl_runtime_resolve @@ -89,25 +91,85 @@ _dl_runtime_profile: /* Actively align the La_x86_64_regs structure. */ andq $0xfffffffffffffff0, %rsp - subq $192, %rsp # sizeof(La_x86_64_regs) +# ifdef HAVE_AVX_SUPPORT + /* sizeof(La_x86_64_regs). Need extra space for 8 SSE registers + to detect if any xmm0-xmm7 registers are changed by audit + module. */ + subq $(LR_SIZE + XMM_SIZE*8), %rsp +# else + subq $LR_SIZE, %rsp # sizeof(La_x86_64_regs) +# endif movq %rsp, 24(%rbx) - movq %rdx, (%rsp) # Fill the La_x86_64_regs structure. - movq %r8, 8(%rsp) - movq %r9, 16(%rsp) - movq %rcx, 24(%rsp) - movq %rsi, 32(%rsp) - movq %rdi, 40(%rsp) - movq %rbp, 48(%rsp) + /* Fill the La_x86_64_regs structure. */ + movq %rdx, LR_RDX_OFFSET(%rsp) + movq %r8, LR_R8_OFFSET(%rsp) + movq %r9, LR_R9_OFFSET(%rsp) + movq %rcx, LR_RCX_OFFSET(%rsp) + movq %rsi, LR_RSI_OFFSET(%rsp) + movq %rdi, LR_RDI_OFFSET(%rsp) + movq %rbp, LR_RBP_OFFSET(%rsp) + leaq 48(%rbx), %rax - movq %rax, 56(%rsp) - movaps %xmm0, 64(%rsp) - movaps %xmm1, 80(%rsp) - movaps %xmm2, 96(%rsp) - movaps %xmm3, 112(%rsp) - movaps %xmm4, 128(%rsp) - movaps %xmm5, 144(%rsp) - movaps %xmm7, 160(%rsp) + movq %rax, LR_RSP_OFFSET(%rsp) + + /* We always store the XMM registers even if AVX is available. + This is to provide backward binary compatility for existing + audit modules. */ + movaps %xmm0, (LR_XMM_OFFSET)(%rsp) + movaps %xmm1, (LR_XMM_OFFSET + XMM_SIZE)(%rsp) + movaps %xmm2, (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp) + movaps %xmm3, (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp) + movaps %xmm4, (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp) + movaps %xmm5, (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp) + movaps %xmm6, (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp) + movaps %xmm7, (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp) + +# ifdef HAVE_AVX_SUPPORT + .data +L(have_avx): + .zero 4 + .size L(have_avx), 4 + .previous + + cmpl $0, L(have_avx)(%rip) + jne 1f + movq %rbx, %r11 # Save rbx + movl $1, %eax + cpuid + movq %r11,%rbx # Restore rbx + movl $1, %eax + testl $(1 << 28), %ecx + jne 2f + negl %eax +2: movl %eax, L(have_avx)(%rip) + cmpl $0, %eax + +1: js L(no_avx1) + + /* This is to support AVX audit modules. */ + vmovdqu %ymm0, (LR_VECTOR_OFFSET)(%rsp) + vmovdqu %ymm1, (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp) + vmovdqu %ymm2, (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp) + vmovdqu %ymm3, (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp) + vmovdqu %ymm4, (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp) + vmovdqu %ymm5, (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp) + vmovdqu %ymm6, (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp) + vmovdqu %ymm7, (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp) + + /* Save xmm0-xmm7 registers to detect if any of them are + changed by audit module. */ + vmovdqa %xmm0, (LR_SIZE)(%rsp) + vmovdqa %xmm1, (LR_SIZE + XMM_SIZE)(%rsp) + vmovdqa %xmm2, (LR_SIZE + XMM_SIZE*2)(%rsp) + vmovdqa %xmm3, (LR_SIZE + XMM_SIZE*3)(%rsp) + vmovdqa %xmm4, (LR_SIZE + XMM_SIZE*4)(%rsp) + vmovdqa %xmm5, (LR_SIZE + XMM_SIZE*5)(%rsp) + vmovdqa %xmm6, (LR_SIZE + XMM_SIZE*6)(%rsp) + vmovdqa %xmm7, (LR_SIZE + XMM_SIZE*7)(%rsp) + +L(no_avx1): +# endif movq %rsp, %rcx # La_x86_64_regs pointer to %rcx. movq 48(%rbx), %rdx # Load return address if needed. @@ -119,27 +181,87 @@ _dl_runtime_profile: movq %rax, %r11 # Save return value. movq 8(%rbx), %rax # Get back register content. - movq (%rsp), %rdx - movq 8(%rsp), %r8 - movq 16(%rsp), %r9 - movaps 64(%rsp), %xmm0 - movaps 80(%rsp), %xmm1 - movaps 96(%rsp), %xmm2 - movaps 112(%rsp), %xmm3 - movaps 128(%rsp), %xmm4 - movaps 144(%rsp), %xmm5 - movaps 160(%rsp), %xmm7 - + movq LR_RDX_OFFSET(%rsp), %rdx + movq LR_R8_OFFSET(%rsp), %r8 + movq LR_R9_OFFSET(%rsp), %r9 + + movaps (LR_XMM_OFFSET)(%rsp), %xmm0 + movaps (LR_XMM_OFFSET + XMM_SIZE)(%rsp), %xmm1 + movaps (LR_XMM_OFFSET + XMM_SIZE*2)(%rsp), %xmm2 + movaps (LR_XMM_OFFSET + XMM_SIZE*3)(%rsp), %xmm3 + movaps (LR_XMM_OFFSET + XMM_SIZE*4)(%rsp), %xmm4 + movaps (LR_XMM_OFFSET + XMM_SIZE*5)(%rsp), %xmm5 + movaps (LR_XMM_OFFSET + XMM_SIZE*6)(%rsp), %xmm6 + movaps (LR_XMM_OFFSET + XMM_SIZE*7)(%rsp), %xmm7 + +# ifdef HAVE_AVX_SUPPORT + cmpl $0, L(have_avx)(%rip) + js L(no_avx2) + + /* Check if any xmm0-xmm7 registers are changed by audit + module. */ + vpcmpeqq (LR_SIZE)(%rsp), %xmm0, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET)(%rsp), %ymm0 + +1: vpcmpeqq (LR_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE)(%rsp), %ymm1 + +1: vpcmpeqq (LR_SIZE + XMM_SIZE*2)(%rsp), %xmm2, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*2)(%rsp), %ymm2 + +1: vpcmpeqq (LR_SIZE + XMM_SIZE*3)(%rsp), %xmm3, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*3)(%rsp), %ymm3 + +1: vpcmpeqq (LR_SIZE + XMM_SIZE*4)(%rsp), %xmm4, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*4)(%rsp), %ymm4 + +1: vpcmpeqq (LR_SIZE + XMM_SIZE*5)(%rsp), %xmm5, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*5)(%rsp), %ymm5 + +1: vpcmpeqq (LR_SIZE + XMM_SIZE*6)(%rsp), %xmm6, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*6)(%rsp), %ymm6 + +1: vpcmpeqq (LR_SIZE + XMM_SIZE*7)(%rsp), %xmm7, %xmm8 + vpmovmskb %xmm8, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu (LR_VECTOR_OFFSET + VECTOR_SIZE*7)(%rsp), %ymm7 + +L(no_avx2): +1: +# endif movq 16(%rbx), %r10 # Anything in framesize? testq %r10, %r10 - jns 1f + jns 3f /* There's nothing in the frame size, so there will be no call to the _dl_call_pltexit. */ - movq 24(%rsp), %rcx # Get back registers content. - movq 32(%rsp), %rsi - movq 40(%rsp), %rdi + /* Get back registers content. */ + movq LR_RCX_OFFSET(%rsp), %rcx + movq LR_RSI_OFFSET(%rsp), %rsi + movq LR_RDI_OFFSET(%rsp), %rdi movq %rbx, %rsp movq (%rsp), %rbx @@ -151,7 +273,7 @@ _dl_runtime_profile: cfi_adjust_cfa_offset(-48) jmp *%r11 # Jump to function address. -1: +3: cfi_adjust_cfa_offset(48) cfi_rel_offset(%rbx, 0) cfi_def_cfa_register(%rbx) @@ -161,7 +283,7 @@ _dl_runtime_profile: temporary buffer of the size specified by the 'framesize' returned from _dl_profile_fixup */ - leaq 56(%rbx), %rsi # stack + leaq LR_RSP_OFFSET(%rbx), %rsi # stack addq $8, %r10 andq $0xfffffffffffffff0, %r10 movq %r10, %rcx @@ -183,31 +305,80 @@ _dl_runtime_profile: _dl_call_pltexit. The La_x86_64_regs is being pointed by rsp now, so we just need to allocate the sizeof(La_x86_64_retval) space on the stack, since the alignment has already been taken care of. */ - - subq $80, %rsp # sizeof(La_x86_64_retval) +# ifdef HAVE_AVX_SUPPORT + /* sizeof(La_x86_64_retval). Need extra space for 2 SSE + registers to detect if xmm0/xmm1 registers are changed + by audit module. */ + subq $(LRV_SIZE + XMM_SIZE*2), %rsp +# else + subq $LRV_SIZE, %rsp # sizeof(La_x86_64_retval) +# endif movq %rsp, %rcx # La_x86_64_retval argument to %rcx. - movq %rax, (%rcx) # Fill in the La_x86_64_retval structure. - movq %rdx, 8(%rcx) - movaps %xmm0, 16(%rcx) - movaps %xmm1, 32(%rcx) - fstpt 48(%rcx) - fstpt 64(%rcx) + /* Fill in the La_x86_64_retval structure. */ + movq %rax, LRV_RAX_OFFSET(%rcx) + movq %rdx, LRV_RDX_OFFSET(%rcx) + + movaps %xmm0, LRV_XMM0_OFFSET(%rcx) + movaps %xmm1, LRV_XMM1_OFFSET(%rcx) + +# ifdef HAVE_AVX_SUPPORT + cmpl $0, L(have_avx)(%rip) + js L(no_avx3) + + /* This is to support AVX audit modules. */ + vmovdqu %ymm0, LRV_VECTOR0_OFFSET(%rcx) + vmovdqu %ymm1, LRV_VECTOR1_OFFSET(%rcx) + + /* Save xmm0/xmm1 registers to detect if they are changed + by audit module. */ + vmovdqa %xmm0, (LRV_SIZE)(%rcx) + vmovdqa %xmm1, (LRV_SIZE + XMM_SIZE)(%rcx) + +L(no_avx3): +# endif + + fstpt LRV_ST0_OFFSET(%rcx) + fstpt LRV_ST1_OFFSET(%rcx) movq 24(%rbx), %rdx # La_x86_64_regs argument to %rdx. movq 40(%rbx), %rsi # Copy args pushed by PLT in register. movq 32(%rbx), %rdi # %rdi: link_map, %rsi: reloc_index call _dl_call_pltexit - movq (%rsp), %rax # Restore return registers. - movq 8(%rsp), %rdx - movaps 16(%rsp), %xmm0 - movaps 32(%rsp), %xmm1 - fldt 64(%rsp) - fldt 48(%rsp) + /* Restore return registers. */ + movq LRV_RAX_OFFSET(%rsp), %rax + movq LRV_RDX_OFFSET(%rsp), %rdx + + movaps LRV_XMM0_OFFSET(%rsp), %xmm0 + movaps LRV_XMM1_OFFSET(%rsp), %xmm1 + +# ifdef HAVE_AVX_SUPPORT + cmpl $0, L(have_avx)(%rip) + js L(no_avx4) + + /* Check if xmm0/xmm1 registers are changed by audit module. */ + vpcmpeqq (LRV_SIZE)(%rsp), %xmm0, %xmm2 + vpmovmskb %xmm2, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu LRV_VECTOR0_OFFSET(%rsp), %ymm0 + +1: vpcmpeqq (LRV_SIZE + XMM_SIZE)(%rsp), %xmm1, %xmm2 + vpmovmskb %xmm2, %esi + cmpl $0xffff, %esi + je 1f + vmovdqu LRV_VECTOR1_OFFSET(%rsp), %ymm1 + +L(no_avx4): +1: +# endif + + fldt LRV_ST1_OFFSET(%rsp) + fldt LRV_ST0_OFFSET(%rsp) movq %rbx, %rsp - movq (%rsp), %rbx + movq (%rsp), %rbx cfi_restore(rbx) cfi_def_cfa_register(%rsp) |