From a6361c78bf774f2b4dbeeaf4147c286cff4ae5a4 Mon Sep 17 00:00:00 2001 From: Igor Tsimbalist Date: Tue, 7 Nov 2017 17:00:24 +0300 Subject: [PATCH 21/22] Add extra field to gtm_jmpbuf on x86 only Expand the gtm_jmpbuf structure by one word field to add Intel CET support further. The code in sjlj.S already allocates more space on the stack then gtm_jmpbuf needs. Use this extra space to absorb the new field. The structure is allocated on the stack in such a way that eip/rsp field is overlapped with return address on the stack. Locate the new field right before eip/rsp so code that accesses buffer fields relative to address of gtm_jmpbuf has its offsets unchanged. The libtool_VERSION is updated for x86 due to extending the gtm_jmpbuf structure. * libitm/config/x86/target.h: Add new field (ssp). * libitm/config/x86/sjlj.S: Change offsets. * libitm/configure.tgt: Update libtool_VERSION. --- libitm/config/x86/sjlj.S | 44 ++++++++++++++++++++++++-------------------- libitm/config/x86/target.h | 2 ++ libitm/configure.tgt | 12 ++++++++++++ 3 files changed, 38 insertions(+), 20 deletions(-) diff --git a/libitm/config/x86/sjlj.S b/libitm/config/x86/sjlj.S index 21ca9d7..1c8597a 100644 --- a/libitm/config/x86/sjlj.S +++ b/libitm/config/x86/sjlj.S @@ -126,20 +126,22 @@ SYM(_ITM_beginTransaction): /* Store edi for future HTM fast path retries. We use a stack slot lower than the jmpbuf so that the jmpbuf's rip field will overlap with the proper return address on the stack. */ - movl %edi, 8(%rsp) + movl %edi, (%rsp) /* Save the jmpbuf for any non-HTM-fastpath execution method. Because rsp-based addressing is 1 byte larger and we've got rax handy, use it. */ - movq %rax, -64(%rax) - movq %rbx, -56(%rax) - movq %rbp, -48(%rax) - movq %r12, -40(%rax) - movq %r13, -32(%rax) - movq %r14, -24(%rax) - movq %r15, -16(%rax) - leaq -64(%rax), %rsi + movq %rax, -72(%rax) + movq %rbx, -64(%rax) + movq %rbp, -56(%rax) + movq %r12, -48(%rax) + movq %r13, -40(%rax) + movq %r14, -32(%rax) + movq %r15, -24(%rax) + xorq %rdx, %rdx + movq %rdx, -16(%rax) + leaq -72(%rax), %rsi call SYM(GTM_begin_transaction) - movl 8(%rsp), %edi + movl (%rsp), %edi addq $72, %rsp cfi_adjust_cfa_offset(-72) #ifdef HAVE_AS_RTM @@ -162,12 +164,14 @@ SYM(_ITM_beginTransaction): movl 4(%esp), %eax subl $28, %esp cfi_def_cfa_offset(32) - movl %ecx, 8(%esp) - movl %ebx, 12(%esp) - movl %esi, 16(%esp) - movl %edi, 20(%esp) - movl %ebp, 24(%esp) - leal 8(%esp), %edx + movl %ecx, 4(%esp) + movl %ebx, 8(%esp) + movl %esi, 12(%esp) + movl %edi, 16(%esp) + movl %ebp, 20(%esp) + xorl %edx, %edx + movl %edx, 24(%eax) + leal 4(%esp), %edx #if defined HAVE_ATTRIBUTE_VISIBILITY || !defined __PIC__ call SYM(GTM_begin_transaction) #elif defined __ELF__ @@ -203,10 +207,10 @@ SYM(GTM_longjmp): movq 48(%rsi), %r15 movl %edi, %eax cfi_def_cfa(%rsi, 0) - cfi_offset(%rip, 56) + cfi_offset(%rip, 64) cfi_register(%rsp, %rcx) movq %rcx, %rsp - jmp *56(%rsi) + jmp *64(%rsi) #else movl (%edx), %ecx movl 4(%edx), %ebx @@ -214,10 +218,10 @@ SYM(GTM_longjmp): movl 12(%edx), %edi movl 16(%edx), %ebp cfi_def_cfa(%edx, 0) - cfi_offset(%eip, 20) + cfi_offset(%eip, 24) cfi_register(%esp, %ecx) movl %ecx, %esp - jmp *20(%edx) + jmp *24(%edx) #endif cfi_endproc diff --git a/libitm/config/x86/target.h b/libitm/config/x86/target.h index 1b79dc0..5a4b820 100644 --- a/libitm/config/x86/target.h +++ b/libitm/config/x86/target.h @@ -39,12 +39,14 @@ typedef struct gtm_jmpbuf unsigned long long r13; unsigned long long r14; unsigned long long r15; + unsigned long long ssp; unsigned long long rip; #else unsigned long ebx; unsigned long esi; unsigned long edi; unsigned long ebp; + unsigned long ssp; unsigned long eip; #endif } gtm_jmpbuf; diff --git a/libitm/configure.tgt b/libitm/configure.tgt index 4ea71c8..49e4109 100644 --- a/libitm/configure.tgt +++ b/libitm/configure.tgt @@ -149,3 +149,15 @@ case "${target}" in UNSUPPORTED=1 ;; esac + +# Update libtool_VERSION since the size of struct gtm_jmpbuf is +# changed for x86. +case "${host}" in + + # For x86, we use slots in the TCB head for most of our TLS. + # The setup of those slots in beginTransaction can afford to + # use the global-dynamic model. + i[456]86-*-* | x86_64-*-*) + libtool_VERSION=2:0:0 + ;; +esac -- 1.8.3.1