[PATCH 1/5] x86: change THREAD_INFO definition to not depend on KERNEL_STACK_OFFSET

From: Denys Vlasenko
Date: Thu Mar 19 2015 - 13:18:28 EST


This changes THREAD_INFO definition and all its callsites
so that they do not count stack position from
(top of stack - KERNEL_STACK_OFFSET), but from top of stack.

Semi-mysterious expressions THREAD_INFO(%rsp,RIP) - "why RIP??"
are now replaced by more logical THREAD_INFO(%rsp,SIZEOF_PTREGS) -
"calculate thread_info's address using information that
rsp is SIZEOF_PTREGS bytes below top of stack".

While at it, replace "(off)-THREAD_SIZE(reg)" with equivalent
"((off)-THREAD_SIZE)(reg)". The form without parentheses
falsely looks like we invoke THREAD_SIZE() macro.

Improve comment atop THREAD_INFO macro definition.

This patch does not change generated code (verified by objdump).

Signed-off-by: Denys Vlasenko <dvlasenk@xxxxxxxxxx>
CC: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
CC: Steven Rostedt <rostedt@xxxxxxxxxxx>
CC: Ingo Molnar <mingo@xxxxxxxxxx>
CC: Borislav Petkov <bp@xxxxxxxxx>
CC: "H. Peter Anvin" <hpa@xxxxxxxxx>
CC: Andy Lutomirski <luto@xxxxxxxxxxxxxx>
CC: Oleg Nesterov <oleg@xxxxxxxxxx>
CC: Frederic Weisbecker <fweisbec@xxxxxxxxx>
CC: Alexei Starovoitov <ast@xxxxxxxxxxxx>
CC: Will Drewry <wad@xxxxxxxxxxxx>
CC: Kees Cook <keescook@xxxxxxxxxxxx>
CC: x86@xxxxxxxxxx
CC: linux-kernel@xxxxxxxxxxxxxxx
---

Changes since last version:
* split this change into a separate patch
* on Borislav's request, improve comment atop THREAD_INFO definition.

arch/x86/ia32/ia32entry.S | 30 +++++++++++++++---------------
arch/x86/include/asm/thread_info.h | 8 +++++---
arch/x86/kernel/entry_64.S | 4 ++--
3 files changed, 22 insertions(+), 20 deletions(-)

diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index ad9efef..50190e1 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -127,7 +127,7 @@ ENTRY(ia32_sysenter_target)
CFI_REL_OFFSET rsp,0
pushfq_cfi
/*CFI_REL_OFFSET rflags,0*/
- movl TI_sysenter_return+THREAD_INFO(%rsp,3*8-KERNEL_STACK_OFFSET),%r10d
+ movl TI_sysenter_return+THREAD_INFO(%rsp,3*8),%r10d
CFI_REGISTER rip,r10
pushq_cfi $__USER32_CS
/*CFI_REL_OFFSET cs,0*/
@@ -159,8 +159,8 @@ ENTRY(ia32_sysenter_target)
jnz sysenter_fix_flags
sysenter_flags_fixed:

- orl $TS_COMPAT,TI_status+THREAD_INFO(%rsp,RIP)
- testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,RIP)
+ orl $TS_COMPAT,TI_status+THREAD_INFO(%rsp,SIZEOF_PTREGS)
+ testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
CFI_REMEMBER_STATE
jnz sysenter_tracesys
cmpq $(IA32_NR_syscalls-1),%rax
@@ -177,10 +177,10 @@ sysenter_dispatch:
movq %rax,RAX(%rsp)
DISABLE_INTERRUPTS(CLBR_NONE)
TRACE_IRQS_OFF
- testl $_TIF_ALLWORK_MASK,TI_flags+THREAD_INFO(%rsp,RIP)
+ testl $_TIF_ALLWORK_MASK,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jnz sysexit_audit
sysexit_from_sys_call:
- andl $~TS_COMPAT,TI_status+THREAD_INFO(%rsp,RIP)
+ andl $~TS_COMPAT,TI_status+THREAD_INFO(%rsp,SIZEOF_PTREGS)
/* clear IF, that popfq doesn't enable interrupts early */
andl $~0x200,EFLAGS(%rsp)
movl RIP(%rsp),%edx /* User %eip */
@@ -225,7 +225,7 @@ sysexit_from_sys_call:
.endm

.macro auditsys_exit exit
- testl $(_TIF_ALLWORK_MASK & ~_TIF_SYSCALL_AUDIT),TI_flags+THREAD_INFO(%rsp,RIP)
+ testl $(_TIF_ALLWORK_MASK & ~_TIF_SYSCALL_AUDIT),TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jnz ia32_ret_from_sys_call
TRACE_IRQS_ON
ENABLE_INTERRUPTS(CLBR_NONE)
@@ -240,7 +240,7 @@ sysexit_from_sys_call:
movl $(_TIF_ALLWORK_MASK & ~_TIF_SYSCALL_AUDIT),%edi
DISABLE_INTERRUPTS(CLBR_NONE)
TRACE_IRQS_OFF
- testl %edi,TI_flags+THREAD_INFO(%rsp,RIP)
+ testl %edi,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jz \exit
CLEAR_RREGS
jmp int_with_check
@@ -262,7 +262,7 @@ sysenter_fix_flags:

sysenter_tracesys:
#ifdef CONFIG_AUDITSYSCALL
- testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags+THREAD_INFO(%rsp,RIP)
+ testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jz sysenter_auditsys
#endif
SAVE_EXTRA_REGS
@@ -346,8 +346,8 @@ ENTRY(ia32_cstar_target)
1: movl (%r8),%r9d
_ASM_EXTABLE(1b,ia32_badarg)
ASM_CLAC
- orl $TS_COMPAT,TI_status+THREAD_INFO(%rsp,RIP)
- testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,RIP)
+ orl $TS_COMPAT,TI_status+THREAD_INFO(%rsp,SIZEOF_PTREGS)
+ testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
CFI_REMEMBER_STATE
jnz cstar_tracesys
cmpq $IA32_NR_syscalls-1,%rax
@@ -364,10 +364,10 @@ cstar_dispatch:
movq %rax,RAX(%rsp)
DISABLE_INTERRUPTS(CLBR_NONE)
TRACE_IRQS_OFF
- testl $_TIF_ALLWORK_MASK,TI_flags+THREAD_INFO(%rsp,RIP)
+ testl $_TIF_ALLWORK_MASK,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jnz sysretl_audit
sysretl_from_sys_call:
- andl $~TS_COMPAT,TI_status+THREAD_INFO(%rsp,RIP)
+ andl $~TS_COMPAT,TI_status+THREAD_INFO(%rsp,SIZEOF_PTREGS)
RESTORE_RSI_RDI_RDX
movl RIP(%rsp),%ecx
CFI_REGISTER rip,rcx
@@ -402,7 +402,7 @@ sysretl_audit:

cstar_tracesys:
#ifdef CONFIG_AUDITSYSCALL
- testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags+THREAD_INFO(%rsp,RIP)
+ testl $(_TIF_WORK_SYSCALL_ENTRY & ~_TIF_SYSCALL_AUDIT),TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jz cstar_auditsys
#endif
xchgl %r9d,%ebp
@@ -469,8 +469,8 @@ ENTRY(ia32_syscall)
this could be a problem. */
ALLOC_PT_GPREGS_ON_STACK
SAVE_C_REGS_EXCEPT_R891011
- orl $TS_COMPAT,TI_status+THREAD_INFO(%rsp,RIP)
- testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,RIP)
+ orl $TS_COMPAT,TI_status+THREAD_INFO(%rsp,SIZEOF_PTREGS)
+ testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jnz ia32_tracesys
cmpq $(IA32_NR_syscalls-1),%rax
ja ia32_badsys
diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h
index ba115eb..0519945 100644
--- a/arch/x86/include/asm/thread_info.h
+++ b/arch/x86/include/asm/thread_info.h
@@ -207,10 +207,12 @@ static inline unsigned long current_stack_pointer(void)
_ASM_SUB $(THREAD_SIZE-KERNEL_STACK_OFFSET),reg ;

/*
- * Same if PER_CPU_VAR(kernel_stack) is, perhaps with some offset, already in
- * a certain register (to be used in assembler memory operands).
+ * ASM operand which evaluates to thread_info address
+ * if it is known that "reg" is exactly "off" bytes below stack top.
+ * Example (fetch thread_info->fieldname):
+ * mov TI_fieldname+THREAD_INFO(reg, off),%eax
*/
-#define THREAD_INFO(reg, off) KERNEL_STACK_OFFSET+(off)-THREAD_SIZE(reg)
+#define THREAD_INFO(reg, off) ((off)-THREAD_SIZE)(reg)

#endif

diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index 0c91256..d785a7d 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -258,7 +258,7 @@ GLOBAL(system_call_after_swapgs)
SAVE_C_REGS_EXCEPT_RAX_RCX_R11
movq $-ENOSYS,RAX(%rsp)
CFI_REL_OFFSET rip,RIP
- testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,RIP)
+ testl $_TIF_WORK_SYSCALL_ENTRY,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jnz tracesys
system_call_fastpath:
#if __SYSCALL_MASK == ~0
@@ -276,7 +276,7 @@ system_call_fastpath:
* Has incompletely filled pt_regs, iret frame is also incomplete.
*/
ret_from_sys_call:
- testl $_TIF_ALLWORK_MASK,TI_flags+THREAD_INFO(%rsp,RIP)
+ testl $_TIF_ALLWORK_MASK,TI_flags+THREAD_INFO(%rsp,SIZEOF_PTREGS)
jnz int_ret_from_sys_call_fixup /* Go the the slow path */

LOCKDEP_SYS_EXIT
--
1.8.1.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/