[PATCH] x86-64: fix register leak in 32-bit syscall audting

[PATCH] x86-64: fix register leak in 32-bit syscall audting

[Jan Beulich]

Restoring %ebp after the call to audit_syscall_exit() is not only
unnecessary (because the register didn't get clobbered), but in the
sysenter case wasn't even doing the right thing: It loaded %ebp from a
location below the top of stack (RBP < ARGOFFSET), i.e. arbitrary
kernel data got passed back to user mode in the register.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: stable@kernel.org 

---
 arch/x86/ia32/ia32entry.S |    5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

--- linux-2.6.32-rc5/arch/x86/ia32/ia32entry.S	2009-10-19 13:12:43.000000000 +0200
+++ 2.6.32-rc5-x86_64-ia32-syscall-audit-ebp/arch/x86/ia32/ia32entry.S	2009-10-19 09:36:39.000000000 +0200
@@ -204,7 +204,7 @@ sysexit_from_sys_call:
 	movl RDI-ARGOFFSET(%rsp),%r8d	/* reload 5th syscall arg */
 	.endm
 
-	.macro auditsys_exit exit,ebpsave=RBP
+	.macro auditsys_exit exit
 	testl $(_TIF_ALLWORK_MASK & ~_TIF_SYSCALL_AUDIT),TI_flags(%r10)
 	jnz ia32_ret_from_sys_call
 	TRACE_IRQS_ON
@@ -217,7 +217,6 @@ sysexit_from_sys_call:
 	call audit_syscall_exit
 	GET_THREAD_INFO(%r10)
 	movl RAX-ARGOFFSET(%rsp),%eax	/* reload syscall return value */
-	movl \ebpsave-ARGOFFSET(%rsp),%ebp /* reload user register value */
 	movl $(_TIF_ALLWORK_MASK & ~_TIF_SYSCALL_AUDIT),%edi
 	cli
 	TRACE_IRQS_OFF
@@ -351,7 +350,7 @@ cstar_auditsys:
 	jmp cstar_dispatch
 
 sysretl_audit:
-	auditsys_exit sysretl_from_sys_call, RCX /* user %ebp in RCX slot */
+	auditsys_exit sysretl_from_sys_call
 #endif
 
 cstar_tracesys:

Re: [PATCH] x86-64: fix register leak in 32-bit syscall audting

[Roland McGrath]

Acked-by: Roland McGrath <roland@redhat.com>