Module Name:    src
Committed By:   maxv
Date:           Sun Mar 24 13:15:43 UTC 2019

Modified Files:
        src/sys/arch/amd64/amd64: machdep.c
        src/sys/compat/linux/arch/amd64: linux_machdep.c

Log Message:
Fix a tiny race in setregs and linux_setregs. Between the moment we set
pcb_flags to zero, and the moment cpu_segregs64_zero resets pcb_gs, we may
be preempted.

If this happens, and if the calling LWP was a 32bit thread, when switching
back to that LWP, the context switcher sees that PCB_COMPAT32 is not set in
pcb_flags and tries to perform a 64bit context switch; but pcb_gs contains
a 32bit GDT descriptor, and not a 64bit GS.base value. The wrmsr therefore
faults because the value is non-canonical, and this fault is fatal.

Rearrange the code so that the update of pcb_flags and pcb_gs/pcb_fs is non
interruptible. This fixes the problem, tested with a reproducer (which
therefore doesn't work anymore).

Likely fixes PR/53993.


To generate a diff of this commit:
cvs rdiff -u -r1.327 -r1.328 src/sys/arch/amd64/amd64/machdep.c
cvs rdiff -u -r1.56 -r1.57 src/sys/compat/linux/arch/amd64/linux_machdep.c

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.

Modified files:

Index: src/sys/arch/amd64/amd64/machdep.c
diff -u src/sys/arch/amd64/amd64/machdep.c:1.327 src/sys/arch/amd64/amd64/machdep.c:1.328
--- src/sys/arch/amd64/amd64/machdep.c:1.327	Sat Mar  9 08:42:25 2019
+++ src/sys/arch/amd64/amd64/machdep.c	Sun Mar 24 13:15:42 2019
@@ -1,4 +1,4 @@
-/*	$NetBSD: machdep.c,v 1.327 2019/03/09 08:42:25 maxv Exp $	*/
+/*	$NetBSD: machdep.c,v 1.328 2019/03/24 13:15:42 maxv Exp $	*/
 
 /*
  * Copyright (c) 1996, 1997, 1998, 2000, 2006, 2007, 2008, 2011
@@ -110,7 +110,7 @@
  */
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: machdep.c,v 1.327 2019/03/09 08:42:25 maxv Exp $");
+__KERNEL_RCSID(0, "$NetBSD: machdep.c,v 1.328 2019/03/24 13:15:42 maxv Exp $");
 
 #include "opt_modular.h"
 #include "opt_user_ldt.h"
@@ -1371,17 +1371,18 @@ setregs(struct lwp *l, struct exec_packa
 
 	fpu_save_area_clear(l, pack->ep_osversion >= 699002600
 	    ? __NetBSD_NPXCW__ : __NetBSD_COMPAT_NPXCW__);
-	pcb->pcb_flags = 0;
 	x86_dbregs_clear(l);
 
+	kpreempt_disable();
+	pcb->pcb_flags = 0;
 	l->l_proc->p_flag &= ~PK_32;
-
 	l->l_md.md_flags = MDL_IRET;
+	cpu_segregs64_zero(l);
+	kpreempt_enable();
 
 	tf = l->l_md.md_regs;
 	tf->tf_ds = GSEL(GUDATA_SEL, SEL_UPL);
 	tf->tf_es = GSEL(GUDATA_SEL, SEL_UPL);
-	cpu_segregs64_zero(l);
 	tf->tf_rdi = 0;
 	tf->tf_rsi = 0;
 	tf->tf_rbp = 0;
@@ -2198,12 +2199,12 @@ cpu_segregs64_zero(struct lwp *l)
 	struct pcb *pcb;
 	uint64_t zero = 0;
 
+	KASSERT(kpreempt_disabled());
 	KASSERT((l->l_proc->p_flag & PK_32) == 0);
 	KASSERT(l == curlwp);
 
 	pcb = lwp_getpcb(l);
 
-	kpreempt_disable();
 	tf->tf_fs = 0;
 	tf->tf_gs = 0;
 	setds(GSEL(GUDATA_SEL, SEL_UPL));
@@ -2223,7 +2224,6 @@ cpu_segregs64_zero(struct lwp *l)
 	pcb->pcb_gs = 0;
 	update_descriptor(&curcpu()->ci_gdt[GUFS_SEL], &zero);
 	update_descriptor(&curcpu()->ci_gdt[GUGS_SEL], &zero);
-	kpreempt_enable();
 }
 
 /*

Index: src/sys/compat/linux/arch/amd64/linux_machdep.c
diff -u src/sys/compat/linux/arch/amd64/linux_machdep.c:1.56 src/sys/compat/linux/arch/amd64/linux_machdep.c:1.57
--- src/sys/compat/linux/arch/amd64/linux_machdep.c:1.56	Mon Jan  1 08:03:43 2018
+++ src/sys/compat/linux/arch/amd64/linux_machdep.c	Sun Mar 24 13:15:43 2019
@@ -1,4 +1,4 @@
-/*	$NetBSD: linux_machdep.c,v 1.56 2018/01/01 08:03:43 maxv Exp $ */
+/*	$NetBSD: linux_machdep.c,v 1.57 2019/03/24 13:15:43 maxv Exp $ */
 
 /*-
  * Copyright (c) 2005 Emmanuel Dreyfus, all rights reserved.
@@ -33,7 +33,7 @@
 
 #include <sys/cdefs.h>
 
-__KERNEL_RCSID(0, "$NetBSD: linux_machdep.c,v 1.56 2018/01/01 08:03:43 maxv Exp $");
+__KERNEL_RCSID(0, "$NetBSD: linux_machdep.c,v 1.57 2019/03/24 13:15:43 maxv Exp $");
 
 #include <sys/param.h>
 #include <sys/types.h>
@@ -91,11 +91,13 @@ linux_setregs(struct lwp *l, struct exec
 #endif
 
 	fpu_save_area_clear(l, __NetBSD_NPXCW__);
-	pcb->pcb_flags = 0;
 
+	kpreempt_disable();
+	pcb->pcb_flags = 0;
 	l->l_proc->p_flag &= ~PK_32;
-
 	l->l_md.md_flags = MDL_IRET;
+	cpu_segregs64_zero(l);
+	kpreempt_enable();
 
 	tf = l->l_md.md_regs;
 	tf->tf_rax = 0;
@@ -120,7 +122,6 @@ linux_setregs(struct lwp *l, struct exec
 	tf->tf_ss = GSEL(GUDATA_SEL, SEL_UPL);
 	tf->tf_ds = GSEL(GUDATA_SEL, SEL_UPL);
 	tf->tf_es = 0;
-	cpu_segregs64_zero(l);
 
 	return;
 }

Reply via email to