Jeremy Fitzhardinge
2011-Aug-17 23:53 UTC
[Xen-devel] ld bug causing stupid CTOR count in mini-os
I''m seeing pvgrub crashing when running CTORs. It appears its because
the magic in the linker script is generating junk. If I get ld to
output a map, I see:
.ctors 0x0000000000097000 0x18
0x0000000000097000 __CTOR_LIST__ = .
0x0000000000097000 0x4 LONG 0x25c04 (((__CTOR_END__ -
__CTOR_LIST__) / 0x4) - 0x2)
*(.ctors)
.ctors 0x0000000000097004 0x10
/home/jeremy/hg/xen/unstable/stubdom/mini-os-x86_32-grub/mini-os.o
0x0000000000097014 0x4 LONG 0x0
0x0000000000097018 __CTOR_END__ = .
In other words, somehow ((0x97018-0x97000) / 4) - 2 = 0x25c04
The specific crash is that the ctor loop tries to call the NULL
sentinel. I''m seeing the same with the DTOR list.
I''m wondering, why not just terminate the loop with the NULL sentinel,
and forget about putting the count there?
J
diff -r b81c0417b901 extras/mini-os/arch/ia64/minios-ia64.lds
--- a/extras/mini-os/arch/ia64/minios-ia64.lds Wed Aug 17 16:08:41 2011 -0700
+++ b/extras/mini-os/arch/ia64/minios-ia64.lds Wed Aug 17 16:34:32 2011 -0700
@@ -55,7 +55,6 @@
.ctors : AT(ADDR(.ctors) - (((5<<(61))+0x100000000) - (1 << 20)))
{
__CTOR_LIST__ = .;
- QUAD((__CTOR_END__ - __CTOR_LIST__) / 8 - 2)
*(.ctors)
CONSTRUCTORS
QUAD(0)
@@ -65,7 +64,6 @@
.dtors : AT(ADDR(.dtors) - (((5<<(61))+0x100000000) - (1 << 20)))
{
__DTOR_LIST__ = .;
- QUAD((__DTOR_END__ - __DTOR_LIST__) / 8 - 2)
*(.dtors)
QUAD(0)
__DTOR_END__ = .;
diff -r b81c0417b901 extras/mini-os/arch/x86/minios-x86_32.lds
--- a/extras/mini-os/arch/x86/minios-x86_32.lds Wed Aug 17 16:08:41 2011 -0700
+++ b/extras/mini-os/arch/x86/minios-x86_32.lds Wed Aug 17 16:34:32 2011 -0700
@@ -30,7 +30,6 @@
.ctors : {
__CTOR_LIST__ = .;
- LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
*(.ctors)
CONSTRUCTORS
LONG(0)
@@ -39,7 +38,6 @@
.dtors : {
__DTOR_LIST__ = .;
- LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
*(.dtors)
LONG(0)
__DTOR_END__ = .;
diff -r b81c0417b901 extras/mini-os/arch/x86/minios-x86_64.lds
--- a/extras/mini-os/arch/x86/minios-x86_64.lds Wed Aug 17 16:08:41 2011 -0700
+++ b/extras/mini-os/arch/x86/minios-x86_64.lds Wed Aug 17 16:34:32 2011 -0700
@@ -30,7 +30,6 @@
.ctors : {
__CTOR_LIST__ = .;
- QUAD((__CTOR_END__ - __CTOR_LIST__) / 8 - 2)
*(.ctors)
CONSTRUCTORS
QUAD(0)
@@ -39,7 +38,6 @@
.dtors : {
__DTOR_LIST__ = .;
- QUAD((__DTOR_END__ - __DTOR_LIST__) / 8 - 2)
*(.dtors)
QUAD(0)
__DTOR_END__ = .;
diff -r b81c0417b901 extras/mini-os/main.c
--- a/extras/mini-os/main.c Wed Aug 17 16:08:41 2011 -0700
+++ b/extras/mini-os/main.c Wed Aug 17 16:34:32 2011 -0700
@@ -153,7 +153,7 @@
__libc_init_array();
environ = envp;
- for (i = 1; i <= __CTOR_LIST__[0]; i++)
+ for (i = 0; __CTOR_LIST__[i] != 0; i++)
((void((*)(void)))__CTOR_LIST__[i]) ();
tzset();
@@ -164,7 +164,7 @@
{
int i;
- for (i = 1; i <= __DTOR_LIST__[0]; i++)
+ for (i = 0; __DTOR_LIST__[i] != 0; i++)
((void((*)(void)))__DTOR_LIST__[i]) ();
close_all_files();
__libc_fini_array();
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
Samuel Thibault
2011-Aug-17 23:57 UTC
[Xen-devel] Re: ld bug causing stupid CTOR count in mini-os
Jeremy Fitzhardinge, le Wed 17 Aug 2011 16:53:06 -0700, a écrit :> I''m wondering, why not just terminate the loop with the NULL sentinel, > and forget about putting the count there?It looks safer than relying on correct size computation indeed. Samuel _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel