Re: [PATCH] SEGV whilst placing sections

[Mark Shinwell <shinwell at codesourcery dot com>]

H. J. Lu wrote:
> On Tue, Jun 20, 2006 at 03:08:52PM +0100, Mark Shinwell wrote:
> > The following C code, when compiled for an ARM target with -fexceptions
> > and linked using the attached linker script, causes SEGV in
> > lang_insert_orphan () when placing the .ARM.extab section.
> >
> > int f(int *);
> >
> > int g(void)
> > {
> >     int a;
> >     return f(&a);
> > }
> >
> > In this case, the "after" parameter to lang_insert_orphan () corresponds
> > to the .rodata section, the last in the linked list of "asection"s.
> > The upshot is that place->section points at the "next" entry of the final
> > section in the list, .rodata.  This "next" pointer is NULL, so "as" ends up
> > NULL, and we fault when "as->prev" is computed.
>
> Can you provide a testcase in assembly so that people can check it
> with a cross binutils?

Attached is the output of compiling the above with -fexceptions.
Now the following link should segfault (linker script attached to previous
mail):

arm-none-linux-gnueabi-as -o nonleaf.o nonleaf.s
arm-none-linux-gnueabi-ld -T vmlinux.lds -o nonleaf nonleaf.o

Alan, is this test case ok for you?

Mark
	.cpu arm10tdmi
	.fpu softvfp
	.eabi_attribute 18, 4
	.eabi_attribute 20, 1
	.eabi_attribute 21, 1
	.eabi_attribute 23, 3
	.eabi_attribute 24, 1
	.eabi_attribute 25, 1
	.eabi_attribute 26, 2
	.eabi_attribute 30, 6
	.file	"nonleaf.c"
	.text
	.align	2
	.global	g
	.type	g, %function
g:
	.fnstart
.LFB2:
	@ args = 0, pretend = 0, frame = 8
	@ frame_needed = 1, uses_anonymous_args = 0
	.movsp ip
	mov	ip, sp
.LCFI0:
	stmfd	sp!, {fp, ip, lr, pc}
	.pad #4
	.save {fp, ip, lr}
.LCFI1:
	.setfp fp, ip, #-4
	sub	fp, ip, #4
.LCFI2:
	.pad #8
	sub	sp, sp, #8
.LCFI3:
	sub	r3, fp, #16
	mov	r0, r3
	bl	f
	mov	r3, r0
	mov	r0, r3
	sub	sp, fp, #12
	ldmfd	sp, {fp, sp, pc}
.LFE2:
	.fnend
	.size	g, .-g
	.ident	"GCC: (GNU) 4.1.0 (CodeSourcery ARM 2006q1-6)"