2008-10-14 20:39:45

by Oliver Bock

Subject: 2.6.27: "array subscript is above array bounds" warnings

Hi all,

FYI, I just came across the following warnings when compiling 2.6.27
(release) using gcc 4.3.2 (Debian 4.3.2-1) on Debian Lenny (i386):

----------
drivers/acpi/tables/tbfadt.c: In function ?acpi_tb_create_local_fadt?:
include/asm/string_32.h:75: warning: array subscript is above array bounds

drivers/usb/core/hcd.c: In function ?usb_hcd_poll_rh_status?:
include/asm/string_32.h:75: warning: array subscript is above array bounds

fs/xfs/xfs_mount.c: In function ?xfs_mod_sb?:
fs/xfs/xfs_mount.c:1456: warning: array subscript is above array bounds
----------


Cheers,
Oliver

2008-10-18 06:51:11

by Andrew Morton

Subject: Re: 2.6.27: "array subscript is above array bounds" warnings

(cc's added)

On Tue, 14 Oct 2008 22:17:36 +0200 Oliver Bock <[email protected]> wrote:

> Hi all,
>
> FYI, I just came across the following warnings when compiling 2.6.27
> (release) using gcc 4.3.2 (Debian 4.3.2-1) on Debian Lenny (i386):
>
> ----------
> drivers/acpi/tables/tbfadt.c: In function ___acpi_tb_create_local_fadt___:
> include/asm/string_32.h:75: warning: array subscript is above array bounds
>
> drivers/usb/core/hcd.c: In function ___usb_hcd_poll_rh_status___:
> include/asm/string_32.h:75: warning: array subscript is above array bounds
>
> fs/xfs/xfs_mount.c: In function ___xfs_mod_sb___:
> fs/xfs/xfs_mount.c:1456: warning: array subscript is above array bounds
> ----------

2008-10-18 15:00:25

by Alan Stern

Subject: Re: 2.6.27: "array subscript is above array bounds" warnings

On Fri, 17 Oct 2008, Andrew Morton wrote:

> (cc's added)
>
> On Tue, 14 Oct 2008 22:17:36 +0200 Oliver Bock <[email protected]> wrote:
>
> > Hi all,
> >
> > FYI, I just came across the following warnings when compiling 2.6.27
> > (release) using gcc 4.3.2 (Debian 4.3.2-1) on Debian Lenny (i386):

...

> > drivers/usb/core/hcd.c: In function ___usb_hcd_poll_rh_status___:
> > include/asm/string_32.h:75: warning: array subscript is above array bounds

Does this refer to the call to memcpy() in usb_hcd_poll_rh_status()?

Line 75 in string_32.h is part of the __constant_memcpy() function.
That function should not be involved here because the length argument
isn't a compile-time constant.

This has to be either a compiler bug or else a bug in the
implementation of memcpy.

Alan Stern