linux-kernel - Re: [PATCH] kconfig: undefined symbols can crash dependency loop detection

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Thu, 10 Feb 2011 23:35:31 +0100
From:	Ulf Magnusson <ulfalizer.lkml@...il.com>
To:	Michal Marek <mmarek@...e.cz>
Cc:	linux-kbuild@...r.kernel.org, zippel@...ux-m68k.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] kconfig: undefined symbols can crash dependency loop
 detection

On Thu, Feb 10, 2011 at 04:23:55PM +0100, Michal Marek wrote:
> On 5.2.2011 00:16, Ulf Magnusson wrote:
> > The Kconfig
> > 
> > config FOO
> > 	bool "FOO" if BAR
> > 	select BAR
> > 
> > has a dependency loop involving an undefined symbol BAR. This causes a
> > segfault in sym_check_print_recursive() as it assumes all symbols have a
> > non-null 'prop', which is not the case for undefined symbols. This is a
> > proposed fix.
> > 
> > Signed-off-by: Ulf Magnusson <ulfalizer.lkml@...il.com>
> > ---
> >  scripts/kconfig/symbol.c |   32 ++++++++++++++++++++++++++++----
> >  1 files changed, 28 insertions(+), 4 deletions(-)
> > 
> > diff --git a/scripts/kconfig/symbol.c b/scripts/kconfig/symbol.c
> > index a796c95..81a217f 100644
> > --- a/scripts/kconfig/symbol.c
> > +++ b/scripts/kconfig/symbol.c
> > @@ -5,6 +5,7 @@
> >  
> >  #include <ctype.h>
> >  #include <stdlib.h>
> > +#include <stdio.h>
> >  #include <string.h>
> >  #include <regex.h>
> >  #include <sys/utsname.h>
> > @@ -972,6 +973,10 @@ static void sym_check_print_recursive(struct symbol *last_sym)
> >  	struct menu *menu = NULL;
> >  	struct property *prop;
> >  	struct dep_stack cv_stack;
> > +	const char *filename;
> > +	int lineno;
> > +	char undef_msg_buf[100];
> > +	int snprintf_size;
> >  
> >  	if (sym_is_choice_value(last_sym)) {
> >  		dep_stack_insert(&cv_stack, last_sym);
> > @@ -1001,18 +1006,37 @@ static void sym_check_print_recursive(struct symbol *last_sym)
> >  					break;
> >  			}
> >  		}
> > +
> > +		if (prop) {
> > +			filename = prop->file->name;
> > +			lineno = prop->lineno;
> > +		} else {
> > +			/* The dependency loop involves an undefined symbol.
> > +			 * Checking sym->name is probably unnecessary here, but
> > +			 * just to be on the safe side. */
> > +			snprintf_size = snprintf(undef_msg_buf,
> > +						 sizeof(undef_msg_buf),
> > +						 "<no file, %s is undefined>",
> > +						 sym->name ? sym->name :
> > +							     "<choice>");
> > +			if (snprintf_size > sizeof(undef_msg_buf))
> > +				undef_msg_buf[sizeof(undef_msg_buf) - 1] = '\0';
> > +			filename = undef_msg_buf;
> > +			lineno = 0;
> 
> Good catch. But as this handles a corner case involving a severely
> borken Kconfig file, I would simply do
> filename = "<unknown>";
> lineno = 0;
> and be done with it.
>
Yeah, it probably isn't very likely to occur in practise, though it
would also happen if a symbol selects an undefined symbol it depends on
via an enclosing menu or if.

Pick whatever version you prefer :)

/Ulf
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/