xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2 0/3] Make sscanf() stricter
@ 2023-06-10 17:07 Demi Marie Obenour
  2023-06-10 17:07 ` [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure Demi Marie Obenour
                   ` (2 more replies)
  0 siblings, 3 replies; 6+ messages in thread
From: Demi Marie Obenour @ 2023-06-10 17:07 UTC (permalink / raw)
  To: Hans de Goede, Mauro Carvalho Chehab, Sakari Ailus,
	Greg Kroah-Hartman, Juergen Gross, Stefano Stabellini,
	Oleksandr Tyshchenko, Lee Jones, Andy Lutomirski,
	Thomas Gleixner, Vincenzo Frascino, Petr Mladek, Steven Rostedt,
	Sergey Senozhatsky, Andy Shevchenko, Rasmus Villemoes
  Cc: Demi Marie Obenour, linux-media, linux-staging, linux-kernel, xen-devel

Roger Pau Monné suggested making xenbus_scanf() stricter instead of
using a custom parser.  Christoph Hellwig asked why the normal vsscanf()
cannot be made stricter.  Richard Weinberger mentioned Linus Torvalds’s
suggestion of using ! to allow overflow.

Changes since v1:

- Better commit messages.
- Use ! to explicitly opt-in to allowing overflow.
- Treat overflow as a conversion failure instead of returning ERANGE.
- Drop the first patch (removal of simple_strtoll()) as it breaks
  bcache.
- Stop skipping spaces in vsscanf() instead of adding a separate
  vsscanf_strict() function.

Demi Marie Obenour (3):
  vsscanf(): Integer overflow is a conversion failure
  vsscanf(): do not skip spaces
  Strict XenStore entry parsing

 .../hive_isp_css_include/platform_support.h   |  1 -
 drivers/xen/xenbus/xenbus_xs.c                | 17 ++--
 include/linux/limits.h                        |  1 +
 include/linux/mfd/wl1273-core.h               |  3 -
 include/vdso/limits.h                         |  3 +
 lib/vsprintf.c                                | 90 +++++++++++++------
 6 files changed, 80 insertions(+), 35 deletions(-)

-- 
Sincerely,
Demi Marie Obenour (she/her/hers)
Invisible Things Lab



^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure
  2023-06-10 17:07 [PATCH v2 0/3] Make sscanf() stricter Demi Marie Obenour
@ 2023-06-10 17:07 ` Demi Marie Obenour
  2023-06-10 19:38   ` kernel test robot
  2023-06-10 19:59   ` Dan Carpenter
  2023-06-10 17:07 ` [PATCH v2 2/3] vsscanf(): do not skip spaces Demi Marie Obenour
  2023-06-10 17:07 ` [PATCH v2 3/3] Strict XenStore entry parsing Demi Marie Obenour
  2 siblings, 2 replies; 6+ messages in thread
From: Demi Marie Obenour @ 2023-06-10 17:07 UTC (permalink / raw)
  To: Hans de Goede, Mauro Carvalho Chehab, Sakari Ailus,
	Greg Kroah-Hartman, Juergen Gross, Stefano Stabellini,
	Oleksandr Tyshchenko, Lee Jones, Andy Lutomirski,
	Thomas Gleixner, Vincenzo Frascino, Petr Mladek, Steven Rostedt,
	Sergey Senozhatsky, Andy Shevchenko, Rasmus Villemoes
  Cc: Demi Marie Obenour, linux-media, linux-staging, linux-kernel,
	xen-devel, Linus Torvalds

sscanf() and friends currently ignore integer overflow, but this is a
bad idea.  It is much better to detect integer overflow errors and
consider this a conversion failure.

This implements Linus's suggestion of using '!' to treat integer
overflow as wrapping.  It does _not_ allow wrapping of unsigned
conversions by default, though, as in at least some cases accepting
negative numbers is _not_ intended.

Suggested-By: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Demi Marie Obenour <demi@invisiblethingslab.com>
---
 .../hive_isp_css_include/platform_support.h   |  1 -
 include/linux/limits.h                        |  1 +
 include/linux/mfd/wl1273-core.h               |  3 -
 include/vdso/limits.h                         |  3 +
 lib/vsprintf.c                                | 82 ++++++++++++++-----
 5 files changed, 67 insertions(+), 23 deletions(-)

diff --git a/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h b/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h
index 0cdef4a5e8b1bed9884133f1a0b9d853d59d43a4..e29b96d8bebf14839f6dd48fdc6c0f8b029ef31d 100644
--- a/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h
+++ b/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h
@@ -27,7 +27,6 @@
 
 #define UINT16_MAX USHRT_MAX
 #define UINT32_MAX UINT_MAX
-#define UCHAR_MAX  (255)
 
 #define CSS_ALIGN(d, a) d __attribute__((aligned(a)))
 
diff --git a/include/linux/limits.h b/include/linux/limits.h
index f6bcc936901071f496e3e85bb6e1d93905b12e32..8f7fd85b41fb46e6992d9e5912da00424119227a 100644
--- a/include/linux/limits.h
+++ b/include/linux/limits.h
@@ -8,6 +8,7 @@
 
 #define SIZE_MAX	(~(size_t)0)
 #define SSIZE_MAX	((ssize_t)(SIZE_MAX >> 1))
+#define SSIZE_MIN	(-SSIZE_MAX - 1)
 #define PHYS_ADDR_MAX	(~(phys_addr_t)0)
 
 #define U8_MAX		((u8)~0U)
diff --git a/include/linux/mfd/wl1273-core.h b/include/linux/mfd/wl1273-core.h
index c28cf76d5c31ee1c94a9319a2e2d318bf00283a6..b81a229135ed9f756c749122a8341816031c8311 100644
--- a/include/linux/mfd/wl1273-core.h
+++ b/include/linux/mfd/wl1273-core.h
@@ -204,9 +204,6 @@
 				 WL1273_IS2_TRI_OPT | \
 				 WL1273_IS2_RATE_48K)
 
-#define SCHAR_MIN (-128)
-#define SCHAR_MAX 127
-
 #define WL1273_FR_EVENT			BIT(0)
 #define WL1273_BL_EVENT			BIT(1)
 #define WL1273_RDS_EVENT		BIT(2)
diff --git a/include/vdso/limits.h b/include/vdso/limits.h
index 0197888ad0e00b2f853d3f25ffa764f61cca7385..0cad0a2490e5efc194d874025eb3e3b846a5c7b4 100644
--- a/include/vdso/limits.h
+++ b/include/vdso/limits.h
@@ -2,6 +2,9 @@
 #ifndef __VDSO_LIMITS_H
 #define __VDSO_LIMITS_H
 
+#define UCHAR_MAX	((unsigned char)~0U)
+#define SCHAR_MAX	((signed char)(UCHAR_MAX >> 1))
+#define SCHAR_MIN	((signed char)(-SCHAR_MAX - 1))
 #define USHRT_MAX	((unsigned short)~0U)
 #define SHRT_MAX	((short)(USHRT_MAX >> 1))
 #define SHRT_MIN	((short)(-SHRT_MAX - 1))
diff --git a/lib/vsprintf.c b/lib/vsprintf.c
index 40f560959b169b4c4ac6154d658cfe76cfd0c5a6..8caccdcda0a2b470cda70c9b3837de37207eb512 100644
--- a/lib/vsprintf.c
+++ b/lib/vsprintf.c
@@ -59,7 +59,7 @@
 bool no_hash_pointers __ro_after_init;
 EXPORT_SYMBOL_GPL(no_hash_pointers);
 
-static noinline unsigned long long simple_strntoull(const char *startp, size_t max_chars, char **endp, unsigned int base)
+static noinline unsigned long long simple_strntoull(const char *startp, size_t max_chars, char **endp, unsigned int base, bool *overflow)
 {
 	const char *cp;
 	unsigned long long result = 0ULL;
@@ -71,6 +71,8 @@ static noinline unsigned long long simple_strntoull(const char *startp, size_t m
 	if (prefix_chars < max_chars) {
 		rv = _parse_integer_limit(cp, base, &result, max_chars - prefix_chars);
 		/* FIXME */
+		if (overflow)
+			*overflow = !!(rv & KSTRTOX_OVERFLOW);
 		cp += (rv & ~KSTRTOX_OVERFLOW);
 	} else {
 		/* Field too short for prefix + digit, skip over without converting */
@@ -94,7 +96,7 @@ static noinline unsigned long long simple_strntoull(const char *startp, size_t m
 noinline
 unsigned long long simple_strtoull(const char *cp, char **endp, unsigned int base)
 {
-	return simple_strntoull(cp, INT_MAX, endp, base);
+	return simple_strntoull(cp, INT_MAX, endp, base, NULL);
 }
 EXPORT_SYMBOL(simple_strtoull);
 
@@ -130,18 +132,22 @@ long simple_strtol(const char *cp, char **endp, unsigned int base)
 EXPORT_SYMBOL(simple_strtol);
 
 static long long simple_strntoll(const char *cp, size_t max_chars, char **endp,
-				 unsigned int base)
+				 unsigned int base, bool *overflow)
 {
+	unsigned long long minand;
+	bool negate;
+
 	/*
 	 * simple_strntoull() safely handles receiving max_chars==0 in the
 	 * case cp[0] == '-' && max_chars == 1.
 	 * If max_chars == 0 we can drop through and pass it to simple_strntoull()
 	 * and the content of *cp is irrelevant.
 	 */
-	if (*cp == '-' && max_chars > 0)
-		return -simple_strntoull(cp + 1, max_chars - 1, endp, base);
-
-	return simple_strntoull(cp, max_chars, endp, base);
+	negate = *cp == '-' && max_chars > 0;
+	minand = simple_strntoull(cp + negate, max_chars - negate, endp, base, overflow);
+	if (minand > (unsigned long long)LONG_MAX + negate)
+		*overflow = true;
+	return negate ? -minand : minand;
 }
 
 /**
@@ -3441,7 +3447,7 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
 		unsigned long long u;
 	} val;
 	s16 field_width;
-	bool is_sign;
+	bool is_sign, overflow, allow_overflow;
 
 	while (*fmt) {
 		/* skip any white space in format */
@@ -3464,6 +3470,9 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
 			break;
 		++fmt;
 
+		allow_overflow = *fmt == '!';
+		fmt += (int)allow_overflow;
+
 		/* skip this conversion.
 		 * advance both strings to next white space
 		 */
@@ -3649,45 +3658,80 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
 		if (is_sign)
 			val.s = simple_strntoll(str,
 						field_width >= 0 ? field_width : INT_MAX,
-						&next, base);
+						&next, base, &overflow);
 		else
 			val.u = simple_strntoull(str,
 						 field_width >= 0 ? field_width : INT_MAX,
-						 &next, base);
+						 &next, base, &overflow);
+		if (unlikely(overflow && !allow_overflow))
+			break;
 
 		switch (qualifier) {
 		case 'H':	/* that's 'hh' in format */
-			if (is_sign)
+			if (is_sign) {
+				if (unlikely(val.s < SCHAR_MIN || val.s > SCHAR_MAX) &&
+				    !allow_overflow)
+					return num;
 				*va_arg(args, signed char *) = val.s;
-			else
+			} else {
+				if (unlikely(val.u > UCHAR_MAX) && !allow_overflow)
+					return num;
 				*va_arg(args, unsigned char *) = val.u;
+			}
 			break;
 		case 'h':
-			if (is_sign)
+			if (is_sign) {
+				if (unlikely(val.s < SHRT_MIN || val.s > SHRT_MAX) &&
+				    !allow_overflow)
+					return num;
 				*va_arg(args, short *) = val.s;
-			else
+			} else {
+				if (unlikely(val.u > USHRT_MAX) && !allow_overflow)
+					return num;
 				*va_arg(args, unsigned short *) = val.u;
+			}
 			break;
 		case 'l':
-			if (is_sign)
+			if (is_sign) {
+				if (unlikely(val.s < LONG_MIN || val.s > LONG_MAX) &&
+				    !allow_overflow)
+					return num;
 				*va_arg(args, long *) = val.s;
-			else
+			} else {
+				if (unlikely(val.u > ULONG_MAX) && !allow_overflow)
+					return num;
 				*va_arg(args, unsigned long *) = val.u;
+			}
 			break;
 		case 'L':
+			/* No overflow check needed */
 			if (is_sign)
 				*va_arg(args, long long *) = val.s;
 			else
 				*va_arg(args, unsigned long long *) = val.u;
 			break;
 		case 'z':
-			*va_arg(args, size_t *) = val.u;
+			if (is_sign) {
+				if (unlikely(val.s < SSIZE_MIN || val.s > SSIZE_MAX))
+					return num;
+				*va_arg(args, ssize_t *) = val.s;
+			} else {
+				if (unlikely(val.u > SIZE_MAX) && !allow_overflow)
+					return num;
+				*va_arg(args, size_t *) = val.u;
+			}
 			break;
 		default:
-			if (is_sign)
+			if (is_sign) {
+				if (unlikely(val.s < INT_MIN || val.s > INT_MAX) &&
+				    !allow_overflow)
+					return num;
 				*va_arg(args, int *) = val.s;
-			else
+			} else {
+				if (unlikely(val.u > UINT_MAX) && !allow_overflow)
+					return num;
 				*va_arg(args, unsigned int *) = val.u;
+			}
 			break;
 		}
 		num++;
-- 
Sincerely,
Demi Marie Obenour (she/her/hers)
Invisible Things Lab



^ permalink raw reply related	[flat|nested] 6+ messages in thread

* [PATCH v2 2/3] vsscanf(): do not skip spaces
  2023-06-10 17:07 [PATCH v2 0/3] Make sscanf() stricter Demi Marie Obenour
  2023-06-10 17:07 ` [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure Demi Marie Obenour
@ 2023-06-10 17:07 ` Demi Marie Obenour
  2023-06-10 17:07 ` [PATCH v2 3/3] Strict XenStore entry parsing Demi Marie Obenour
  2 siblings, 0 replies; 6+ messages in thread
From: Demi Marie Obenour @ 2023-06-10 17:07 UTC (permalink / raw)
  To: Hans de Goede, Mauro Carvalho Chehab, Sakari Ailus,
	Greg Kroah-Hartman, Juergen Gross, Stefano Stabellini,
	Oleksandr Tyshchenko, Lee Jones, Andy Lutomirski,
	Thomas Gleixner, Vincenzo Frascino, Petr Mladek, Steven Rostedt,
	Sergey Senozhatsky, Andy Shevchenko, Rasmus Villemoes
  Cc: Demi Marie Obenour, linux-media, linux-staging, linux-kernel,
	xen-devel, Christoph Hellwig

Passing spaces before e.g. an integer is usually
not intended.  This was suggested by Christoph in
https://lore.kernel.org/lkml/ZIQrohcizoj4bZWx@infradead.org/.

Suggested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Demi Marie Obenour <demi@invisiblethingslab.com>
---
 lib/vsprintf.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/lib/vsprintf.c b/lib/vsprintf.c
index 8caccdcda0a2b470cda70c9b3837de37207eb512..d2d795d1aff0fd5cc50fb65ffb067067ab9f0dc6 100644
--- a/lib/vsprintf.c
+++ b/lib/vsprintf.c
@@ -3547,8 +3547,6 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
 			char *s = (char *)va_arg(args, char *);
 			if (field_width == -1)
 				field_width = SHRT_MAX;
-			/* first, skip leading white space in buffer */
-			str = skip_spaces(str);
 
 			/* now copy until next white space */
 			while (*str && !isspace(*str) && field_width--)
@@ -3635,11 +3633,7 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
 			return num;
 		}
 
-		/* have some sort of integer conversion.
-		 * first, skip white space in buffer.
-		 */
-		str = skip_spaces(str);
-
+		/* have some sort of integer conversion. */
 		digit = *str;
 		if (is_sign && digit == '-') {
 			if (field_width == 1)
-- 
Sincerely,
Demi Marie Obenour (she/her/hers)
Invisible Things Lab



^ permalink raw reply related	[flat|nested] 6+ messages in thread

* [PATCH v2 3/3] Strict XenStore entry parsing
  2023-06-10 17:07 [PATCH v2 0/3] Make sscanf() stricter Demi Marie Obenour
  2023-06-10 17:07 ` [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure Demi Marie Obenour
  2023-06-10 17:07 ` [PATCH v2 2/3] vsscanf(): do not skip spaces Demi Marie Obenour
@ 2023-06-10 17:07 ` Demi Marie Obenour
  2 siblings, 0 replies; 6+ messages in thread
From: Demi Marie Obenour @ 2023-06-10 17:07 UTC (permalink / raw)
  To: Hans de Goede, Mauro Carvalho Chehab, Sakari Ailus,
	Greg Kroah-Hartman, Juergen Gross, Stefano Stabellini,
	Oleksandr Tyshchenko, Lee Jones, Andy Lutomirski,
	Thomas Gleixner, Vincenzo Frascino, Petr Mladek, Steven Rostedt,
	Sergey Senozhatsky, Andy Shevchenko, Rasmus Villemoes
  Cc: Demi Marie Obenour, linux-media, linux-staging, linux-kernel, xen-devel

This rejects bogus xenstore node values that include e.g. leading spaces
or overflow the expected integer type.

Signed-off-by: Demi Marie Obenour <demi@invisiblethingslab.com>
---
 drivers/xen/xenbus/xenbus_xs.c | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/drivers/xen/xenbus/xenbus_xs.c b/drivers/xen/xenbus/xenbus_xs.c
index 12e02eb01f5991b31db451cc57037205359b347f..7cb2a22a7698ac40c81add23476594d9f27de8d0 100644
--- a/drivers/xen/xenbus/xenbus_xs.c
+++ b/drivers/xen/xenbus/xenbus_xs.c
@@ -569,16 +569,20 @@ int xenbus_scanf(struct xenbus_transaction t,
 		 const char *dir, const char *node, const char *fmt, ...)
 {
 	va_list ap;
-	int ret;
+	int ret = 0;
+	unsigned int len;
 	char *val;
 
-	val = xenbus_read(t, dir, node, NULL);
+	val = xenbus_read(t, dir, node, &len);
 	if (IS_ERR(val))
 		return PTR_ERR(val);
+	if (strlen(val) != len)
+		goto bad;
 
 	va_start(ap, fmt);
 	ret = vsscanf(val, fmt, ap);
 	va_end(ap);
+bad:
 	kfree(val);
 	/* Distinctive errno. */
 	if (ret == 0)
@@ -636,15 +640,18 @@ int xenbus_gather(struct xenbus_transaction t, const char *dir, ...)
 	while (ret == 0 && (name = va_arg(ap, char *)) != NULL) {
 		const char *fmt = va_arg(ap, char *);
 		void *result = va_arg(ap, void *);
+		unsigned len;
 		char *p;
 
-		p = xenbus_read(t, dir, name, NULL);
+		p = xenbus_read(t, dir, name, &len);
 		if (IS_ERR(p)) {
 			ret = PTR_ERR(p);
 			break;
 		}
-		if (fmt) {
-			if (sscanf(p, fmt, result) == 0)
+		if (strlen(p) != len)
+			ret = -EINVAL;
+		else if (fmt) {
+			if (sscanf(p, fmt, result) <= 0)
 				ret = -EINVAL;
 			kfree(p);
 		} else
-- 
Sincerely,
Demi Marie Obenour (she/her/hers)
Invisible Things Lab



^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure
  2023-06-10 17:07 ` [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure Demi Marie Obenour
@ 2023-06-10 19:38   ` kernel test robot
  2023-06-10 19:59   ` Dan Carpenter
  1 sibling, 0 replies; 6+ messages in thread
From: kernel test robot @ 2023-06-10 19:38 UTC (permalink / raw)
  To: Demi Marie Obenour, Hans de Goede, Mauro Carvalho Chehab,
	Sakari Ailus, Greg Kroah-Hartman, Juergen Gross,
	Stefano Stabellini, Oleksandr Tyshchenko, Lee Jones,
	Andy Lutomirski, Thomas Gleixner, Vincenzo Frascino, Petr Mladek,
	Steven Rostedt, Sergey Senozhatsky, Andy Shevchenko,
	Rasmus Villemoes
  Cc: oe-kbuild-all, linux-media, Demi Marie Obenour, linux-staging,
	linux-kernel, xen-devel

Hi Demi,

kernel test robot noticed the following build errors:

[auto build test ERROR on media-tree/master]
[also build test ERROR on lee-mfd/for-mfd-next xen-tip/linux-next linus/master lee-mfd/for-mfd-fixes v6.4-rc5 next-20230609]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Demi-Marie-Obenour/vsscanf-Integer-overflow-is-a-conversion-failure/20230611-010926
base:   git://linuxtv.org/media_tree.git master
patch link:    https://lore.kernel.org/r/20230610170743.2510-2-demi%40invisiblethingslab.com
patch subject: [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure
config: arc-randconfig-r033-20230611 (https://download.01.org/0day-ci/archive/20230611/202306110319.q5StyX6B-lkp@intel.com/config)
compiler: arceb-elf-gcc (GCC) 12.3.0
reproduce (this is a W=1 build):
        mkdir -p ~/bin
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        git remote add media-tree git://linuxtv.org/media_tree.git
        git fetch media-tree master
        git checkout media-tree/master
        b4 shazam https://lore.kernel.org/r/20230610170743.2510-2-demi@invisiblethingslab.com
        # save the config file
        mkdir build_dir && cp config build_dir/.config
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.3.0 ~/bin/make.cross W=1 O=build_dir ARCH=arc olddefconfig
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.3.0 ~/bin/make.cross W=1 O=build_dir ARCH=arc SHELL=/bin/bash

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202306110319.q5StyX6B-lkp@intel.com/

All errors (new ones prefixed by >>):

   lib/vsprintf.c: In function 'simple_strtoll':
>> lib/vsprintf.c:163:16: error: too few arguments to function 'simple_strntoll'
     163 |         return simple_strntoll(cp, INT_MAX, endp, base);
         |                ^~~~~~~~~~~~~~~
   lib/vsprintf.c:134:18: note: declared here
     134 | static long long simple_strntoll(const char *cp, size_t max_chars, char **endp,
         |                  ^~~~~~~~~~~~~~~
   lib/vsprintf.c: In function 'va_format':
   lib/vsprintf.c:1687:9: warning: function 'va_format' might be a candidate for 'gnu_printf' format attribute [-Wsuggest-attribute=format]
    1687 |         buf += vsnprintf(buf, end > buf ? end - buf : 0, va_fmt->fmt, va);
         |         ^~~
   lib/vsprintf.c: In function 'simple_strtoll':
   lib/vsprintf.c:164:1: error: control reaches end of non-void function [-Werror=return-type]
     164 | }
         | ^
   cc1: some warnings being treated as errors


vim +/simple_strntoll +163 lib/vsprintf.c

900fdc4573766d Richard Fitzgerald 2021-05-14  152  
^1da177e4c3f41 Linus Torvalds     2005-04-16  153  /**
^1da177e4c3f41 Linus Torvalds     2005-04-16  154   * simple_strtoll - convert a string to a signed long long
^1da177e4c3f41 Linus Torvalds     2005-04-16  155   * @cp: The start of the string
^1da177e4c3f41 Linus Torvalds     2005-04-16  156   * @endp: A pointer to the end of the parsed string will be placed here
^1da177e4c3f41 Linus Torvalds     2005-04-16  157   * @base: The number base to use
462e471107624f Eldad Zack         2012-12-17  158   *
e8cc2b97ca5aa1 Andy Shevchenko    2020-02-21  159   * This function has caveats. Please use kstrtoll instead.
^1da177e4c3f41 Linus Torvalds     2005-04-16  160   */
^1da177e4c3f41 Linus Torvalds     2005-04-16  161  long long simple_strtoll(const char *cp, char **endp, unsigned int base)
^1da177e4c3f41 Linus Torvalds     2005-04-16  162  {
900fdc4573766d Richard Fitzgerald 2021-05-14 @163  	return simple_strntoll(cp, INT_MAX, endp, base);
^1da177e4c3f41 Linus Torvalds     2005-04-16  164  }
98d5ce0d004466 Hans Verkuil       2010-04-23  165  EXPORT_SYMBOL(simple_strtoll);
^1da177e4c3f41 Linus Torvalds     2005-04-16  166  

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure
  2023-06-10 17:07 ` [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure Demi Marie Obenour
  2023-06-10 19:38   ` kernel test robot
@ 2023-06-10 19:59   ` Dan Carpenter
  1 sibling, 0 replies; 6+ messages in thread
From: Dan Carpenter @ 2023-06-10 19:59 UTC (permalink / raw)
  To: Demi Marie Obenour
  Cc: Hans de Goede, Mauro Carvalho Chehab, Sakari Ailus,
	Greg Kroah-Hartman, Juergen Gross, Stefano Stabellini,
	Oleksandr Tyshchenko, Lee Jones, Andy Lutomirski,
	Thomas Gleixner, Vincenzo Frascino, Petr Mladek, Steven Rostedt,
	Sergey Senozhatsky, Andy Shevchenko, Rasmus Villemoes,
	linux-media, linux-staging, linux-kernel, xen-devel,
	Linus Torvalds

On Sat, Jun 10, 2023 at 01:07:41PM -0400, Demi Marie Obenour wrote:
> ---
>  .../hive_isp_css_include/platform_support.h   |  1 -
>  include/linux/limits.h                        |  1 +
>  include/linux/mfd/wl1273-core.h               |  3 -
>  include/vdso/limits.h                         |  3 +
>  lib/vsprintf.c                                | 82 ++++++++++++++-----
>  5 files changed, 67 insertions(+), 23 deletions(-)
> 
> diff --git a/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h b/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h
> index 0cdef4a5e8b1bed9884133f1a0b9d853d59d43a4..e29b96d8bebf14839f6dd48fdc6c0f8b029ef31d 100644
> --- a/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h
> +++ b/drivers/staging/media/atomisp/pci/hive_isp_css_include/platform_support.h
> @@ -27,7 +27,6 @@
>  
>  #define UINT16_MAX USHRT_MAX
>  #define UINT32_MAX UINT_MAX
> -#define UCHAR_MAX  (255)
>  
>  #define CSS_ALIGN(d, a) d __attribute__((aligned(a)))
>  
> diff --git a/include/linux/limits.h b/include/linux/limits.h
> index f6bcc936901071f496e3e85bb6e1d93905b12e32..8f7fd85b41fb46e6992d9e5912da00424119227a 100644
> --- a/include/linux/limits.h
> +++ b/include/linux/limits.h
> @@ -8,6 +8,7 @@
>  
>  #define SIZE_MAX	(~(size_t)0)
>  #define SSIZE_MAX	((ssize_t)(SIZE_MAX >> 1))
> +#define SSIZE_MIN	(-SSIZE_MAX - 1)
>  #define PHYS_ADDR_MAX	(~(phys_addr_t)0)
>  
>  #define U8_MAX		((u8)~0U)
> diff --git a/include/linux/mfd/wl1273-core.h b/include/linux/mfd/wl1273-core.h
> index c28cf76d5c31ee1c94a9319a2e2d318bf00283a6..b81a229135ed9f756c749122a8341816031c8311 100644
> --- a/include/linux/mfd/wl1273-core.h
> +++ b/include/linux/mfd/wl1273-core.h
> @@ -204,9 +204,6 @@
>  				 WL1273_IS2_TRI_OPT | \
>  				 WL1273_IS2_RATE_48K)
>  
> -#define SCHAR_MIN (-128)
> -#define SCHAR_MAX 127
> -
>  #define WL1273_FR_EVENT			BIT(0)
>  #define WL1273_BL_EVENT			BIT(1)
>  #define WL1273_RDS_EVENT		BIT(2)
> diff --git a/include/vdso/limits.h b/include/vdso/limits.h
> index 0197888ad0e00b2f853d3f25ffa764f61cca7385..0cad0a2490e5efc194d874025eb3e3b846a5c7b4 100644
> --- a/include/vdso/limits.h
> +++ b/include/vdso/limits.h
> @@ -2,6 +2,9 @@
>  #ifndef __VDSO_LIMITS_H
>  #define __VDSO_LIMITS_H
>  
> +#define UCHAR_MAX	((unsigned char)~0U)
> +#define SCHAR_MAX	((signed char)(UCHAR_MAX >> 1))
> +#define SCHAR_MIN	((signed char)(-SCHAR_MAX - 1))
>  #define USHRT_MAX	((unsigned short)~0U)
>  #define SHRT_MAX	((short)(USHRT_MAX >> 1))
>  #define SHRT_MIN	((short)(-SHRT_MAX - 1))

It looks like you're going to have to redo these patches anyway for
various reasons.  Could you pull the U/SCHAR_MAX changes out into a
separate patch?

> diff --git a/lib/vsprintf.c b/lib/vsprintf.c
> index 40f560959b169b4c4ac6154d658cfe76cfd0c5a6..8caccdcda0a2b470cda70c9b3837de37207eb512 100644
> --- a/lib/vsprintf.c
> +++ b/lib/vsprintf.c
> @@ -59,7 +59,7 @@
>  bool no_hash_pointers __ro_after_init;
>  EXPORT_SYMBOL_GPL(no_hash_pointers);
>  
> -static noinline unsigned long long simple_strntoull(const char *startp, size_t max_chars, char **endp, unsigned int base)
> +static noinline unsigned long long simple_strntoull(const char *startp, size_t max_chars, char **endp, unsigned int base, bool *overflow)
>  {
>  	const char *cp;
>  	unsigned long long result = 0ULL;
> @@ -71,6 +71,8 @@ static noinline unsigned long long simple_strntoull(const char *startp, size_t m
>  	if (prefix_chars < max_chars) {
>  		rv = _parse_integer_limit(cp, base, &result, max_chars - prefix_chars);
>  		/* FIXME */

It's not clear what this FIXME is for, but probably it should go next to
the cp += (rv & ~KSTRTOX_OVERFLOW); line?  Does anyone know what it
means?  Maybe just delete it.

> +		if (overflow)
> +			*overflow = !!(rv & KSTRTOX_OVERFLOW);
>  		cp += (rv & ~KSTRTOX_OVERFLOW);
>  	} else {

*overflow isn't initialized on the else path.

>  		/* Field too short for prefix + digit, skip over without converting */
> @@ -94,7 +96,7 @@ static noinline unsigned long long simple_strntoull(const char *startp, size_t m
>  noinline
>  unsigned long long simple_strtoull(const char *cp, char **endp, unsigned int base)
>  {
> -	return simple_strntoull(cp, INT_MAX, endp, base);
> +	return simple_strntoull(cp, INT_MAX, endp, base, NULL);
>  }
>  EXPORT_SYMBOL(simple_strtoull);
>  
> @@ -130,18 +132,22 @@ long simple_strtol(const char *cp, char **endp, unsigned int base)
>  EXPORT_SYMBOL(simple_strtol);
>  
>  static long long simple_strntoll(const char *cp, size_t max_chars, char **endp,
> -				 unsigned int base)
> +				 unsigned int base, bool *overflow)
>  {
> +	unsigned long long minand;
> +	bool negate;
> +
>  	/*
>  	 * simple_strntoull() safely handles receiving max_chars==0 in the
>  	 * case cp[0] == '-' && max_chars == 1.
>  	 * If max_chars == 0 we can drop through and pass it to simple_strntoull()
>  	 * and the content of *cp is irrelevant.
>  	 */
> -	if (*cp == '-' && max_chars > 0)
> -		return -simple_strntoull(cp + 1, max_chars - 1, endp, base);
> -
> -	return simple_strntoull(cp, max_chars, endp, base);
> +	negate = *cp == '-' && max_chars > 0;
> +	minand = simple_strntoull(cp + negate, max_chars - negate, endp, base, overflow);
> +	if (minand > (unsigned long long)LONG_MAX + negate)
> +		*overflow = true;
> +	return negate ? -minand : minand;
>  }
>  
>  /**
> @@ -3441,7 +3447,7 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
>  		unsigned long long u;
>  	} val;
>  	s16 field_width;
> -	bool is_sign;
> +	bool is_sign, overflow, allow_overflow;
>  
>  	while (*fmt) {
>  		/* skip any white space in format */
> @@ -3464,6 +3470,9 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
>  			break;
>  		++fmt;
>  
> +		allow_overflow = *fmt == '!';
> +		fmt += (int)allow_overflow;
> +
>  		/* skip this conversion.
>  		 * advance both strings to next white space
>  		 */
> @@ -3649,45 +3658,80 @@ int vsscanf(const char *buf, const char *fmt, va_list args)
>  		if (is_sign)
>  			val.s = simple_strntoll(str,
>  						field_width >= 0 ? field_width : INT_MAX,
> -						&next, base);
> +						&next, base, &overflow);
>  		else
>  			val.u = simple_strntoull(str,
>  						 field_width >= 0 ? field_width : INT_MAX,
> -						 &next, base);
> +						 &next, base, &overflow);
> +		if (unlikely(overflow && !allow_overflow))

So that means that *overflow can be uninitialized here.

> +			break;
>  
>  		switch (qualifier) {

regards,
dan carpenter



^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2023-06-10 19:59 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-06-10 17:07 [PATCH v2 0/3] Make sscanf() stricter Demi Marie Obenour
2023-06-10 17:07 ` [PATCH v2 1/3] vsscanf(): Integer overflow is a conversion failure Demi Marie Obenour
2023-06-10 19:38   ` kernel test robot
2023-06-10 19:59   ` Dan Carpenter
2023-06-10 17:07 ` [PATCH v2 2/3] vsscanf(): do not skip spaces Demi Marie Obenour
2023-06-10 17:07 ` [PATCH v2 3/3] Strict XenStore entry parsing Demi Marie Obenour

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).