Skip to contentSkip to conversation

Fix buffer overflow in config parser

PATCH In review Thomas Jarosch 5 participants 10 messages Jun 15, 2016

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 Fix buffer overflow in config parser

When interpreting a config value, the config parser reads in 1+ space character(s) and puts -one- space character in the buffer as soon as the first non-space character is encountered (if not inside quotes).

Unfortunately the buffer size check lacks the extra space character which gets inserted at the next non-space character, resulting in a crash with a specially crafted config entry.

---
 config.c                |    2 +-
 t/t1303-wacky-config.sh |    9 ++++++++-
 2 files changed, 9 insertions(+), 2 deletions(-)
config.c +1 -1
index b76fe4c..2d70398 100644
@@ -51,7 +51,7 @@ static char *parse_value(void)
 
 	for (;;) {
 		int c = get_next_char();
-		if (len >= sizeof(value))
+		if (len >= sizeof(value) - 1)
 			return NULL;
 		if (c == '\n') {
 			if (quote)
t/t1303-wacky-config.sh +8 -1
index 1983076..a7d8d25 100755
@@ -10,7 +10,7 @@ setup() {
 
 check() {
 	echo "$2" >expected
-	git config --get "$1" >actual
+	git config --get "$1" >actual 2>&1
 	test_cmp actual expected
 }
 
@@ -40,4 +40,11 @@ test_expect_success 'make sure git config escapes section names properly' '
 	check "$SECTION" bar
 '
 
+LONG_VALUE=`perl -e 'print "x" x 1023," a"'`
+test_expect_success 'do not crash on special long config line' '
+	setup &&
+	git config section.key "$LONG_VALUE" &&
+	check section.key "fatal: bad config file line 2 in .git/config"
+'
+
 test_done
Signature · 1 line
1.6.1.3
#1 lore raw

Hi,

Quoted 1 line · Thomas Jarosch
 t/t1303-wacky-config.sh |    9 ++++++++-

I like the name!

Quoted 1 line

+LONG_VALUE=perl -e 'print "x" x 1023," a"'

But should it not be guarded against NO_PERL?

Ciao,
Dscho

#2 lore raw
Quoted 3 lines · Johannes Schindelin
Quoted 1 line

+LONG_VALUE=perl -e 'print "x" x 1023," a"'

But should it not be guarded against NO_PERL?

Hmm, lots of other tests like "t4200-rerere.sh" use perl and it don't see any special guard around the perl usage.

If it's needed, just add it while applying :-)

Thomas

#3 lore raw
Quoted 7 lines · Thomas Jarosch
Quoted 3 lines · Johannes Schindelin
Quoted 1 line

+LONG_VALUE=perl -e 'print "x" x 1023," a"'

But should it not be guarded against NO_PERL?

Hmm, lots of other tests like "t4200-rerere.sh" use perl and it don't see any special guard around the perl usage.

Right. There are really two types of perl usage in the tests:

1. Testing git programs which use perl.

2. Tests which happen to require perl as part of the testing.

Right now, only instances of (1) are marked with NO_PERL. This means that you can build with NO_PERL, test and install the result, and then uninstall perl (or make a binary package for perl-less people), and have it work fine. But you can't currently run the full test suite without any perl.

Your usage falls into (2), none of which are marked (and if they are to be marked, then all of them should be, since there is otherwise no point).

-Peff

#4 lore raw
Quoted 5 lines · Jeff King

2. Tests which happen to require perl as part of the testing.
[...]
Your usage falls into (2), none of which are marked (and if they are to be marked, then all of them should be, since there is otherwise no point).

I meant to add: it is still nice to avoid the use of perl in the test scripts if we can (and I think we can in this instance, as other responses have noted). If a patch were made to skip tests which use perl, then the fewer tests we have to skip, the better.

-Peff

#5 lore raw
In reply to Johannes Schindelin
Quoted 11 lines · Johannes Schindelin

Hi,

Quoted 1 line · Thomas Jarosch
 t/t1303-wacky-config.sh |    9 ++++++++-

I like the name!

Quoted 1 line

+LONG_VALUE=perl -e 'print "x" x 1023," a"'

But should it not be guarded against NO_PERL?

The right question to ask is a rhetorical "do we need perl to do this?"

#6 lore raw
Quoted 6 lines · Junio C Hamano
Quoted 4 lines · Johannes Schindelin

Hi,

On Tue, 14 Apr 2009, Thomas Jarosch wrote:

Quoted 7 lines
Quoted 5 lines
Quoted 1 line
 t/t1303-wacky-config.sh |    9 ++++++++-

I like the name!

Quoted 1 line

+LONG_VALUE=perl -e 'print "x" x 1023," a"'

But should it not be guarded against NO_PERL?

The right question to ask is a rhetorical "do we need perl to do this?"

LONG_VALUE=$(printf "x%0.1021dx a", 7)

-- Hannes

#7 lore raw
Quoted 7 lines · Johannes Sixt
Quoted 6 lines · Junio C Hamano
Quoted 4 lines · Johannes Schindelin

Hi,

On Tue, 14 Apr 2009, Thomas Jarosch wrote:

Quoted 8 lines
Quoted 6 lines
Quoted 5 lines
Quoted 1 line
 t/t1303-wacky-config.sh |    9 ++++++++-

I like the name!

Quoted 1 line

+LONG_VALUE=perl -e 'print "x" x 1023," a"'

But should it not be guarded against NO_PERL?

The right question to ask is a rhetorical "do we need perl to do this?"

LONG_VALUE=$(printf "x%0.1021dx a", 7)

Oops! Make this

LONG_VALUE=$(printf "x%01021dx a" 7)

-- Hannes

#8 lore raw

PATCH v3 1/1 Fix buffer overflow in config parser

When interpreting a config value, the config parser reads in 1+ space character(s) and puts -one- space character in the buffer as soon as the first non-space character is encountered (if not inside quotes).

Unfortunately the buffer size check lacks the extra space character which gets inserted at the next non-space character, resulting in a crash with a specially crafted config entry.

The unit test now uses Java to compile a platform independent .NET framework to output the test string in C# :o) Read:
Thanks to Johannes Sixt for the correct printf call which replaces the perl invocation.

---
 config.c                |    2 +-
 t/t1303-wacky-config.sh |    9 ++++++++-
 2 files changed, 9 insertions(+), 2 deletions(-)
config.c +1 -1
index b76fe4c..2d70398 100644
@@ -51,7 +51,7 @@ static char *parse_value(void)
 
 	for (;;) {
 		int c = get_next_char();
-		if (len >= sizeof(value))
+		if (len >= sizeof(value) - 1)
 			return NULL;
 		if (c == '\n') {
 			if (quote)
t/t1303-wacky-config.sh +8 -1
index 1983076..a7d8d25 100755
@@ -10,7 +10,7 @@ setup() {
 
 check() {
 	echo "$2" >expected
-	git config --get "$1" >actual
+	git config --get "$1" >actual 2>&1
 	test_cmp actual expected
 }
 
@@ -40,4 +40,11 @@ test_expect_success 'make sure git config escapes section names properly' '
 	check "$SECTION" bar
 '
 
+LONG_VALUE=$(printf "x%01021dx a" 7)
+test_expect_success 'do not crash on special long config line' '
+	setup &&
+	git config section.key "$LONG_VALUE" &&
+	check section.key "fatal: bad config file line 2 in .git/config"
+'
+
 test_done
Signature · 1 line
1.6.1.3
#9 lore raw

Hi,

Quoted 4 lines · Thomas Jarosch

When interpreting a config value, the config parser reads in 1+ space character(s) and puts -one- space character in the buffer as soon as the first non-space character is encountered (if not inside quotes).

Quoted 8 lines

Unfortunately the buffer size check lacks the extra space character which gets inserted at the next non-space character, resulting in a crash with a specially crafted config entry.

The unit test now uses Java to compile a platform independent .NET framework to output the test string in C# :o) Read:
Thanks to Johannes Sixt for the correct printf call which replaces the perl invocation.

LOL!

Thanks,
Dscho