lib/string_helpers.c: fix infinite loop in string_get_size()
Subsystems (2)
v3 cover lib/string_helpers.c: fix infinite loop in string_get_size()
This patch series prevents string_get_size() from entering an infinite loop on some inputs and adds several basic tests for string_get_size() including one to test for the issue.
Changes since v2:
- Add a comment explaining the fix to string_get_size() [James Bottomley]
- Add string_get_size() tests [Andy Shevchenko]
Vitaly Kuznetsov (2):
lib/string_helpers.c: fix infinite loop in string_get_size() lib/test-string_helpers.c: add string_get_size() tests
lib/string_helpers.c | 6 +++++- lib/test-string_helpers.c | 27 +++++++++++++++++++++++++++ 2 files changed, 32 insertions(+), 1 deletion(-)
Signature · 1 line
2.4.3
PATCH v3 1/2 lib/string_helpers.c: fix infinite loop in string_get_size()
Some string_get_size() calls (e.g.:
string_get_size(1, 512, STRING_UNITS_10, ..., ...) string_get_size(15, 64, STRING_UNITS_10, ..., ...)
) result in an infinite loop. The problem is that if size is equal to divisor[units]/blk_size and is smaller than divisor[units] we'll end up with size == 0 when we start doing sf_cap calculations:
For string_get_size(1, 512, STRING_UNITS_10, ..., ...) case:
...
remainder = do_div(size, divisor[units]); -> size is 0, remainder is 1 remainder *= blk_size; -> remainder is 512
...
size *= blk_size; -> size is still 0
size += remainder / divisor[units]; -> size is still 0
The caller causing the issue is sd_read_capacity(), the problem was noticed on Hyper-V, such weird size was reported by host when scanning collides with device removal. This is probably a separate issue worth fixing, this patch is intended to prevent the library routine from infinite looping.
--- lib/string_helpers.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
lib/string_helpers.c +5 -1
@@ -59,7 +59,11 @@ void string_get_size(u64 size, u64 blk_size, const enum string_size_units units, } exp = divisor[units] / (u32 )blk_size; -if (size >= exp) { +/* +* size must be strictly greater than exp here to ensure that remainder +* is greater than divisor[units] coming out of the if below. +*/ +if (size > exp) { remainder =do_div (size, divisor[units]); remainder *= blk_size; i++;
Signature · 1 line
2.4.3
PATCH v3 2/2 lib/test-string_helpers.c: add string_get_size() tests
In reply to Vitaly KuznetsovAdd a couple of simple tests for string_get_size(). The last one will hang the kernel without the 'lib/string_helpers.c: fix infinite loop in string_get_size()' fix.
--- lib/test-string_helpers.c | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+)
lib/test-string_helpers.c +27 -0
@@ -326,6 +326,30 @@ out:kfree (out_test); } +static __init void test_string_get_size_one (u64 size,u64 blk_size, +const enum string_size_units units, +const char *exp_result) +{ +char buf[256 ]; + +string_get_size (size, blk_size, units, buf,sizeof (buf)); +if (!strncmp (buf, exp_result,min (sizeof (buf),strlen (exp_result)))) +return ; + +pr_warn ("Test 'test_string_get_size_one' failed!\n" ); +pr_warn ("string_get_size(size = %llu, blk_size = %llu, units = %d\n" , + size, blk_size, units); +pr_warn ("expected: %s, got %s\n" , exp_result, buf); + +} + +static __init void test_string_get_size (void ) +{ +test_string_get_size_one (16384 ,512 ,STRING_UNITS_2 ,"8.00 MiB" ); +test_string_get_size_one (8192 ,4096 ,STRING_UNITS_10 ,"32.7 MB" ); +test_string_get_size_one (1 ,512 ,STRING_UNITS_10 ,"512 B" ); +} +static int __init test_string_helpers_init (void ) {unsigned int i;
@@ -344,6 +368,9 @@ static int __init test_string_helpers_init(void)for (i =0 ; i < (ESCAPE_ANY_NP |ESCAPE_HEX ) +1 ; i++)test_string_escape ("escape 1" , escape1, i,TEST_STRING_2_DICT_1 ); +/* Test string_get_size() */ +test_string_get_size (); +return -EINVAL ; }module_init (test_string_helpers_init);
Signature · 1 line
2.4.3
Re: [PATCH v3 2/2] lib/test-string_helpers.c: add string_get_size() tests
lib/test-string_helpers.c out: · line 337
Quoted 9 lines · Vitaly Kuznetsov
lib/test-string_helpers.c +9
-0+static __init void test_string_get_size_one (u64 size,u64 blk_size, +const enum string_size_units units, +const char *exp_result) +{ +char buf[256 ]; + +string_get_size (size, blk_size, units, buf,sizeof (buf)); +if (!strncmp (buf, exp_result,min (sizeof (buf),strlen (exp_result)))) +return ;
Nits: It probably makes sense to also test that string_get_size '\0'-terminates the buffer, so I'd spell this
if (!memcmp(buf, exp_result, min(sizeof(buf), strlen(exp_result)+1)))
With a generous stack buffer, that min() will always evaluate to the strlen(exp_result)+1. On that note: Maybe 256 is a bit excessive. I don't think this will run very deep in the kernel stack, but the code might get copy-pasted somewhere else. 16 should be plenty.
lib/test-string_helpers.c out: · line 341
Quoted 3 lines
lib/test-string_helpers.c +3
-0+pr_warn ("Test 'test_string_get_size_one' failed!\n" ); +pr_warn ("string_get_size(size = %llu, blk_size = %llu, units = %d\n" , + size, blk_size, units);
[There's probably no pretty way of getting from units to a text representation, but it's slightly annoying to have to check the source for the enum definition to figure out what units=0 or units=1 means.]
lib/test-string_helpers.c out: · line 342
Quoted 1 line
lib/test-string_helpers.c +1
-0+pr_warn ("expected: %s, got %s\n" , exp_result, buf);
In case we failed to '\0'-terminate buf, we might want to print it with "%.*s", (int)sizeof(buf), buf. But maybe I'm just overly paranoid.
Rasmus
Re: [PATCH v3 2/2] lib/test-string_helpers.c: add string_get_size() tests
Quoted 2 lines · Rasmus Villemoes
On Mon, Sep 14 2015, Vitaly Kuznetsov <[email protected]> wrote:
Vitaly, thanks for the test cases. My comments below.
Quoted 17 lines
lib/test-string_helpers.c out: · line 337
Quoted 13 lines
+static
__initvoid test_string_get_size_one(u64 size, u64blk_size,(quoted diff) +1
-0+ const enum
string_size_unitsunits,(quoted diff) +1
-0+ const char
*exp_result)lib/test-string_helpers.c +5
-0+{ + char buf[256]; + + string_get_size(size, blk_size, units, buf, sizeof(buf)); + if (!strncmp(buf, exp_result, min(sizeof(buf),
strlen(exp_result))))lib/test-string_helpers.c +1
-0+return ;Nits: It probably makes sense to also test that
string_get_size'\0'-terminates the buffer, so I'd spell this
Quoted 8 lines
if (!memcmp(buf,
exp_result, min(sizeof(buf),
strlen(exp_result)+1)))With a generous stack buffer, that
min()will always evaluate to the strlen(exp_result)+1. On that note: Maybe 256 is a bit excessive. I don't think this will run very deep in the kernel stack, but the code might
get copy-pasted somewhere else. 16 should be plenty.
Agree with Rasmus.
And just to make a side note that useless use of min() since we have strnlen() :-)
Quoted 11 lines
lib/test-string_helpers.c out: · line 341
Quoted 5 lines
(quoted diff) +2
-0+ pr_warn("Test 'test_string_get_size_one' failed!\n"); + pr_warn("string_get_size(size = %llu, blk_size = %llu,units = %d\n",
lib/test-string_helpers.c +1
-0+ size, blk_size, units);[There's probably no pretty way of getting from units to a text representation, but it's slightly annoying to have to check the source
for the enum definition to figure out what units=0 or units=1 means.]
Quoted 5 lines
lib/test-string_helpers.c out: · line 342
Quoted 1 line
lib/test-string_helpers.c +1
-0+pr_warn ("expected: %s, got %s\n" , exp_result, buf);In case we failed to '\0'-terminate buf, we might want to print it with
"%.*s", (int)sizeof(buf), buf. But maybe I'm just overly paranoid.
I prefer to put '\0' at the position after we expected have an actual '\0'. In this case we always be NULL terminated. I did this for hexdump test cases.
Signature · 2 lines
Andy Shevchenko <[email protected]> Intel Finland Oy
Re: [PATCH v3 2/2] lib/test-string_helpers.c: add string_get_size() tests
Quoted 54 lines · Andy Shevchenko
Quoted 2 lines · Rasmus Villemoes
On Mon, Sep 14 2015, Vitaly Kuznetsov <[email protected]> wrote:
Vitaly, thanks for the test cases. My comments below.
Quoted 17 lines
lib/test-string_helpers.c out: · line 337
Quoted 13 lines
+static
__initvoid test_string_get_size_one(u64 size, u64blk_size,(quoted diff) +1
-0+ const enum
string_size_unitsunits,(quoted diff) +1
-0+ const char
*exp_result)lib/test-string_helpers.c +5
-0+{ + char buf[256]; + + string_get_size(size, blk_size, units, buf, sizeof(buf)); + if (!strncmp(buf, exp_result, min(sizeof(buf),
strlen(exp_result))))lib/test-string_helpers.c +1
-0+return ;Nits: It probably makes sense to also test that
string_get_size'\0'-terminates the buffer, so I'd spell thisQuoted 8 lines
if (!memcmp(buf,
exp_result, min(sizeof(buf),
strlen(exp_result)+1)))With a generous stack buffer, that
min()will always evaluate to the strlen(exp_result)+1. On that note: Maybe 256 is a bit excessive. I don't think this will run very deep in the kernel stack, but the code might
get copy-pasted somewhere else. 16 should be plenty.Agree with Rasmus.
And just to make a side note that useless use of
min()since we havestrnlen():-)Quoted 11 lines
lib/test-string_helpers.c out: · line 341
Quoted 5 lines
(quoted diff) +2
-0+ pr_warn("Test 'test_string_get_size_one' failed!\n"); + pr_warn("string_get_size(size = %llu, blk_size = %llu,units = %d\n",
lib/test-string_helpers.c +1
-0+ size, blk_size, units);[There's probably no pretty way of getting from units to a text representation, but it's slightly annoying to have to check the source
for the enum definition to figure out what units=0 or units=1 means.]Quoted 5 lines
lib/test-string_helpers.c out: · line 342
Quoted 1 line
lib/test-string_helpers.c +1
-0+pr_warn ("expected: %s, got %s\n" , exp_result, buf);In case we failed to '\0'-terminate buf, we might want to print it with
"%.*s", (int)sizeof(buf), buf. But maybe I'm just overly paranoid.
Quoted 3 lines
I prefer to put '\0' at the position after we expected have an actual '\0'. In this case we always be NULL terminated. I did this for hexdump test cases.
Just to check I got your suggestions right:
...
+ if (!memcmp(buf, exp_result, strnlen(exp_result, sizeof(buf) - 1) + 1))
+ return;
+
+ /* NULL terminate buf right after the expected '\0' */
+ buf[strnlen(exp_result, sizeof(buf) - 2) + 1] = '\0';
...
Alternatively, we could have avoided strnlen() by asserting strlen(exp_result) < sizeof(buf) - 1 at the very beginning.
Signature · 1 line
Vitaly
Re: [PATCH v3 2/2] lib/test-string_helpers.c: add string_get_size() tests
Quoted 19 lines · Vitaly Kuznetsov
Quoted 5 lines · Andy Shevchenko
I prefer to put '\0' at the position after we expected have an actual
'\0'. In this case we always be NULL terminated. I did this for hexdump
test cases.Just to check I got your suggestions right:
...
(quoted diff) +5
-1+ if (!memcmp(buf, exp_result, strnlen(exp_result, sizeof(buf) - 1) + 1)) + return; + + /* NULL terminate buf right after the expected '\0' */ + buf[strnlen(exp_result, sizeof(buf) - 2) + 1] = '\0';
Quoted 3 lines
Alternatively, we could have avoided
strnlen()by assertingstrlen(exp_result)< sizeof(buf) - 1 at the very beginning.
Just
buf[sizeof(buf) - 1] = '\0';
should be enough after you called the string_get_size().
And minimize the buffer to something like 16 (whatever is the biggest possible length + '\0' + some space for the wrong algorithm aligned to let's say 4.
Signature · 2 lines
Andy Shevchenko <[email protected]> Intel Finland Oy