I wrote some scripts to do some tests about irqs.Sometimes the input from user may cause an unexpected result.
Could you please provide specific example?
just like __bitmap_parse, we return -EINVAL if there is no avaiable digit in each
parsing procedures.
Signed-off-by: Pan Xinhui <xinhuix.pan@xxxxxxxxx>
Hello, Pan.
(Adding Alexey Klimov, Rasmus Villemoes)
---
lib/bitmap.c | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/lib/bitmap.c b/lib/bitmap.c
index 64c0926..995fca2 100644
--- a/lib/bitmap.c
+++ b/lib/bitmap.c
@@ -504,7 +504,7 @@ static int __bitmap_parselist(const char *buf, unsigned int buflen,
int nmaskbits)
{
unsigned a, b;
- int c, old_c, totaldigits;
+ int c, old_c, totaldigits, ndigits;
const char __user __force *ubuf = (const char __user __force *)buf;
int exp_digit, in_range;
@@ -514,6 +514,7 @@ static int __bitmap_parselist(const char *buf, unsigned int buflen,
exp_digit = 1;
in_range = 0;
a = b = 0;
+ ndigits = 0;
/* Get the next cpu# or a range of cpu#'s */
while (buflen) {
@@ -555,8 +556,10 @@ static int __bitmap_parselist(const char *buf, unsigned int buflen,
if (!in_range)
a = b;
exp_digit = 0;
- totaldigits++;
+ ndigits++; totaldigits++;
I'm not happy with joining two statements to a single line.
Maybe sometimes it's OK for loop iterators like
while (a[i][j]) {
i++; j++;
}
But here it looks nasty. Anyway, it's minor.
}
+ if (ndigits == 0)
+ return -EINVAL;
You can avoid in-loop incrementation of ndigits if you'll
save current totaldigits to ndigits before loop, and check
ndigits against totaldigits after the loop:
ndigits = totaldigits;
while (...) {
...
totaldigits++;
}
if (ndigits == totaldigits)
return -EINVAL;
Maybe it's a good point to rework initial __bitmap_parse() similar way...
--if (!(a <= b))
return -EINVAL;
if (b >= nmaskbits)
--
1.9.1