* [PATCH 5/9] staging: panel: check put_user() return value
@ 2010-07-30 11:08 Kulikov Vasiliy
2010-07-30 21:15 ` Willy Tarreau
0 siblings, 1 reply; 3+ messages in thread
From: Kulikov Vasiliy @ 2010-07-30 11:08 UTC (permalink / raw)
To: kernel-janitors
Cc: Greg Kroah-Hartman, Willy Tarreau, Peter Huewe, Andy Shevchenko,
Julia Lawall, devel, linux-kernel
put_user() may fail, if so return -EFAULT and do not update driver
pointers.
Signed-off-by: Kulikov Vasiliy <segooon@gmail.com>
---
drivers/staging/panel/panel.c | 17 +++++++++++------
1 files changed, 11 insertions(+), 6 deletions(-)
diff --git a/drivers/staging/panel/panel.c b/drivers/staging/panel/panel.c
index f58da32..57f4946 100644
--- a/drivers/staging/panel/panel.c
+++ b/drivers/staging/panel/panel.c
@@ -1589,25 +1589,30 @@ void lcd_init(void)
static ssize_t keypad_read(struct file *file,
char *buf, size_t count, loff_t *ppos)
{
-
+ int buflen = keypad_buflen;
unsigned i = *ppos;
char *tmp = buf;
+ int start = keypad_start;
- if (keypad_buflen == 0) {
+ if (buflen == 0) {
if (file->f_flags & O_NONBLOCK)
return -EAGAIN;
interruptible_sleep_on(&keypad_read_wait);
if (signal_pending(current))
return -EINTR;
+ buflen = keypad_buflen;
}
- for (; count-- > 0 && (keypad_buflen > 0);
- ++i, ++tmp, --keypad_buflen) {
- put_user(keypad_buffer[keypad_start], tmp);
- keypad_start = (keypad_start + 1) % KEYPAD_BUFFER;
+ for (; count-- > 0 && (buflen > 0);
+ ++i, ++tmp, --buflen) {
+ if (put_user(keypad_buffer[start], tmp))
+ return -EFAULT;
+ start = (start + 1) % KEYPAD_BUFFER;
}
*ppos = i;
+ keypad_buflen = buflen;
+ keypad_start = start;
return tmp - buf;
}
--
1.7.0.4
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH 5/9] staging: panel: check put_user() return value
2010-07-30 11:08 [PATCH 5/9] staging: panel: check put_user() return value Kulikov Vasiliy
@ 2010-07-30 21:15 ` Willy Tarreau
2010-07-31 8:39 ` walter harms
0 siblings, 1 reply; 3+ messages in thread
From: Willy Tarreau @ 2010-07-30 21:15 UTC (permalink / raw)
To: Kulikov Vasiliy
Cc: kernel-janitors, Greg Kroah-Hartman, Peter Huewe, Andy Shevchenko,
Julia Lawall, devel, linux-kernel
On Fri, Jul 30, 2010 at 03:08:42PM +0400, Kulikov Vasiliy wrote:
> diff --git a/drivers/staging/panel/panel.c b/drivers/staging/panel/panel.c
> index f58da32..57f4946 100644
> --- a/drivers/staging/panel/panel.c
> +++ b/drivers/staging/panel/panel.c
> @@ -1589,25 +1589,30 @@ void lcd_init(void)
> static ssize_t keypad_read(struct file *file,
> char *buf, size_t count, loff_t *ppos)
> {
> -
> + int buflen = keypad_buflen;
> unsigned i = *ppos;
> char *tmp = buf;
> + int start = keypad_start;
>
> - if (keypad_buflen == 0) {
> + if (buflen == 0) {
> if (file->f_flags & O_NONBLOCK)
> return -EAGAIN;
>
> interruptible_sleep_on(&keypad_read_wait);
> if (signal_pending(current))
> return -EINTR;
> + buflen = keypad_buflen;
> }
Not sure what the intent was here, I think you're re-adjusting
the buffer's length in case something else was read. But then
I don't understand why buflen it not simply assigned after the
if() block.
The rest below looks fine otherwise.
>
> - for (; count-- > 0 && (keypad_buflen > 0);
> - ++i, ++tmp, --keypad_buflen) {
> - put_user(keypad_buffer[keypad_start], tmp);
> - keypad_start = (keypad_start + 1) % KEYPAD_BUFFER;
> + for (; count-- > 0 && (buflen > 0);
> + ++i, ++tmp, --buflen) {
> + if (put_user(keypad_buffer[start], tmp))
> + return -EFAULT;
> + start = (start + 1) % KEYPAD_BUFFER;
> }
> *ppos = i;
> + keypad_buflen = buflen;
> + keypad_start = start;
>
> return tmp - buf;
> }
Regards,
Willy
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH 5/9] staging: panel: check put_user() return value
2010-07-30 21:15 ` Willy Tarreau
@ 2010-07-31 8:39 ` walter harms
0 siblings, 0 replies; 3+ messages in thread
From: walter harms @ 2010-07-31 8:39 UTC (permalink / raw)
To: Willy Tarreau
Cc: Kulikov Vasiliy, kernel-janitors, Greg Kroah-Hartman, Peter Huewe,
Andy Shevchenko, Julia Lawall, devel, linux-kernel
Willy Tarreau schrieb:
> On Fri, Jul 30, 2010 at 03:08:42PM +0400, Kulikov Vasiliy wrote:
>> diff --git a/drivers/staging/panel/panel.c b/drivers/staging/panel/panel.c
>> index f58da32..57f4946 100644
>> --- a/drivers/staging/panel/panel.c
>> +++ b/drivers/staging/panel/panel.c
>> @@ -1589,25 +1589,30 @@ void lcd_init(void)
>> static ssize_t keypad_read(struct file *file,
>> char *buf, size_t count, loff_t *ppos)
>> {
>> -
>> + int buflen = keypad_buflen;
>> unsigned i = *ppos;
>> char *tmp = buf;
>> + int start = keypad_start;
>>
>> - if (keypad_buflen == 0) {
>> + if (buflen == 0) {
>> if (file->f_flags & O_NONBLOCK)
>> return -EAGAIN;
>>
>> interruptible_sleep_on(&keypad_read_wait);
>> if (signal_pending(current))
>> return -EINTR;
>> + buflen = keypad_buflen;
>> }
>
> Not sure what the intent was here, I think you're re-adjusting
> the buffer's length in case something else was read. But then
> I don't understand why buflen it not simply assigned after the
> if() block.
>
> The rest below looks fine otherwise.
>
>>
>> - for (; count-- > 0 && (keypad_buflen > 0);
>> - ++i, ++tmp, --keypad_buflen) {
>> - put_user(keypad_buffer[keypad_start], tmp);
>> - keypad_start = (keypad_start + 1) % KEYPAD_BUFFER;
>> + for (; count-- > 0 && (buflen > 0);
>> + ++i, ++tmp, --buflen) {
>> + if (put_user(keypad_buffer[start], tmp))
>> + return -EFAULT;
>> + start = (start + 1) % KEYPAD_BUFFER;
>> }
>> *ppos = i;
>> + keypad_buflen = buflen;
>> + keypad_start = start;
>>
>> return tmp - buf;
>> }
>
>
IMHO opinion the for() construct breaks the rule of "no surprise please".
perhaps a while() would improve readability.
just my two cents,
re,
wh
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2010-07-31 8:39 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-07-30 11:08 [PATCH 5/9] staging: panel: check put_user() return value Kulikov Vasiliy
2010-07-30 21:15 ` Willy Tarreau
2010-07-31 8:39 ` walter harms
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox