2019-07-30 04:14:40

by Zhang Yi

Subject: [PATCH v2] aio: add timeout validity check for io_[p]getevents

io_[p]getevents syscall should return -EINVAL if timeout is out of
range, add this validity check.

Signed-off-by: zhangyi (F) <[email protected]>
Reviewed-by: Jeff Moyer <[email protected]>
Cc: Jeff Moyer <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Deepa Dinamani <[email protected]>
---
Changes since v1:
- add Reviewed-by and Cc tags.

fs/aio.c | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/fs/aio.c b/fs/aio.c
index 01e0fb9..dd967a0 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -2031,10 +2031,17 @@ static long do_io_getevents(aio_context_t ctx_id,
struct io_event __user *events,
struct timespec64 *ts)
{
- ktime_t until = ts ? timespec64_to_ktime(*ts) : KTIME_MAX;
- struct kioctx *ioctx = lookup_ioctx(ctx_id);
+ ktime_t until = KTIME_MAX;
+ struct kioctx *ioctx = NULL;
long ret = -EINVAL;

+ if (ts) {
+ if (!timespec64_valid(ts))
+ return ret;
+ until = timespec64_to_ktime(*ts);
+ }
+
+ ioctx = lookup_ioctx(ctx_id);
if (likely(ioctx)) {
if (likely(min_nr <= nr && min_nr >= 0))
ret = read_events(ioctx, min_nr, nr, events, until);
--
2.7.4

2019-07-30 10:29:32

by Arnd Bergmann

Subject: Re: [PATCH v2] aio: add timeout validity check for io_[p]getevents

On Tue, Jul 30, 2019 at 3:46 AM zhangyi (F) <[email protected]> wrote:

> {
> - ktime_t until = ts ? timespec64_to_ktime(*ts) : KTIME_MAX;
> - struct kioctx *ioctx = lookup_ioctx(ctx_id);
> + ktime_t until = KTIME_MAX;
> + struct kioctx *ioctx = NULL;
> long ret = -EINVAL;
>
> + if (ts) {
> + if (!timespec64_valid(ts))
> + return ret;
> + until = timespec64_to_ktime(*ts);
> + }

The man page should probably get updated as well to reflect that this
will now return -EINVAL for a negative timeout or malformed
nanoseconds.

Arnd

2019-07-30 17:26:02

by Zhang Yi

Subject: Re: [PATCH v2] aio: add timeout validity check for io_[p]getevents

On 2019/7/30 15:11, Arnd Bergmann Wrote:
> On Tue, Jul 30, 2019 at 3:46 AM zhangyi (F) <[email protected]> wrote:
>
>> {
>> - ktime_t until = ts ? timespec64_to_ktime(*ts) : KTIME_MAX;
>> - struct kioctx *ioctx = lookup_ioctx(ctx_id);
>> + ktime_t until = KTIME_MAX;
>> + struct kioctx *ioctx = NULL;
>> long ret = -EINVAL;
>>
>> + if (ts) {
>> + if (!timespec64_valid(ts))
>> + return ret;
>> + until = timespec64_to_ktime(*ts);
>> + }
>
> The man page should probably get updated as well to reflect that this
> will now return -EINVAL for a negative timeout or malformed
> nanoseconds.
>

Thanks for your suggestion, I will add a patch to update the man page.

Thanks,
Yi.