2013-05-25 12:04:14

by Manfred Spraul

Subject: Re: [PATCH 06/10] ipc: Don't allocate a copy larger than max

Hi Peter,

You wrote:
> When MSG_COPY is set, a duplicate message must be allocated for
> the copy before locking the queue. However, the copy could
> not be larger than was sent which is limited to msg_ctlmax.
>
> Signed-off-by: Peter Hurley <[email protected]>
> ---
> ipc/msg.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/ipc/msg.c b/ipc/msg.c
> index 950572f..31cd1bf 100644
> --- a/ipc/msg.c
> +++ b/ipc/msg.c
> @@ -820,15 +820,17 @@ long do_msgrcv(int msqid, void __user *buf, size_t bufsz, long msgtyp,
> struct msg_msg *copy = NULL;
> unsigned long copy_number = 0;
>
> + ns = current->nsproxy->ipc_ns;
> +
> if (msqid < 0 || (long) bufsz < 0)
> return -EINVAL;
> if (msgflg & MSG_COPY) {
> - copy = prepare_copy(buf, bufsz, msgflg, &msgtyp, &copy_number);
> + copy = prepare_copy(buf, min_t(size_t, bufsz, ns->msg_ctlmax),
> + msgflg, &msgtyp, &copy_number);

What about:
- increase msg_ctlmax
- send message
- reduce msg_ctlmax

The side effects of the patch are odd:
- without MSG_COPY, a message can be read regardsless of the size.
The user could check for E2BIG and increase the buffer size until
msgrcv succeeds.
- with MSG_COPY, something else would happen.
As far as I can see, it would oops: msg_ctlmax bytes are allocated,
then the E2BIG test is against bufsz, and copy_msg() doesn't check
the size of the target buffer.

I.e.: I would propose to revert the patch.

--
Manfred

2013-05-26 10:25:09

by Peter Hurley

Subject: Re: [PATCH 06/10] ipc: Don't allocate a copy larger than max

On 05/25/2013 08:04 AM, Manfred Spraul wrote:
> Hi Peter,
>
> You wrote:
>> When MSG_COPY is set, a duplicate message must be allocated for
>> the copy before locking the queue. However, the copy could
>> not be larger than was sent which is limited to msg_ctlmax.
>>
>> Signed-off-by: Peter Hurley <[email protected]>
>> ---
>> ipc/msg.c | 6 ++++--
>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/ipc/msg.c b/ipc/msg.c
>> index 950572f..31cd1bf 100644
>> --- a/ipc/msg.c
>> +++ b/ipc/msg.c
>> @@ -820,15 +820,17 @@ long do_msgrcv(int msqid, void __user *buf, size_t bufsz, long msgtyp,
>> struct msg_msg *copy = NULL;
>> unsigned long copy_number = 0;
>> + ns = current->nsproxy->ipc_ns;
>> +
>> if (msqid < 0 || (long) bufsz < 0)
>> return -EINVAL;
>> if (msgflg & MSG_COPY) {
>> - copy = prepare_copy(buf, bufsz, msgflg, &msgtyp, &copy_number);
>> + copy = prepare_copy(buf, min_t(size_t, bufsz, ns->msg_ctlmax),
>> + msgflg, &msgtyp, &copy_number);
>
> What about:
> - increase msg_ctlmax
> - send message
> - reduce msg_ctlmax
>
> The side effects of the patch are odd:
> - without MSG_COPY, a message can be read regardsless of the size.
> The user could check for E2BIG and increase the buffer size until
> msgrcv succeeds.

The patch does not change the behavior of non-MSG_COPY msg receive.

> - with MSG_COPY, something else would happen.
> As far as I can see, it would oops: msg_ctlmax bytes are allocated,
> then the E2BIG test is against bufsz, and copy_msg() doesn't check
> the size of the target buffer.

I assume you are using 3.9

Current mainline returns EINVAL.

Regards,
Peter Hurley