It would be nice, if you could send the test cases you used to test these
fixes or
their outputs before and after the fix, showing the problem and then the
problem
going away.
Thanks,
Balbir Singh.
Mingming cao wrote:
> Hello linus & Alan,
>
> This patch fixes a bug in msgsnd. If a message is sent via
> pipelined_send(), the q_lrpid field of the associated message queue
> (last pid receive) is not updated correctly. The existing code
> (apparently by mistake) updates the lspid field instead. The second
> problem is that, for the pipelinesend case, q_lstime(last msgsnd time)
> is set AFTER q_lrtime(last msgrcv time), yielding a negative time delta.
>
> This patch is against kernel 2.4.9 and has been tested. Please
> apply.
>
> Please CC your feedback to me also, thanks.
>
> --
> Mingming Cao
> IBM Linux Technology Center
>
> diff -urN -X dontdiff linux/ipc/msg.c linux-tk/ipc/msg.c
> --- linux/ipc/msg.c Tue Sep 11 16:21:42 2001
> +++ linux-tk/ipc/msg.c Tue Sep 11 16:21:31 2001
> @@ -613,7 +613,7 @@
> wake_up_process(msr->r_tsk);
> } else {
> msr->r_msg = msg;
> - msq->q_lspid = msr->r_tsk->pid;
> + msq->q_lrpid = msr->r_tsk->pid;
> msq->q_rtime = CURRENT_TIME;
> wake_up_process(msr->r_tsk);
> return 1;
> @@ -683,6 +683,9 @@
> goto retry;
> }
>
> + msq->q_lspid = current->pid;
> + msq->q_stime = CURRENT_TIME;
> +
> if(!pipelined_send(msq,msg)) {
> /* noone is waiting for this message, enqueue it */
> list_add_tail(&msg->m_list,&msq->q_messages);
> @@ -694,8 +697,6 @@
>
> err = 0;
> msg = NULL;
> - msq->q_lspid = current->pid;
> - msq->q_stime = CURRENT_TIME;
>
> out_unlock_free:
> msg_unlock(msqid);
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Sep 15 2001 - 21:00:45 EST