Hi,
I came across a bug with a panic because of a 'blocking' call to
sock_alloc_send_skb( ) called from ndisc code [2.4].
I had a look at the patch from:
http://www.linux-ipv6.org/ml/usagi-users/msg02309.html
where all these calls are changed from 'blocking' to 'non-blocking'.
I was just curious to know if instead of doing this, could we have a
check in sock_alloc_send_skb( ) by a call to in_interrupt( ) and send an
error message if we are in interrupt or bottom half context? Because,
this is a generic function and would prevent the panics if any of the
upper layers is trying to block in interrupt or bhp context. [and thus
preventing code changes from blocking to non-blocking at all relevant
locations of code]
The code change could be in net/core/sock.c:
err = -EAGAIN;
if (!timeo || in_interrupt( )) <<<<<<<
goto failure;
if (signal_pending(current)
goto interrupted;
Please correct me if i am wrong.
Prashanth
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
