c96356a9ba
The old code did not check the return value of skb_clone(). The extra skb_clone() is not needed at all, if using skb_realloc_headroom() instead, which gives us a private copy with enough headroom as well. We need to requeue the original skb if the call failed, because we cannot inform upper layers about the data loss. Restructure the code to minimise rollback effort if it happens. This fix kernel bug #86091 Thanks to Insu Yun <wuninsu@gmail.com> to remind me on this issue. Signed-off-by: Karsten Keil <keil@b1-systems.de> Signed-off-by: David S. Miller <davem@davemloft.net> |
||
|---|---|---|
| .. | ||
| act2000 | ||
| capi | ||
| divert | ||
| gigaset | ||
| hardware | ||
| hisax | ||
| hysdn | ||
| i4l | ||
| icn | ||
| isdnloop | ||
| mISDN | ||
| pcbit | ||
| sc | ||
| Kconfig | ||
| Makefile | ||