I wrote :
> I just noticed something at the end of process_input_packet :
> In the normal case, skb is given to the next stage and ap->rpkt is reset,
> but in the error case, skb is kept, ap->rpkt is not reset, so we keep
> the skb with skb->data aligned for one message and we put another one
> int
Paul Mackerras wrote :
> Philippe De Muyter writes:
>
> > Actually, that's probably the case I had, but my fix gets the ip adresses
> > 4byte aligned in my case : I had verified the address of the saddr field,
> > and I needed to shift the buffer by 3, not 1, to get it 4byte aligned.
>
> Please o
Philippe De Muyter writes:
> Actually, that's probably the case I had, but my fix gets the ip adresses
> 4byte aligned in my case : I had verified the address of the saddr field,
> and I needed to shift the buffer by 3, not 1, to get it 4byte aligned.
Please outline the code flow that leads to th
Paul Mackerras wrote :
> Philippe De Muyter writes:
>
> > > This patch seems a bit strange and/or incomplete. Are we trying to
> > > get 2-byte alignment or 4-byte alignment of the payload? It seems
> >
> > Actually, we try to get a 4n+2 alignment for skb->data, to get the
> > ip-addresses
> >
Philippe De Muyter writes:
> > This patch seems a bit strange and/or incomplete. Are we trying to
> > get 2-byte alignment or 4-byte alignment of the payload? It seems
>
> Actually, we try to get a 4n+2 alignment for skb->data, to get the
> ip-addresses
> field 4bytes aligned.
> I think the on
Paul Mackerras wrote :
> Jeff Garzik writes:
>
> > From: "Philippe De Muyter" <[EMAIL PROTECTED]>
> >
> > Avoid ppp-generated kernel crashes on machines where unaligned accesses are
> > forbidden (ie: 68000-based CPUs)
>
> This patch seems a bit strange and/or incomplete. Are we trying to
> get
Paul Mackerras wrote:
> Can we assume that dev_alloc_skb will give us a 4-byte aligned
> skb->data? If we can then I suggest we change 3 to 1 in the skb_put
> and memmove above, and get rid of the if (since its condition will
> always be false).
The usual kmalloc alignment rules hold for skb->hea
Jeff Garzik writes:
> From: "Philippe De Muyter" <[EMAIL PROTECTED]>
>
> Avoid ppp-generated kernel crashes on machines where unaligned accesses are
> forbidden (ie: 68000-based CPUs)
This patch seems a bit strange and/or incomplete. Are we trying to
get 2-byte alignment or 4-byte alignment of
