Re: CVS commit: src/sys

[Roy Marples <roy%marples.name@localhost>]

Hi Christos

On 14/02/2021 20:58, Christos Zoulas wrote:
> Module Name:	src
> Committed By:	christos
> Date:		Sun Feb 14 20:58:35 UTC 2021
>
> Modified Files:
> 	src/sys/net: if_arp.h if_bridge.c
> 	src/sys/netinet: icmp_private.h if_arp.c igmp_var.h in_l2tp.c ip_flow.c
> 	    ip_input.c ip_private.h tcp_input.c tcp_private.h udp_private.h
> 	    udp_usrreq.c
> 	src/sys/netinet6: icmp6.c in6_l2tp.c ip6_flow.c ip6_input.c
> 	    ip6_private.h udp6_usrreq.c
> 	src/sys/sys: mbuf.h param.h
>
> Log Message:
> - centralize header align and pullup into a single inline function
> - use a single macro to align pointers and expose the alignment, instead
>    of hard-coding 3 in 1/2 the macros.
> - fix an issue in the ipv6 lt2p where it was aligning for ipv4 and pulling
>    for ipv6.

-#ifdef __NO_STRICT_ALIGNMENT
-#define	IP_HDR_ALIGNED_P(ip)	1
-#else
-#define	IP_HDR_ALIGNED_P(ip)	((((vaddr_t) (ip)) & 3) == 0)
-#endif
+#define	IP_HDR_ALIGNMENT	3
 #endif /* _KERNEL */

While this is a like for like change, I feel that the meaning of 
IP_HDR_ALIGNMENT is no longer clear as 3 without context makes no sense at all.
We know it should be aligned to 4 bytes.

I suggest we change POINTER_ALIGNED_P to accept the alignment value we want 
rather than the bitwise test we supply, like so:

#define	POINTER_ALIGNED_P(p, a)		(((uintptr_t)(p) & ((a) - 1)) == 0)

Roy