On 2023/03/07 13:46, lijiang wrote:
>     > +     char str[INET6_ADDRSTRLEN + 1] = {0};
>     > +     char buffer[INET6_ADDRSTRLEN + 4] = {0};
>
>     What are the +1 and +4 for?
>
>
> I noticed that the size of INET6_ADDRSTRLEN is 48 in kernel code as below:
>
> --- a/include/linux/inet.h
> +++ b/include/linux/inet.h
> +/*
> + * These mimic similar macros defined in user-space for inet_ntop(3).
> + * See /usr/include/netinet/in.h .
> + */
> +#define INET_ADDRSTRLEN                (16)
> +#define INET6_ADDRSTRLEN       (48)
>
>
> And, the size of INET6_ADDRSTRLEN is 46 in my machine as below:
>
> # cat /usr/include/netinet/in.h |grep INET6_ADDRSTRLEN
> #define INET6_ADDRSTRLEN 46
 
I'm not sure why the kernel one is 48 (for a multiple of 8?), but
the INET6_ADDRSTRLEN in glibc header should mean the longest result
of inet_ntop() in glibc, and the latest glibc also has the same 46.
So I don't think we need to consider the kernel one.
>
>
> The rfc2460 said that the IPv6 increases the IP address size from 32 bits to 128
bits.
>
> crash> struct in6_addr
> struct in6_addr {
>     union {
>         __u8 u6_addr8[16];
>         __be16 u6_addr16[8];
>         __be32 u6_addr32[4];
>     } in6_u;
> }
> SIZE: 16
>
> Given that, maybe they should be defined like this?
>
> +     char str[INET6_ADDRSTRLEN + 2] = {0};
> +     char buffer[INET6_ADDRSTRLEN + 2 + 2] = {0};
>
>  Not sure what's the best way for this case.
>
>     Looking at the example in the man page of inet_pton(3), INET6_ADDRSTRLEN
>     seems enough for the str and contains a null char.  The buffer can have
>     a comma and a space (", ") so +2 is enough?  i.e.
>
>        char str[INET6_ADDRSTRLEN] = {0};
>        char buffer[INET6_ADDRSTRLEN + 2] = {0};
>
>
>     > +     uint len = 0;
>     > +
>     > +     buf = *bufp;
>     > +     pos = strlen(buf);
>
>     Ah nicely done :)
>
>     > +
>     > +     readmem(devaddr + OFFSET(net_device_ip6_ptr), KVADDR,
>     > +             &ip6_ptr, sizeof(ulong), "ip6_ptr",
FAULT_ON_ERROR);
>     > +
>     > +     if (!ip6_ptr)
>     > +             return;
>     > +
>     > +     if (VALID_MEMBER(inet6_ifaddr_if_list)) {
>     > +             struct list_data list_data, *ld;
>     > +             ulong cnt = 0, i;
>     > +
>     > +             ld = &list_data;
>     > +             BZERO(ld, sizeof(struct list_data));
>     > +             ld->flags |= LIST_ALLOCATE;
>     > +             ld->start = ip6_ptr + OFFSET(inet6_dev_addr_list);
>     > +             cnt = do_list(ld);
>     > +
>     > +             for (i = 1; i < cnt; i++) {
>     > +
>     > +                     addr = ld->list_ptr[i] +
OFFSET(inet6_ifaddr_addr);
>
>     > +                     addr -= OFFSET(inet6_ifaddr_if_list);
>
>
> The above code is easy to understand, because it actually imitates the
container_of().
>
> But if you would like to have the same style as show_net_devices_v2()
and show_net_devices_v3(), that's also fine to me.
 
Yes, please.  I've not seen the above style (do_list and minus OFFSET)
in crash before.  Although we may go with the above, I think generally
a new style tends to cause an unexpected result.  so I'd like to use
the common style if possible.
Thanks,
Kazu