2022-04-21 17:16:09

by David Laight

[permalink] [raw]
Subject: RE: [PATCH v2 bpf 07/11] samples/bpf: fix uin64_t format literals

From: Alexander Lobakin
> Sent: 21 April 2022 01:40
>
> There's a couple places where uin64_t is being passed as an %lu
> format argument. That type is defined as unsigned long on 64-bit
> systems and as unsigned long long on 32-bit, so neither %lu nor
> %llu are not universal.
> One of the options is %PRIu64, but since it's always 8-byte long,
> just cast it to the _proper_ __u64 and print as %llu.

Is __u64 guaranteed to be 'unsigned long long' ? No reason why it should be.
I think you need to cast to (unsigned long long).

David

> Fixes: 51570a5ab2b7 ("A Sample of using socket cookie and uid for traffic monitoring")
> Fixes: 00f660eaf378 ("Sample program using SO_COOKIE")
> Signed-off-by: Alexander Lobakin <[email protected]>
> ---
> samples/bpf/cookie_uid_helper_example.c | 12 ++++++------
> samples/bpf/lwt_len_hist_user.c | 7 ++++---
> 2 files changed, 10 insertions(+), 9 deletions(-)
>
> diff --git a/samples/bpf/cookie_uid_helper_example.c b/samples/bpf/cookie_uid_helper_example.c
> index f0df3dda4b1f..269fac58fd5c 100644
> --- a/samples/bpf/cookie_uid_helper_example.c
> +++ b/samples/bpf/cookie_uid_helper_example.c
> @@ -207,9 +207,9 @@ static void print_table(void)
> error(1, errno, "fail to get entry value of Key: %u\n",
> curN);
> } else {
> - printf("cookie: %u, uid: 0x%x, Packet Count: %lu,"
> - " Bytes Count: %lu\n", curN, curEntry.uid,
> - curEntry.packets, curEntry.bytes);
> + printf("cookie: %u, uid: 0x%x, Packet Count: %llu, Bytes Count: %llu\n",
> + curN, curEntry.uid, (__u64)curEntry.packets,
> + (__u64)curEntry.bytes);
> }
> }
> }
> @@ -265,9 +265,9 @@ static void udp_client(void)
> if (res < 0)
> error(1, errno, "lookup sk stat failed, cookie: %lu\n",
> cookie);
> - printf("cookie: %lu, uid: 0x%x, Packet Count: %lu,"
> - " Bytes Count: %lu\n\n", cookie, dataEntry.uid,
> - dataEntry.packets, dataEntry.bytes);
> + printf("cookie: %llu, uid: 0x%x, Packet Count: %llu, Bytes Count: %llu\n\n",
> + (__u64)cookie, dataEntry.uid, (__u64)dataEntry.packets,
> + (__u64)dataEntry.bytes);
> }
> close(s_send);
> close(s_rcv);
> diff --git a/samples/bpf/lwt_len_hist_user.c b/samples/bpf/lwt_len_hist_user.c
> index 430a4b7e353e..c682faa75a2b 100644
> --- a/samples/bpf/lwt_len_hist_user.c
> +++ b/samples/bpf/lwt_len_hist_user.c
> @@ -44,7 +44,8 @@ int main(int argc, char **argv)
>
> while (bpf_map_get_next_key(map_fd, &key, &next_key) == 0) {
> if (next_key >= MAX_INDEX) {
> - fprintf(stderr, "Key %lu out of bounds\n", next_key);
> + fprintf(stderr, "Key %llu out of bounds\n",
> + (__u64)next_key);
> continue;
> }
>
> @@ -66,8 +67,8 @@ int main(int argc, char **argv)
>
> for (i = 1; i <= max_key + 1; i++) {
> stars(starstr, data[i - 1], max_value, MAX_STARS);
> - printf("%8ld -> %-8ld : %-8ld |%-*s|\n",
> - (1l << i) >> 1, (1l << i) - 1, data[i - 1],
> + printf("%8ld -> %-8ld : %-8lld |%-*s|\n",
> + (1l << i) >> 1, (1l << i) - 1, (__u64)data[i - 1],
> MAX_STARS, starstr);
> }
>
> --
> 2.36.0
>

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)


2022-04-22 19:10:27

by Alexander Lobakin

[permalink] [raw]
Subject: Re: [PATCH v2 bpf 07/11] samples/bpf: fix uin64_t format literals

From: David Laight <[email protected]>
Date: Thu, 21 Apr 2022 07:46:05 +0000

> From: Alexander Lobakin
> > Sent: 21 April 2022 01:40
> >
> > There's a couple places where uin64_t is being passed as an %lu
> > format argument. That type is defined as unsigned long on 64-bit
> > systems and as unsigned long long on 32-bit, so neither %lu nor
> > %llu are not universal.
> > One of the options is %PRIu64, but since it's always 8-byte long,
> > just cast it to the _proper_ __u64 and print as %llu.
>
> Is __u64 guaranteed to be 'unsigned long long' ? No reason why it should be.
> I think you need to cast to (unsigned long long).

__u64 can be unsigned long only if an architecture uses int-l64.h
instead of int-ll64.h. This is currently possible for Alpha and
PPC64 when __SANE_USERSPACE_TYPES__ is not defined -- I guess you
know what that flag does.
I messed up a bit and didn't notice that samples/bpf/Makefile
defines this flag only for MIPS. IMO it should be defined in here
unconditionally, but I guess it's out of bpf-fixes scope, so I'll
go with unsigned long long in v3 (got to resend with no PGP crap
anyway lol).

>
> David

--- 8< ---

> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
> Registration No: 1397386 (Wales)

Thanks,
Al