Philipp Hortmann <[email protected]> writes:
> I do like this example:
> https://lwn.net/Articles/745261/
>
> ...
> pr_err("%llx%llx\n", (u64) (val >> 64), (u64) val);
> pr_err("%llx%llx\n", (u64) (v >> 64), (u64) v);
> ...
>
This will produce the wrong output unless at least one of bit 60 - 63 is
set and have a confusing leading zero unless one of the upper 64 bits is
set.
Running this test:
#include <stdio.h>
#include <stdlib.h>
int main(int argc, char **argv)
{
__uint128_t foo;
if (argc < 3)
return -1;
foo = (__uint128_t)strtoull(argv[1], NULL, 0) << 64 | strtoull(argv[2],
NULL, 0);
fprintf(stderr, "foo=0x%llx%llx\n", (__uint64_t) (foo >> 64),
(__uint64_t) foo);
fprintf(stderr, "foo=0x%016llx%016llx\n", (__uint64_t) (foo >> 64),
(__uint64_t) foo);
return 0;
}
Produces:
bjorn@miraculix:~$ /tmp/x 1 0
foo=0x10
foo=0x00000000000000010000000000000000
bjorn@miraculix:~$ /tmp/x 1 2
foo=0x12
foo=0x00000000000000010000000000000002
bjorn@miraculix:~$ /tmp/x 0 2
foo=0x02
foo=0x00000000000000000000000000000002
Bjørn
_______________________________________________
Kernelnewbies mailing list
[email protected]
https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
