Re: [Qemu-devel] [PATCH 1/2] tests: Use real size for iov tests

[Juan Quintela <quintela@redhat.com>]

Thomas Huth <thuth@redhat.com> wrote:
> On 23.08.2017 10:39, Juan Quintela wrote:
>> We were using -1 instead of the real size because the functions check
>> what is bigger, size in bytes or the size of the iov.  Recent gcc's
>> barf at this.
>> 
>> Signed-off-by: Juan Quintela <quintela@redhat.com>
>> ---
>>  tests/test-iov.c | 8 ++++----
>>  1 file changed, 4 insertions(+), 4 deletions(-)
>
> While you're at it, could you maybe also adjust the comments in
> include/qemu/iov.h ? It currently says:
>
>  * It is okay to use very large value for `bytes' since we're
>  * limited by the size of the iovec anyway, provided that the
>  * buffer pointed to by buf has enough space.  One possible
>  * such "large" value is -1 (sinice size_t is unsigned),
>  * so specifying `-1' as `bytes' means 'up to the end of iovec'.

This is for the _full() versions, and still work.  the same for
iov_memset().  

> ... and apparently -1 is not working anymore as expected. Maybe SIZE_MAX
> from stdint.h is a better choice?


static inline size_t
iov_from_buf(const struct iovec *iov, unsigned int iov_cnt,
             size_t offset, const void *buf, size_t bytes)
{
    if (__builtin_constant_p(bytes) && iov_cnt &&
        offset <= iov[0].iov_len && bytes <= iov[0].iov_len - offset) {
        memcpy(iov[0].iov_base + offset, buf, bytes);
        return bytes;
    } else {
        return iov_from_buf_full(iov, iov_cnt, offset, buf, bytes);
    }
}

This optimization don't work very well if we used bytes = -1,
furthermore, we return the wrong value.

And no, I don't understand how the (bytes <= iov[0].iov_len - offset)
test pass when bytes == -1.

Later, Juan.