latencystats: fix missing f suffix on literal float
Checks
Commit Message
Add missing f suffix to floating point literal to avoid warning about
truncation from double to float.
Fixes: 5cd3cac9ed22 ("latency: added new library for latency stats")
Cc: reshma.pattan@intel.com
Cc: stable@dpdk.org
Signed-off-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
---
lib/latencystats/rte_latencystats.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Comments
On Mon, 15 Apr 2024 14:32:26 -0700
Tyler Retzlaff <roretzla@linux.microsoft.com> wrote:
> Add missing f suffix to floating point literal to avoid warning about
> truncation from double to float.
>
> Fixes: 5cd3cac9ed22 ("latency: added new library for latency stats")
> Cc: reshma.pattan@intel.com
> Cc: stable@dpdk.org
>
> Signed-off-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
> ---
> lib/latencystats/rte_latencystats.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/lib/latencystats/rte_latencystats.c b/lib/latencystats/rte_latencystats.c
> index 4ea9b0d..cae4b50 100644
> --- a/lib/latencystats/rte_latencystats.c
> +++ b/lib/latencystats/rte_latencystats.c
> @@ -166,7 +166,7 @@ struct latency_stats_nameoff {
> * a constant smoothing factor between 0 and 1. The value
> * is used below for measuring average latency.
> */
> - const float alpha = 0.2;
> + const float alpha = 0.2f;
>
> now = rte_rdtsc();
> for (i = 0; i < nb_pkts; i++) {
The use of floating point in fast path here is unnecessary.
Even without doing the full per-core values, this could just change
to an alpha of .25 and use fast fixed point math.
On Mon, Apr 15, 2024 at 03:17:38PM -0700, Stephen Hemminger wrote:
> On Mon, 15 Apr 2024 14:32:26 -0700
> Tyler Retzlaff <roretzla@linux.microsoft.com> wrote:
>
> > Add missing f suffix to floating point literal to avoid warning about
> > truncation from double to float.
> >
> > Fixes: 5cd3cac9ed22 ("latency: added new library for latency stats")
> > Cc: reshma.pattan@intel.com
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
> > ---
> > lib/latencystats/rte_latencystats.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/lib/latencystats/rte_latencystats.c b/lib/latencystats/rte_latencystats.c
> > index 4ea9b0d..cae4b50 100644
> > --- a/lib/latencystats/rte_latencystats.c
> > +++ b/lib/latencystats/rte_latencystats.c
> > @@ -166,7 +166,7 @@ struct latency_stats_nameoff {
> > * a constant smoothing factor between 0 and 1. The value
> > * is used below for measuring average latency.
> > */
> > - const float alpha = 0.2;
> > + const float alpha = 0.2f;
> >
> > now = rte_rdtsc();
> > for (i = 0; i < nb_pkts; i++) {
>
> The use of floating point in fast path here is unnecessary.
> Even without doing the full per-core values, this could just change
> to an alpha of .25 and use fast fixed point math.
agree, though my thoughts are make this code correct optimization is for
the maintainer. if they submit an alternate series i can withdraw this
patch. otherwise i think we should take it as-is.
@@ -166,7 +166,7 @@ struct latency_stats_nameoff {
* a constant smoothing factor between 0 and 1. The value
* is used below for measuring average latency.
*/
- const float alpha = 0.2;
+ const float alpha = 0.2f;
now = rte_rdtsc();
for (i = 0; i < nb_pkts; i++) {