On 11/11/2015 16:28, Stephen Hemminger wrote:
> Also Linux has more robust mechanism in the watchdog timers, why not use that?
Keepalive was made with tight intra-process reporting intervals of
5-10ms in mind, whereas Linux watchdog appears to be an inter-process
system that operates with granu
2015-11-05 11:32, Remy Horton:
> --- /dev/null
> +++ b/lib/librte_eal/common/include/rte_keepalive.h
Please add it in doxygen configuration and index.
> --- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
> +++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
> @@ -133,5 +133,9 @@ DPDK_2.2 {
>
On 10/11/2015 14:02, Thomas Monjalon wrote:
> Why ptr_timer is unused?
Use as rte_timer callback requires the parameter be present, but
responsibility for setting this up is delegated to the application.
..Remy
In general, try not to introduce new thinks and avoid extra code.
Also Linux has more robust mechanism in the watchdog timers, why not use that?
Could you use RTE_PER_LCORE some how?
> +#ifdef KEEPALIVE_DEBUG_MSGS
Any #ifdef must have a config option to enable it.
> +static void
> +print_tra
Hi,
2015-11-05 11:32, Remy Horton:
> +/**
> + * @param *ptr_timer Triggering timer (unused)
> + * @param *ptr_data Data pointer (keepalive structure)
> + */
> +void rte_keepalive_dispatch_pings(void *ptr_timer, void *ptr_data);
There is no description for this function.
Why ptr_timer is unused?
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Remy Horton
> Sent: Thursday, November 5, 2015 11:33 AM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH v4 1/3] rte: add keep alive functionality
>
> Adds functions for detecting and reporting the live-ness of LCor
Adds functions for detecting and reporting the live-ness of LCores,
the primary requirement of which is minimal overheads for the
core(s) being checked. Core failures are notified via an application
defined callback.
Signed-off-by: Remy Horton
---
lib/librte_eal/bsdapp/eal/Makefile
7 matches
Mail list logo
