Date: Sat, 28 Nov 2020 08:52:58 -0700 From: Ian Lepore <ian@freebsd.org> To: "freebsd-arch@FreeBSD.org" <freebsd-arch@FreeBSD.org> Subject: sbintime_t in userland Message-ID: <567389bbf8882800c86a9cd8a502051af48cc4ec.camel@freebsd.org>
next in thread | raw e-mail | index | archive | help
I notice everything related to sbintime_t in sys/time.h is in the public part of the header file (not protected by #ifdef _KERNEL). As near as I can tell, no existing userland software uses this type, and no userland<->kernel APIs involve sbintime_t. I have a situation where I need to timestamp events relative to each other in the kernel and report those timestamps to userland. sbintime_t would be an ideal compact format for the data, and sbinuptime() would be a fine way to obtain the timestamps in the kernel. Before breaking new ground like this I thought I'd solicit opinions: is it a bad idea to export sbintime_t values from the kernel like this? If it's considered okay, a second question would be: is it okay to document that they are on the 'uptime timescale' (meaning userland could translate them to UTC timestamps by determining the offset between CLOCK_REALTIME and CLOCK_MONOTONIC)? -- Ian
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?567389bbf8882800c86a9cd8a502051af48cc4ec.camel>