Re: [PATCH 1/2] vfs: add tracepoints in inode_set_ctime_deleg

From: Steven Rostedt
Date: Tue Jul 22 2025 - 21:11:09 EST


On Tue, 22 Jul 2025 14:52:27 -0400
Jeff Layton <jlayton@xxxxxxxxxx> wrote:

> + TP_fast_assign(
> + __entry->dev = inode->i_sb->s_dev;
> + __entry->ino = inode->i_ino;
> + __entry->gen = inode->i_generation;
> + __entry->old_s = old->tv_sec;
> + __entry->req_s = req->tv_sec;
> + __entry->now_s = now ? now->tv_sec : 0;
> + __entry->old_ns = old->tv_nsec;
> + __entry->req_ns = req->tv_nsec;
> + __entry->now_ns = now ? now->tv_nsec : 0;
> + ),
> +
> + TP_printk("ino=%d:%d:%ld:%u old=%lld.%u req=%lld.%u now=%lld.%u",

Hmm, wouldn't you want the above to be:

TP_printk("ino=%d:%d:%ld:%u old=%lld.%09u req=%lld.%09u now=%lld.%09u",

Otherwise the nanosecond part is going to look confusing if it's less that 100,000,000.

-- Steve


> + MAJOR(__entry->dev), MINOR(__entry->dev), __entry->ino, __entry->gen,
> + __entry->old_s, __entry->old_ns,
> + __entry->req_s, __entry->req_ns,
> + __entry->now_s, __entry->now_ns
> + )