Re: [PATCH v3] tracing: Record trace_clock and recover when reboot

On Wed, 14 May 2025 10:14:31 +0900
Masami Hiramatsu (Google) [off-list ref] wrote:

quoted

In case the tscratch->clock_id gets corrupted, we better make sure it
doesn't overflow:

		if (tscratch->clock_id >= ARRAY_SIZE(trace_clocks) ||
		    tracing_set_clock(tr, trace_clocks[tscratch->clock_id].name) < 0) {  

Good catch! BTW, don't we have any check about scratch area?
(e.g. adding a checksum)

We could, but it would require updating every time we update the
scratch area.

Currently, the scratch area is only considered valid if the ring buffer
is proven to be valid. But if this becomes a problem, we probably
should add a checksum or something.

-- Steve