On 21/02/17 12:34, Mike Leach wrote:
Hi Sudeep
-----Original Message----- From: CoreSight [mailto:coresight-bounces@lists.linaro.org] On Behalf Of Sudeep Holla Sent: 21 February 2017 11:03 To: Mike Leach; linux-arm-kernel@lists.infradead.org Cc: devicetree@vger.kernel.org; coresight@lists.linaro.org; robh+dt@kernel.org; Sudeep Holla Subject: Re: [PATCH] arm64: dts: juno: update definition for programmable replicator.
Hi Mike,
On 17/02/17 19:13, Mike Leach wrote:
Juno platforms have a programmable replicator splitting the trace output to TPIU and ETR. Currently this is not being programmed as it is being treated as a none-programmable replicator - which is the default operational mode for these devices. The TPIU in the system is enabled by default, and this combination is causing back-pressure in the trace system resulting in overflows at the source.
Replaces the existing definition with one that defines the programmable replicator, using the "qcom,coresight-replicator1x" driver that provides the correct functionality for CoreSight programmable
replicators.
I assume this is just enhancement and not a fix.
I guess it depends on your point of view - with this update the trace overflows I was seeing disappear, as the trace path to TPIU is blocked. So it affects the quality of collected trace using the ETR rather than a binary didn't work / works now change.
OK, could you add the warning overflow messages if you get, so that it's very clear from the change log. Then I can see if it can be pushed as fix.