lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CANLsYkz-3i4wnZ2s1ns0zjkrtJ_5UgYOFf31QPpmPuANHqz_tQ@mail.gmail.com>
Date:	Thu, 21 Apr 2016 16:00:22 -0600
From:	Mathieu Poirier <mathieu.poirier@...aro.org>
To:	Suzuki K Poulose <Suzuki.Poulose@....com>
Cc:	"linux-arm-kernel@...ts.infradead.org" 
	<linux-arm-kernel@...ts.infradead.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH V2 14/15] coresight: tmc: implementing TMC-ETR AUX space API

On 21 April 2016 at 10:10, Suzuki K Poulose <Suzuki.Poulose@....com> wrote:
> On 12/04/16 18:54, Mathieu Poirier wrote:
>>
>> This patch implement the AUX area interfaces required to
>> use the TMC (configured as an ETR) from the Perf sub-system.
>>
>> The heuristic is heavily borrowed from the ETB10 and TMC-ETF
>> implementation.
>>
>> Signed-off-by: Mathieu Poirier <mathieu.poirier@...aro.org>
>
>
>> +static void tmc_update_etr_buffer(struct coresight_device *csdev,
>> +                                 struct perf_output_handle *handle,
>> +                                 void *sink_config)
>> +{
>> +       struct cs_etr_buffers *buf = sink_config;
>> +
>> +       /*
>> +        * An ETR configured to work in contiguous memory mode works the
>> same
>> +        * was as an ETB or ETF.
>> +        */
>> +       tmc_update_etf_buffer(csdev, handle, &buf->tmc);
>
>
> Really ? I thought the ETR stores the data to the allocated System RAM and
> can
> be read directly from the memory than using the RRD ?

Using an ETR in contiguous mode is inefficient to start with.
Regardless of the approach taken trace data has to be copied to perf's
non-contiguous ring buffer pages.  It would be more efficient to copy
bigger chunks of data to the ring buffer (because we can), but I
rather spend my time enabling scatter-gather mode than optimizing a
sub-optimal mode of operation.

This currently work and enables people to use the IP block while
scatter-gather mode is in flight.  I should probably make that clear
in the commit log.

Thanks,
Mathieu

>
> Suzuki
>
>
>

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ