linux-kernel - Re: [PATCH v9 5/6] Coresight: Add Coresight TMC Control Unit driver

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <44bd0d5e-a232-49c3-ba2f-e63e2f2c99be@linaro.org>
Date: Tue, 28 Jan 2025 11:55:51 +0000
From: James Clark <james.clark@...aro.org>
To: Jie Gan <quic_jiegan@...cinc.com>
Cc: Tingwei Zhang <quic_tingweiz@...cinc.com>,
 Jinlong Mao <quic_jinlmao@...cinc.com>, coresight@...ts.linaro.org,
 linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
 devicetree@...r.kernel.org, linux-arm-msm@...r.kernel.org,
 linux-stm32@...md-mailman.stormreply.com,
 Suzuki K Poulose <suzuki.poulose@....com>, Mike Leach
 <mike.leach@...aro.org>,
 Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
 Maxime Coquelin <mcoquelin.stm32@...il.com>,
 Alexandre Torgue <alexandre.torgue@...s.st.com>,
 Rob Herring <robh@...nel.org>, Krzysztof Kozlowski <krzk+dt@...nel.org>,
 Conor Dooley <conor+dt@...nel.org>, Bjorn Andersson <andersson@...nel.org>,
 Konrad Dybcio <konradybcio@...nel.org>
Subject: Re: [PATCH v9 5/6] Coresight: Add Coresight TMC Control Unit driver



On 24/01/2025 7:25 am, Jie Gan wrote:
> The Coresight TMC Control Unit hosts miscellaneous configuration registers
> which control various features related to TMC ETR sink.
> 
> Based on the trace ID, which is programmed in the related CTCU ATID
> register of a specific ETR, trace data with that trace ID gets into
> the ETR buffer, while other trace data gets dropped.
> 
> Enabling source device sets one bit of the ATID register based on
> source device's trace ID.
> Disabling source device resets the bit according to the source
> device's trace ID.
> 
> Signed-off-by: Jie Gan <quic_jiegan@...cinc.com>
> ---
>   drivers/hwtracing/coresight/Kconfig          |  12 +
>   drivers/hwtracing/coresight/Makefile         |   1 +
>   drivers/hwtracing/coresight/coresight-ctcu.c | 276 +++++++++++++++++++
>   drivers/hwtracing/coresight/coresight-ctcu.h |  30 ++
>   include/linux/coresight.h                    |   3 +-
>   5 files changed, 321 insertions(+), 1 deletion(-)
>   create mode 100644 drivers/hwtracing/coresight/coresight-ctcu.c
>   create mode 100644 drivers/hwtracing/coresight/coresight-ctcu.h
 >

[...]

> +/*
> + * ctcu_set_etr_traceid: Retrieve the ATID offset and trace ID.
> + *
> + * Returns 0 indicates success. None-zero result means failure.
> + */
> +static int ctcu_set_etr_traceid(struct coresight_device *csdev, struct coresight_path *cs_path,
> +				bool enable)
> +{
> +	struct coresight_device *sink = coresight_get_sink(cs_path->path);
> +	struct ctcu_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);
> +	u8 trace_id = cs_path->trace_id;
> +	int port_num;
> +
> +	if ((sink == NULL) || !IS_VALID_CS_TRACE_ID(trace_id) || IS_ERR_OR_NULL(drvdata)) {
> +		dev_err(&csdev->dev, "Invalid parameters\n");
> +		return -EINVAL;
> +	}
> +
> +	port_num = ctcu_get_active_port(sink, csdev);
> +	if (port_num < 0)
> +		return -EINVAL;
> +
> +	/*
> +	 * Skip the disable session if more than one TPDM device that
> +	 * connected to the same TPDA device has been enabled.
> +	 */
> +	if (enable)
> +		atomic_inc(&drvdata->traceid_refcnt[port_num][trace_id]);
> +	else {
> +		if (atomic_dec_return(&drvdata->traceid_refcnt[port_num][trace_id]) > 0) {
> +			dev_dbg(&csdev->dev, "Skip the disable session\n");
> +			return 0;
> +		}
> +	}
> +
> +	dev_dbg(&csdev->dev, "traceid is %d\n", cs_path->trace_id);
> +
> +	return __ctcu_set_etr_traceid(csdev, trace_id, port_num, enable);

Hi Jie,

Using atomic_dec_return() here doesn't prevent __ctcu_set_etr_traceid() 
from running concurrent enable and disables. Once you pass the 
atomic_dec_return() a second call to enable it will mess it up.

I think you need a spinlock around the whole thing and then the 
refcounts don't need to be atomics.