[PATCH v13 7/8] coresight: tmc: integrate byte-cntr's sysfs_ops with tmc sysfs file_ops
Jie Gan
jie.gan at oss.qualcomm.com
Fri Mar 6 02:21:50 PST 2026
On 3/6/2026 5:44 PM, Mike Leach wrote:
>
> Hi,
>> -----Original Message-----
>> From: Jie Gan <jie.gan at oss.qualcomm.com>
>> Sent: Monday, February 23, 2026 6:56 AM
>> To: Suzuki Poulose <Suzuki.Poulose at arm.com>; Mike Leach
>> <Mike.Leach at arm.com>; James Clark <james.clark at linaro.org>; Alexander
>> Shishkin <alexander.shishkin at linux.intel.com>; Rob Herring
>> <robh at kernel.org>; Krzysztof Kozlowski <krzk+dt at kernel.org>; Conor Dooley
>> <conor+dt at kernel.org>; Tingwei Zhang
>> <tingwei.zhang at oss.qualcomm.com>; Mao Jinlong
>> <jinlong.mao at oss.qualcomm.com>; Bjorn Andersson
>> <andersson at kernel.org>; Konrad Dybcio <konradybcio at kernel.org>
>> Cc: coresight at lists.linaro.org; linux-arm-kernel at lists.infradead.org; linux-
>> kernel at vger.kernel.org; linux-arm-msm at vger.kernel.org;
>> devicetree at vger.kernel.org; Jie Gan <jie.gan at oss.qualcomm.com>
>> Subject: [PATCH v13 7/8] coresight: tmc: integrate byte-cntr's sysfs_ops with
>> tmc sysfs file_ops
>>
>> Add code logic to invoke byte-cntr's tmc_sysfs_ops if the byte-cntr
>> is enabled.
>>
>> Signed-off-by: Jie Gan <jie.gan at oss.qualcomm.com>
>> ---
>> drivers/hwtracing/coresight/coresight-tmc-core.c | 53
>> +++++++++++++++++++++++-
>> 1 file changed, 52 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/hwtracing/coresight/coresight-tmc-core.c
>> b/drivers/hwtracing/coresight/coresight-tmc-core.c
>> index 32ca2ec994de..6486bdafdddc 100644
>> --- a/drivers/hwtracing/coresight/coresight-tmc-core.c
>> +++ b/drivers/hwtracing/coresight/coresight-tmc-core.c
>> @@ -31,6 +31,7 @@
>>
>> #include "coresight-priv.h"
>> #include "coresight-tmc.h"
>> +#include "coresight-ctcu.h"
>>
>> DEFINE_CORESIGHT_DEVLIST(etb_devs, "tmc_etb");
>> DEFINE_CORESIGHT_DEVLIST(etf_devs, "tmc_etf");
>> @@ -228,15 +229,47 @@ static int tmc_prepare_crashdata(struct
>> tmc_drvdata *drvdata)
>> return 0;
>> }
>>
>> +/* Return the byte-cntr's tmc_sysfs_ops if in using */
>> +static const struct tmc_sysfs_ops *tmc_get_byte_cntr_sysfs_ops(struct
>> tmc_drvdata *drvdata)
>> +{
>> + struct ctcu_byte_cntr *byte_cntr_data;
>> + struct ctcu_drvdata *ctcu_drvdata;
>> + struct coresight_device *ctcu;
>> + int port;
>> +
>> + ctcu = tmc_etr_get_ctcu_device(drvdata);
>> + if (!ctcu)
>> + return NULL;
>> +
>> + port = coresight_get_in_port(drvdata->csdev, ctcu);
>> + if (port < 0)
>> + return NULL;
>> +
>> + ctcu_drvdata = dev_get_drvdata(ctcu->dev.parent);
>> + byte_cntr_data = &ctcu_drvdata->byte_cntr_data[port];
>> + if (byte_cntr_data && byte_cntr_data->thresh_val)
>> + return ctcu_drvdata->byte_cntr_sysfs_ops;
>> +
>> + return NULL;
>> +}
>> +
>
> Should be in a CTCU source file, not part of the common tmc code
>
>> static int tmc_read_prepare(struct tmc_drvdata *drvdata)
>> {
>> + const struct tmc_sysfs_ops *byte_cntr_sysfs_ops;
>> int ret = 0;
>>
>> + byte_cntr_sysfs_ops = tmc_get_byte_cntr_sysfs_ops(drvdata);
>> + if (byte_cntr_sysfs_ops) {
>> + ret = byte_cntr_sysfs_ops->read_prepare(drvdata);
>> + goto out;
>> + }
>> +
>> if (drvdata->sysfs_ops)
>> ret = drvdata->sysfs_ops->read_prepare(drvdata);
>> else
>> ret = -EINVAL;
>>
>
> I understand ctcu usage is per session & per device, but at the start of the session would it not be better to have a function in the ctcu code that takes the drvdata->sysfs_ops and substitutes the callback directly, restoring it at the end.
>
Hi Mike,
Thanks for the suggestion. That would be a better solution, and this
patch is no longer needed.
I will export etr_sysfs_ops in tmc header file so the CTCU driver can
retrieve the pointer and restore to the etr_sysfs_ops.
With this solution, sysfs_ops will be switched to byte_cntr_sysfs_ops in
ctcu_enable and restored in ctcu_disable.
Thanks,
Jie
>
>> +out:
>> if (!ret)
>> dev_dbg(&drvdata->csdev->dev, "TMC read start\n");
>>
>> @@ -245,13 +278,21 @@ static int tmc_read_prepare(struct tmc_drvdata
>> *drvdata)
>>
>> static int tmc_read_unprepare(struct tmc_drvdata *drvdata)
>> {
>> + const struct tmc_sysfs_ops *byte_cntr_sysfs_ops;
>> int ret = 0;
>>
>> + byte_cntr_sysfs_ops = tmc_get_byte_cntr_sysfs_ops(drvdata);
>> + if (byte_cntr_sysfs_ops) {
>> + ret = byte_cntr_sysfs_ops->read_unprepare(drvdata);
>> + goto out;
>> + }
>> +
>> if (drvdata->sysfs_ops)
>> ret = drvdata->sysfs_ops->read_unprepare(drvdata);
>> else
>> ret = -EINVAL;
>>
>
> Again override / restore over the session.
>
> Regards
>
> Mike
>
>> +out:
>> if (!ret)
>> dev_dbg(&drvdata->csdev->dev, "TMC read end\n");
>>
>> @@ -277,6 +318,12 @@ static int tmc_open(struct inode *inode, struct file
>> *file)
>> static ssize_t tmc_get_sysfs_trace(struct tmc_drvdata *drvdata, loff_t pos,
>> size_t len,
>> char **bufpp)
>> {
>> + const struct tmc_sysfs_ops *byte_cntr_sysfs_ops;
>> +
>> + byte_cntr_sysfs_ops = tmc_get_byte_cntr_sysfs_ops(drvdata);
>> + if (byte_cntr_sysfs_ops)
>> + return byte_cntr_sysfs_ops->get_trace_data(drvdata, pos,
>> len, bufpp);
>> +
>> return drvdata->sysfs_ops->get_trace_data(drvdata, pos, len, bufpp);
>> }
>>
>> @@ -297,7 +344,11 @@ static ssize_t tmc_read(struct file *file, char __user
>> *data, size_t len,
>> return -EFAULT;
>> }
>>
>> - *ppos += actual;
>> + if (drvdata->reading_node)
>> + drvdata->reading_node->pos += actual;
>> + else
>> + *ppos += actual;
>> +
>> dev_dbg(&drvdata->csdev->dev, "%zu bytes copied\n", actual);
>>
>> return actual;
>>
>> --
>> 2.34.1
>
More information about the linux-arm-kernel
mailing list