* Re: [PATCH -next 3/4] spi: mockup: Add runtime device tree overlay interface
2022-08-29 21:29 ` [PATCH -next 3/4] spi: mockup: Add runtime device tree overlay interface Frank Rowand
@ 2022-08-30 3:05 ` weiyongjun (A)
0 siblings, 0 replies; 2+ messages in thread
From: weiyongjun (A) @ 2022-08-30 3:05 UTC (permalink / raw)
To: Frank Rowand, Mark Brown, Steven Rostedt, Ingo Molnar,
Rob Herring
Cc: linux-kernel, linux-spi, devicetree@vger.kernel.org
On 2022/8/30 5:29, Frank Rowand wrote:
> comment inline below, plus adding to cc: and to: list
>
> On 8/26/22 09:43, Wei Yongjun wrote:
>> Add a runtime device tree overlay interface for device need dts file.
>> With it its possible to use device tree overlays without having to use
>> a per-platform overlay manager.
>
> Why is an overlay needed? The documentation in patch 4 shows providing
> a dtb as an argument to the qemu-system-x86_64 command, which should be
> sufficient to supply the appropriate dtb.
Fist, we are aimed to build a device driver regression testsuit. Without
overlay, lots of the dtb fragments must be combined to a very big one, it
may different to maintainence. overlay makes easily to attach/detech as
the device's requirement.
Seconds, overlay dts may be used to found or verification issues related
to of_node_put or others, such as patches like:
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/log/?qt=grep&q=node_put
many of them may found by static check tool, which may leak of tesing.
The documentation I post does not use overlay dts, but when overlay
dts[1] is used, we can found the of node leak:
$ python3 -m kddv.cmds.mock -b spi -d mchp23k256
create spi device mchp23k256 success!
$ ls /sys/bus/spi/drivers/mchp23k256/spi0.0/mtd/
mtd0 mtd0ro
$ python3 -m kddv.cmds.mock -b spi -d mchp23k256 -r
[ 233.508565] Deleting MTD partitions on "spi0.0":
[ 233.516784] OF: ERROR: memory leak, expected refcount 1 instead of 2,
of_node_get()/of_node_put() unbalanced - destroy cset entry: attach
overlay node /spi/spi-sram@0
remove spi device mchp23k256 success!
Not sure it is worth to fix issues like this, but at least, we can
tesing them now.
-[1]---------------------------------------------------------
/dts-v1/;
/plugin/;
&{/spi} {
spi-sram@0 {
#address-cells = <1>;
#size-cells = <1>;
compatible = "microchip,mchp23k256";
reg = <0>;
spi-max-frequency = <20000000>;
};
};
-------------------------------------------------------
Regards,
Wei Yongjun
>
> -Frank
>
>>
>> Add a new device by command:
>> $ cat test.dtbo > /sys/class/spi_master/spi0/overlay_fdto
>>
>> Remove the device by command:
>> $ echo remove > /sys/class/spi_master/spi0/overlay_fdto
>>
>> Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
>> ---
>> drivers/spi/Kconfig | 2 ++
>> drivers/spi/spi-mockup.c | 48 ++++++++++++++++++++++++++++++++++++++++
>> 2 files changed, 50 insertions(+)
>>
>> diff --git a/drivers/spi/Kconfig b/drivers/spi/Kconfig
>> index e0f0fa2746ad..4b7c84ddb367 100644
>> --- a/drivers/spi/Kconfig
>> +++ b/drivers/spi/Kconfig
>> @@ -1161,6 +1161,8 @@ config SPI_TLE62X0
>> config SPI_MOCKUP
>> tristate "SPI controller Testing Driver"
>> depends on OF
>> + select OF_EARLY_FLATTREE
>> + select OF_RESOLVE
>> select BPF_EVENTS
>> help
>> This enables SPI controller testing driver, which provides a way to
>> diff --git a/drivers/spi/spi-mockup.c b/drivers/spi/spi-mockup.c
>> index 7a93b194ee53..404ad821bf6a 100644
>> --- a/drivers/spi/spi-mockup.c
>> +++ b/drivers/spi/spi-mockup.c
>> @@ -21,6 +21,9 @@
>> struct mockup_spi {
>> struct mutex lock;
>> struct spi_device *devs[MOCKUP_CHIPSELECT_MAX];
>> +
>> + void *fdto;
>> + int ovcs_id;
>> };
>>
>> static struct spi_master *to_spi_master(struct device *dev)
>> @@ -145,9 +148,53 @@ delete_device_store(struct device *dev, struct device_attribute *attr,
>> }
>> static DEVICE_ATTR_WO(delete_device);
>>
>> +static ssize_t
>> +overlay_fdto_store(struct device *dev, struct device_attribute *attr,
>> + const char *buf, size_t count)
>> +{
>> + struct spi_master *master = to_spi_master(dev);
>> + struct mockup_spi *mock = spi_master_get_devdata(master);
>> + int ret;
>> +
>> + mutex_lock(&mock->lock);
>> +
>> + if (strncmp(buf, "remove\n", count) == 0) {
>> + if (mock->ovcs_id < 0) {
>> + ret = -ENOENT;
>> + goto out_unlock;
>> + }
>> + of_overlay_remove(&mock->ovcs_id);
>> + kfree(mock->fdto);
>> + mock->ovcs_id = -1;
>> + mock->fdto = NULL;
>> + } else {
>> + if (mock->ovcs_id >= 0) {
>> + ret = -EINVAL;
>> + goto out_unlock;
>> + }
>> + mock->fdto = kmemdup(buf, count, GFP_KERNEL);
>> + if (!mock->fdto) {
>> + ret = -ENOMEM;
>> + goto out_unlock;
>> + }
>> + ret = of_overlay_fdt_apply(mock->fdto, count, &mock->ovcs_id);
>> + if (ret < 0)
>> + goto out_unlock;
>> + }
>> +
>> + mutex_unlock(&mock->lock);
>> + return count;
>> +
>> +out_unlock:
>> + mutex_unlock(&mock->lock);
>> + return ret;
>> +}
>> +static DEVICE_ATTR_WO(overlay_fdto);
>> +
>> static struct attribute *spi_mockup_attrs[] = {
>> &dev_attr_new_device.attr,
>> &dev_attr_delete_device.attr,
>> + &dev_attr_overlay_fdto.attr,
>> NULL
>> };
>> ATTRIBUTE_GROUPS(spi_mockup);
>> @@ -227,6 +274,7 @@ static int spi_mockup_probe(struct platform_device *pdev)
>>
>> mock = spi_master_get_devdata(master);
>> mutex_init(&mock->lock);
>> + mock->ovcs_id = -1;
>>
>> ret = spi_register_master(master);
>> if (ret) {
>
^ permalink raw reply [flat|nested] 2+ messages in thread