diff mbox series

[v9,1/1] gpio: add sloppy logic analyzer using polling

Message ID 20240610112700.80819-2-wsa+renesas@sang-engineering.com
State New
Headers show
Series [v9,1/1] gpio: add sloppy logic analyzer using polling | expand

Commit Message

Wolfram Sang June 10, 2024, 11:26 a.m. UTC
This is a sloppy logic analyzer using GPIOs. It comes with a script to
isolate a CPU for polling. While this is definitely not a production
level analyzer, it can be a helpful first view when remote debugging.
Read the documentation for details.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
---

Changes since v8:

* analyzer: gave 'lock' a more descriptive name
* script: handles now SI suffixes somewhat, e.g. '2G'
* script: add '--list-instances' option
* script: add comment why stall_detector gets disabled
* script: improvements to help text
* script: added '.sh' extension to filename
* script, docs: added SPDX license identifiers


 .../dev-tools/gpio-sloppy-logic-analyzer.rst  |  93 +++++
 Documentation/dev-tools/index.rst             |   1 +
 drivers/gpio/Kconfig                          |  17 +
 drivers/gpio/Makefile                         |   1 +
 drivers/gpio/gpio-sloppy-logic-analyzer.c     | 340 ++++++++++++++++++
 tools/gpio/gpio-sloppy-logic-analyzer.sh      | 246 +++++++++++++
 6 files changed, 698 insertions(+)
 create mode 100644 Documentation/dev-tools/gpio-sloppy-logic-analyzer.rst
 create mode 100644 drivers/gpio/gpio-sloppy-logic-analyzer.c
 create mode 100755 tools/gpio/gpio-sloppy-logic-analyzer.sh

Comments

Bartosz Golaszewski June 11, 2024, 8:03 p.m. UTC | #1
On Mon, Jun 10, 2024 at 1:27 PM Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
>
> This is a sloppy logic analyzer using GPIOs. It comes with a script to
> isolate a CPU for polling. While this is definitely not a production
> level analyzer, it can be a helpful first view when remote debugging.
> Read the documentation for details.
>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> ---

I really dislike drivers being called in an ambiguous way like
"simple" or - in this case "sloppy". I understand why it is - in fact
- sloppy but can we call it anything else? Like
"gpio-logic-analyzer.c"?

> +Setup
> +=====
> +
> +Your kernel must have CONFIG_DEBUG_FS and CONFIG_CPUSETS enabled. Ideally, your
> +runtime environment does not utilize cpusets otherwise, then isolation of a CPU
> +core is easiest. If you do need cpusets, check that helper script for the
> +sloppy logic analyzer does not interfere with your other settings.
> +
> +Tell the kernel which GPIOs are used as probes. For a Device Tree based system,
> +you need to use the following bindings. Because these bindings are only for
> +debugging, there is no official schema::
> +
> +    i2c-analyzer {
> +            compatible = "gpio-sloppy-logic-analyzer";
> +            probe-gpios = <&gpio6 21 GPIO_OPEN_DRAIN>, <&gpio6 4 GPIO_OPEN_DRAIN>;
> +            probe-names = "SCL", "SDA";
> +    };
> +
> +Note that you must provide a name for every GPIO specified. Currently a
> +maximum of 8 probes are supported. 32 are likely possible but are not
> +implemented yet.
> +

What happens on non-DT systems? Can you still create an analyzer in a
different way? Can I maybe interest you in configfs for the purpose of
device configuration like what gpio-sim and the upcoming gpio-virtuser
does?

Bart
Wolfram Sang June 12, 2024, 10:03 a.m. UTC | #2
Hi Andi,


> > +#include <linux/delay.h>
> 
> + device.h
> + err.h

OK about the includes.

> > +	mutex_lock(&priv->blob_lock);
> 
> guard() (from cleanup.h)?

If you insist. I doesn't save an exit path, so I don't think this will
improve readability of the code. But I don't mind...

> > +static const struct file_operations fops_trigger = {
> > +	.owner = THIS_MODULE,
> > +	.open = trigger_open,
> > +	.write = trigger_write,
> > +	.llseek = no_llseek,
> > +	.release = single_release,
> > +};
> 
> Wondering if you can use DEFINE_SHOW_STORE_ATTRIBUTE(), or if it makes sense.
> It might be that it requires to use DEFINE_SHOW_ATTRIBUTE() for the sake of
> consistency, but I don't remember if there is a difference WRT debugfs usage.

Well, then we can just leave it for now and change it later, if desired.

> > +	mutex_init(&priv->blob_lock);
> 
> devm_mutex_init()

OK.

> > +		new_meta = devm_krealloc(dev, meta, meta_len + add_len, GFP_KERNEL);
> 
> Can it be rewritten to use devm_krealloc_array()?

'meta' is not an array?

> > +	dev_info(dev, "initialized");
> 
> Do we need this? Existence of folder in debugfs is enough indication of
> success, no?

Since the script can now list instances easily, this can be argued.
Still, I don't think this matters much for a debug-only device.

> > +static const struct of_device_id gpio_la_poll_of_match[] = {
> > +	{ .compatible = GPIO_LA_NAME, },
> 
> Redundant inner comma.

Yes.

> > +late_initcall(gpio_la_poll_init);
> 
> Why? Can we add a comment?

Sure.

> Btw, have you tried `shellcheck` against your script?

We did this in one of the 8 previous iterations.

All the best,

   Wolfram
Wolfram Sang June 12, 2024, 4:03 p.m. UTC | #3
Hi Bart,

> I really dislike drivers being called in an ambiguous way like
> "simple" or - in this case "sloppy". I understand why it is - in fact
> - sloppy but can we call it anything else? Like
> "gpio-logic-analyzer.c"?

Sure, we can if you prefer. I named it like this to make the limitations
super-clear. And even with that in place, I still got a private email
where someone wanted to build a 400MHz-RPi-based logic analyzer device
with it. Which would not only have the latency problems, but also
likely have a max sampling speed of whopping 400kHz.

> > +Note that you must provide a name for every GPIO specified. Currently a
> > +maximum of 8 probes are supported. 32 are likely possible but are not
> > +implemented yet.
> > +
> 
> What happens on non-DT systems? Can you still create an analyzer in a
> different way? Can I maybe interest you in configfs for the purpose of
> device configuration like what gpio-sim and the upcoming gpio-virtuser
> does?

Frankly, I'd like to leave this to the person needing it. I've been
working on this for way too long already and am not up to major changes
anymore. Minor stuff, okay, I'll go one or two more rounds.

The GPIO analyzer is a debug tool aimed for development boards in remote
labs, and all boards I have access to use DT. Furthermore, debugfs is nice
because it is clear there is no stable ABI. It has been useful as-is in
the past. That's what I am offering. If that's not enough, no hard
feelings, but someone else needs to continue then.

All the best,

   Wolfram
Bartosz Golaszewski June 13, 2024, 8:17 a.m. UTC | #4
On Wed, Jun 12, 2024 at 6:03 PM Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
>
> Hi Bart,
>
> > I really dislike drivers being called in an ambiguous way like
> > "simple" or - in this case "sloppy". I understand why it is - in fact
> > - sloppy but can we call it anything else? Like
> > "gpio-logic-analyzer.c"?
>
> Sure, we can if you prefer. I named it like this to make the limitations
> super-clear. And even with that in place, I still got a private email
> where someone wanted to build a 400MHz-RPi-based logic analyzer device
> with it. Which would not only have the latency problems, but also
> likely have a max sampling speed of whopping 400kHz.
>
> > > +Note that you must provide a name for every GPIO specified. Currently a
> > > +maximum of 8 probes are supported. 32 are likely possible but are not
> > > +implemented yet.
> > > +
> >
> > What happens on non-DT systems? Can you still create an analyzer in a
> > different way? Can I maybe interest you in configfs for the purpose of
> > device configuration like what gpio-sim and the upcoming gpio-virtuser
> > does?
>
> Frankly, I'd like to leave this to the person needing it. I've been
> working on this for way too long already and am not up to major changes
> anymore. Minor stuff, okay, I'll go one or two more rounds.
>
> The GPIO analyzer is a debug tool aimed for development boards in remote
> labs, and all boards I have access to use DT. Furthermore, debugfs is nice
> because it is clear there is no stable ABI. It has been useful as-is in
> the past. That's what I am offering. If that's not enough, no hard
> feelings, but someone else needs to continue then.
>
> All the best,

I see. I don't want to stop it from going upstream. On second thought
though: are you sure drivers/gpio/ is the right place for it? This
directory is for GPIO *providers* - mostly drivers for real HW and
some "virtual" GPIO providers (gpio-aggregator, gpio-sim, gpio-latch).
In general: modules that export their own gpio_chip. This module is a
consumer of GPIOs exclusively. May I suggest moving it over to
drivers/misc/? I think it's a better place for it and I wouldn't mind
keeping the "sloppy" in the name then.

Bart
Linus Walleij June 13, 2024, 8:27 a.m. UTC | #5
On Thu, Jun 13, 2024 at 10:17 AM Bartosz Golaszewski <brgl@bgdev.pl> wrote:

> On second thought
> though: are you sure drivers/gpio/ is the right place for it?

Actually that is something I requested.

I think it fits in drivers/gpio as it is such a clear cut usage of GPIO
lines, and it doesn't really fit into any other subsystem.

> May I suggest moving it over to drivers/misc/?

Misc is a bit...
messy. I remember Arnd being very sceptical about putting stuff there
rather than creating new subsystems, so since I've tried to avoid it,
albeit recently more and more stuff gets merged there again :/

Yours,
Linus Walleij
Linus Walleij June 13, 2024, 8:30 a.m. UTC | #6
On Wed, Jun 12, 2024 at 6:03 PM Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:

> [Bart]
> > I really dislike drivers being called in an ambiguous way like
> > "simple" or - in this case "sloppy". I understand why it is - in fact
> > - sloppy but can we call it anything else? Like
> > "gpio-logic-analyzer.c"?
>
> Sure, we can if you prefer. I named it like this to make the limitations
> super-clear. And even with that in place, I still got a private email
> where someone wanted to build a 400MHz-RPi-based logic analyzer device
> with it. Which would not only have the latency problems, but also
> likely have a max sampling speed of whopping 400kHz.

What about "gpio-low-fidelity-logic-analyzer.c"

(+/- Kconfig etc adjusted accordingly)

It says what it is, not really sloppy but really low-fi.

Yours,
Linus Walleij
Arnd Bergmann June 13, 2024, 8:50 a.m. UTC | #7
On Thu, Jun 13, 2024, at 10:27, Linus Walleij wrote:
> On Thu, Jun 13, 2024 at 10:17 AM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
>
>> On second thought
>> though: are you sure drivers/gpio/ is the right place for it?
>
> Actually that is something I requested.
>
> I think it fits in drivers/gpio as it is such a clear cut usage of GPIO
> lines, and it doesn't really fit into any other subsystem.
>
>> May I suggest moving it over to drivers/misc/?
>
> Misc is a bit...
> messy. I remember Arnd being very sceptical about putting stuff there
> rather than creating new subsystems, so since I've tried to avoid it,
> albeit recently more and more stuff gets merged there again :/

Right, and that is mostly to avoid having code in there because
there is no other place for it. Some parts of drivers/misc should
have been a separate subsystem, some should have use an existing
subsystem, and other parts should have never been merged.

The parts of drivers/misc that make the most sense to me are
those that expose a one-of-a-kind piece of hardware as a
single character device.

This one would probably fit into drivers/misc/ better than
some other drivers we have in there, but leaving it in
drivers/gpio/ also seems fine.

I could also imagine the functionality being exposed
through drivers/iio/ in a way that is similar to an
adc, but I don't know if that would work in practice or
how much of a rewrite that would be.

      Arnd
Bartosz Golaszewski June 13, 2024, 9:43 a.m. UTC | #8
On Thu, Jun 13, 2024 at 10:50 AM Arnd Bergmann <arnd@arndb.de> wrote:
>
> On Thu, Jun 13, 2024, at 10:27, Linus Walleij wrote:
> > On Thu, Jun 13, 2024 at 10:17 AM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
> >
> >> On second thought
> >> though: are you sure drivers/gpio/ is the right place for it?
> >
> > Actually that is something I requested.
> >
> > I think it fits in drivers/gpio as it is such a clear cut usage of GPIO
> > lines, and it doesn't really fit into any other subsystem.
> >
> >> May I suggest moving it over to drivers/misc/?
> >
> > Misc is a bit...
> > messy. I remember Arnd being very sceptical about putting stuff there
> > rather than creating new subsystems, so since I've tried to avoid it,
> > albeit recently more and more stuff gets merged there again :/
>
> Right, and that is mostly to avoid having code in there because
> there is no other place for it. Some parts of drivers/misc should
> have been a separate subsystem, some should have use an existing
> subsystem, and other parts should have never been merged.
>
> The parts of drivers/misc that make the most sense to me are
> those that expose a one-of-a-kind piece of hardware as a
> single character device.
>
> This one would probably fit into drivers/misc/ better than
> some other drivers we have in there, but leaving it in
> drivers/gpio/ also seems fine.
>

This is my point. This really is a one-of-a-kind module that also
doesn't register with any particular subsystem. If anything fits into
drivers/misc/ then it's this.

To prove this point, I even moved the gpio-virtuser driver I'm working
on to drivers/misc/ too as it isn't a GPIO provider either and merely
a GPIO consumer with a one-shot user-space interface not conforming to
any standards.

> I could also imagine the functionality being exposed
> through drivers/iio/ in a way that is similar to an
> adc, but I don't know if that would work in practice or
> how much of a rewrite that would be.
>

I could see it using configfs instead of DT for configuration and iio
for presenting the output but - from what Wolfram said - insisting on
this will simply result in this development being dropped entirely.

Bart
Wolfram Sang June 13, 2024, 11:45 a.m. UTC | #9
> It says what it is, not really sloppy but really low-fi.

I think it is sloppy because it cannot guarantee equi-distant sample
points.
Wolfram Sang June 13, 2024, 11:48 a.m. UTC | #10
> > I could also imagine the functionality being exposed
> > through drivers/iio/ in a way that is similar to an
> > adc, but I don't know if that would work in practice or
> > how much of a rewrite that would be.
> >
> 
> I could see it using configfs instead of DT for configuration and iio
> for presenting the output but - from what Wolfram said - insisting on
> this will simply result in this development being dropped entirely.

Despite that, I'd be afraid that the analyzer looks more trustworthy
than it actually is then. debugfs makes that clearer IMO.
Linus Walleij June 13, 2024, 1:47 p.m. UTC | #11
On Thu, Jun 13, 2024 at 11:43 AM Bartosz Golaszewski <brgl@bgdev.pl> wrote:

> To prove this point, I even moved the gpio-virtuser driver I'm working
> on to drivers/misc/ too as it isn't a GPIO provider either and merely
> a GPIO consumer with a one-shot user-space interface not conforming to
> any standards.

We *could* just create drivers/gpio/consumers/* and an entry into the
top-level drivers/Kconfig to have those appear right under the GPIO
providers...

Yours,
Linus Walleij
Bartosz Golaszewski June 13, 2024, 1:51 p.m. UTC | #12
On Thu, Jun 13, 2024 at 3:47 PM Linus Walleij <linus.walleij@linaro.org> wrote:
>
> On Thu, Jun 13, 2024 at 11:43 AM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
>
> > To prove this point, I even moved the gpio-virtuser driver I'm working
> > on to drivers/misc/ too as it isn't a GPIO provider either and merely
> > a GPIO consumer with a one-shot user-space interface not conforming to
> > any standards.
>
> We *could* just create drivers/gpio/consumers/* and an entry into the
> top-level drivers/Kconfig to have those appear right under the GPIO
> providers...
>
> Yours,
> Linus Walleij

That would just add to confusion. GPIO consumers are all over the tree
after all.

Whatever, let's keep it in drivers/gpio/. Greg KH just shot down my
idea of putting gpio-virtuser in drivers/misc/.

Bart
Arnd Bergmann June 13, 2024, 2:20 p.m. UTC | #13
On Thu, Jun 13, 2024, at 15:51, Bartosz Golaszewski wrote:
> On Thu, Jun 13, 2024 at 3:47 PM Linus Walleij <linus.walleij@linaro.org> wrote:
>>
>> On Thu, Jun 13, 2024 at 11:43 AM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
>>
>> > To prove this point, I even moved the gpio-virtuser driver I'm working
>> > on to drivers/misc/ too as it isn't a GPIO provider either and merely
>> > a GPIO consumer with a one-shot user-space interface not conforming to
>> > any standards.
>>
>> We *could* just create drivers/gpio/consumers/* and an entry into the
>> top-level drivers/Kconfig to have those appear right under the GPIO
>> providers...
>>
>> Yours,
>> Linus Walleij
>
> That would just add to confusion. GPIO consumers are all over the tree
> after all.
>
> Whatever, let's keep it in drivers/gpio/. Greg KH just shot down my
> idea of putting gpio-virtuser in drivers/misc/.

I could imagine treating both gpio-virtuser and this code as
a gpiolib extension rather than a consumer (which is usually
part of some other subsystem's driver).

It would also make sense to me to separate gpio providers
from gpiolib in a way, moving one or both of them into a
subdirectory of drivers/gpio/.

It's probably not worth the pain of moving files, but at
least in Kconfig and filenames, they could be named
gpiolib-virtuser.c and gpiolib-sloppy-logic-analyzer.c
to make it clear that these are not gpio provider drivers
but something else, more along the lines of gpiolib-cdev.c
and gpiolib-sysfs.c.

     Arnd
Geert Uytterhoeven June 13, 2024, 3:04 p.m. UTC | #14
Hi Arnd,

On Thu, Jun 13, 2024 at 10:54 AM Arnd Bergmann <arnd@arndb.de> wrote:
> On Thu, Jun 13, 2024, at 10:27, Linus Walleij wrote:
> > On Thu, Jun 13, 2024 at 10:17 AM Bartosz Golaszewski <brgl@bgdev.pl> wrote:
> >
> >> On second thought
> >> though: are you sure drivers/gpio/ is the right place for it?
> >
> > Actually that is something I requested.
> >
> > I think it fits in drivers/gpio as it is such a clear cut usage of GPIO
> > lines, and it doesn't really fit into any other subsystem.
> >
> >> May I suggest moving it over to drivers/misc/?
> >
> > Misc is a bit...
> > messy. I remember Arnd being very sceptical about putting stuff there
> > rather than creating new subsystems, so since I've tried to avoid it,
> > albeit recently more and more stuff gets merged there again :/
>
> Right, and that is mostly to avoid having code in there because
> there is no other place for it. Some parts of drivers/misc should
> have been a separate subsystem, some should have use an existing
> subsystem, and other parts should have never been merged.
>
> The parts of drivers/misc that make the most sense to me are
> those that expose a one-of-a-kind piece of hardware as a
> single character device.
>
> This one would probably fit into drivers/misc/ better than
> some other drivers we have in there, but leaving it in
> drivers/gpio/ also seems fine.
>
> I could also imagine the functionality being exposed
> through drivers/iio/ in a way that is similar to an
> adc, but I don't know if that would work in practice or
> how much of a rewrite that would be.

Hmm, I like the iio idea.
Sorry, Wolfram ;-)

Gr{oetje,eeting}s,

                        Geert
Wolfram Sang June 13, 2024, 4:33 p.m. UTC | #15
> Hmm, I like the iio idea.
> Sorry, Wolfram ;-)

Ah, no worries, Geert. I am happy you want to take over :)
Wolfram Sang June 14, 2024, 10:03 a.m. UTC | #16
Hi Arnd, everyone,

> I could imagine treating both gpio-virtuser and this code as
> a gpiolib extension rather than a consumer (which is usually
> part of some other subsystem's driver).

I have difficulties seeing this. For the analyzer, at least. It does not
extend gpiolib in a way another consumer could make use of it?

> It would also make sense to me to separate gpio providers
> from gpiolib in a way, moving one or both of them into a
> subdirectory of drivers/gpio/.

I'd also like 'drivers/gpio/providers' and leave the core stuff (incl.
the analyzer) in 'drivers/gpio'. But I am biased, I2C looks like this :)
And yes, this is some churn and git-history spoiling.

> gpiolib-virtuser.c and gpiolib-sloppy-logic-analyzer.c

'gpio-tool-sloppy-logic-analyzer.c' ? Based on what gets added to
Kconfig with this patch:

	+menu "GPIO hardware hacking tools"

Happy hacking,

   Wolfram
Linus Walleij June 14, 2024, 12:14 p.m. UTC | #17
On Fri, Jun 14, 2024 at 1:59 PM Arnd Bergmann <arnd@arndb.de> wrote:

> Of course we don't want to make it a first-class interface
> because of the reasons you explain in the cover letter,
> but it would totally make sense to me to call it a
> debugging feature of libgpio instead of calling it a
> standalone driver.

I second this opinion. The logic analyzer does in my mind
classify as a GPIO debugging feature. Surely someone
debugging anything connected to GPIO, such as a key or
MMC card detect or whatever could use this feature to see
what is going on on that line.

Yours,
Linus Walleij
Wolfram Sang June 16, 2024, 4:04 p.m. UTC | #18
> I second this opinion. The logic analyzer does in my mind
> classify as a GPIO debugging feature. Surely someone
> debugging anything connected to GPIO, such as a key or
> MMC card detect or whatever could use this feature to see
> what is going on on that line.

Okay, with that picture I can see where your argument is coming from.
However, making it a gpiolib debugging feature will surely raise
expectations. And it is not only the non-equi-distant sampling. The
script trying to isolate a CPU core tries really hard but is still hacky
IMO. It has to disable the RCU stall detector and will likely interfere
with your CPUSET configuration if you have one. As I always said, it is
last resort debugging.

As I write this, I start to wonder if this should be really upstream or
if I just keep it as a branch in my repo. Maybe it is just too hackish
to be supported mainline?
Bartosz Golaszewski June 18, 2024, 12:50 p.m. UTC | #19
On Tue, Jun 18, 2024 at 11:23 AM Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
>
> Hi Bart,
>
> > I could see it using configfs instead of DT for configuration and iio
> > for presenting the output but - from what Wolfram said - insisting on
> > this will simply result in this development being dropped entirely.
>
> How do you assign a GPIO via debugfs? I only found the out-of-tree
> pwm-gpio driver[1] which uses a GPIO number. But those are deprecated
> these days, or? Any other driver doing this you can point me to?
>

You mean configfs? Please take a look at my proposed gpio-virtuser
module, it should do what you want:
https://lore.kernel.org/linux-gpio/20240613092830.15761-1-brgl@bgdev.pl/

Bart

> Thanks and happy hacking,
>
>    Wolfram
>
> [1] https://lore.kernel.org/lkml/1301630392-20793-3-git-send-email-bgat@billgatliff.com/raw
Wolfram Sang June 18, 2024, 4:09 p.m. UTC | #20
> You mean configfs? Please take a look at my proposed gpio-virtuser

Oops, yes, configfs.

> module, it should do what you want:
> https://lore.kernel.org/linux-gpio/20240613092830.15761-1-brgl@bgdev.pl/

Thank you! Exactly what I was looking for. What I like is the symmetry
of instantiating GPIOs in either DT or configfs. Very elegant for the
user. But boy, this is a lot of non-trivial code for it.
Geert Uytterhoeven June 18, 2024, 6:58 p.m. UTC | #21
Hi Wolfram,

On Tue, Jun 18, 2024 at 12:54 PM Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
> > I could see it using configfs instead of DT for configuration and iio
> > for presenting the output but - from what Wolfram said - insisting on
> > this will simply result in this development being dropped entirely.
>
> How do you assign a GPIO via debugfs? I only found the out-of-tree
> pwm-gpio driver[1] which uses a GPIO number. But those are deprecated
> these days, or? Any other driver doing this you can point me to?

Do you really need debugfs (or configfs)?
I guess you can just write GPIO line names or GPIO chip
labels + offsets to the new_device file, like gpio-aggregator does?

Documentation/admin-guide/gpio/gpio-aggregator.rst

Gr{oetje,eeting}s,

                        Geert
Bartosz Golaszewski June 18, 2024, 7:46 p.m. UTC | #22
On Tue, Jun 18, 2024 at 8:58 PM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
> Hi Wolfram,
>
> On Tue, Jun 18, 2024 at 12:54 PM Wolfram Sang
> <wsa+renesas@sang-engineering.com> wrote:
> > > I could see it using configfs instead of DT for configuration and iio
> > > for presenting the output but - from what Wolfram said - insisting on
> > > this will simply result in this development being dropped entirely.
> >
> > How do you assign a GPIO via debugfs? I only found the out-of-tree
> > pwm-gpio driver[1] which uses a GPIO number. But those are deprecated
> > these days, or? Any other driver doing this you can point me to?
>
> Do you really need debugfs (or configfs)?
> I guess you can just write GPIO line names or GPIO chip
> labels + offsets to the new_device file, like gpio-aggregator does?
>
> Documentation/admin-guide/gpio/gpio-aggregator.rst
>

IMO that adds a lot of custom string parsing in kernel for no reason.
TBH Today I'd NAK this interface and propose configfs instead as well.

Bart

> Gr{oetje,eeting}s,
>
>                         Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
>                                 -- Linus Torvalds
diff mbox series

Patch

diff --git a/Documentation/dev-tools/gpio-sloppy-logic-analyzer.rst b/Documentation/dev-tools/gpio-sloppy-logic-analyzer.rst
new file mode 100644
index 000000000000..d69f24c0d9e1
--- /dev/null
+++ b/Documentation/dev-tools/gpio-sloppy-logic-analyzer.rst
@@ -0,0 +1,93 @@ 
+.. SPDX-License-Identifier: GPL-2.0
+
+=============================================
+Linux Kernel GPIO based sloppy logic analyzer
+=============================================
+
+:Author: Wolfram Sang
+
+Introduction
+============
+
+This document briefly describes how to run the GPIO based in-kernel sloppy
+logic analyzer running on an isolated CPU.
+
+The sloppy logic analyzer will utilize a few GPIO lines in input mode on a
+system to rapidly sample these digital lines, which will, if the Nyquist
+criteria is met, result in a time series log with approximate waveforms as they
+appeared on these lines. One way to use it is to analyze external traffic
+connected to these GPIO lines with wires (i.e. digital probes), acting as a
+common logic analyzer.
+
+Another feature is to snoop on on-chip peripherals if the I/O cells of these
+peripherals can be used in GPIO input mode at the same time as they are being
+used as inputs or outputs for the peripheral. That means you could e.g. snoop
+I2C traffic without any wiring (if your hardware supports it). In the pin
+control subsystem such pin controllers are called "non-strict": a certain pin
+can be used with a certain peripheral and as a GPIO input line at the same
+time.
+
+Note that this is a last resort analyzer which can be affected by latencies,
+non-deterministic code paths and non-maskable interrupts. It is called 'sloppy'
+for a reason. However, for e.g. remote development, it may be useful to get a
+first view and aid further debugging.
+
+Setup
+=====
+
+Your kernel must have CONFIG_DEBUG_FS and CONFIG_CPUSETS enabled. Ideally, your
+runtime environment does not utilize cpusets otherwise, then isolation of a CPU
+core is easiest. If you do need cpusets, check that helper script for the
+sloppy logic analyzer does not interfere with your other settings.
+
+Tell the kernel which GPIOs are used as probes. For a Device Tree based system,
+you need to use the following bindings. Because these bindings are only for
+debugging, there is no official schema::
+
+    i2c-analyzer {
+            compatible = "gpio-sloppy-logic-analyzer";
+            probe-gpios = <&gpio6 21 GPIO_OPEN_DRAIN>, <&gpio6 4 GPIO_OPEN_DRAIN>;
+            probe-names = "SCL", "SDA";
+    };
+
+Note that you must provide a name for every GPIO specified. Currently a
+maximum of 8 probes are supported. 32 are likely possible but are not
+implemented yet.
+
+Usage
+=====
+
+The logic analyzer is configurable via files in debugfs. However, it is
+strongly recommended to not use them directly, but to use the script
+``tools/gpio/gpio-sloppy-logic-analyzer``. Besides checking parameters more
+extensively, it will isolate the CPU core so you will have the least
+disturbance while measuring.
+
+The script has a help option explaining the parameters. For the above DT
+snippet which analyzes an I2C bus at 400kHz on a Renesas Salvator-XS board, the
+following settings are used: The isolated CPU shall be CPU1 because it is a big
+core in a big.LITTLE setup. Because CPU1 is the default, we don't need a
+parameter. The bus speed is 400kHz. So, the sampling theorem says we need to
+sample at least at 800kHz. However, falling edges of both signals in an I2C
+start condition happen faster, so we need a higher sampling frequency, e.g.
+``-s 1500000`` for 1.5MHz. Also, we don't want to sample right away but wait
+for a start condition on an idle bus. So, we need to set a trigger to a falling
+edge on SDA while SCL stays high, i.e. ``-t 1H+2F``. Last is the duration, let
+us assume 15ms here which results in the parameter ``-d 15000``. So,
+altogether::
+
+    gpio-sloppy-logic-analyzer -s 1500000 -t 1H+2F -d 15000
+
+Note that the process will return you back to the prompt but a sub-process is
+still sampling in the background. Unless this has finished, you will not find a
+result file in the current or specified directory. For the above example, we
+will then need to trigger I2C communication::
+
+    i2cdetect -y -r <your bus number>
+
+Result is a .sr file to be consumed with PulseView or sigrok-cli from the free
+`sigrok`_ project. It is a zip file which also contains the binary sample data
+which may be consumed by other software. The filename is the logic analyzer
+instance name plus a since-epoch timestamp.
+
+.. _sigrok: https://sigrok.org/
diff --git a/Documentation/dev-tools/index.rst b/Documentation/dev-tools/index.rst
index efa49cdc8e2e..6971ed581c08 100644
--- a/Documentation/dev-tools/index.rst
+++ b/Documentation/dev-tools/index.rst
@@ -32,6 +32,7 @@  Documentation/dev-tools/testing-overview.rst
    kunit/index
    ktap
    checkuapi
+   gpio-sloppy-logic-analyzer
 
 
 .. only::  subproject and html
diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
index 1c28a48915bb..44869de5e3fc 100644
--- a/drivers/gpio/Kconfig
+++ b/drivers/gpio/Kconfig
@@ -1891,4 +1891,21 @@  config GPIO_SIM
 
 endmenu
 
+menu "GPIO hardware hacking tools"
+
+config GPIO_SLOPPY_LOGIC_ANALYZER
+	tristate "Sloppy GPIO logic analyzer"
+	depends on (GPIOLIB || COMPILE_TEST) && CPUSETS && DEBUG_FS && EXPERT
+	help
+	  This option enables support for a sloppy logic analyzer using polled
+	  GPIOs. Use the 'tools/gpio/gpio-sloppy-logic-analyzer' script with
+	  this driver. The script will make it easier to use and will also
+	  isolate a CPU for the polling task. Note that this is a last resort
+	  analyzer which can be affected by latencies, non-deterministic code
+	  paths, or NMIs. However, for e.g. remote development, it may be useful
+	  to get a first view and aid further debugging.
+
+	  If this driver is built as a module it will be called
+	  'gpio-sloppy-logic-analyzer'.
+endmenu
 endif
diff --git a/drivers/gpio/Makefile b/drivers/gpio/Makefile
index e2a53013780e..9889f5b962bf 100644
--- a/drivers/gpio/Makefile
+++ b/drivers/gpio/Makefile
@@ -150,6 +150,7 @@  obj-$(CONFIG_GPIO_SIFIVE)		+= gpio-sifive.o
 obj-$(CONFIG_GPIO_SIM)			+= gpio-sim.o
 obj-$(CONFIG_GPIO_SIOX)			+= gpio-siox.o
 obj-$(CONFIG_GPIO_SL28CPLD)		+= gpio-sl28cpld.o
+obj-$(CONFIG_GPIO_SLOPPY_LOGIC_ANALYZER) += gpio-sloppy-logic-analyzer.o
 obj-$(CONFIG_GPIO_SODAVILLE)		+= gpio-sodaville.o
 obj-$(CONFIG_GPIO_SPEAR_SPICS)		+= gpio-spear-spics.o
 obj-$(CONFIG_GPIO_SPRD)			+= gpio-sprd.o
diff --git a/drivers/gpio/gpio-sloppy-logic-analyzer.c b/drivers/gpio/gpio-sloppy-logic-analyzer.c
new file mode 100644
index 000000000000..4f61a1852ab0
--- /dev/null
+++ b/drivers/gpio/gpio-sloppy-logic-analyzer.c
@@ -0,0 +1,340 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Sloppy logic analyzer using GPIOs (to be run on an isolated CPU)
+ *
+ * Use the 'gpio-sloppy-logic-analyzer' script in the 'tools/gpio' folder for
+ * easier usage and further documentation. Note that this is a last resort
+ * analyzer which can be affected by latencies and non-deterministic code
+ * paths. However, for e.g. remote development, it may be useful to get a first
+ * view and aid further debugging.
+ *
+ * Copyright (C) Wolfram Sang <wsa@sang-engineering.com>
+ * Copyright (C) Renesas Electronics Corporation
+ */
+
+#include <linux/ctype.h>
+#include <linux/debugfs.h>
+#include <linux/delay.h>
+#include <linux/gpio/consumer.h>
+#include <linux/init.h>
+#include <linux/ktime.h>
+#include <linux/mod_devicetable.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/platform_device.h>
+#include <linux/property.h>
+#include <linux/slab.h>
+#include <linux/sizes.h>
+#include <linux/timekeeping.h>
+#include <linux/vmalloc.h>
+
+#define GPIO_LA_NAME "gpio-sloppy-logic-analyzer"
+#define GPIO_LA_DEFAULT_BUF_SIZE SZ_256K
+/* can be increased but then we need to extend the u8 buffers */
+#define GPIO_LA_MAX_PROBES 8
+#define GPIO_LA_NUM_TESTS 1024
+
+struct gpio_la_poll_priv {
+	struct mutex blob_lock; /* serialize access to the blob (data) */
+	u32 buf_idx;
+	struct gpio_descs *descs;
+	unsigned long delay_ns;
+	unsigned long acq_delay;
+	struct debugfs_blob_wrapper blob;
+	struct dentry *debug_dir;
+	struct dentry *blob_dent;
+	struct debugfs_blob_wrapper meta;
+	struct device *dev;
+	unsigned int trig_len;
+	u8 *trig_data;
+};
+
+static struct dentry *gpio_la_poll_debug_dir;
+
+static __always_inline int gpio_la_get_array(struct gpio_descs *d, unsigned long *sptr)
+{
+	int ret;
+
+	ret = gpiod_get_array_value(d->ndescs, d->desc, d->info, sptr);
+	if (ret == 0 && fatal_signal_pending(current))
+		ret = -EINTR;
+
+	return ret;
+}
+
+static int fops_capture_set(void *data, u64 val)
+{
+	struct gpio_la_poll_priv *priv = data;
+	u8 *la_buf = priv->blob.data;
+	unsigned long state = 0; /* zeroed because GPIO arrays are bitfields */
+	unsigned long delay;
+	ktime_t start_time;
+	unsigned int i;
+	int ret;
+
+	if (!val)
+		return 0;
+
+	if (!la_buf)
+		return -ENOMEM;
+
+	if (!priv->delay_ns)
+		return -EINVAL;
+
+	mutex_lock(&priv->blob_lock);
+	if (priv->blob_dent) {
+		debugfs_remove(priv->blob_dent);
+		priv->blob_dent = NULL;
+	}
+
+	priv->buf_idx = 0;
+
+	local_irq_disable();
+	preempt_disable_notrace();
+
+	/* Measure delay of reading GPIOs */
+	start_time = ktime_get();
+	for (i = 0; i < GPIO_LA_NUM_TESTS; i++) {
+		ret = gpio_la_get_array(priv->descs, &state);
+		if (ret)
+			goto out;
+	}
+
+	priv->acq_delay = ktime_sub(ktime_get(), start_time) / GPIO_LA_NUM_TESTS;
+	if (priv->delay_ns < priv->acq_delay) {
+		ret = -ERANGE;
+		goto out;
+	}
+
+	delay = priv->delay_ns - priv->acq_delay;
+
+	/* Wait for triggers */
+	for (i = 0; i < priv->trig_len; i += 2) {
+		do {
+			ret = gpio_la_get_array(priv->descs, &state);
+			if (ret)
+				goto out;
+
+			ndelay(delay);
+		} while ((state & priv->trig_data[i]) != priv->trig_data[i + 1]);
+	}
+
+	/* With triggers, final state is also the first sample */
+	if (priv->trig_len)
+		la_buf[priv->buf_idx++] = state;
+
+	/* Sample */
+	while (priv->buf_idx < priv->blob.size) {
+		ret = gpio_la_get_array(priv->descs, &state);
+		if (ret)
+			goto out;
+
+		la_buf[priv->buf_idx++] = state;
+		ndelay(delay);
+	}
+out:
+	preempt_enable_notrace();
+	local_irq_enable();
+	if (ret)
+		dev_err(priv->dev, "couldn't read GPIOs: %d\n", ret);
+
+	kfree(priv->trig_data);
+	priv->trig_data = NULL;
+	priv->trig_len = 0;
+
+	priv->blob_dent = debugfs_create_blob("sample_data", 0400, priv->debug_dir, &priv->blob);
+	mutex_unlock(&priv->blob_lock);
+
+	return ret;
+}
+DEFINE_DEBUGFS_ATTRIBUTE(fops_capture, NULL, fops_capture_set, "%llu\n");
+
+static int fops_buf_size_get(void *data, u64 *val)
+{
+	struct gpio_la_poll_priv *priv = data;
+
+	*val = priv->blob.size;
+
+	return 0;
+}
+
+static int fops_buf_size_set(void *data, u64 val)
+{
+	struct gpio_la_poll_priv *priv = data;
+	int ret = 0;
+	void *p;
+
+	if (!val)
+		return -EINVAL;
+
+	mutex_lock(&priv->blob_lock);
+
+	vfree(priv->blob.data);
+	p = vzalloc(val);
+	if (!p) {
+		val = 0;
+		ret = -ENOMEM;
+	}
+
+	priv->blob.data = p;
+	priv->blob.size = val;
+
+	mutex_unlock(&priv->blob_lock);
+	return ret;
+}
+DEFINE_DEBUGFS_ATTRIBUTE(fops_buf_size, fops_buf_size_get, fops_buf_size_set, "%llu\n");
+
+static int trigger_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, NULL, inode->i_private);
+}
+
+static ssize_t trigger_write(struct file *file, const char __user *ubuf,
+			     size_t count, loff_t *offset)
+{
+	struct seq_file *m = file->private_data;
+	struct gpio_la_poll_priv *priv = m->private;
+	char *buf;
+
+	/* upper limit is arbitrary but should be less than PAGE_SIZE */
+	if (count > 2048 || count & 1)
+		return -EINVAL;
+
+	buf = memdup_user(ubuf, count);
+	if (IS_ERR(buf))
+		return PTR_ERR(buf);
+
+	priv->trig_data = buf;
+	priv->trig_len = count;
+
+	return count;
+}
+
+static const struct file_operations fops_trigger = {
+	.owner = THIS_MODULE,
+	.open = trigger_open,
+	.write = trigger_write,
+	.llseek = no_llseek,
+	.release = single_release,
+};
+
+static int gpio_la_poll_probe(struct platform_device *pdev)
+{
+	struct gpio_la_poll_priv *priv;
+	struct device *dev = &pdev->dev;
+	const char *devname = dev_name(dev);
+	const char *gpio_names[GPIO_LA_MAX_PROBES];
+	char *meta = NULL;
+	unsigned int i, meta_len = 0;
+	int ret;
+
+	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	mutex_init(&priv->blob_lock);
+
+	fops_buf_size_set(priv, GPIO_LA_DEFAULT_BUF_SIZE);
+
+	priv->descs = devm_gpiod_get_array(dev, "probe", GPIOD_IN);
+	if (IS_ERR(priv->descs))
+		return PTR_ERR(priv->descs);
+
+	/* artificial limit to keep 1 byte per sample for now */
+	if (priv->descs->ndescs > GPIO_LA_MAX_PROBES)
+		return -EFBIG;
+
+	ret = device_property_read_string_array(dev, "probe-names", gpio_names,
+						priv->descs->ndescs);
+	if (ret >= 0 && ret != priv->descs->ndescs)
+		ret = -EBADR;
+	if (ret < 0)
+		return dev_err_probe(dev, ret, "error naming the GPIOs");
+
+	for (i = 0; i < priv->descs->ndescs; i++) {
+		unsigned int add_len;
+		char *new_meta, *consumer_name;
+
+		if (gpiod_cansleep(priv->descs->desc[i]))
+			return -EREMOTE;
+
+		consumer_name = kasprintf(GFP_KERNEL, "%s: %s", devname, gpio_names[i]);
+		if (!consumer_name)
+			return -ENOMEM;
+		gpiod_set_consumer_name(priv->descs->desc[i], consumer_name);
+		kfree(consumer_name);
+
+		/* '10' is length of 'probe00=\n\0' */
+		add_len = strlen(gpio_names[i]) + 10;
+
+		new_meta = devm_krealloc(dev, meta, meta_len + add_len, GFP_KERNEL);
+		if (!new_meta)
+			return -ENOMEM;
+
+		meta = new_meta;
+		meta_len += snprintf(meta + meta_len, add_len, "probe%02u=%s\n",
+				     i + 1, gpio_names[i]);
+	}
+
+	platform_set_drvdata(pdev, priv);
+	priv->dev = dev;
+
+	priv->meta.data = meta;
+	priv->meta.size = meta_len;
+	priv->debug_dir = debugfs_create_dir(devname, gpio_la_poll_debug_dir);
+	debugfs_create_blob("meta_data", 0400, priv->debug_dir, &priv->meta);
+	debugfs_create_ulong("delay_ns", 0600, priv->debug_dir, &priv->delay_ns);
+	debugfs_create_ulong("delay_ns_acquisition", 0400, priv->debug_dir, &priv->acq_delay);
+	debugfs_create_file_unsafe("buf_size", 0600, priv->debug_dir, priv, &fops_buf_size);
+	debugfs_create_file_unsafe("capture", 0200, priv->debug_dir, priv, &fops_capture);
+	debugfs_create_file_unsafe("trigger", 0200, priv->debug_dir, priv, &fops_trigger);
+
+	dev_info(dev, "initialized");
+	return 0;
+}
+
+static int gpio_la_poll_remove(struct platform_device *pdev)
+{
+	struct gpio_la_poll_priv *priv = platform_get_drvdata(pdev);
+
+	mutex_lock(&priv->blob_lock);
+	debugfs_remove_recursive(priv->debug_dir);
+	mutex_unlock(&priv->blob_lock);
+	mutex_destroy(&priv->blob_lock);
+
+	return 0;
+}
+
+static const struct of_device_id gpio_la_poll_of_match[] = {
+	{ .compatible = GPIO_LA_NAME, },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, gpio_la_poll_of_match);
+
+static struct platform_driver gpio_la_poll_device_driver = {
+	.probe = gpio_la_poll_probe,
+	.remove = gpio_la_poll_remove,
+	.driver = {
+		.name = GPIO_LA_NAME,
+		.of_match_table = gpio_la_poll_of_match,
+	}
+};
+
+static int __init gpio_la_poll_init(void)
+{
+	gpio_la_poll_debug_dir = debugfs_create_dir(GPIO_LA_NAME, NULL);
+
+	return platform_driver_register(&gpio_la_poll_device_driver);
+}
+late_initcall(gpio_la_poll_init);
+
+static void __exit gpio_la_poll_exit(void)
+{
+	platform_driver_unregister(&gpio_la_poll_device_driver);
+	debugfs_remove_recursive(gpio_la_poll_debug_dir);
+}
+module_exit(gpio_la_poll_exit);
+
+MODULE_AUTHOR("Wolfram Sang <wsa@sang-engineering.com>");
+MODULE_DESCRIPTION("Sloppy logic analyzer using GPIOs");
+MODULE_LICENSE("GPL");
diff --git a/tools/gpio/gpio-sloppy-logic-analyzer.sh b/tools/gpio/gpio-sloppy-logic-analyzer.sh
new file mode 100755
index 000000000000..ed21a110df5e
--- /dev/null
+++ b/tools/gpio/gpio-sloppy-logic-analyzer.sh
@@ -0,0 +1,246 @@ 
+#!/bin/sh -eu
+# SPDX-License-Identifier: GPL-2.0
+#
+# Helper script for the Linux Kernel GPIO sloppy logic analyzer
+#
+# Copyright (C) Wolfram Sang <wsa@sang-engineering.com>
+# Copyright (C) Renesas Electronics Corporation
+
+samplefreq=1000000
+numsamples=250000
+cpusetdefaultdir='/sys/fs/cgroup'
+cpusetprefix='cpuset.'
+debugdir='/sys/kernel/debug'
+ladirname='gpio-sloppy-logic-analyzer'
+outputdir="$PWD"
+neededcmds='taskset zip'
+max_chans=8
+duration=
+initcpu=
+listinstances=0
+lainstance=
+lasysfsdir=
+triggerdat=
+trigger_bindat=
+progname="${0##*/}"
+print_help()
+{
+	cat << EOF
+$progname - helper script for the Linux Kernel Sloppy GPIO Logic Analyzer
+Available options:
+	-c|--cpu <n>: which CPU to isolate for sampling. Only needed once. Default <1>.
+		      Remember that a more powerful CPU gives you higher sampling speeds.
+		      Also CPU0 is not recommended as it usually does extra bookkeeping.
+	-d|--duration-us <SI-n>: number of microseconds to sample. Overrides -n, no default value.
+	-h|--help: print this help
+	-i|--instance <str>: name of the logic analyzer in case you have multiple instances. Default
+			     to first instance found
+	-k|--kernel-debug-dir <str>: path to the kernel debugfs mountpoint. Default: <$debugdir>
+	-l|--list-instances: list all available instances
+	-n|--num_samples <SI-n>: number of samples to acquire. Default <$numsamples>
+	-o|--output-dir <str>: directory to put the result files. Default: current dir
+	-s|--sample_freq <SI-n>: desired sampling frequency. Might be capped if too large.
+				 Default: <1000000>
+	-t|--trigger <str>: pattern to use as trigger. <str> consists of two-char pairs. First
+			    char is channel number starting at "1". Second char is trigger level:
+			    "L" - low; "H" - high; "R" - rising; "F" - falling
+			    These pairs can be combined with "+", so "1H+2F" triggers when probe 1
+			    is high while probe 2 has a falling edge. You can have multiple triggers
+			    combined with ",". So, "1H+2F,1H+2R" is like the example before but it
+			    waits for a rising edge on probe 2 while probe 1 is still high after the
+			    first trigger has been met.
+			    Trigger data will only be used for the next capture and then be erased.
+
+<SI-n> is an integer value where SI units "T", "G", "M", "K" are recognized, e.g. '1M500K' is 1500000.
+
+Examples:
+Samples $numsamples values at 1MHz with an already prepared CPU or automatically prepares CPU1 if needed,
+use the first logic analyzer instance found:
+	'$progname'
+Samples 50us at 2MHz waiting for a falling edge on channel 2. CPU and instance as above:
+	'$progname -d 50 -s 2M -t "2F"'
+
+Note that the process exits after checking all parameters but a sub-process still works in
+the background. The result is only available once the sub-process finishes.
+
+Result is a .sr file to be consumed with PulseView from the free Sigrok project. It is
+a zip file which also contains the binary sample data which may be consumed by others.
+The filename is the logic analyzer instance name plus a since-epoch timestamp.
+EOF
+}
+
+fail()
+{
+	echo "$1"
+	exit 1
+}
+
+parse_si()
+{
+	conv_si="$(printf $1 | sed 's/[tT]+\?/*1000G+/g; s/[gG]+\?/*1000M+/g; s/[mM]+\?/*1000K+/g; s/[kK]+\?/*1000+/g; s/+$//')"
+	si_val="$((conv_si))"
+}
+set_newmask()
+{
+	for f in $(find "$1" -iname "$2"); do echo "$newmask" > "$f" 2>/dev/null || true; done
+}
+
+init_cpu()
+{
+	isol_cpu="$1"
+
+	[ -d "$lacpusetdir" ] || mkdir "$lacpusetdir"
+
+	cur_cpu=$(cat "${lacpusetfile}cpus")
+	[ "$cur_cpu" = "$isol_cpu" ] && return
+	[ -z "$cur_cpu" ] || fail "CPU$isol_cpu requested but CPU$cur_cpu already isolated"
+
+	echo "$isol_cpu" > "${lacpusetfile}cpus" || fail "Could not isolate CPU$isol_cpu. Does it exist?"
+	echo 1 > "${lacpusetfile}cpu_exclusive"
+	echo 0 > "${lacpusetfile}mems"
+
+	oldmask=$(cat /proc/irq/default_smp_affinity)
+	newmask=$(printf "%x" $((0x$oldmask & ~(1 << isol_cpu))))
+
+	set_newmask '/proc/irq' '*smp_affinity'
+	set_newmask '/sys/devices/virtual/workqueue/' 'cpumask'
+
+	# Move tasks away from isolated CPU
+	for p in $(ps -o pid | tail -n +2); do
+		mask=$(taskset -p "$p") || continue
+		# Ignore tasks with a custom mask, i.e. not equal $oldmask
+		[ "${mask##*: }" = "$oldmask" ] || continue
+		taskset -p "$newmask" "$p" || continue
+	done 2>/dev/null >/dev/null
+
+	# Big hammer! Working with 'rcu_momentary_dyntick_idle()' for a more fine-grained solution
+	# still printed warnings. Same for re-enabling the stall detector after sampling.
+	echo 1 > /sys/module/rcupdate/parameters/rcu_cpu_stall_suppress
+
+	cpufreqgov="/sys/devices/system/cpu/cpu$isol_cpu/cpufreq/scaling_governor"
+	[ -w "$cpufreqgov" ] && echo 'performance' > "$cpufreqgov" || true
+}
+
+parse_triggerdat()
+{
+	oldifs="$IFS"
+	IFS=','; for trig in $1; do
+		mask=0; val1=0; val2=0
+		IFS='+'; for elem in $trig; do
+			chan=${elem%[lhfrLHFR]}
+			mode=${elem#$chan}
+			# Check if we could parse something and the channel number fits
+			[ "$chan" != "$elem" ] && [ "$chan" -le $max_chans ] || fail "Trigger syntax error: $elem"
+			bit=$((1 << (chan - 1)))
+			mask=$((mask | bit))
+			case $mode in
+				[hH]) val1=$((val1 | bit)); val2=$((val2 | bit));;
+				[fF]) val1=$((val1 | bit));;
+				[rR]) val2=$((val2 | bit));;
+			esac
+		done
+		trigger_bindat="$trigger_bindat$(printf '\\%o\\%o' $mask $val1)"
+		[ $val1 -ne $val2 ] && trigger_bindat="$trigger_bindat$(printf '\\%o\\%o' $mask $val2)"
+	done
+	IFS="$oldifs"
+}
+
+do_capture()
+{
+	taskset "$1" echo 1 > "$lasysfsdir"/capture || fail "Capture error! Check kernel log"
+
+	srtmp=$(mktemp -d)
+	echo 1 > "$srtmp"/version
+	cp "$lasysfsdir"/sample_data "$srtmp"/logic-1-1
+	cat > "$srtmp"/metadata << EOF
+[global]
+sigrok version=0.2.0
+
+[device 1]
+capturefile=logic-1
+total probes=$(wc -l < "$lasysfsdir"/meta_data)
+samplerate=${samplefreq}Hz
+unitsize=1
+EOF
+	cat "$lasysfsdir"/meta_data >> "$srtmp"/metadata
+
+	zipname="$outputdir/${lasysfsdir##*/}-$(date +%s).sr"
+	zip -jq "$zipname" "$srtmp"/*
+	rm -rf "$srtmp"
+	delay_ack=$(cat "$lasysfsdir"/delay_ns_acquisition)
+	[ "$delay_ack" -eq 0 ] && delay_ack=1
+	echo "Logic analyzer done. Saved '$zipname'"
+	echo "Max sample frequency this time: $((1000000000 / delay_ack))Hz."
+}
+
+rep=$(getopt -a -l cpu:,duration-us:,help,instance:,list-instances,kernel-debug-dir:,num_samples:,output-dir:,sample_freq:,trigger: -o c:d:hi:k:ln:o:s:t: -- "$@") || exit 1
+eval set -- "$rep"
+while true; do
+	case "$1" in
+	-c|--cpu) initcpu="$2"; shift;;
+	-d|--duration-us) parse_si $2; duration=$si_val; shift;;
+	-h|--help) print_help; exit 0;;
+	-i|--instance) lainstance="$2"; shift;;
+	-k|--kernel-debug-dir) debugdir="$2"; shift;;
+	-l|--list-instances) listinstances=1;;
+	-n|--num_samples) parse_si $2; numsamples=$si_val; shift;;
+	-o|--output-dir) outputdir="$2"; shift;;
+	-s|--sample_freq) parse_si $2; samplefreq=$si_val; shift;;
+	-t|--trigger) triggerdat="$2"; shift;;
+	--) break;;
+	*) fail "error parsing command line: $*";;
+	esac
+	shift
+done
+
+for f in $neededcmds; do
+	command -v "$f" >/dev/null || fail "Command '$f' not found"
+done
+
+# print cpuset mountpoint if any, errorcode > 0 if noprefix option was found
+cpusetdir=$(awk '$3 == "cgroup" && $4 ~ /cpuset/ { print $2; exit (match($4, /noprefix/) > 0) }' /proc/self/mounts) || cpusetprefix=''
+if [ -z "$cpusetdir" ]; then
+	cpusetdir="$cpusetdefaultdir"
+	[ -d $cpusetdir ] || mkdir $cpusetdir
+	mount -t cgroup -o cpuset none $cpusetdir || fail "Couldn't mount cpusets. Not in kernel or already in use?"
+fi
+
+lacpusetdir="$cpusetdir/$ladirname"
+lacpusetfile="$lacpusetdir/$cpusetprefix"
+sysfsdir="$debugdir/$ladirname"
+
+[ "$samplefreq" -ne 0 ] || fail "Invalid sample frequency"
+
+[ -d "$sysfsdir" ] || fail "Could not find logic analyzer root dir '$sysfsdir'. Module loaded?"
+[ -x "$sysfsdir" ] || fail "Could not access logic analyzer root dir '$sysfsdir'. Need root?"
+
+[ $listinstances -gt 0 ] && find "$sysfsdir" -mindepth 1 -type d | sed 's|.*/||' && exit 0
+
+if [ -n "$lainstance" ]; then
+	lasysfsdir="$sysfsdir/$lainstance"
+else
+	lasysfsdir=$(find "$sysfsdir" -mindepth 1 -type d -print -quit)
+fi
+[ -d "$lasysfsdir" ] || fail "Logic analyzer directory '$lasysfsdir' not found!"
+[ -d "$outputdir" ] || fail "Output directory '$outputdir' not found!"
+
+[ -n "$initcpu" ] && init_cpu "$initcpu"
+[ -d "$lacpusetdir" ] || { echo "Auto-Isolating CPU1"; init_cpu 1; }
+
+ndelay=$((1000000000 / samplefreq))
+echo "$ndelay" > "$lasysfsdir"/delay_ns
+
+[ -n "$duration" ] && numsamples=$((samplefreq * duration / 1000000))
+echo $numsamples > "$lasysfsdir"/buf_size
+
+if [ -n "$triggerdat" ]; then
+	parse_triggerdat "$triggerdat"
+	printf "$trigger_bindat" > "$lasysfsdir"/trigger 2>/dev/null || fail "Trigger data '$triggerdat' rejected"
+fi
+
+workcpu=$(cat "${lacpusetfile}effective_cpus")
+[ -n "$workcpu" ] || fail "No isolated CPU found"
+cpumask=$(printf '%x' $((1 << workcpu)))
+instance=${lasysfsdir##*/}
+echo "Setting up '$instance': $numsamples samples at ${samplefreq}Hz with ${triggerdat:-no} trigger using CPU$workcpu"
+do_capture "$cpumask" &