Hi Wolfram,
On Mon, Dec 20, 2021 at 10:07 AM Wolfram Sang
<wsa+renesas@xxxxxxxxxxxxxxxxxxxx> 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@xxxxxxxxxxxxxxxxxxxx>
Thanks for your patch!
> --- /dev/null
> +++ b/tools/gpio/gpio-sloppy-logic-analyzer
> @@ -0,0 +1,221 @@
> +#!/bin/sh -eu
> +# Helper script for the Linux Kernel GPIO sloppy logic analyzer
> +#
> +# Copyright (C) Wolfram Sang <wsa@xxxxxxxxxxxxxxxxxxxx>
> +# Copyright (C) Renesas Electronics Corporation
> +#
> +# TODO: support SI units in command line parameters?
> +
> +samplefreq=1000000
> +numsamples=250000
> +cpusetdir='/dev/cpuset'
> +debugdir='/sys/kernel/debug'
> +ladirname='gpio-sloppy-logic-analyzer'
> +outputdir="$PWD"
> +neededcmds='taskset zip'
> +max_chans=8
> +duration=
> +initcpu=
> +lainstance=
> +lasysfsdir=
> +triggerdat=
> +trigger_bindat=
> +progname="${0##*/}"
> +print_help()
> +{
> + cat << EOF
> +$progname - helper script for the Linux Kernel Sloppy GPIO Logic Analyzer
> +Available options:
> + -c|--cpu <n>: which CPU to isolate for sampling. Only needed once. Default <1>.
> + Remember that a more powerful CPU gives you higher sampling speeds.
> + Also CPU0 is not recommended as it usually does extra bookkeeping.
> + -d|--duration-us <n>: number of microseconds to sample. Overrides -n, no default value.
> + -h|--help: print this help
> + -i|--instance <str>: name of the logic analyzer in case you have multiple instances. Default
> + to first instance found
> + -k|--kernel-debug-dir: path to the kernel debugfs mountpoint. Default: <$debugdir>
> + -n|--num_samples <n>: number of samples to acquire. Default <$numsamples>
> + -o|--output-dir <str>: directory to put the result files. Default: current dir
> + -s|--sample_freq <n>: desired sampling frequency. Might be capped if too large. Default: 1MHz.
> + -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.
> +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 2000000 -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
> +}
> +
> +set_newmask()
> +{
> + for f in $(find "$1" -iname "$2"); do echo "$newmask" > "$f" 2>/dev/null || true; done
> +}
> +
> +init_cpu()
> +{
> + isol_cpu="$1"
> + [ -d $cpusetdir ] || mkdir $cpusetdir
> + mount | grep -q $cpusetdir || mount -t cpuset cpuset $cpusetdir
This needs CONFIG_CPUSETS=y, so you may want to document that.
> + [ -d "$lacpusetdir" ] || mkdir "$lacpusetdir"
> +
> + cur_cpu="$(cat "$lacpusetdir"/cpus)"
cat: /dev/cpuset/gpio-sloppy-logic-analyzer/cpus: No such file or directory
I do have a "cpuset.cpus" file.
> + [ "$cur_cpu" = "$isol_cpu" ] && return
> + [ -z "$cur_cpu" ] || fail "CPU$isol_cpu requested but CPU$cur_cpu already isolated"
> +
> + echo "$isol_cpu" > "$lacpusetdir"/cpus || fail "Could not isolate CPU$isol_cpu. Does it exist?"
> + echo 1 > "$lacpusetdir"/cpu_exclusive
> + echo 0 > "$lacpusetdir"/mems
No complaint, although the real files have a "cpuset."-prefix again.
[...]
> +workcpu=$(cat "$lacpusetdir"/effective_cpus)
cat: /dev/cpuset/gpio-sloppy-logic-analyzer/effective_cpus: No such
file or directory
I do have a "cpuset.effective_cpus" file.
> +[ -n "$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" &
Anyone with a clue? Thanks!
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
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
