[9front] mnt reform trackball stops responding

[9front] mnt reform trackball stops responding

[sl]

this just started happening in the past few days. it never happened before. 
twice tonight my trackball has stopped responding. soft reboot (fshalt -r) 
did not restore it. powering off completely and powering back on did.

this is what i saw in /dev/kmesg immediately after the trackball stopped 
responding:

nusb/kb: hid: /dev/usb/ep9.1: read: USB Transaction Error
nusb/kb: fatal: no report descriptor

but nothing still running seemed broken:

; broke
; ps ax|grep kb
sl               68    0:00   0:00      448K Pread    kbdfs
sl               69    0:00   0:00      448K Rendez   kbdfs
sl               70    0:00   0:00      448K Rendez   kbdfs
sl               71    0:00   0:00      448K Pread    kbdfs
sl               72    0:00   0:00      448K Rendez   kbdfs
sl               91    0:01   0:00      512K Pread    kb
sl               92    0:00   0:00      512K Pread    kb
sl               93    0:00   0:00      512K Pread    kb
sl              100    0:00   0:00      384K Pread    kb
sl              101    0:00   0:00      384K Pread    kb
sl              453    0:00   0:00      512K Rendez   kb
sl              454    0:00   0:00      512K Sleep    kb

**update:**

as i was typing this message the trackball stopped working again.
i unplugged my external usb mouse to see if this would have any
effect, and then the keyboard stopped responding as well. i didn't
have an external keyboard at hand so i had to circle-r reboot the
machine. both trackball and keyboard worked again after reboot.

note: my 9front is fully updated.

sl

Re: [9front] mnt reform trackball stops responding

[sl]

quasi-reproducible failure:

this doesn't happen every single time, but more often than not.

- boot with external usb mouse plugged in
- unplug mouse
- plug mouse into different usb port -> trackball and/or keyboard stops working

typical output from /dev/kprint:

nusb/kb: hid: /dev/usb/ep5.1: read: USB Transaction Error
nusb/kb: fatal: device is detached
nusb/kb: hid: /dev/usb/ep9.1: read: USB Transaction Error
nusb/kb: hid: /dev/usb/ep8.1: read: USB Transaction Error
nusb/kb: hid: /dev/usb/ep8.3: read: USB Transaction Error
nusb/kb: hid: /dev/usb/ep9.1: read: USB Transaction Error
nusb/kb: fatal: device is detached
nusb/kb: fatal: no report descriptor
nusb/kb: fatal: no report descriptor

sl

Re: [9front] mnt reform trackball stops responding

[Philip Silva]

Actually I had the same problem (but also problems with the screen) and this was done: https://community.mnt.re/t/trackball-v2-with-rp2040-requires-a-trace-cut/1473

Although it was the other way around for me, I usually connected any external mouse after the internal trackball stopped working

Re: [9front] mnt reform trackball stops responding

[Stanley Lieber]

On July 20, 2023 12:36:47 PM EDT, Philip Silva <philip.silva@protonmail.com> wrote:
>Actually I had the same problem (but also problems with the screen) and this was done: https://community.mnt.re/t/trackball-v2-with-rp2040-requires-a-trace-cut/1473
>
>Although it was the other way around for me, I usually connected any external mouse after the internal trackball stopped working
>
>

thanks! my trackball came from crowd supply just a couple of weeks ago. i believe it is a 2.0, but i'll have to check to make sure.

yeah, there seems to be some general usb flakiness when things are plugged/unplugged in/out of multiple ports. i don't recall having this problem on other systems, but then again, none of my thinkpads' keyboards/trackpoints are usb internally, so maybe i just never triggered it or never noticed.

sl

Re: [9front] mnt reform trackball stops responding

[sl]

>> Actually I had the same problem (but also problems with the screen)
> > and this was done:
>> https://community.mnt.re/t/trackball-v2-with-rp2040-requires-a-trace-cut/1473
>> Although it was the other way around for me, I usually connected
>> any external mouse after the internal trackball stopped working >>
> 
> thanks!  my trackball came from crowd supply just a couple of weeks
> ago.  i believe it is a 2.0, but i'll have to check to make sure.

i checked. it is a 2.0.


> yeah, there seems to be some general usb flakiness when things are
> plugged/unplugged in/out of multiple ports.  i don't recall having
> this problem on other systems, but then again, none of my thinkpads'
> keyboards/trackpoints are usb internally, so maybe i just never
> triggered it or never noticed.

one notable aspect of my specific experience is that nothing goes wrong
so long as i don't mess with the usb ports. every instance of the
trackball and/or keyboard stopping working was in conjunction with
something being plugged into or unplugged from a usb port. i didn't
realize the correlation was so consistent until after writing my
original message and performing additional tests.

sl

Re: [9front] mnt reform trackball stops responding

[Philip Silva]

Ah ok wow, but at least that's that then! I had the trackball stopping to work at random times even when not sitting directly at the laptop. (Not often but every now and then) Other problems got worse with time, there was also a problem with static electricity which created glitches on the screen until it would eventually turn off for a while. (One time it was also flickering when being with my hand close to a USB port) But maybe one doesn't always run into that... Although now all that seems gone.

Re: [9front] mnt reform trackball stops responding

[Stanley Lieber]

On July 22, 2023 12:31:19 PM EDT, Philip Silva <philip.silva@protonmail.com> wrote:
>Ah ok wow, but at least that's that then! I had the trackball stopping to work at random times even when not sitting directly at the laptop. (Not often but every now and then) Other problems got worse with time, there was also a problem with static electricity which created glitches on the screen until it would eventually turn off for a while. (One time it was also flickering when being with my hand close to a USB port) But maybe one doesn't always run into that... Although now all that seems gone.
>
>

haha, weird! i've never observed any induction glitching. there is a thing where the screen will occasionally momentarily go wobbly for a second, but iirc this is a known software problem, and it has happened on both units i've used. it seems logical that no two reforms would meet precisely the same tolerances since they're all assembled by hand, either by mnt personnel or the owner. maybe something internal on yours is or was not quite right.

the usb stuff may not be limited to the reform, but since the reform leans so heavily on usb for devices that are constantly in use, it may be triggering some more general bug. fwiw, my trackball hasn't stopped responding since the day of my original post. the keyboard not responding has been even less frequent, but it did happen more than once without me trying to make it happen on purpose. i thought it was interesting i was able to induce the failures (though not entirely consistently) just by plugging/unplugging things into different usb ports.

sl

Re: [9front] mnt reform trackball stops responding

[Sigrid Solveig Haflínudóttir]

I just pushed a small change to the kernel ('imx8: usb: enable host IN auto retry'),
could you please test with that? I managed to repro an *extra* usb device consistently
falling off as long as I used it even just a bit - in this case nusb/ptp-mounted Android
phone. With the change it works fine. Wonder if it helps in your case.

Re: [9front] mnt reform trackball stops responding

[cinap_lenrek]

working fine for me...

--
cinap

[9front] 9front Boots Into "No Config" -- The Most Bizarre Error Of Them All

[Jay F. Shachter]

Esteemed Colleagues:

My recent attempts to install a working 9front system on my laptop
computer have been astonishingly frustrating.  At first, I could not
perform an installation because the installer could not keep its
sanity while connected to a USB keyboard that had USB ports on it.
After disconnecting all USB devices except the one that contained the
installation medium, I successfully made it thru an installation.
Then I could not run the installed system, because I had made it too
small (I thought 5G was enough, I was wrong).  I prepared a 20G slice
of disk, and successfully installed a working 9front system onto it.

Now for the most bizarre and frustrating error of them all: the system
has stopped working.  It now boots into a "no config" screen.  I have
done nothing to the system, since I successfully installed it.  I do
not know 9front well enough to have done anything to it.  I have
installed other operating systems, but I have done nothing to the
9front system.  It no longer boots.  Moreover, I zeroed out the slice
of disk on which it resides, and re-installed it.  It still no longer
boots.  Moreover, I designated a different 20G slice of disk to be the
one with the 0x39 description, installed onto that one, and that one
doesn't boot, either.  Instead, I see the following (there may be
typographical errors, I am copying manually from the screen):

  pbs.................ok
  no config
  *e820=1 0x0000000000000000 0x000000000009d400 2 0x000000000009d400 0x00000000000
  a0000 2 0x00000000000e0000 0x0000000000100000 1 0x0000000000100000 0x00000000200
  00000 2 0x0000000020000000 0x0000000020200000 1 0x0000000020200000 0x00000000400
  04000 2 0x0000000040004000 0x0000000040005000 1 0x0000000040005000 0x00000000c7f
  0a000 2 0x00000000c7f0a000 0x00000000c8000000 1 0x00000000c8000000 0x00000000c87
  53000 2 0x00000000c8753000 0x00000000c8800000 1 0x00000000c8800000 0x00000000c8f
  b2000 3 0x00000000c8fb2000 0x00000000c9000000 1 0x00000000c9000000 0x00000000ca6
  fc000 4 0x00000000ca6fc000 0x00000000ca800000 1 0x00000000ca800000 0x00000000cbb
  0f000 2 0x00000000cbb0f000 0x00000000cc7a7000 4 0x00000000cc7a7000 0x00000000cc7
  ea000 1 0x00000000cc7ea000 0x00000000ccab0000 2 0x00000000ccab0000 0x00000000ccf
  f0000 1 0x00000000ccff0000 0x00000000cd000000 2 0x00000000cd800000 0x00000000cfa
  00000 2 0x00000000f8000000 0x00000000fc000000 2 0x00000000fec00000 0x00000000fec
  01000 2 0x00000000fed00000 0x00000000fed04000 2 0x00000000fed1c000 0x00000000fed
  20000 2 0x00000000fee00000 0x00000000fee01000 2 0x00000000ff000000 0x00000001000
  00000 1 0x0000000010000000 0x000000042e600000
  >

with a blinking underline cursor immediately to the right of the >
symbol.  I think this is how 9front always starts, except for the "no
config" error and the > prompt, but I don't know for certain because
when the system boots successfully the initial output flies off the
screen too fast.

When I installed 9front sucessfully, I accepted all the defaults,
except for the machine name, which I changed from "circo" to "m5", and
the timezone, which I changed from "US_Eastern" to "US_Central".  I
said "no" when asked whether to overwrite the MBR, and I said "no"
when asked whether to make the partition bootable.  When I reinstalled
9front, after my functioning system mysteriously stopped working, I
did the same.  It still didn't work.  I reinstalled again and said
"yes" to the bootable partition question.  It still didn't work.

How does a functioning 9front system suddenly stop working, with a "no
config" error and an angle-bracket prompt?  More important, how do I
get it to work?

As always, thank you in advance for any and all replies.


                        Jay F. Shachter
                        6424 North Whipple Street
                        Chicago IL  60645-4111
                                (1-773)7613784   landline
                                (1-410)9964737   GoogleVoice
                                jay@m5.chicago.il.us
                                http://m5.chicago.il.us

                        "Quidquid latine dictum sit, altum videtur"

Re: [9front] mnt reform trackball stops responding

[sl]

> I just pushed a small change to the kernel ('imx8: usb: enable host IN auto retry'),
> could you please test with that? I managed to repro an *extra* usb device consistently

thanks!

i can still reliably induce the trackball and keyboard to stop responding by doing this:

- boot with external usb mouse plugged in
- roll trackball
- unplug usb mouse, plug it back into a different usb port
- roll trackball

at this point, the trackball will roll for a second and then the mouse pointer stops,
with both the trackball and keyboard being completely non-responsive until reboot.

sl

Re: [9front] 9front Boots Into "No Config" -- The Most Bizarre Error Of Them All

[Jacob Moody]

On 7/26/23 18:48, Jay F. Shachter wrote:
> 
> Esteemed Colleagues:
> 
> My recent attempts to install a working 9front system on my laptop
> computer have been astonishingly frustrating.  At first, I could not
> perform an installation because the installer could not keep its
> sanity while connected to a USB keyboard that had USB ports on it.
> After disconnecting all USB devices except the one that contained the
> installation medium, I successfully made it thru an installation.
> Then I could not run the installed system, because I had made it too
> small (I thought 5G was enough, I was wrong).  I prepared a 20G slice
> of disk, and successfully installed a working 9front system onto it.
> 
> Now for the most bizarre and frustrating error of them all: the system
> has stopped working.  It now boots into a "no config" screen.  I have
> done nothing to the system, since I successfully installed it.  I do
> not know 9front well enough to have done anything to it.  I have
> installed other operating systems, but I have done nothing to the
> 9front system.  It no longer boots.  Moreover, I zeroed out the slice
> of disk on which it resides, and re-installed it.  It still no longer
> boots.  Moreover, I designated a different 20G slice of disk to be the
> one with the 0x39 description, installed onto that one, and that one
> doesn't boot, either.  Instead, I see the following (there may be
> typographical errors, I am copying manually from the screen):
> 
>   pbs.................ok
>   no config
>   *e820=1 0x0000000000000000 0x000000000009d400 2 0x000000000009d400 0x00000000000
>   a0000 2 0x00000000000e0000 0x0000000000100000 1 0x0000000000100000 0x00000000200
>   00000 2 0x0000000020000000 0x0000000020200000 1 0x0000000020200000 0x00000000400
>   04000 2 0x0000000040004000 0x0000000040005000 1 0x0000000040005000 0x00000000c7f
>   0a000 2 0x00000000c7f0a000 0x00000000c8000000 1 0x00000000c8000000 0x00000000c87
>   53000 2 0x00000000c8753000 0x00000000c8800000 1 0x00000000c8800000 0x00000000c8f
>   b2000 3 0x00000000c8fb2000 0x00000000c9000000 1 0x00000000c9000000 0x00000000ca6
>   fc000 4 0x00000000ca6fc000 0x00000000ca800000 1 0x00000000ca800000 0x00000000cbb
>   0f000 2 0x00000000cbb0f000 0x00000000cc7a7000 4 0x00000000cc7a7000 0x00000000cc7
>   ea000 1 0x00000000cc7ea000 0x00000000ccab0000 2 0x00000000ccab0000 0x00000000ccf
>   f0000 1 0x00000000ccff0000 0x00000000cd000000 2 0x00000000cd800000 0x00000000cfa
>   00000 2 0x00000000f8000000 0x00000000fc000000 2 0x00000000fec00000 0x00000000fec
>   01000 2 0x00000000fed00000 0x00000000fed04000 2 0x00000000fed1c000 0x00000000fed
>   20000 2 0x00000000fee00000 0x00000000fee01000 2 0x00000000ff000000 0x00000001000
>   00000 1 0x0000000010000000 0x000000042e600000
>   >
> 
> with a blinking underline cursor immediately to the right of the >
> symbol.  I think this is how 9front always starts, except for the "no
> config" error and the > prompt, but I don't know for certain because
> when the system boots successfully the initial output flies off the
> screen too fast.
> 
> When I installed 9front sucessfully, I accepted all the defaults,
> except for the machine name, which I changed from "circo" to "m5", and
> the timezone, which I changed from "US_Eastern" to "US_Central".  I
> said "no" when asked whether to overwrite the MBR, and I said "no"
> when asked whether to make the partition bootable.  When I reinstalled
> 9front, after my functioning system mysteriously stopped working, I
> did the same.  It still didn't work.  I reinstalled again and said
> "yes" to the bootable partition question.  It still didn't work.
> 
> How does a functioning 9front system suddenly stop working, with a "no
> config" error and an angle-bracket prompt?  More important, how do I
> get it to work?
> 
> As always, thank you in advance for any and all replies.
> 
> 
>                         Jay F. Shachter
>                         6424 North Whipple Street
>                         Chicago IL  60645-4111
>                                 (1-773)7613784   landline
>                                 (1-410)9964737   GoogleVoice
>                                 jay@m5.chicago.il.us
>                                 http://m5.chicago.il.us
> 
>                         "Quidquid latine dictum sit, altum videtur"
> 


"no config" is the boot loader telling you that it can not find the plan9.ini
file on the 9fat partition. You must have some old (corrupted?) 9fat partition sitting around.
I don't know why you choose to partition the disk from linux, I think that is doing
something incorrect in the installation process. I would just using a new blank disk and give
9front all of it to avoid such issues. Or leaving the space entirely unallocated and doing the
partitioning from 9front may assist you.

Re: [9front] mnt reform trackball stops responding

[sl]

> working fine for me...

frantic usb hot swapping seems to be the trigger.

sl

Re: [9front] mnt reform trackball stops responding

[Sigrid Solveig Haflínudóttir]

Could reproduce with v1 trackball, though it's not very easy to repro - took me many tries.

Re: [9front] mnt reform trackball stops responding

[cinap_lenrek]

i pushed a commit this weekend...

i could reproduce the trackball and the keyboard stop working
when i plug a external mouse in the middle usb port.

basically the keyboard and trackball just get a single
transaction error from where they never recover.

it seems to be a quirk of the internal usb hub or electrical
issue for me. it never happens with the port to the front,
which is on the same hub (shared with keyboard and trackball)...
there should be zero difference...

also works fine if i use a powered hub.

in any case, i found a issue in the way nusb/kb tries to
handle these kind of errors that would not work with xhci.

basically, we now immediately issue a port reset when
we encounter a error reading the interrupt endpoints,
which makes usbd detach the device and then attempt
a new attach.

i cannot avoid this glitch, but at least it recovers quickly.

--
cinap

Re: [9front] mnt reform trackball stops responding

[cinap_lenrek]

maybe some more info.

i investigated the glitch, instrumenting usbd to check
hub status and port status and it is *NOT* indicating
any overcurrent or power supply issues...

so that is that.

unrelated:

also funny thing, the internal hub supports per-port
power switching, which is rare but the reform team
has actually put the switches to support it.

--
cinap

Re: [9front] mnt reform trackball stops responding

[sl]

> i pushed a commit this weekend...

after updating i am no longer able to trigger the trackball or keyboard
to stop responding. maybe i am and it just recovers so fast i don't notice,
but in any case, this saves me having to reboot.

thanks!

sl

Re: [9front] mnt reform trackball stops responding

[ieliedonge]

Oh, cool! You're pushing the needle further toward me picking up one of these
machines. What specs does your mntre have, by the way?

sl@stanleylieber.com wrote:
> > i pushed a commit this weekend...
> 
> after updating i am no longer able to trigger the trackball or keyboard
> to stop responding. maybe i am and it just recovers so fast i don't notice,
> but in any case, this saves me having to reboot.
> 
> thanks!
> 
> sl

Re: [9front] mnt reform trackball stops responding

[sl]

> Oh, cool! You're pushing the needle further toward me picking up one of these
> machines. What specs does your mntre have, by the way?

http://plan9.stanleylieber.com/hardware/mnt/reform/000158/

sl

Re: [9front] mnt reform trackball stops responding

[cinap_lenrek]

> maybe i am and it just recovers so fast i don't notice,
> but in any case, this saves me having to reboot.

exactly, it just recovers quickly.

--
cinap