Problem with Logitech Mx900 again - psm strikes again - SOLVED

[Vladimir Grebenschikov]

В вт, 21/06/2005 в 09:45 -0700, Maksim Yevmenkin пишет: 
> Vladimir,

> however, it could not establish HID interrupt channel on psm 19 (0x13 - 
> HID interrupt). instead bthidd(8) tried to connect on psm 25 (0x19)! and 
> of course mouse rejects the connection.
> 
> 1119333385.549430 < L2CAP(s): Connect req: psm 25 scid 0x0043
> 1119333385.580204 > L2CAP(s): Connect rsp: dcid 0x0000 scid 0x0043 
> result 2 stat
> 1119333385.580438 < L2CAP(s): Disconn req: dcid 0x0042 scid 0x0042
> 1119333385.638242 > L2CAP(s): Disconn rsp: dcid 0x0042 scid 0x0042
> 
> i remember you had similar problem before due to incorrect 
> "interrupt_psm" value in the bthidd.conf file. in your first email you 
> have posted your bthidd.conf file and it had correct values for 
> "interrupt_psm" and "control_psm".
> 
> so, could you please double check your /etc/bluetooth/bthidd.conf file 
> and make sure bthidd(8) uses the correct config.

It strikes again. I was pretty sure, that bug was fixed last time. 
Manual changing psm in bthidd.conf fix my problem.

I remember I've tested code with your patch and even send you ok notice
(PR misc/76107).

I was not touch bthidd.conf because I was pretty sure that it has valid
values.
I am sure that trying to make mouse work I've done
# bthidcontrol -a mouse > /etc/bluetooth/bthidd.conf.tmp
# diff /etc/bluetooth/bthidd.conf{,.tmp}

and got no difference.

But it strikes again.

Ok, I will try all cycle from scratch.

Just reconnect mouse to cradle - Ok
Then reset mouse and run bthidcontrol query -> psm 0x13
Then reconnect directly to bthidd - Ok

I have only idea why it happens - I have used very old binary of
bthidcontrol installed while compare/rebuild bthidd.conf (with bug,
unfortunately it was just overwritten by fresh build).
/probably one should attach build+install of bthid* binaries to world ?/

Looks like there was hardware glitch (when win and freebsd connection
was broken over night, probably out of power in battaries, it was)
After that I've tried to restore mouse and replace bthidd.conf with
broken version and it stop working.

> i suspect there might be another bug in bthidd(8), so if you could 
> please try to run bthidd(8) under debugger and try to examine internal 
> configuration structures after bthidd(8) parses the config file.

Looks like bthidd is ok in this case.

> i also attached a patch that renames global variable to avoid name 
> collision with the local variables. i do not expect it to fix your 
> problem, but please try it anyway.

I have tried - it works ok (without it too), only issue - first chunk of patch - should be removed before apply.

> > It looks like hardware problem for me (over one night both OSes stop
> > stop seen mouse) no reset was initiated, cradle was far-far away from
> > place where it happens. But, probably there is software solution.
> 
> lets not blame the hardware just yet :)

Yes, it was software.

But still It is not clean for me two issues:
 a) why mouse stops work under WinXP (it works before, and stop to work at the same night as under freebsd)
 b) why it happens (no suspend/hibernate was initiated)

> thanks,
> max

-- 
Vladimir B. Grebenschikov
vova at fbsd.ru