hostap and sleeping mode.

[Pavel Roskin]

On Wed, 21 Apr 2004, Andrea G. Forte wrote:

> ok here are some new issues about hostap and sleeping mode. Now hostap
> works when resuming from sleeping also in host_roaming 2. I have two
> different problems with hostap 0.0.4 and 0.1.3. With the first,
> everytime I resume from sleeping, the hostap driver does not get
> unloaded and with the "lsmod" command the hostap_cs module appears to be
> used 3,4,5,6, etc times (once for each time that the laptop went to
> sleep). With hostap 0.1.3 the same thing happens, but this time the
> number is negative -2, -3, -4, etc. You guys have any clue why this
> would happen??? How do I fix it? Sounds like it could be a cardmgr
> issue.

Please don't worry too much about correctness of old software.  I tried
and could not reproduce this problem with CVS hostap and Linux 2.6.6-rc2.
I used ACPI to go to the S1 state by writing "standby" to
/sys/power/state.  After I pressed a key, the system was restored to the
original state, and the use count for hostap_cs was 1, just like it was
before the suspend.

If you still want to debug this problem use "cardctl suspend" and "cardctl
resume".  This would suspend the card without suspending the whole
system.  Then see where the use count changes in the driver.  Look for
MOD_INC_USE_COUNT and MOD_DEC_USE_COUNT in the sources.

> The second question I have is about hostap 0.0.4. In the file hostap.c
> there are two pointers: dev->stop and dev->open. Who uses these two
> pointers? Where are they called? I need to know this because in order to
> make hostap resume from sleeping, I had to comment out the dev->stop
> pointer assignment. (The card would close right after the sleeping event
> was captured so that nothing else would work for resuming). I suspect
> the two pointers are used by the card manager, but I could not find
> where.

Those functions are called by the network level when the network device is
brought up (dev->open) and down (dev->stop).

-- 
Regards,
Pavel Roskin