[ECOS] Does ethernet driver "deliver" method need to be re-entrant?

[ECOS] Does ethernet driver "deliver" method need to be re-entrant?

[Grant Edwards]

I'm having intermittent problems with an Ethernet driver.  One
possible explanation would be if the driver's "deliver" method
was being called "simultaneously" from two different contexts.

Can this happen, or are calls to deliver guaranteed to be
serialized?

-- 
Grant Edwards                   grante             Yow!  I want to dress you
                                  at               up as TALLULAH BANKHEAD and
                               visi.com            cover you with VASELINE and
                                                   WHEAT THINS...


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Re: Does ethernet driver "deliver" method need to be re-entrant?

[Grant Edwards]

On 2007-04-01, Grant Edwards <grante@visi.com> wrote:

> I'm having intermittent problems with an Ethernet driver.  One
> possible explanation would be if the driver's "deliver" method
> was being called "simultaneously" from two different contexts.
>
> Can this happen, or are calls to deliver guaranteed to be
> serialized?

The problem caused by a race condition in a different section
of the driver.  I'm not sure why the author of the driver
didn't put in mutexes.  The sample driver we gave him to follow
had them.

Anywy, I still need to know if deliver needs to be re-entrant
or not.  It looks to me like not -- it is called only from a
DSR context by an alarm handler in the stack, right?

It would be nice if the driver API documentation state whether
each driver method needed to be re-entrant or not...

-- 
Grant Edwards                   grante             Yow!  Imagine--a WORLD
                                  at               without POODLES...
                               visi.com            


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Re: Does ethernet driver "deliver" method need to be re-entrant?

[Gary Thomas]

Grant Edwards wrote:
> On 2007-04-01, Grant Edwards <grante@visi.com> wrote:
> 
>> I'm having intermittent problems with an Ethernet driver.  One
>> possible explanation would be if the driver's "deliver" method
>> was being called "simultaneously" from two different contexts.
>>
>> Can this happen, or are calls to deliver guaranteed to be
>> serialized?
> 
> The problem caused by a race condition in a different section
> of the driver.  I'm not sure why the author of the driver
> didn't put in mutexes.  The sample driver we gave him to follow
> had them.
> 
> Anywy, I still need to know if deliver needs to be re-entrant
> or not.  It looks to me like not -- it is called only from a
> DSR context by an alarm handler in the stack, right?
> 
> It would be nice if the driver API documentation state whether
> each driver method needed to be re-entrant or not...
> 

There is no expectation for any of the ethernet driver functions
to be reentrant.  Serialization is handled by the other layers
that call these functions.


-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Re: Does ethernet driver "deliver" method need to be re-entrant?

[Grant Edwards]

On 2007-04-03, Gary Thomas <gary@mlbassoc.com> wrote:

> There is no expectation for any of the ethernet driver functions
> to be reentrant.  Serialization is handled by the other layers
> that call these functions.

Thanks.  That's what I thought, but I wasn't sure I had
correctly found my way throught he network stack.

-- 
Grant Edwards                   grante             Yow!  Vote for ME
                                  at               -- I'm well-tapered,
                               visi.com            half-cocked, ill-conceived
                                                   and TAX-DEFERRED!


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss