All of lore.kernel.org
 help / color / mirror / Atom feed
From: Denis Kenzior <denkenz@gmail.com>
To: ofono@ofono.org
Subject: Re: [PATCH 4/4] isigen: make number of PDP contexts configurable
Date: Thu, 11 Nov 2010 09:36:43 -0600	[thread overview]
Message-ID: <4CDC0D8B.4060404@gmail.com> (raw)
In-Reply-To: <C358A26273CF2948B8BCDBA661A581782BE95B62A6@NOK-EUMSG-03.mgdnok.nokia.com>

[-- Attachment #1: Type: text/plain, Size: 1913 bytes --]

On 11/11/2010 02:16 AM, Mika.Liljeberg(a)nokia.com wrote:
> Hi Marcel,
> 
>> why to do you bother making this a configurable option? What isthe
>> benefit here?
> 
> The maximum context count is a compile time option for the ISI modem. Having this option in oFono makes it possible to optimize the APE side resource usage instead of overallocating drivers and contexts. Not a huge benefit, I guess, but the cost is not huge either.
> 
> Not a big deal, though. I'll try to see if the context limit could be probed somehow.
> 

The gprs context structure is about the size of 5 pointers, so there's
not much to be saved here.  If you're worried about space usage of the
isi specific data, you can always allocate it during the activation stage.

>> Personally I think that always enabling 4 context if the hardware
>> supports it should be enough. If you do support more then just enable
>> more all the time. There are no real resources used in context of ISI
>> anyway. The AT command based modems have a different problem since for
>> most of them we need an extra TTY/DLC and an extra GAtChat object, but
>> ISI does not have that problem.
> 
> I don't see the difference. The AT modem resources should be allocated dynamically as well, at context activation time. (Maybe that's the case already, I didn't really check.) In any case, I believe oFono has to allow things like external AT command processors and vendor specific modem tools to access the TTY's directly bypassing oFono. This means the mux channels should not be preallocated.

Hah, I know some people who vehemently disagree ;)  Of course I'm not
one of them...  Strictly speaking what you say is possible even with
today's architecture, however almost no AT modem we have supports more
than about 3-4 contexts.  So doing this to save 300-400 bytes / context
is simply not worth it at this point.

Regards,
-Denis

      parent reply	other threads:[~2010-11-11 15:36 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-11-10 12:22 [PATCH 0/4] bugfixes + revectored isimodem GPRS context driver Mika Liljeberg
2010-11-10 12:22 ` [PATCH 1/4] gisi: fix crash bug in g_isi_remove_subscription Mika Liljeberg
2010-11-11  8:13   ` Aki Niemi
2010-11-10 12:22 ` [PATCH 2/4] gisi: return NULL instead of asserting Mika Liljeberg
2010-11-11  8:13   ` Aki Niemi
2010-11-10 12:22 ` [PATCH 3/4] isimodem: revector GPRS context driver Mika Liljeberg
2010-11-11  8:15   ` Aki Niemi
2010-11-10 12:22 ` [PATCH 4/4] isigen: make number of PDP contexts configurable Mika Liljeberg
2010-11-10 17:52   ` Marcel Holtmann
2010-11-11  8:16     ` Mika.Liljeberg
2010-11-11  9:25       ` Marcel Holtmann
2010-11-11 15:36       ` Denis Kenzior [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4CDC0D8B.4060404@gmail.com \
    --to=denkenz@gmail.com \
    --cc=ofono@ofono.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.